mirror
/
openwrt
mirror of https://git.openwrt.org/openwrt/openwrt.git
1
0
Fork 0
Code Releases Activity

kernel: remove linux 3.18 support 

No targets are using it anymore

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
This commit is contained in:
Koen Vandeputte 2018-08-08 09:54:51 +02:00
parent 5742a2ba35
commit e3ab280e47
260 changed files with 1 additions and 41440 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
include/kernel-version.mk
View File

@ -2,12 +2,10 @@
LINUX_RELEASE?=1
LINUX_VERSION-3.18 = .71
LINUX_VERSION-4.4 = .121
LINUX_VERSION-4.9 = .117
LINUX_VERSION-4.14 = .60
LINUX_KERNEL_HASH-3.18.71 = 5abc9778ad44ce02ed6c8ab52ece8a21c6d20d21f6ed8a19287b4a38a50c1240
LINUX_KERNEL_HASH-4.4.121 = 44a88268b5088dc326b30c9b9133ac35a9a200b636b7268d08f32abeae6ca729
LINUX_KERNEL_HASH-4.9.117 = 8232ea326ac5784d71baa09c0cd7806a2674f8449c0be73e2ae341e6880279b0
LINUX_KERNEL_HASH-4.14.60 = 6ce143258fba2e1f543ff56802e047d7d3a741f746b9c947dd088d4a2873b295

2
include/target.mk
View File

@ -51,7 +51,7 @@ else
endif
endif
ifneq ($(filter 3.18 4.4 4.9,$(KERNEL_PATCHVER)),)
ifneq ($(filter 4.4 4.9,$(KERNEL_PATCHVER)),)
DEFAULT_PACKAGES.router:=$(filter-out kmod-ipt-offload,$(DEFAULT_PACKAGES.router))
endif

129
target/linux/ar7/config-3.18
View File

@ -1,129 +0,0 @@
CONFIG_ADM6996_PHY=y
CONFIG_AR7=y
CONFIG_AR7_TI=y
# CONFIG_AR7_TYPE_AC49X is not set
CONFIG_AR7_TYPE_TI=y
CONFIG_AR7_WDT=y
CONFIG_ARCH_BINFMT_ELF_RANDOMIZE_PIE=y
CONFIG_ARCH_DISCARD_MEMBLOCK=y
CONFIG_ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE=y
CONFIG_ARCH_HAVE_CUSTOM_GPIO_H=y
CONFIG_ARCH_HIBERNATION_POSSIBLE=y
CONFIG_ARCH_MIGHT_HAVE_PC_PARPORT=y
CONFIG_ARCH_MIGHT_HAVE_PC_SERIO=y
CONFIG_ARCH_REQUIRE_GPIOLIB=y
CONFIG_ARCH_SUSPEND_POSSIBLE=y
CONFIG_ARCH_WANT_IPC_PARSE_VERSION=y
CONFIG_BOOT_ELF32=y
CONFIG_CEVT_R4K=y
CONFIG_CLONE_BACKWARDS=y
CONFIG_CMDLINE="rootfstype=squashfs,jffs2"
CONFIG_CMDLINE_BOOL=y
# CONFIG_CMDLINE_OVERRIDE is not set
CONFIG_CPMAC=y
CONFIG_CPU_GENERIC_DUMP_TLB=y
CONFIG_CPU_HAS_PREFETCH=y
CONFIG_CPU_HAS_SYNC=y
CONFIG_CPU_LITTLE_ENDIAN=y
CONFIG_CPU_MIPS32=y
CONFIG_CPU_MIPS32_R1=y
CONFIG_CPU_MIPSR1=y
CONFIG_CPU_R4K_CACHE_TLB=y
CONFIG_CPU_R4K_FPU=y
CONFIG_CPU_SUPPORTS_32BIT_KERNEL=y
CONFIG_CPU_SUPPORTS_HIGHMEM=y
CONFIG_CSRC_R4K=y
CONFIG_DMA_NONCOHERENT=y
CONFIG_EARLY_PRINTK=y
CONFIG_ETHERNET_PACKET_MANGLE=y
CONFIG_FIXED_PHY=y
CONFIG_GENERIC_ATOMIC64=y
CONFIG_GENERIC_CLOCKEVENTS=y
CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
CONFIG_GENERIC_CMOS_UPDATE=y
CONFIG_GENERIC_IO=y
CONFIG_GENERIC_IRQ_SHOW=y
CONFIG_GENERIC_PCI_IOMAP=y
CONFIG_GENERIC_SMP_IDLE_THREAD=y
CONFIG_GPIOLIB=y
CONFIG_GPIO_DEVRES=y
CONFIG_HARDWARE_WATCHPOINTS=y
CONFIG_HAS_DMA=y
CONFIG_HAS_IOMEM=y
CONFIG_HAS_IOPORT=y
# CONFIG_HAVE_64BIT_ALIGNED_ACCESS is not set
CONFIG_HAVE_ARCH_JUMP_LABEL=y
CONFIG_HAVE_ARCH_KGDB=y
CONFIG_HAVE_ARCH_TRACEHOOK=y
# CONFIG_HAVE_BOOTMEM_INFO_NODE is not set
CONFIG_HAVE_CC_STACKPROTECTOR=y
CONFIG_HAVE_CLK=y
CONFIG_HAVE_CONTEXT_TRACKING=y
CONFIG_HAVE_C_RECORDMCOUNT=y
CONFIG_HAVE_DEBUG_KMEMLEAK=y
CONFIG_HAVE_DEBUG_STACKOVERFLOW=y
CONFIG_HAVE_DMA_API_DEBUG=y
CONFIG_HAVE_DMA_ATTRS=y
CONFIG_HAVE_DYNAMIC_FTRACE=y
CONFIG_HAVE_FTRACE_MCOUNT_RECORD=y
CONFIG_HAVE_FUNCTION_GRAPH_TRACER=y
CONFIG_HAVE_FUNCTION_TRACER=y
CONFIG_HAVE_FUNCTION_TRACE_MCOUNT_TEST=y
CONFIG_HAVE_GENERIC_DMA_COHERENT=y
CONFIG_HAVE_IDE=y
CONFIG_HAVE_KERNEL_BZIP2=y
CONFIG_HAVE_KERNEL_GZIP=y
CONFIG_HAVE_KERNEL_LZ4=y
CONFIG_HAVE_KERNEL_LZMA=y
CONFIG_HAVE_KERNEL_LZO=y
CONFIG_HAVE_KERNEL_XZ=y
CONFIG_HAVE_MEMBLOCK=y
CONFIG_HAVE_MEMBLOCK_NODE_MAP=y
CONFIG_HAVE_MOD_ARCH_SPECIFIC=y
CONFIG_HAVE_NET_DSA=y
CONFIG_HAVE_OPROFILE=y
CONFIG_HAVE_PERF_EVENTS=y
CONFIG_HAVE_SYSCALL_TRACEPOINTS=y
CONFIG_HW_RANDOM=y
CONFIG_HZ_PERIODIC=y
CONFIG_INITRAMFS_SOURCE=""
CONFIG_IP17XX_PHY=y
CONFIG_IRQ_CPU=y
CONFIG_IRQ_FORCED_THREADING=y
CONFIG_IRQ_WORK=y
CONFIG_KALLSYMS=y
CONFIG_LEDS_GPIO=y
CONFIG_LEDS_TRIGGER_HEARTBEAT=y
CONFIG_MDIO_BOARDINFO=y
CONFIG_MIPS=y
# CONFIG_MIPS_HUGE_TLB_SUPPORT is not set
CONFIG_MIPS_L1_CACHE_SHIFT=5
# CONFIG_MIPS_MACHINE is not set
CONFIG_MIPS_MT_DISABLED=y
CONFIG_MODULES_USE_ELF_REL=y
# CONFIG_MTD_AC49X_PARTS is not set
CONFIG_MTD_CFI_STAA=y
CONFIG_MTD_PHYSMAP=y
CONFIG_MVSWITCH_PHY=y
CONFIG_NEED_DMA_MAP_STATE=y
CONFIG_NEED_PER_CPU_KM=y
CONFIG_NO_EXCEPT_FILL=y
CONFIG_PAGEFLAGS_EXTENDED=y
CONFIG_PERF_USE_VMALLOC=y
CONFIG_PHYLIB=y
# CONFIG_PREEMPT_RCU is not set
# CONFIG_RCU_STALL_COMMON is not set
# CONFIG_SCSI_DMA is not set
CONFIG_SWAP_IO_SPACE=y
CONFIG_SWCONFIG=y
CONFIG_SYS_HAS_CPU_MIPS32_R1=y
CONFIG_SYS_HAS_EARLY_PRINTK=y
CONFIG_SYS_SUPPORTS_32BIT_KERNEL=y
CONFIG_SYS_SUPPORTS_ARBIT_HZ=y
CONFIG_SYS_SUPPORTS_LITTLE_ENDIAN=y
CONFIG_SYS_SUPPORTS_ZBOOT=y
CONFIG_SYS_SUPPORTS_ZBOOT_UART16550=y
CONFIG_TICK_CPU_ACCOUNTING=y
CONFIG_VLYNQ=y
# CONFIG_VLYNQ_DEBUG is not set
CONFIG_ZONE_DMA_FLAG=0

23
target/linux/ar7/patches-3.18/001-mips-ar7-fix-serial.patch
View File

@ -1,23 +0,0 @@
From 443ab715a40881d6c9ba11b027ba154bac904cb0 Mon Sep 17 00:00:00 2001
From: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Date: Sat, 10 May 2014 23:19:08 +0200
Subject: [PATCH] MIPS/AR7: ensure that serial ports are properly set up
without UPF_FIXED_TYPE, the data from the PORT_AR7 uart_config entry is
never copied, resulting in a dead port.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
---
arch/mips/ar7/platform.c | 1 +
1 file changed, 1 insertion(+)
--- a/arch/mips/ar7/platform.c
+++ b/arch/mips/ar7/platform.c
@@ -581,6 +581,7 @@ static int __init ar7_register_uarts(voi
uart_port.type = PORT_AR7;
uart_port.uartclk = clk_get_rate(bus_clk) / 2;
uart_port.iotype = UPIO_MEM32;
+ uart_port.flags = UPF_FIXED_TYPE;
uart_port.regshift = 2;
uart_port.line = 0;

48
target/linux/ar7/patches-3.18/002-MIPS-AR7-ensure-the-port-type-s-FCR-value-is-used.patch
View File

@ -1,48 +0,0 @@
From ee6c9d41de084b2cefd90e5e0c9f30a35f6d3967 Mon Sep 17 00:00:00 2001
From: Jonas Gorski <jonas.gorski@gmail.com>
Date: Sun, 29 Oct 2017 15:50:42 +0100
Subject: [PATCH RFC 3/3] MIPS: AR7: ensure the port type's FCR value is used
Since commit aef9a7bd9b67 ("serial/uart/8250: Add tunable RX interrupt
trigger I/F of FIFO buffers"), the port's default FCR value isn't used
in serial8250_do_set_termios anymore, but copied over once in
serial8250_config_port and then modified as needed.
Unfortunately, serial8250_config_port will never be called if the port
is shared between kernel and userspace, and the port's flag doesn't have
UPF_BOOT_AUTOCONF, which would trigger a serial8250_config_port as well.
This causes garbled output from userspace:
[ 5.220000] random: procd urandom read with 49 bits of entropy available
ers
[kee
Fix this by forcing it to be configured on boot, resulting in the
expected output:
[ 5.250000] random: procd urandom read with 50 bits of entropy available
Press the [f] key and hit [enter] to enter failsafe mode
Press the [1], [2], [3] or [4] key and hit [enter] to select the debug level
Fixes: aef9a7bd9b67 ("serial/uart/8250: Add tunable RX interrupt trigger I/F of FIFO buffers")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
---
I'm not sure if this is just AR7's issue, or if this points to a general
issue for UARTs used as kernel console and login console with the "fixed"
commit.
arch/mips/ar7/platform.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/arch/mips/ar7/platform.c
+++ b/arch/mips/ar7/platform.c
@@ -581,7 +581,7 @@ static int __init ar7_register_uarts(voi
uart_port.type = PORT_AR7;
uart_port.uartclk = clk_get_rate(bus_clk) / 2;
uart_port.iotype = UPIO_MEM32;
- uart_port.flags = UPF_FIXED_TYPE;
+ uart_port.flags = UPF_FIXED_TYPE | UPF_BOOT_AUTOCONF;
uart_port.regshift = 2;
uart_port.line = 0;

11
target/linux/ar7/patches-3.18/100-fix-highmem-offset.patch
View File

@ -1,11 +0,0 @@
--- a/arch/mips/include/asm/mach-ar7/spaces.h
+++ b/arch/mips/include/asm/mach-ar7/spaces.h
@@ -20,6 +20,8 @@
#define UNCAC_BASE _AC(0xb4000000, UL) /* 0xa0000000 + PHYS_OFFSET */
#define IO_BASE UNCAC_BASE
+#define HIGHMEM_START _AC(0x20000000, UL)
+
#include <asm/mach-generic/spaces.h>
#endif /* __ASM_AR7_SPACES_H */

45
target/linux/ar7/patches-3.18/101-MIPS-AR7-allow-NULL-clock-for-clk_get_rate.patch
View File

@ -1,45 +0,0 @@
From patchwork Tue Jul 18 10:17:26 2017
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: [5/9] MIPS: AR7: allow NULL clock for clk_get_rate
X-Patchwork-Submitter: Jonas Gorski <jonas.gorski@gmail.com>
X-Patchwork-Id: 16775
Message-Id: <20170718101730.2541-6-jonas.gorski@gmail.com>
To: unlisted-recipients:; (no To-header on input)
Cc: Ralf Baechle <ralf@linux-mips.org>,
Paul Gortmaker <paul.gortmaker@windriver.com>,
James Hogan <james.hogan@imgtec.com>,
linux-mips@linux-mips.org, linux-kernel@vger.kernel.org
Date: Tue, 18 Jul 2017 12:17:26 +0200
From: Jonas Gorski <jonas.gorski@gmail.com>
List-Id: linux-mips <linux-mips.eddie.linux-mips.org>
Make the behaviour of clk_get_rate consistent with common clk's
clk_get_rate by accepting NULL clocks as parameter. Some device
drivers rely on this, and will cause an OOPS otherwise.
Fixes: 780019ddf02f ("MIPS: AR7: Implement clock API")
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Reported-by: Mathias Kresin <dev@kresin.me>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
---
arch/mips/ar7/clock.c | 3 +++
1 file changed, 3 insertions(+)
--- a/arch/mips/ar7/clock.c
+++ b/arch/mips/ar7/clock.c
@@ -430,6 +430,9 @@ EXPORT_SYMBOL(clk_disable);
unsigned long clk_get_rate(struct clk *clk)
{
+ if (!clk)
+ return 0;
+
return clk->rate;
}
EXPORT_SYMBOL(clk_get_rate);

22
target/linux/ar7/patches-3.18/110-flash.patch
View File

@ -1,22 +0,0 @@
--- a/drivers/mtd/Makefile
+++ b/drivers/mtd/Makefile
@@ -12,7 +12,7 @@ obj-$(CONFIG_MTD_OF_PARTS) += ofpart.o
obj-$(CONFIG_MTD_REDBOOT_PARTS) += redboot.o
obj-$(CONFIG_MTD_CMDLINE_PARTS) += cmdlinepart.o
obj-$(CONFIG_MTD_AFS_PARTS) += afs.o
-obj-$(CONFIG_MTD_AR7_PARTS) += ar7part.o
+obj-$(CONFIG_MTD_AR7_PARTS) += ar7part.o titanpart.o
obj-$(CONFIG_MTD_BCM63XX_PARTS) += bcm63xxpart.o
obj-$(CONFIG_MTD_BCM47XX_PARTS) += bcm47xxpart.o
obj-$(CONFIG_MTD_MYLOADER_PARTS) += myloader.o
--- a/arch/mips/ar7/platform.c
+++ b/arch/mips/ar7/platform.c
@@ -199,7 +199,7 @@ static struct resource physmap_flash_res
.name = "mem",
.flags = IORESOURCE_MEM,
.start = 0x10000000,
- .end = 0x107fffff,
+ .end = 0x11ffffff,
};
static const char *ar7_probe_types[] = { "ar7part", NULL };

300
target/linux/ar7/patches-3.18/160-vlynq_try_remote_first.patch
View File

@ -1,300 +0,0 @@
--- a/drivers/vlynq/vlynq.c
+++ b/drivers/vlynq/vlynq.c
@@ -119,20 +119,40 @@ static int vlynq_linked(struct vlynq_dev
return 0;
}
+static volatile int vlynq_delay_value_new = 0;
+
+static void vlynq_delay_wait(u32 count)
+{
+ /* Code adopted from original vlynq driver */
+ int i = 0;
+ volatile int *ptr = &vlynq_delay_value_new;
+ *ptr = 0;
+
+ /* We are assuming that the each cycle takes about
+ * 23 assembly instructions. */
+ for(i = 0; i < (count + 23)/23; i++)
+ *ptr = *ptr + 1;
+}
+
static void vlynq_reset(struct vlynq_device *dev)
{
+ u32 rtm = readl(&dev->local->revision);
+
+ rtm = rtm < 0x00010205 || readl(&dev->local->status) & 0x800 == 0 ?
+ 0 : 0x600000;
+
writel(readl(&dev->local->control) | VLYNQ_CTRL_RESET,
&dev->local->control);
/* Wait for the devices to finish resetting */
- msleep(5);
+ vlynq_delay_wait(0xffffff);
/* Remove reset bit */
- writel(readl(&dev->local->control) & ~VLYNQ_CTRL_RESET,
+ writel(readl(&dev->local->control) & ~VLYNQ_CTRL_RESET | rtm,
&dev->local->control);
/* Give some time for the devices to settle */
- msleep(5);
+ vlynq_delay_wait(0xffffff);
}
static void vlynq_irq_unmask(struct irq_data *d)
@@ -379,6 +399,61 @@ void vlynq_unregister_driver(struct vlyn
}
EXPORT_SYMBOL(vlynq_unregister_driver);
+enum vlynq_clk_src {
+ vlynq_clk_external,
+ vlynq_clk_local,
+ vlynq_clk_remote,
+ vlynq_clk_invalid,
+};
+
+static int __vlynq_set_clocks(struct vlynq_device *dev,
+ enum vlynq_clk_src clk_dir,
+ int lclk_div, int rclk_div)
+{
+ u32 reg;
+
+ if (clk_dir == vlynq_clk_invalid) {
+ printk(KERN_ERR "%s: attempt to set invalid clocking\n",
+ dev_name(&dev->dev));
+ return -EINVAL;
+ }
+
+ reg = readl(&dev->local->control);
+ if (readl(&dev->local->revision) < 0x00010205) {
+ if (clk_dir & vlynq_clk_local)
+ reg |= VLYNQ_CTRL_CLOCK_INT;
+ else
+ reg &= ~VLYNQ_CTRL_CLOCK_INT;
+ }
+ reg &= ~VLYNQ_CTRL_CLOCK_MASK;
+ reg |= VLYNQ_CTRL_CLOCK_DIV(lclk_div);
+ writel(reg, &dev->local->control);
+
+ if (!vlynq_linked(dev))
+ return -ENODEV;
+
+ printk(KERN_INFO "%s: local VLYNQ protocol rev. is 0x%08x\n",
+ dev_name(&dev->dev), readl(&dev->local->revision));
+ printk(KERN_INFO "%s: remote VLYNQ protocol rev. is 0x%08x\n",
+ dev_name(&dev->dev), readl(&dev->remote->revision));
+
+ reg = readl(&dev->remote->control);
+ if (readl(&dev->remote->revision) < 0x00010205) {
+ if (clk_dir & vlynq_clk_remote)
+ reg |= VLYNQ_CTRL_CLOCK_INT;
+ else
+ reg &= ~VLYNQ_CTRL_CLOCK_INT;
+ }
+ reg &= ~VLYNQ_CTRL_CLOCK_MASK;
+ reg |= VLYNQ_CTRL_CLOCK_DIV(rclk_div);
+ writel(reg, &dev->remote->control);
+
+ if (!vlynq_linked(dev))
+ return -ENODEV;
+
+ return 0;
+}
+
/*
* A VLYNQ remote device can clock the VLYNQ bus master
* using a dedicated clock line. In that case, both the
@@ -392,29 +467,16 @@ static int __vlynq_try_remote(struct vly
int i;
vlynq_reset(dev);
- for (i = dev->dev_id ? vlynq_rdiv2 : vlynq_rdiv8; dev->dev_id ?
- i <= vlynq_rdiv8 : i >= vlynq_rdiv2;
- dev->dev_id ? i++ : i--) {
+ for (i = 0; i <= 7; i++) {
if (!vlynq_linked(dev))
break;
- writel((readl(&dev->remote->control) &
- ~VLYNQ_CTRL_CLOCK_MASK) |
- VLYNQ_CTRL_CLOCK_INT |
- VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1),
- &dev->remote->control);
- writel((readl(&dev->local->control)
- & ~(VLYNQ_CTRL_CLOCK_INT |
- VLYNQ_CTRL_CLOCK_MASK)) |
- VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1),
- &dev->local->control);
-
- if (vlynq_linked(dev)) {
- printk(KERN_DEBUG
- "%s: using remote clock divisor %d\n",
- dev_name(&dev->dev), i - vlynq_rdiv1 + 1);
- dev->divisor = i;
+ if (!__vlynq_set_clocks(dev, vlynq_clk_remote, i, i)) {
+ printk(KERN_INFO
+ "%s: using remote clock divisor %d\n",
+ dev_name(&dev->dev), i + 1);
+ dev->divisor = i + vlynq_rdiv1;
return 0;
} else {
vlynq_reset(dev);
@@ -433,25 +495,17 @@ static int __vlynq_try_remote(struct vly
*/
static int __vlynq_try_local(struct vlynq_device *dev)
{
- int i;
+ int i, dir = !dev->dev_id;
vlynq_reset(dev);
- for (i = dev->dev_id ? vlynq_ldiv2 : vlynq_ldiv8; dev->dev_id ?
- i <= vlynq_ldiv8 : i >= vlynq_ldiv2;
- dev->dev_id ? i++ : i--) {
-
- writel((readl(&dev->local->control) &
- ~VLYNQ_CTRL_CLOCK_MASK) |
- VLYNQ_CTRL_CLOCK_INT |
- VLYNQ_CTRL_CLOCK_DIV(i - vlynq_ldiv1),
- &dev->local->control);
-
- if (vlynq_linked(dev)) {
- printk(KERN_DEBUG
- "%s: using local clock divisor %d\n",
- dev_name(&dev->dev), i - vlynq_ldiv1 + 1);
- dev->divisor = i;
+ for (i = dir ? 7 : 0; dir ? i >= 0 : i <= 7; dir ? i-- : i++) {
+
+ if (!__vlynq_set_clocks(dev, vlynq_clk_local, i, 0)) {
+ printk(KERN_INFO
+ "%s: using local clock divisor %d\n",
+ dev_name(&dev->dev), i + 1);
+ dev->divisor = i + vlynq_ldiv1;
return 0;
} else {
vlynq_reset(dev);
@@ -473,18 +527,10 @@ static int __vlynq_try_external(struct v
if (!vlynq_linked(dev))
return -ENODEV;
- writel((readl(&dev->remote->control) &
- ~VLYNQ_CTRL_CLOCK_INT),
- &dev->remote->control);
-
- writel((readl(&dev->local->control) &
- ~VLYNQ_CTRL_CLOCK_INT),
- &dev->local->control);
-
- if (vlynq_linked(dev)) {
- printk(KERN_DEBUG "%s: using external clock\n",
- dev_name(&dev->dev));
- dev->divisor = vlynq_div_external;
+ if (!__vlynq_set_clocks(dev, vlynq_clk_external, 0, 0)) {
+ printk(KERN_INFO "%s: using external clock\n",
+ dev_name(&dev->dev));
+ dev->divisor = vlynq_div_external;
return 0;
}
@@ -501,24 +547,16 @@ static int __vlynq_enable_device(struct
return result;
switch (dev->divisor) {
- case vlynq_div_external:
case vlynq_div_auto:
/* When the device is brought from reset it should have clock
* generation negotiated by hardware.
* Check which device is generating clocks and perform setup
* accordingly */
- if (vlynq_linked(dev) && readl(&dev->remote->control) &
- VLYNQ_CTRL_CLOCK_INT) {
- if (!__vlynq_try_remote(dev) ||
- !__vlynq_try_local(dev) ||
- !__vlynq_try_external(dev))
- return 0;
- } else {
- if (!__vlynq_try_external(dev) ||
- !__vlynq_try_local(dev) ||
- !__vlynq_try_remote(dev))
- return 0;
- }
+ if (!__vlynq_try_remote(dev) || !__vlynq_try_local(dev))
+ return 0;
+ case vlynq_div_external:
+ if (!__vlynq_try_external(dev))
+ return 0;
break;
case vlynq_ldiv1:
case vlynq_ldiv2:
@@ -528,15 +566,12 @@ static int __vlynq_enable_device(struct
case vlynq_ldiv6:
case vlynq_ldiv7:
case vlynq_ldiv8:
- writel(VLYNQ_CTRL_CLOCK_INT |
- VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
- vlynq_ldiv1), &dev->local->control);
- writel(0, &dev->remote->control);
- if (vlynq_linked(dev)) {
- printk(KERN_DEBUG
- "%s: using local clock divisor %d\n",
- dev_name(&dev->dev),
- dev->divisor - vlynq_ldiv1 + 1);
+ if (!__vlynq_set_clocks(dev, vlynq_clk_local, dev->divisor -
+ vlynq_ldiv1, 0)) {
+ printk(KERN_INFO
+ "%s: using local clock divisor %d\n",
+ dev_name(&dev->dev),
+ dev->divisor - vlynq_ldiv1 + 1);
return 0;
}
break;
@@ -548,20 +583,17 @@ static int __vlynq_enable_device(struct
case vlynq_rdiv6:
case vlynq_rdiv7:
case vlynq_rdiv8:
- writel(0, &dev->local->control);
- writel(VLYNQ_CTRL_CLOCK_INT |
- VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
- vlynq_rdiv1), &dev->remote->control);
- if (vlynq_linked(dev)) {
- printk(KERN_DEBUG
- "%s: using remote clock divisor %d\n",
- dev_name(&dev->dev),
- dev->divisor - vlynq_rdiv1 + 1);
+ if (!__vlynq_set_clocks(dev, vlynq_clk_remote, 0,
+ dev->divisor - vlynq_rdiv1)) {
+ printk(KERN_INFO
+ "%s: using remote clock divisor %d\n",
+ dev_name(&dev->dev),
+ dev->divisor - vlynq_rdiv1 + 1);
return 0;
}
break;
}
-
+ vlynq_reset(dev);
ops->off(dev);
return -ENODEV;
}
@@ -732,14 +764,14 @@ static int vlynq_probe(struct platform_d
platform_set_drvdata(pdev, dev);
printk(KERN_INFO "%s: regs 0x%p, irq %d, mem 0x%p\n",
- dev_name(&dev->dev), (void *)dev->regs_start, dev->irq,
- (void *)dev->mem_start);
+ dev_name(&dev->dev), (void *)dev->regs_start,
+ dev->irq, (void *)dev->mem_start);
dev->dev_id = 0;
dev->divisor = vlynq_div_auto;
- result = __vlynq_enable_device(dev);
- if (result == 0) {
+ if (!__vlynq_enable_device(dev)) {
dev->dev_id = readl(&dev->remote->chip);
+ vlynq_reset(dev);
((struct plat_vlynq_ops *)(dev->dev.platform_data))->off(dev);
}
if (dev->dev_id)

15
target/linux/ar7/patches-3.18/200-free-mem-below-kernel-offset.patch
View File

@ -1,15 +0,0 @@
--- a/arch/mips/ar7/memory.c
+++ b/arch/mips/ar7/memory.c
@@ -66,5 +66,11 @@ void __init prom_meminit(void)
void __init prom_free_prom_memory(void)
{
- /* Nothing to free */
+ /* adapted from arch/mips/txx9/generic/setup.c */
+ unsigned long saddr = PHYS_OFFSET + PAGE_SIZE;
+ unsigned long eaddr = __pa_symbol(&_text);
+
+ /* free memory between prom-record and kernel _text base */
+ if (saddr < eaddr)
+ free_init_pages("prom memory", saddr, eaddr);
}

85
target/linux/ar7/patches-3.18/300-add-ac49x-platform.patch
View File

@ -1,85 +0,0 @@
--- a/arch/mips/ar7/Platform
+++ b/arch/mips/ar7/Platform
@@ -3,4 +3,9 @@
#
platform-$(CONFIG_AR7) += ar7/
cflags-$(CONFIG_AR7) += -I$(srctree)/arch/mips/include/asm/mach-ar7
-load-$(CONFIG_AR7) += 0xffffffff94100000
+load-$(CONFIG_AR7_TI) += 0xffffffff94100000
+
+#
+# AudioCodes AC49x
+#
+load-$(CONFIG_AR7_AC49X) += 0xffffffff945ca000
--- a/arch/mips/ar7/setup.c
+++ b/arch/mips/ar7/setup.c
@@ -69,6 +69,10 @@ const char *get_system_type(void)
return "TI AR7 (TNETV1056)";
case TITAN_CHIP_1060:
return "TI AR7 (TNETV1060)";
+ case TITAN_CHIP_AC495:
+ return "AudioCodes AC495";
+ case TITAN_CHIP_AC496:
+ return "AudioCodes AC496";
}
default:
return "TI AR7 (unknown)";
--- a/arch/mips/include/asm/mach-ar7/ar7.h
+++ b/arch/mips/include/asm/mach-ar7/ar7.h
@@ -92,6 +92,8 @@
#define TITAN_CHIP_1055 0x0e
#define TITAN_CHIP_1056 0x0d
#define TITAN_CHIP_1060 0x07
+#define TITAN_CHIP_AC495 0x00
+#define TITAN_CHIP_AC496 0x02
/* Interrupts */
#define AR7_IRQ_UART0 15
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -94,7 +94,7 @@ config AR7
select HAVE_CLK
help
Support for the Texas Instruments AR7 System-on-a-Chip
- family: TNETD7100, 7200 and 7300.
+ family: TI TNETD7100, 7200, 7300 and AudioCodes AC49x.
config ATH79
bool "Atheros AR71XX/AR724X/AR913X based boards"
@@ -835,6 +835,7 @@ config MIPS_PARAVIRT
endchoice
source "arch/mips/alchemy/Kconfig"
+source "arch/mips/ar7/Kconfig"
source "arch/mips/ath79/Kconfig"
source "arch/mips/bcm47xx/Kconfig"
source "arch/mips/bcm63xx/Kconfig"
--- /dev/null
+++ b/arch/mips/ar7/Kconfig
@@ -0,0 +1,26 @@
+if AR7
+
+config AR7_TI
+ bool
+
+config AR7_AC49X
+ bool
+
+choice
+ prompt "AR7 SoC family selection"
+ default AR7_TYPE_TI
+ depends on AR7
+ help
+ Select AR7 MIPS SoC implementation.
+
+ config AR7_TYPE_TI
+ bool "Texas Instruments AR7"
+ select AR7_TI
+
+ config AR7_TYPE_AC49X
+ bool "AudioCodes AC49X"
+ select AR7_AC49X
+
+endchoice
+
+endif

20
target/linux/ar7/patches-3.18/310-ac49x-prom-support.patch
View File

@ -1,20 +0,0 @@
--- a/arch/mips/ar7/prom.c
+++ b/arch/mips/ar7/prom.c
@@ -70,6 +70,7 @@ struct psbl_rec {
};
static const char psp_env_version[] __initconst = "TIENV0.8";
+static const char psp_env_version_ac49x[] __initconst = "MaxENV0.2";
struct psp_env_chunk {
u8 num;
@@ -186,7 +187,8 @@ static void __init ar7_init_env(struct e
struct psbl_rec *psbl = (struct psbl_rec *)(KSEG1ADDR(0x14000300));
void *psp_env = (void *)KSEG1ADDR(psbl->env_base);
- if (strcmp(psp_env, psp_env_version) == 0) {
+ if (strcmp(psp_env, psp_env_version) == 0 ||
+ strcmp(psp_env, psp_env_version_ac49x) == 0) {
parse_psp_env(psp_env);
} else {
for (i = 0; i < MAX_ENTRY; i++, env++)

35
target/linux/ar7/patches-3.18/320-ac49x-mtd-partitions.patch
View File

@ -1,35 +0,0 @@
--- a/drivers/mtd/Kconfig
+++ b/drivers/mtd/Kconfig
@@ -164,6 +164,11 @@ config MTD_OF_PARTS
the partition map from the children of the flash node,
as described in Documentation/devicetree/booting-without-of.txt.
+config MTD_AC49X_PARTS
+ tristate "AudioCodes AC49X partitioning support"
+ ---help---
+ AudioCodes AC49X partitioning support
+
config MTD_AR7_PARTS
tristate "TI AR7 partitioning support"
---help---
--- a/drivers/mtd/Makefile
+++ b/drivers/mtd/Makefile
@@ -11,6 +11,7 @@ obj-$(CONFIG_MTD_SPLIT) += mtdsplit/
obj-$(CONFIG_MTD_OF_PARTS) += ofpart.o
obj-$(CONFIG_MTD_REDBOOT_PARTS) += redboot.o
obj-$(CONFIG_MTD_CMDLINE_PARTS) += cmdlinepart.o
+obj-$(CONFIG_MTD_AC49X_PARTS) += ac49xpart.o
obj-$(CONFIG_MTD_AFS_PARTS) += afs.o
obj-$(CONFIG_MTD_AR7_PARTS) += ar7part.o titanpart.o
obj-$(CONFIG_MTD_BCM63XX_PARTS) += bcm63xxpart.o
--- a/arch/mips/ar7/platform.c
+++ b/arch/mips/ar7/platform.c
@@ -202,7 +202,7 @@ static struct resource physmap_flash_res
.end = 0x11ffffff,
};
-static const char *ar7_probe_types[] = { "ar7part", NULL };
+static const char *ar7_probe_types[] = { "ac49xpart", "ar7part", NULL };
static struct physmap_flash_data physmap_flash_data = {
.width = 2,

118
target/linux/ar7/patches-3.18/920-ar7part.patch
View File

@ -1,118 +0,0 @@
--- a/drivers/mtd/ar7part.c
+++ b/drivers/mtd/ar7part.c
@@ -30,11 +30,14 @@
#include <uapi/linux/magic.h>
+#include <asm/mach-ar7/prom.h>
+
#define AR7_PARTS 4
#define ROOT_OFFSET 0xe0000
#define LOADER_MAGIC1 le32_to_cpu(0xfeedfa42)
#define LOADER_MAGIC2 le32_to_cpu(0xfeed1281)
+#define LOADER_MAGIC3 le32_to_cpu(0x434d4d4c)
struct ar7_bin_rec {
unsigned int checksum;
@@ -42,12 +45,16 @@ struct ar7_bin_rec {
unsigned int address;
};
+int create_titan_partitions(struct mtd_info *master,
+ struct mtd_partition **pparts,
+ struct mtd_part_parser_data *data);
+
static int create_mtd_partitions(struct mtd_info *master,
struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct ar7_bin_rec header;
- unsigned int offset;
+ unsigned int offset, mtd_start, mtd_end;
size_t len;
unsigned int pre_size = master->erasesize, post_size = 0;
unsigned int root_offset = ROOT_OFFSET;
@@ -55,6 +62,16 @@ static int create_mtd_partitions(struct
int retries = 10;
struct mtd_partition *ar7_parts;
+ const char *prom_str = prom_getenv("ProductID");
+ char mtd_name[] = "mtd1";
+ if(prom_str &&
+ (strcmp(prom_str, "CYWL")==0 ||
+ strcmp(prom_str, "CYWM")==0 ||
+ strcmp(prom_str, "CYLM")==0 ||
+ strcmp(prom_str, "CYLL")==0)){
+ return create_titan_partitions(master, pparts, data);
+ }
+
ar7_parts = kzalloc(sizeof(*ar7_parts) * AR7_PARTS, GFP_KERNEL);
if (!ar7_parts)
return -ENOMEM;
@@ -83,34 +100,39 @@ static int create_mtd_partitions(struct
pre_size = offset;
- if (!ar7_parts[1].offset) {
- ar7_parts[1].offset = master->size - master->erasesize;
- post_size = master->erasesize;
- }
-
switch (header.checksum) {
- case LOADER_MAGIC1:
- while (header.length) {
- offset += sizeof(header) + header.length;
- mtd_read(master, offset, sizeof(header), &len,
- (uint8_t *)&header);
- }
- root_offset = offset + sizeof(header) + 4;
- break;
case LOADER_MAGIC2:
+ for (retries = 0; retries <= 9; retries++) {
+ mtd_name[3] = '0' + retries;
+ prom_str = prom_getenv(mtd_name);
+ if (prom_str == NULL)
+ continue;
+ sscanf(prom_str, "%i,%i", &mtd_start, &mtd_end);
+ if (pre_size == (mtd_start & 0x1ffffff)) {
+ ar7_parts[1].offset = mtd_end &= 0x1ffffff;
+ ar7_parts[1].size = post_size = master->size - mtd_end;
+ break;
+ }
+ }
+ case LOADER_MAGIC1:
+ root_offset = (header.checksum == LOADER_MAGIC1) ? 4 : 0;
while (header.length) {
offset += sizeof(header) + header.length;
mtd_read(master, offset, sizeof(header), &len,
(uint8_t *)&header);
}
- root_offset = offset + sizeof(header) + 4 + 0xff;
- root_offset &= ~(uint32_t)0xff;
+ root_offset += offset + sizeof(header);
break;
default:
printk(KERN_WARNING "Unknown magic: %08x\n", header.checksum);
break;
}
+ if (!ar7_parts[1].offset) {
+ post_size = master->erasesize;
+ ar7_parts[1].offset = master->size - post_size;
+ }
+
mtd_read(master, root_offset, sizeof(header), &len, (u8 *)&header);
if (header.checksum != SQUASHFS_MAGIC) {
root_offset += master->erasesize - 1;
--- a/drivers/mtd/titanpart.c
+++ b/drivers/mtd/titanpart.c
@@ -149,7 +149,7 @@ static void titan_add_partition(char * e
}
int create_titan_partitions(struct mtd_info *master,
struct mtd_partition **pparts,
- unsigned long origin)
+ struct mtd_part_parser_data *data)
{
struct nsp_img_hdr_head hdr;
struct nsp_img_hdr_section_info sect_info;

95
target/linux/ar7/patches-3.18/925-actiontec_leds.patch
View File

@ -1,95 +0,0 @@
--- a/arch/mips/ar7/platform.c
+++ b/arch/mips/ar7/platform.c
@@ -465,31 +465,22 @@ static struct gpio_led fb_fon_leds[] = {
},
};
-static struct gpio_led gt701_leds[] = {
+static struct gpio_led actiontec_leds[] = {
{
.name = "inet:green",
.gpio = 13,
- .active_low = 1,
- },
- {
- .name = "usb",
- .gpio = 12,
- .active_low = 1,
},
{
.name = "inet:red",
.gpio = 9,
- .active_low = 1,
},
{
- .name = "power:red",
+ .name = "power:green",
.gpio = 7,
- .active_low = 1,
},
{
- .name = "power:green",
+ .name = "power:red",
.gpio = 8,
- .active_low = 1,
.default_trigger = "default-on",
},
{
@@ -497,6 +488,44 @@ static struct gpio_led gt701_leds[] = {
.gpio = 10,
.active_low = 1,
},
+ {
+ .name = "wifi",
+ .gpio = 6,
+ .active_low = 1,
+ },
+ {
+ .name = "wifi:red",
+ .gpio = 3,
+ },
+ {
+ .name = "standby",
+ .gpio = 4,
+ },
+ {
+ .name = "wps",
+ .gpio = 16,
+ .active_low = 1,
+ },
+ {
+ .name = "usb",
+ .gpio = 12,
+ .active_low = 1,
+ },
+ {
+ .name = "voip",
+ .gpio = 15,
+ .active_low = 1,
+ },
+ {
+ .name = "line1",
+ .gpio = 23,
+ .active_low = 1,
+ },
+ {
+ .name = "line2",
+ .gpio = 25,
+ .active_low = 1,
+ },
};
static struct gpio_led_platform_data ar7_led_data;
@@ -540,9 +569,9 @@ static void __init detect_leds(void)
} else if (strstr(prid, "CYWM") || strstr(prid, "CYWL")) {
ar7_led_data.num_leds = ARRAY_SIZE(titan_leds);
ar7_led_data.leds = titan_leds;
- } else if (strstr(prid, "GT701")) {
- ar7_led_data.num_leds = ARRAY_SIZE(gt701_leds);
- ar7_led_data.leds = gt701_leds;
+ } else if (strstr(prid, "GT7") || strstr(prid, "PK5000")) {
+ ar7_led_data.num_leds = ARRAY_SIZE(actiontec_leds);
+ ar7_led_data.leds = actiontec_leds;
}
}

52
target/linux/ar7/patches-3.18/950-cpmac_titan.patch
View File

@ -1,52 +0,0 @@
--- a/drivers/net/ethernet/ti/cpmac.c
+++ b/drivers/net/ethernet/ti/cpmac.c
@@ -1147,6 +1147,8 @@ static int cpmac_probe(struct platform_d
goto out;
}
+ ar7_device_reset(pdata->reset_bit);
+
dev->irq = platform_get_irq_byname(pdev, "irq");
dev->netdev_ops = &cpmac_netdev_ops;
@@ -1228,7 +1230,7 @@ int cpmac_init(void)
cpmac_mii->reset = cpmac_mdio_reset;
cpmac_mii->irq = mii_irqs;
- cpmac_mii->priv = ioremap(AR7_REGS_MDIO, 256);
+ cpmac_mii->priv = ioremap(ar7_is_titan() ? TITAN_REGS_MDIO : AR7_REGS_MDIO, 256);
if (!cpmac_mii->priv) {
pr_err("Can't ioremap mdio registers\n");
@@ -1239,10 +1241,16 @@ int cpmac_init(void)
/* FIXME: unhardcode gpio&reset bits */
ar7_gpio_disable(26);
ar7_gpio_disable(27);
- ar7_device_reset(AR7_RESET_BIT_CPMAC_LO);
- ar7_device_reset(AR7_RESET_BIT_CPMAC_HI);
+
+ if (!ar7_is_titan()) {
+ ar7_device_reset(AR7_RESET_BIT_CPMAC_LO);
+ ar7_device_reset(AR7_RESET_BIT_CPMAC_HI);
+ }
ar7_device_reset(AR7_RESET_BIT_EPHY);
+ if (ar7_is_titan())
+ ar7_device_reset(TITAN_RESET_BIT_EPHY1);
+
cpmac_mii->reset(cpmac_mii);
for (i = 0; i < 300; i++) {
@@ -1259,7 +1267,11 @@ int cpmac_init(void)
mask = 0;
}
- cpmac_mii->phy_mask = ~(mask | 0x80000000);
+ if (ar7_is_titan())
+ cpmac_mii->phy_mask = ~(mask | 0x80000000 | 0x40000000);
+ else
+ cpmac_mii->phy_mask = ~(mask | 0x80000000);
+
snprintf(cpmac_mii->id, MII_BUS_ID_SIZE, "cpmac-1");
res = mdiobus_register(cpmac_mii);

4657
target/linux/generic/config-3.18
View File

File diff suppressed because it is too large Load Diff

142
target/linux/generic/pending-3.18/001-mtdsplit_backport.patch
View File

@ -1,142 +0,0 @@
--- a/drivers/mtd/mtdsplit/mtdsplit_brnimage.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_brnimage.c
@@ -27,7 +27,7 @@
#define BRNIMAGE_MAX_OVERHEAD (BRNIMAGE_ALIGN_BYTES + BRNIMAGE_FOOTER_SIZE)
static int mtdsplit_parse_brnimage(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct mtd_partition *parts;
--- a/drivers/mtd/mtdsplit/mtdsplit_eva.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_eva.c
@@ -29,7 +29,7 @@ struct eva_image_header {
};
static int mtdsplit_parse_eva(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct mtd_partition *parts;
--- a/drivers/mtd/mtdsplit/mtdsplit_fit.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_fit.c
@@ -45,8 +45,7 @@ struct fdt_header {
};
static int
-mtdsplit_fit_parse(struct mtd_info *mtd,
- const struct mtd_partition **pparts,
+mtdsplit_fit_parse(struct mtd_info *mtd, struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct fdt_header hdr;
--- a/drivers/mtd/mtdsplit/mtdsplit_lzma.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_lzma.c
@@ -28,7 +28,7 @@ struct lzma_header {
};
static int mtdsplit_parse_lzma(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct lzma_header hdr;
--- a/drivers/mtd/mtdsplit/mtdsplit_seama.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_seama.c
@@ -30,7 +30,7 @@ struct seama_header {
};
static int mtdsplit_parse_seama(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct seama_header hdr;
--- a/drivers/mtd/mtdsplit/mtdsplit_squashfs.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_squashfs.c
@@ -23,7 +23,7 @@
static int
mtdsplit_parse_squashfs(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct mtd_partition *part;
--- a/drivers/mtd/mtdsplit/mtdsplit_tplink.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_tplink.c
@@ -83,8 +83,8 @@ struct tplink_fw_header {
};
static int mtdsplit_parse_tplink(struct mtd_info *master,
- const struct mtd_partition **pparts,
- struct mtd_part_parser_data *data)
+ struct mtd_partition **pparts,
+ struct mtd_part_parser_data *data)
{
struct tplink_fw_header hdr;
size_t hdr_len, retlen, kernel_size;
--- a/drivers/mtd/mtdsplit/mtdsplit_trx.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_trx.c
@@ -56,7 +56,7 @@ read_trx_header(struct mtd_info *mtd, si
static int
mtdsplit_parse_trx(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct mtd_partition *parts;
--- a/drivers/mtd/mtdsplit/mtdsplit_uimage.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_uimage.c
@@ -81,7 +81,7 @@ read_uimage_header(struct mtd_info *mtd,
* of a valid uImage header if found
*/
static int __mtdsplit_parse_uimage(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data,
ssize_t (*find_header)(u_char *buf, size_t len))
{
@@ -232,7 +232,7 @@ static ssize_t uimage_verify_default(u_c
static int
mtdsplit_uimage_parse_generic(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
return __mtdsplit_parse_uimage(master, pparts, data,
@@ -289,7 +289,7 @@ static ssize_t uimage_verify_wndr3700(u_
static int
mtdsplit_uimage_parse_netgear(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
return __mtdsplit_parse_uimage(master, pparts, data,
@@ -331,7 +331,7 @@ static ssize_t uimage_find_edimax(u_char
static int
mtdsplit_uimage_parse_edimax(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
return __mtdsplit_parse_uimage(master, pparts, data,
--- a/drivers/mtd/mtdsplit/mtdsplit_wrgg.c
+++ b/drivers/mtd/mtdsplit/mtdsplit_wrgg.c
@@ -40,8 +40,8 @@ struct wrgg03_header {
static int mtdsplit_parse_wrgg(struct mtd_info *master,
- const struct mtd_partition **pparts,
- struct mtd_part_parser_data *data)
+ struct mtd_partition **pparts,
+ struct mtd_part_parser_data *data)
{
struct wrgg03_header hdr;
size_t hdr_len, retlen, kernel_ent_size;

552
target/linux/generic/pending-3.18/002-phy_drivers_backport.patch
View File

@ -1,552 +0,0 @@
--- a/drivers/net/phy/adm6996.c
+++ b/drivers/net/phy/adm6996.c
@@ -289,7 +289,7 @@ static u16
adm6996_read_mii_reg(struct adm6996_priv *priv, enum admreg reg)
{
struct phy_device *phydev = priv->priv;
- struct mii_bus *bus = phydev->mdio.bus;
+ struct mii_bus *bus = phydev->bus;
return bus->read(bus, PHYADDR(reg));
}
@@ -298,7 +298,7 @@ static void
adm6996_write_mii_reg(struct adm6996_priv *priv, enum admreg reg, u16 val)
{
struct phy_device *phydev = priv->priv;
- struct mii_bus *bus = phydev->mdio.bus;
+ struct mii_bus *bus = phydev->bus;
bus->write(bus, PHYADDR(reg), val);
}
@@ -1050,13 +1050,13 @@ static int adm6996_config_init(struct ph
pdev->supported = ADVERTISED_100baseT_Full;
pdev->advertising = ADVERTISED_100baseT_Full;
- if (pdev->mdio.addr != 0) {
+ if (pdev->addr != 0) {
pr_info ("%s: PHY overlaps ADM6996, providing fixed PHY 0x%x.\n"
- , pdev->attached_dev->name, pdev->mdio.addr);
+ , pdev->attached_dev->name, pdev->addr);
return 0;
}
- priv = devm_kzalloc(&pdev->mdio.dev, sizeof(struct adm6996_priv), GFP_KERNEL);
+ priv = devm_kzalloc(&pdev->dev, sizeof(struct adm6996_priv), GFP_KERNEL);
if (!priv)
return -ENOMEM;
@@ -1076,7 +1076,7 @@ static int adm6996_config_init(struct ph
}
/*
- * Warning: phydev->priv is NULL if phydev->mdio.addr != 0
+ * Warning: phydev->priv is NULL if phydev->addr != 0
*/
static int adm6996_read_status(struct phy_device *phydev)
{
@@ -1092,7 +1092,7 @@ static int adm6996_read_status(struct ph
}
/*
- * Warning: phydev->priv is NULL if phydev->mdio.addr != 0
+ * Warning: phydev->priv is NULL if phydev->addr != 0
*/
static int adm6996_config_aneg(struct phy_device *phydev)
{
@@ -1101,11 +1101,11 @@ static int adm6996_config_aneg(struct ph
static int adm6996_fixup(struct phy_device *dev)
{
- struct mii_bus *bus = dev->mdio.bus;
+ struct mii_bus *bus = dev->bus;
u16 reg;
/* Our custom registers are at PHY addresses 0-10. Claim those. */
- if (dev->mdio.addr > 10)
+ if (dev->addr > 10)
return 0;
/* look for the switch on the bus */
@@ -1152,6 +1152,7 @@ static struct phy_driver adm6996_phy_dri
.config_aneg = &adm6996_config_aneg,
.read_status = &adm6996_read_status,
.soft_reset = adm6996_soft_reset,
+ .driver = { .owner = THIS_MODULE,},
};
static int adm6996_gpio_probe(struct platform_device *pdev)
@@ -1220,7 +1221,7 @@ static int __init adm6996_init(void)
int err;
phy_register_fixup_for_id(PHY_ANY_ID, adm6996_fixup);
- err = phy_driver_register(&adm6996_phy_driver, THIS_MODULE);
+ err = phy_driver_register(&adm6996_phy_driver);
if (err)
return err;
--- a/drivers/net/phy/ar8216.c
+++ b/drivers/net/phy/ar8216.c
@@ -177,7 +177,7 @@ ar8xxx_phy_check_aneg(struct phy_device
if (ret & BMCR_ANENABLE)
return 0;
- dev_info(&phydev->mdio.dev, "ANEG disabled, re-enabling ...\n");
+ dev_info(&phydev->dev, "ANEG disabled, re-enabling ...\n");
ret |= BMCR_ANENABLE | BMCR_ANRESTART;
return phy_write(phydev, MII_BMCR, ret);
}
@@ -2007,7 +2007,7 @@ ar8xxx_phy_config_init(struct phy_device
priv->phy = phydev;
- if (phydev->mdio.addr != 0) {
+ if (phydev->addr != 0) {
if (chip_is_ar8316(priv)) {
/* switch device has been initialized, reinit */
priv->dev.ports = (AR8216_NUM_PORTS - 1);
@@ -2055,7 +2055,7 @@ ar8xxx_check_link_states(struct ar8xxx_p
/* flush ARL entries for this port if it went down*/
if (!link_new)
priv->chip->atu_flush_port(priv, i);
- dev_info(&priv->phy->mdio.dev, "Port %d is %s\n",
+ dev_info(&priv->phy->dev, "Port %d is %s\n",
i, link_new ? "up" : "down");
}
@@ -2074,10 +2074,10 @@ ar8xxx_phy_read_status(struct phy_device
if (phydev->state == PHY_CHANGELINK)
ar8xxx_check_link_states(priv);
- if (phydev->mdio.addr != 0)
+ if (phydev->addr != 0)
return genphy_read_status(phydev);
- ar8216_read_port_link(priv, phydev->mdio.addr, &link);
+ ar8216_read_port_link(priv, phydev->addr, &link);
phydev->link = !!link.link;
if (!phydev->link)
return 0;
@@ -2107,7 +2107,7 @@ ar8xxx_phy_read_status(struct phy_device
static int
ar8xxx_phy_config_aneg(struct phy_device *phydev)
{
- if (phydev->mdio.addr == 0)
+ if (phydev->addr == 0)
return 0;
return genphy_config_aneg(phydev);
@@ -2162,15 +2162,15 @@ ar8xxx_phy_probe(struct phy_device *phyd
int ret;
/* skip PHYs at unused adresses */
- if (phydev->mdio.addr != 0 && phydev->mdio.addr != 4)
+ if (phydev->addr != 0 && phydev->addr != 4)
return -ENODEV;
- if (!ar8xxx_is_possible(phydev->mdio.bus))
+ if (!ar8xxx_is_possible(phydev->bus))
return -ENODEV;
mutex_lock(&ar8xxx_dev_list_lock);
list_for_each_entry(priv, &ar8xxx_dev_list, list)
- if (priv->mii_bus == phydev->mdio.bus)
+ if (priv->mii_bus == phydev->bus)
goto found;
priv = ar8xxx_create();
@@ -2179,7 +2179,7 @@ ar8xxx_phy_probe(struct phy_device *phyd
goto unlock;
}
- priv->mii_bus = phydev->mdio.bus;
+ priv->mii_bus = phydev->bus;
ret = ar8xxx_probe_switch(priv);
if (ret)
@@ -2200,7 +2200,7 @@ ar8xxx_phy_probe(struct phy_device *phyd
found:
priv->use_count++;
- if (phydev->mdio.addr == 0) {
+ if (phydev->addr == 0) {
if (ar8xxx_has_gige(priv)) {
phydev->supported = SUPPORTED_1000baseT_Full;
phydev->advertising = ADVERTISED_1000baseT_Full;
@@ -2288,21 +2288,33 @@ ar8xxx_phy_soft_reset(struct phy_device
return 0;
}
-static struct phy_driver ar8xxx_phy_driver[] = {
- {
- .phy_id = 0x004d0000,
- .name = "Atheros AR8216/AR8236/AR8316",
- .phy_id_mask = 0xffff0000,
- .features = PHY_BASIC_FEATURES,
- .probe = ar8xxx_phy_probe,
- .remove = ar8xxx_phy_remove,
- .detach = ar8xxx_phy_detach,
- .config_init = ar8xxx_phy_config_init,
- .config_aneg = ar8xxx_phy_config_aneg,
- .read_status = ar8xxx_phy_read_status,
- .soft_reset = ar8xxx_phy_soft_reset,
- }
+static struct phy_driver ar8xxx_phy_driver = {
+ .phy_id = 0x004d0000,
+ .name = "Atheros AR8216/AR8236/AR8316",
+ .phy_id_mask = 0xffff0000,
+ .features = PHY_BASIC_FEATURES,
+ .probe = ar8xxx_phy_probe,
+ .remove = ar8xxx_phy_remove,
+ .detach = ar8xxx_phy_detach,
+ .config_init = ar8xxx_phy_config_init,
+ .config_aneg = ar8xxx_phy_config_aneg,
+ .read_status = ar8xxx_phy_read_status,
+ .soft_reset = ar8xxx_phy_soft_reset,
+ .driver = { .owner = THIS_MODULE },
};
-module_phy_driver(ar8xxx_phy_driver);
+int __init
+ar8xxx_init(void)
+{
+ return phy_driver_register(&ar8xxx_phy_driver);
+}
+
+void __exit
+ar8xxx_exit(void)
+{
+ phy_driver_unregister(&ar8xxx_phy_driver);
+}
+
+module_init(ar8xxx_init);
+module_exit(ar8xxx_exit);
MODULE_LICENSE("GPL");
--- a/drivers/net/phy/ar8327.c
+++ b/drivers/net/phy/ar8327.c
@@ -627,11 +627,11 @@ ar8327_hw_init(struct ar8xxx_priv *priv)
if (!priv->chip_data)
return -ENOMEM;
- if (priv->phy->mdio.dev.of_node)
- ret = ar8327_hw_config_of(priv, priv->phy->mdio.dev.of_node);
+ if (priv->phy->dev.of_node)
+ ret = ar8327_hw_config_of(priv, priv->phy->dev.of_node);
else
ret = ar8327_hw_config_pdata(priv,
- priv->phy->mdio.dev.platform_data);
+ priv->phy->dev.platform_data);
if (ret)
return ret;
--- a/drivers/net/phy/ip17xx.c
+++ b/drivers/net/phy/ip17xx.c
@@ -1273,7 +1273,7 @@ static int ip17xx_probe(struct phy_devic
int err;
/* We only attach to PHY 0, but use all available PHYs */
- if (pdev->mdio.addr != 0)
+ if (pdev->addr != 0)
return -ENODEV;
state = kzalloc(sizeof(*state), GFP_KERNEL);
@@ -1283,7 +1283,7 @@ static int ip17xx_probe(struct phy_devic
dev = &state->dev;
pdev->priv = state;
- state->mii_bus = pdev->mdio.bus;
+ state->mii_bus = pdev->bus;
err = get_model(state);
if (err < 0)
@@ -1295,7 +1295,7 @@ static int ip17xx_probe(struct phy_devic
dev->name = state->regs->NAME;
dev->ops = &ip17xx_ops;
- pr_info("IP17xx: Found %s at %s\n", dev->name, dev_name(&pdev->mdio.dev));
+ pr_info("IP17xx: Found %s at %s\n", dev->name, dev_name(&pdev->dev));
return 0;
error:
@@ -1353,25 +1353,58 @@ static int ip17xx_read_status(struct phy
return 0;
}
-static struct phy_driver ip17xx_driver[] = {
- {
- .name = "IC+ IP17xx",
- .phy_id = 0x02430c00,
- .phy_id_mask = 0x0ffffc00,
- .features = PHY_BASIC_FEATURES,
- .probe = ip17xx_probe,
- .remove = ip17xx_remove,
- .config_init = ip17xx_config_init,
- .config_aneg = ip17xx_config_aneg,
- .aneg_done = ip17xx_aneg_done,
- .update_link = ip17xx_update_link,
- .read_status = ip17xx_read_status,
- }
+static struct phy_driver ip17xx_driver = {
+ .name = "IC+ IP17xx",
+ .phy_id = 0x02430c00,
+ .phy_id_mask = 0x0ffffc00,
+ .features = PHY_BASIC_FEATURES,
+ .probe = ip17xx_probe,
+ .remove = ip17xx_remove,
+ .config_init = ip17xx_config_init,
+ .config_aneg = ip17xx_config_aneg,
+ .aneg_done = ip17xx_aneg_done,
+ .update_link = ip17xx_update_link,
+ .read_status = ip17xx_read_status,
+ .driver = { .owner = THIS_MODULE },
};
-module_phy_driver(ip17xx_driver);
+static struct phy_driver ip175a_driver = {
+ .name = "IC+ IP175A",
+ .phy_id = 0x02430c50,
+ .phy_id_mask = 0x0ffffff0,
+ .features = PHY_BASIC_FEATURES,
+ .probe = ip17xx_probe,
+ .remove = ip17xx_remove,
+ .config_init = ip17xx_config_init,
+ .config_aneg = ip17xx_config_aneg,
+ .aneg_done = ip17xx_aneg_done,
+ .update_link = ip17xx_update_link,
+ .read_status = ip17xx_read_status,
+ .driver = { .owner = THIS_MODULE },
+};
+
+
+int __init ip17xx_init(void)
+{
+ int ret;
+
+ ret = phy_driver_register(&ip175a_driver);
+ if (ret < 0)
+ return ret;
+
+ return phy_driver_register(&ip17xx_driver);
+}
+
+void __exit ip17xx_exit(void)
+{
+ phy_driver_unregister(&ip17xx_driver);
+ phy_driver_unregister(&ip175a_driver);
+}
MODULE_AUTHOR("Patrick Horn <patrick.horn@gmail.com>");
MODULE_AUTHOR("Felix Fietkau <nbd@nbd.name>");
MODULE_AUTHOR("Martin Mares <mj@ucw.cz>");
MODULE_LICENSE("GPL");
+
+module_init(ip17xx_init);
+module_exit(ip17xx_exit);
--- a/drivers/net/phy/mvswitch.c
+++ b/drivers/net/phy/mvswitch.c
@@ -50,17 +50,13 @@ struct mvswitch_priv {
static inline u16
r16(struct phy_device *phydev, int addr, int reg)
{
- struct mii_bus *bus = phydev->mdio.bus;
-
- return bus->read(bus, addr, reg);
+ return phydev->bus->read(phydev->bus, addr, reg);
}
static inline void
w16(struct phy_device *phydev, int addr, int reg, u16 val)
{
- struct mii_bus *bus = phydev->mdio.bus;
-
- bus->write(bus, addr, reg, val);
+ phydev->bus->write(phydev->bus, addr, reg, val);
}
@@ -398,13 +394,12 @@ mvswitch_probe(struct phy_device *pdev)
static int
mvswitch_fixup(struct phy_device *dev)
{
- struct mii_bus *bus = dev->mdio.bus;
u16 reg;
- if (dev->mdio.addr != 0x10)
+ if (dev->addr != 0x10)
return 0;
- reg = bus->read(bus, MV_PORTREG(IDENT, 0)) & MV_IDENT_MASK;
+ reg = dev->bus->read(dev->bus, MV_PORTREG(IDENT, 0)) & MV_IDENT_MASK;
if (reg != MV_IDENT_VALUE)
return 0;
@@ -425,13 +420,14 @@ static struct phy_driver mvswitch_driver
.config_aneg = &mvswitch_config_aneg,
.aneg_done = &mvswitch_aneg_done,
.read_status = &mvswitch_read_status,
+ .driver = { .owner = THIS_MODULE,},
};
static int __init
mvswitch_init(void)
{
phy_register_fixup_for_id(PHY_ANY_ID, mvswitch_fixup);
- return phy_driver_register(&mvswitch_driver, THIS_MODULE);
+ return phy_driver_register(&mvswitch_driver);
}
static void __exit
--- a/drivers/net/phy/psb6970.c
+++ b/drivers/net/phy/psb6970.c
@@ -70,16 +70,12 @@ struct psb6970_priv {
static u16 psb6970_mii_read(struct phy_device *phydev, int reg)
{
- struct mii_bus *bus = phydev->mdio.bus;
-
- return bus->read(bus, PHYADDR(reg));
+ return phydev->bus->read(phydev->bus, PHYADDR(reg));
}
static void psb6970_mii_write(struct phy_device *phydev, int reg, u16 val)
{
- struct mii_bus *bus = phydev->mdio.bus;
-
- bus->write(bus, PHYADDR(reg), val);
+ phydev->bus->write(phydev->bus, PHYADDR(reg), val);
}
static int
@@ -316,11 +312,11 @@ static int psb6970_config_init(struct ph
priv->phy = pdev;
- if (pdev->mdio.addr == 0)
+ if (pdev->addr == 0)
printk(KERN_INFO "%s: psb6970 switch driver attached.\n",
pdev->attached_dev->name);
- if (pdev->mdio.addr != 0) {
+ if (pdev->addr != 0) {
kfree(priv);
return 0;
}
@@ -388,14 +384,14 @@ static void psb6970_remove(struct phy_de
if (!priv)
return;
- if (pdev->mdio.addr == 0)
+ if (pdev->addr == 0)
unregister_switch(&priv->dev);
kfree(priv);
}
static int psb6970_fixup(struct phy_device *dev)
{
- struct mii_bus *bus = dev->mdio.bus;
+ struct mii_bus *bus = dev->bus;
u16 reg;
/* look for the switch on the bus */
@@ -419,12 +415,13 @@ static struct phy_driver psb6970_driver
.config_init = &psb6970_config_init,
.config_aneg = &psb6970_config_aneg,
.read_status = &psb6970_read_status,
+ .driver = {.owner = THIS_MODULE},
};
int __init psb6970_init(void)
{
phy_register_fixup_for_id(PHY_ANY_ID, psb6970_fixup);
- return phy_driver_register(&psb6970_driver, THIS_MODULE);
+ return phy_driver_register(&psb6970_driver);
}
module_init(psb6970_init);
--- a/drivers/net/phy/rtl8306.c
+++ b/drivers/net/phy/rtl8306.c
@@ -877,7 +877,7 @@ rtl8306_config_init(struct phy_device *p
int err;
/* Only init the switch for the primary PHY */
- if (pdev->mdio.addr != 0)
+ if (pdev->addr != 0)
return 0;
val.value.i = 1;
@@ -887,7 +887,7 @@ rtl8306_config_init(struct phy_device *p
priv->dev.ops = &rtl8306_ops;
priv->do_cpu = 0;
priv->page = -1;
- priv->bus = pdev->mdio.bus;
+ priv->bus = pdev->bus;
chipid = rtl_get(dev, RTL_REG_CHIPID);
chipver = rtl_get(dev, RTL_REG_CHIPVER);
@@ -933,13 +933,13 @@ rtl8306_fixup(struct phy_device *pdev)
u16 chipid;
/* Attach to primary LAN port and WAN port */
- if (pdev->mdio.addr != 0 && pdev->mdio.addr != 4)
+ if (pdev->addr != 0 && pdev->addr != 4)
return 0;
memset(&priv, 0, sizeof(priv));
priv.fixup = true;
priv.page = -1;
- priv.bus = pdev->mdio.bus;
+ priv.bus = pdev->bus;
chipid = rtl_get(&priv.dev, RTL_REG_CHIPID);
if (chipid == 0x5988)
pdev->phy_id = RTL8306_MAGIC;
@@ -957,14 +957,14 @@ rtl8306_probe(struct phy_device *pdev)
* share one rtl_priv instance between virtual phy
* devices on the same bus
*/
- if (priv->bus == pdev->mdio.bus)
+ if (priv->bus == pdev->bus)
goto found;
}
priv = kzalloc(sizeof(struct rtl_priv), GFP_KERNEL);
if (!priv)
return -ENOMEM;
- priv->bus = pdev->mdio.bus;
+ priv->bus = pdev->bus;
found:
pdev->priv = priv;
@@ -985,7 +985,7 @@ rtl8306_config_aneg(struct phy_device *p
struct rtl_priv *priv = pdev->priv;
/* Only for WAN */
- if (pdev->mdio.addr == 0)
+ if (pdev->addr == 0)
return 0;
/* Restart autonegotiation */
@@ -1001,7 +1001,7 @@ rtl8306_read_status(struct phy_device *p
struct rtl_priv *priv = pdev->priv;
struct switch_dev *dev = &priv->dev;
- if (pdev->mdio.addr == 4) {
+ if (pdev->addr == 4) {
/* WAN */
pdev->speed = rtl_get(dev, RTL_PORT_REG(4, SPEED)) ? SPEED_100 : SPEED_10;
pdev->duplex = rtl_get(dev, RTL_PORT_REG(4, DUPLEX)) ? DUPLEX_FULL : DUPLEX_HALF;
@@ -1044,6 +1044,7 @@ static struct phy_driver rtl8306_driver
.config_init = &rtl8306_config_init,
.config_aneg = &rtl8306_config_aneg,
.read_status = &rtl8306_read_status,
+ .driver = { .owner = THIS_MODULE,},
};
@@ -1051,7 +1052,7 @@ static int __init
rtl_init(void)
{
phy_register_fixup_for_id(PHY_ANY_ID, rtl8306_fixup);
- return phy_driver_register(&rtl8306_driver, THIS_MODULE);
+ return phy_driver_register(&rtl8306_driver);
}
static void __exit

11
target/linux/generic/pending-3.18/003-myloader_backport.patch
View File

@ -1,11 +0,0 @@
--- a/drivers/mtd/myloader.c
+++ b/drivers/mtd/myloader.c
@@ -33,7 +33,7 @@ struct part_data {
};
static int myloader_parse_partitions(struct mtd_info *master,
- const struct mtd_partition **pparts,
+ struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
{
struct part_data *buf;

134
target/linux/generic/pending-3.18/020-ssb_update.patch
View File

@ -1,134 +0,0 @@
--- a/drivers/ssb/pcihost_wrapper.c
+++ b/drivers/ssb/pcihost_wrapper.c
@@ -11,15 +11,17 @@
* Licensed under the GNU/GPL. See COPYING for details.
*/
+#include <linux/pm.h>
#include <linux/pci.h>
#include <linux/export.h>
#include <linux/slab.h>
#include <linux/ssb/ssb.h>
-#ifdef CONFIG_PM
-static int ssb_pcihost_suspend(struct pci_dev *dev, pm_message_t state)
+#ifdef CONFIG_PM_SLEEP
+static int ssb_pcihost_suspend(struct device *d)
{
+ struct pci_dev *dev = to_pci_dev(d);
struct ssb_bus *ssb = pci_get_drvdata(dev);
int err;
@@ -28,17 +30,23 @@ static int ssb_pcihost_suspend(struct pc
return err;
pci_save_state(dev);
pci_disable_device(dev);
- pci_set_power_state(dev, pci_choose_state(dev, state));
+
+ /* if there is a wakeup enabled child device on ssb bus,
+ enable pci wakeup posibility. */
+ device_set_wakeup_enable(d, d->power.wakeup_path);
+
+ pci_prepare_to_sleep(dev);
return 0;
}
-static int ssb_pcihost_resume(struct pci_dev *dev)
+static int ssb_pcihost_resume(struct device *d)
{
+ struct pci_dev *dev = to_pci_dev(d);
struct ssb_bus *ssb = pci_get_drvdata(dev);
int err;
- pci_set_power_state(dev, PCI_D0);
+ pci_back_from_sleep(dev);
err = pci_enable_device(dev);
if (err)
return err;
@@ -49,10 +57,12 @@ static int ssb_pcihost_resume(struct pci
return 0;
}
-#else /* CONFIG_PM */
-# define ssb_pcihost_suspend NULL
-# define ssb_pcihost_resume NULL
-#endif /* CONFIG_PM */
+
+static const struct dev_pm_ops ssb_pcihost_pm_ops = {
+ SET_SYSTEM_SLEEP_PM_OPS(ssb_pcihost_suspend, ssb_pcihost_resume)
+};
+
+#endif /* CONFIG_PM_SLEEP */
static int ssb_pcihost_probe(struct pci_dev *dev,
const struct pci_device_id *id)
@@ -115,8 +125,9 @@ int ssb_pcihost_register(struct pci_driv
{
driver->probe = ssb_pcihost_probe;
driver->remove = ssb_pcihost_remove;
- driver->suspend = ssb_pcihost_suspend;
- driver->resume = ssb_pcihost_resume;
+#ifdef CONFIG_PM_SLEEP
+ driver->driver.pm = &ssb_pcihost_pm_ops;
+#endif
return pci_register_driver(driver);
}
--- a/drivers/ssb/driver_pcicore.c
+++ b/drivers/ssb/driver_pcicore.c
@@ -357,6 +357,16 @@ static void ssb_pcicore_init_hostmode(st
pcicore_write32(pc, SSB_PCICORE_SBTOPCI2,
SSB_PCICORE_SBTOPCI_MEM | SSB_PCI_DMA);
+ /*
+ * Accessing PCI config without a proper delay after devices reset (not
+ * GPIO reset) was causing reboots on WRT300N v1.0 (BCM4704).
+ * Tested delay 850 us lowered reboot chance to 50-80%, 1000 us fixed it
+ * completely. Flushing all writes was also tested but with no luck.
+ * The same problem was reported for WRT350N v1 (BCM4705), so we just
+ * sleep here unconditionally.
+ */
+ usleep_range(1000, 2000);
+
/* Enable PCI bridge BAR0 prefetch and burst */
val = PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY;
ssb_extpci_write_config(pc, 0, 0, 0, PCI_COMMAND, &val, 2);
--- a/drivers/ssb/main.c
+++ b/drivers/ssb/main.c
@@ -90,25 +90,6 @@ found:
}
#endif /* CONFIG_SSB_PCMCIAHOST */
-#ifdef CONFIG_SSB_SDIOHOST
-struct ssb_bus *ssb_sdio_func_to_bus(struct sdio_func *func)
-{
- struct ssb_bus *bus;
-
- ssb_buses_lock();
- list_for_each_entry(bus, &buses, list) {
- if (bus->bustype == SSB_BUSTYPE_SDIO &&
- bus->host_sdio == func)
- goto found;
- }
- bus = NULL;
-found:
- ssb_buses_unlock();
-
- return bus;
-}
-#endif /* CONFIG_SSB_SDIOHOST */
-
int ssb_for_each_bus_call(unsigned long data,
int (*func)(struct ssb_bus *bus, unsigned long data))
{
@@ -1154,6 +1135,8 @@ static u32 ssb_tmslow_reject_bitmask(str
case SSB_IDLOW_SSBREV_25: /* TODO - find the proper REJECT bit */
case SSB_IDLOW_SSBREV_27: /* same here */
return SSB_TMSLOW_REJECT; /* this is a guess */
+ case SSB_IDLOW_SSBREV:
+ break;
default:
WARN(1, KERN_INFO "ssb: Backplane Revision 0x%.8X\n", rev);
}

32
target/linux/generic/pending-3.18/021-ssb_sprom.patch
View File

@ -1,32 +0,0 @@
--- a/include/linux/ssb/ssb.h
+++ b/include/linux/ssb/ssb.h
@@ -29,10 +29,13 @@ struct ssb_sprom {
u8 il0mac[6] __aligned(sizeof(u16)); /* MAC address for 802.11b/g */
u8 et0mac[6] __aligned(sizeof(u16)); /* MAC address for Ethernet */
u8 et1mac[6] __aligned(sizeof(u16)); /* MAC address for 802.11a */
+ u8 et2mac[6] __aligned(sizeof(u16)); /* MAC address for extra Ethernet */
u8 et0phyaddr; /* MII address for enet0 */
u8 et1phyaddr; /* MII address for enet1 */
+ u8 et2phyaddr; /* MII address for enet2 */
u8 et0mdcport; /* MDIO for enet0 */
u8 et1mdcport; /* MDIO for enet1 */
+ u8 et2mdcport; /* MDIO for enet2 */
u16 dev_id; /* Device ID overriding e.g. PCI ID */
u16 board_rev; /* Board revision number from SPROM. */
u16 board_num; /* Board number from SPROM. */
@@ -88,11 +91,14 @@ struct ssb_sprom {
u32 ofdm5glpo; /* 5.2GHz OFDM power offset */
u32 ofdm5gpo; /* 5.3GHz OFDM power offset */
u32 ofdm5ghpo; /* 5.8GHz OFDM power offset */
+ u32 boardflags;
+ u32 boardflags2;
+ u32 boardflags3;
+ /* TODO: Switch all drivers to new u32 fields and drop below ones */
u16 boardflags_lo; /* Board flags (bits 0-15) */
u16 boardflags_hi; /* Board flags (bits 16-31) */
u16 boardflags2_lo; /* Board flags (bits 32-47) */
u16 boardflags2_hi; /* Board flags (bits 48-63) */
- /* TODO store board flags in a single u64 */
struct ssb_sprom_core_pwr_info core_pwr_info[4];

286
target/linux/generic/pending-3.18/025-bcma_backport.patch
View File

@ -1,286 +0,0 @@
--- a/drivers/bcma/bcma_private.h
+++ b/drivers/bcma/bcma_private.h
@@ -22,6 +22,7 @@ struct bcma_bus;
/* main.c */
bool bcma_wait_value(struct bcma_device *core, u16 reg, u32 mask, u32 value,
int timeout);
+void bcma_prepare_core(struct bcma_bus *bus, struct bcma_device *core);
int bcma_bus_register(struct bcma_bus *bus);
void bcma_bus_unregister(struct bcma_bus *bus);
int __init bcma_bus_early_register(struct bcma_bus *bus,
--- a/drivers/bcma/driver_chipcommon.c
+++ b/drivers/bcma/driver_chipcommon.c
@@ -339,7 +339,7 @@ void bcma_chipco_serial_init(struct bcma
return;
}
- irq = bcma_core_irq(cc->core);
+ irq = bcma_core_irq(cc->core, 0);
/* Determine the registers of the UARTs */
cc->nr_serial_ports = (cc->capabilities & BCMA_CC_CAP_NRUART);
--- a/drivers/bcma/driver_gpio.c
+++ b/drivers/bcma/driver_gpio.c
@@ -152,7 +152,7 @@ static int bcma_gpio_irq_domain_init(str
handle_simple_irq);
}
- hwirq = bcma_core_irq(cc->core);
+ hwirq = bcma_core_irq(cc->core, 0);
err = request_irq(hwirq, bcma_gpio_irq_handler, IRQF_SHARED, "gpio",
cc);
if (err)
@@ -183,7 +183,7 @@ static void bcma_gpio_irq_domain_exit(st
return;
bcma_cc_mask32(cc, BCMA_CC_IRQMASK, ~BCMA_CC_IRQ_GPIO);
- free_irq(bcma_core_irq(cc->core), cc);
+ free_irq(bcma_core_irq(cc->core, 0), cc);
for (gpio = 0; gpio < chip->ngpio; gpio++) {
int irq = irq_find_mapping(cc->irq_domain, gpio);
--- a/drivers/bcma/driver_mips.c
+++ b/drivers/bcma/driver_mips.c
@@ -115,7 +115,7 @@ static u32 bcma_core_mips_irqflag(struct
* If disabled, 5 is returned.
* If not supported, 6 is returned.
*/
-static unsigned int bcma_core_mips_irq(struct bcma_device *dev)
+unsigned int bcma_core_mips_irq(struct bcma_device *dev)
{
struct bcma_device *mdev = dev->bus->drv_mips.core;
u32 irqflag;
@@ -133,13 +133,6 @@ static unsigned int bcma_core_mips_irq(s
return 5;
}
-unsigned int bcma_core_irq(struct bcma_device *dev)
-{
- unsigned int mips_irq = bcma_core_mips_irq(dev);
- return mips_irq <= 4 ? mips_irq + 2 : 0;
-}
-EXPORT_SYMBOL(bcma_core_irq);
-
static void bcma_core_mips_set_irq(struct bcma_device *dev, unsigned int irq)
{
unsigned int oldirq = bcma_core_mips_irq(dev);
@@ -423,7 +416,7 @@ void bcma_core_mips_init(struct bcma_drv
break;
default:
list_for_each_entry(core, &bus->cores, list) {
- core->irq = bcma_core_irq(core);
+ core->irq = bcma_core_irq(core, 0);
}
bcma_err(bus,
"Unknown device (0x%x) found, can not configure IRQs\n",
--- a/drivers/bcma/driver_pci_host.c
+++ b/drivers/bcma/driver_pci_host.c
@@ -593,7 +593,7 @@ int bcma_core_pci_plat_dev_init(struct p
pr_info("PCI: Fixing up device %s\n", pci_name(dev));
/* Fix up interrupt lines */
- dev->irq = bcma_core_irq(pc_host->pdev->core);
+ dev->irq = bcma_core_irq(pc_host->pdev->core, 0);
pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq);
readrq = pcie_get_readrq(dev);
@@ -617,6 +617,6 @@ int bcma_core_pci_pcibios_map_irq(const
pc_host = container_of(dev->bus->ops, struct bcma_drv_pci_host,
pci_ops);
- return bcma_core_irq(pc_host->pdev->core);
+ return bcma_core_irq(pc_host->pdev->core, 0);
}
EXPORT_SYMBOL(bcma_core_pci_pcibios_map_irq);
--- a/drivers/bcma/main.c
+++ b/drivers/bcma/main.c
@@ -11,6 +11,7 @@
#include <linux/bcma/bcma.h>
#include <linux/slab.h>
#include <linux/of_address.h>
+#include <linux/of_irq.h>
MODULE_DESCRIPTION("Broadcom's specific AMBA driver");
MODULE_LICENSE("GPL");
@@ -153,6 +154,46 @@ static struct device_node *bcma_of_find_
return NULL;
}
+static int bcma_of_irq_parse(struct platform_device *parent,
+ struct bcma_device *core,
+ struct of_phandle_args *out_irq, int num)
+{
+ __be32 laddr[1];
+ int rc;
+
+ if (core->dev.of_node) {
+ rc = of_irq_parse_one(core->dev.of_node, num, out_irq);
+ if (!rc)
+ return rc;
+ }
+
+ out_irq->np = parent->dev.of_node;
+ out_irq->args_count = 1;
+ out_irq->args[0] = num;
+
+ laddr[0] = cpu_to_be32(core->addr);
+ return of_irq_parse_raw(laddr, out_irq);
+}
+
+static unsigned int bcma_of_get_irq(struct platform_device *parent,
+ struct bcma_device *core, int num)
+{
+ struct of_phandle_args out_irq;
+ int ret;
+
+ if (!parent || !parent->dev.of_node)
+ return 0;
+
+ ret = bcma_of_irq_parse(parent, core, &out_irq, num);
+ if (ret) {
+ bcma_debug(core->bus, "bcma_of_get_irq() failed with rc=%d\n",
+ ret);
+ return 0;
+ }
+
+ return irq_create_of_mapping(&out_irq);
+}
+
static void bcma_of_fill_device(struct platform_device *parent,
struct bcma_device *core)
{
@@ -161,18 +202,47 @@ static void bcma_of_fill_device(struct p
node = bcma_of_find_child_device(parent, core);
if (node)
core->dev.of_node = node;
+
+ core->irq = bcma_of_get_irq(parent, core, 0);
}
#else
static void bcma_of_fill_device(struct platform_device *parent,
struct bcma_device *core)
{
}
+static inline unsigned int bcma_of_get_irq(struct platform_device *parent,
+ struct bcma_device *core, int num)
+{
+ return 0;
+}
#endif /* CONFIG_OF */
-static void bcma_register_core(struct bcma_bus *bus, struct bcma_device *core)
+unsigned int bcma_core_irq(struct bcma_device *core, int num)
{
- int err;
+ struct bcma_bus *bus = core->bus;
+ unsigned int mips_irq;
+
+ switch (bus->hosttype) {
+ case BCMA_HOSTTYPE_PCI:
+ return bus->host_pci->irq;
+ case BCMA_HOSTTYPE_SOC:
+ if (bus->drv_mips.core && num == 0) {
+ mips_irq = bcma_core_mips_irq(core);
+ return mips_irq <= 4 ? mips_irq + 2 : 0;
+ }
+ if (bus->host_pdev)
+ return bcma_of_get_irq(bus->host_pdev, core, num);
+ return 0;
+ case BCMA_HOSTTYPE_SDIO:
+ return 0;
+ }
+ return 0;
+}
+EXPORT_SYMBOL(bcma_core_irq);
+
+void bcma_prepare_core(struct bcma_bus *bus, struct bcma_device *core)
+{
core->dev.release = bcma_release_core_dev;
core->dev.bus = &bcma_bus_type;
dev_set_name(&core->dev, "bcma%d:%d", bus->num, core->core_index);
@@ -196,6 +266,11 @@ static void bcma_register_core(struct bc
case BCMA_HOSTTYPE_SDIO:
break;
}
+}
+
+static void bcma_register_core(struct bcma_bus *bus, struct bcma_device *core)
+{
+ int err;
err = device_register(&core->dev);
if (err) {
--- a/drivers/bcma/scan.c
+++ b/drivers/bcma/scan.c
@@ -505,6 +505,7 @@ int bcma_bus_scan(struct bcma_bus *bus)
bus->nr_cores++;
other_core = bcma_find_core_reverse(bus, core->id.id);
core->core_unit = (other_core == NULL) ? 0 : other_core->core_unit + 1;
+ bcma_prepare_core(bus, core);
bcma_info(bus, "Core %d found: %s (manuf 0x%03X, id 0x%03X, rev 0x%02X, class 0x%X)\n",
core->core_index, bcma_device_name(&core->id),
--- a/include/linux/bcma/bcma.h
+++ b/include/linux/bcma/bcma.h
@@ -448,4 +448,6 @@ extern u32 bcma_chipco_pll_read(struct b
#define BCMA_DMA_TRANSLATION_DMA64_CMT 0x80000000 /* Client Mode Translation for 64-bit DMA */
extern u32 bcma_core_dma_translation(struct bcma_device *core);
+extern unsigned int bcma_core_irq(struct bcma_device *core, int num);
+
#endif /* LINUX_BCMA_H_ */
--- a/include/linux/bcma/bcma_driver_mips.h
+++ b/include/linux/bcma/bcma_driver_mips.h
@@ -43,12 +43,12 @@ struct bcma_drv_mips {
extern void bcma_core_mips_init(struct bcma_drv_mips *mcore);
extern void bcma_core_mips_early_init(struct bcma_drv_mips *mcore);
-extern unsigned int bcma_core_irq(struct bcma_device *core);
+extern unsigned int bcma_core_mips_irq(struct bcma_device *dev);
#else
static inline void bcma_core_mips_init(struct bcma_drv_mips *mcore) { }
static inline void bcma_core_mips_early_init(struct bcma_drv_mips *mcore) { }
-static inline unsigned int bcma_core_irq(struct bcma_device *core)
+static inline unsigned int bcma_core_mips_irq(struct bcma_device *dev)
{
return 0;
}
--- a/Documentation/devicetree/bindings/bus/bcma.txt
+++ b/Documentation/devicetree/bindings/bus/bcma.txt
@@ -8,6 +8,11 @@ Required properties:
The cores on the AXI bus are automatically detected by bcma with the
memory ranges they are using and they get registered afterwards.
+Automatic detection of the IRQ number is not working on
+BCM47xx/BCM53xx ARM SoCs. To assign IRQ numbers to the cores, provide
+them manually through device tree. Use an interrupt-map to specify the
+IRQ used by the devices on the bus. The first address is just an index,
+because we do not have any special register.
The top-level axi bus may contain children representing attached cores
(devices). This is needed since some hardware details can't be auto
@@ -22,6 +27,22 @@ Example:
ranges = <0x00000000 0x18000000 0x00100000>;
#address-cells = <1>;
#size-cells = <1>;
+ #interrupt-cells = <1>;
+ interrupt-map-mask = <0x000fffff 0xffff>;
+ interrupt-map =
+ /* Ethernet Controller 0 */
+ <0x00024000 0 &gic GIC_SPI 147 IRQ_TYPE_LEVEL_HIGH>,
+
+ /* Ethernet Controller 1 */
+ <0x00025000 0 &gic GIC_SPI 148 IRQ_TYPE_LEVEL_HIGH>;
+
+ /* PCIe Controller 0 */
+ <0x00012000 0 &gic GIC_SPI 126 IRQ_TYPE_LEVEL_HIGH>,
+ <0x00012000 1 &gic GIC_SPI 127 IRQ_TYPE_LEVEL_HIGH>,
+ <0x00012000 2 &gic GIC_SPI 128 IRQ_TYPE_LEVEL_HIGH>,
+ <0x00012000 3 &gic GIC_SPI 129 IRQ_TYPE_LEVEL_HIGH>,
+ <0x00012000 4 &gic GIC_SPI 130 IRQ_TYPE_LEVEL_HIGH>,
+ <0x00012000 5 &gic GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
chipcommon {
reg = <0x00000000 0x1000>;

527
target/linux/generic/pending-3.18/026-bcma-from-3.20.patch
View File

@ -1,527 +0,0 @@
--- a/drivers/bcma/bcma_private.h
+++ b/drivers/bcma/bcma_private.h
@@ -23,22 +23,18 @@ struct bcma_bus;
bool bcma_wait_value(struct bcma_device *core, u16 reg, u32 mask, u32 value,
int timeout);
void bcma_prepare_core(struct bcma_bus *bus, struct bcma_device *core);
+void bcma_init_bus(struct bcma_bus *bus);
int bcma_bus_register(struct bcma_bus *bus);
void bcma_bus_unregister(struct bcma_bus *bus);
-int __init bcma_bus_early_register(struct bcma_bus *bus,
- struct bcma_device *core_cc,
- struct bcma_device *core_mips);
+int __init bcma_bus_early_register(struct bcma_bus *bus);
#ifdef CONFIG_PM
int bcma_bus_suspend(struct bcma_bus *bus);
int bcma_bus_resume(struct bcma_bus *bus);
#endif
/* scan.c */
+void bcma_detect_chip(struct bcma_bus *bus);
int bcma_bus_scan(struct bcma_bus *bus);
-int __init bcma_bus_scan_early(struct bcma_bus *bus,
- struct bcma_device_id *match,
- struct bcma_device *core);
-void bcma_init_bus(struct bcma_bus *bus);
/* sprom.c */
int bcma_sprom_get(struct bcma_bus *bus);
@@ -109,6 +105,14 @@ extern int bcma_chipco_watchdog_register
#ifdef CONFIG_BCMA_DRIVER_PCI_HOSTMODE
bool bcma_core_pci_is_in_hostmode(struct bcma_drv_pci *pc);
void bcma_core_pci_hostmode_init(struct bcma_drv_pci *pc);
+#else
+static inline bool bcma_core_pci_is_in_hostmode(struct bcma_drv_pci *pc)
+{
+ return false;
+}
+static inline void bcma_core_pci_hostmode_init(struct bcma_drv_pci *pc)
+{
+}
#endif /* CONFIG_BCMA_DRIVER_PCI_HOSTMODE */
#ifdef CONFIG_BCMA_DRIVER_GPIO
--- a/drivers/bcma/driver_chipcommon.c
+++ b/drivers/bcma/driver_chipcommon.c
@@ -79,7 +79,9 @@ static int bcma_chipco_watchdog_ticks_pe
if (cc->capabilities & BCMA_CC_CAP_PMU) {
if (bus->chipinfo.id == BCMA_CHIP_ID_BCM4706)
- /* 4706 CC and PMU watchdogs are clocked at 1/4 of ALP clock */
+ /* 4706 CC and PMU watchdogs are clocked at 1/4 of ALP
+ * clock
+ */
return bcma_chipco_get_alp_clock(cc) / 4000;
else
/* based on 32KHz ILP clock */
@@ -97,7 +99,8 @@ int bcma_chipco_watchdog_register(struct
wdt.driver_data = cc;
wdt.timer_set = bcma_chipco_watchdog_timer_set_wdt;
wdt.timer_set_ms = bcma_chipco_watchdog_timer_set_ms_wdt;
- wdt.max_timer_ms = bcma_chipco_watchdog_get_max_timer(cc) / cc->ticks_per_ms;
+ wdt.max_timer_ms =
+ bcma_chipco_watchdog_get_max_timer(cc) / cc->ticks_per_ms;
pdev = platform_device_register_data(NULL, "bcm47xx-wdt",
cc->core->bus->num, &wdt,
@@ -175,7 +178,6 @@ void bcma_core_chipcommon_init(struct bc
u32 bcma_chipco_watchdog_timer_set(struct bcma_drv_cc *cc, u32 ticks)
{
u32 maxt;
- enum bcma_clkmode clkmode;
maxt = bcma_chipco_watchdog_get_max_timer(cc);
if (cc->capabilities & BCMA_CC_CAP_PMU) {
@@ -185,8 +187,13 @@ u32 bcma_chipco_watchdog_timer_set(struc
ticks = maxt;
bcma_cc_write32(cc, BCMA_CC_PMU_WATCHDOG, ticks);
} else {
- clkmode = ticks ? BCMA_CLKMODE_FAST : BCMA_CLKMODE_DYNAMIC;
- bcma_core_set_clockmode(cc->core, clkmode);
+ struct bcma_bus *bus = cc->core->bus;
+
+ if (bus->chipinfo.id != BCMA_CHIP_ID_BCM4707 &&
+ bus->chipinfo.id != BCMA_CHIP_ID_BCM53018)
+ bcma_core_set_clockmode(cc->core,
+ ticks ? BCMA_CLKMODE_FAST : BCMA_CLKMODE_DYNAMIC);
+
if (ticks > maxt)
ticks = maxt;
/* instant NMI */
@@ -335,7 +342,8 @@ void bcma_chipco_serial_init(struct bcma
| BCMA_CC_CORECTL_UARTCLKEN);
}
} else {
- bcma_err(cc->core->bus, "serial not supported on this device ccrev: 0x%x\n", ccrev);
+ bcma_err(cc->core->bus, "serial not supported on this device ccrev: 0x%x\n",
+ ccrev);
return;
}
--- a/drivers/bcma/driver_pci.c
+++ b/drivers/bcma/driver_pci.c
@@ -145,6 +145,47 @@ static u16 bcma_pcie_mdio_writeread(stru
}
/**************************************************
+ * Early init.
+ **************************************************/
+
+static void bcma_core_pci_fixcfg(struct bcma_drv_pci *pc)
+{
+ struct bcma_device *core = pc->core;
+ u16 val16, core_index;
+ uint regoff;
+
+ regoff = BCMA_CORE_PCI_SPROM(BCMA_CORE_PCI_SPROM_PI_OFFSET);
+ core_index = (u16)core->core_index;
+
+ val16 = pcicore_read16(pc, regoff);
+ if (((val16 & BCMA_CORE_PCI_SPROM_PI_MASK) >> BCMA_CORE_PCI_SPROM_PI_SHIFT)
+ != core_index) {
+ val16 = (core_index << BCMA_CORE_PCI_SPROM_PI_SHIFT) |
+ (val16 & ~BCMA_CORE_PCI_SPROM_PI_MASK);
+ pcicore_write16(pc, regoff, val16);
+ }
+}
+
+/*
+ * Apply some early fixes required before accessing SPROM.
+ * See also si_pci_fixcfg.
+ */
+void bcma_core_pci_early_init(struct bcma_drv_pci *pc)
+{
+ if (pc->early_setup_done)
+ return;
+
+ pc->hostmode = bcma_core_pci_is_in_hostmode(pc);
+ if (pc->hostmode)
+ goto out;
+
+ bcma_core_pci_fixcfg(pc);
+
+out:
+ pc->early_setup_done = true;
+}
+
+/**************************************************
* Workarounds.
**************************************************/
@@ -175,24 +216,6 @@ static void bcma_pcicore_serdes_workarou
tmp & ~BCMA_CORE_PCI_PLL_CTRL_FREQDET_EN);
}
-static void bcma_core_pci_fixcfg(struct bcma_drv_pci *pc)
-{
- struct bcma_device *core = pc->core;
- u16 val16, core_index;
- uint regoff;
-
- regoff = BCMA_CORE_PCI_SPROM(BCMA_CORE_PCI_SPROM_PI_OFFSET);
- core_index = (u16)core->core_index;
-
- val16 = pcicore_read16(pc, regoff);
- if (((val16 & BCMA_CORE_PCI_SPROM_PI_MASK) >> BCMA_CORE_PCI_SPROM_PI_SHIFT)
- != core_index) {
- val16 = (core_index << BCMA_CORE_PCI_SPROM_PI_SHIFT) |
- (val16 & ~BCMA_CORE_PCI_SPROM_PI_MASK);
- pcicore_write16(pc, regoff, val16);
- }
-}
-
/* Fix MISC config to allow coming out of L2/L3-Ready state w/o PRST */
/* Needs to happen when coming out of 'standby'/'hibernate' */
static void bcma_core_pci_config_fixup(struct bcma_drv_pci *pc)
@@ -216,7 +239,6 @@ static void bcma_core_pci_config_fixup(s
static void bcma_core_pci_clientmode_init(struct bcma_drv_pci *pc)
{
- bcma_core_pci_fixcfg(pc);
bcma_pcicore_serdes_workaround(pc);
bcma_core_pci_config_fixup(pc);
}
@@ -226,13 +248,11 @@ void bcma_core_pci_init(struct bcma_drv_
if (pc->setup_done)
return;
-#ifdef CONFIG_BCMA_DRIVER_PCI_HOSTMODE
- pc->hostmode = bcma_core_pci_is_in_hostmode(pc);
+ bcma_core_pci_early_init(pc);
+
if (pc->hostmode)
bcma_core_pci_hostmode_init(pc);
-#endif /* CONFIG_BCMA_DRIVER_PCI_HOSTMODE */
-
- if (!pc->hostmode)
+ else
bcma_core_pci_clientmode_init(pc);
}
--- a/drivers/bcma/host_pci.c
+++ b/drivers/bcma/host_pci.c
@@ -13,10 +13,12 @@
static void bcma_host_pci_switch_core(struct bcma_device *core)
{
+ int win2 = core->bus->host_is_pcie2 ?
+ BCMA_PCIE2_BAR0_WIN2 : BCMA_PCI_BAR0_WIN2;
+
pci_write_config_dword(core->bus->host_pci, BCMA_PCI_BAR0_WIN,
core->addr);
- pci_write_config_dword(core->bus->host_pci, BCMA_PCI_BAR0_WIN2,
- core->wrap);
+ pci_write_config_dword(core->bus->host_pci, win2, core->wrap);
core->bus->mapped_core = core;
bcma_debug(core->bus, "Switched to core: 0x%X\n", core->id.id);
}
--- a/drivers/bcma/host_soc.c
+++ b/drivers/bcma/host_soc.c
@@ -193,7 +193,7 @@ int __init bcma_host_soc_init(struct bcm
int err;
/* Scan bus and initialize it */
- err = bcma_bus_early_register(bus, &soc->core_cc, &soc->core_mips);
+ err = bcma_bus_early_register(bus);
if (err)
iounmap(bus->mmio);
--- a/drivers/bcma/main.c
+++ b/drivers/bcma/main.c
@@ -268,6 +268,18 @@ void bcma_prepare_core(struct bcma_bus *
}
}
+void bcma_init_bus(struct bcma_bus *bus)
+{
+ mutex_lock(&bcma_buses_mutex);
+ bus->num = bcma_bus_next_num++;
+ mutex_unlock(&bcma_buses_mutex);
+
+ INIT_LIST_HEAD(&bus->cores);
+ bus->nr_cores = 0;
+
+ bcma_detect_chip(bus);
+}
+
static void bcma_register_core(struct bcma_bus *bus, struct bcma_device *core)
{
int err;
@@ -356,12 +368,19 @@ static void bcma_unregister_cores(struct
struct bcma_device *core, *tmp;
list_for_each_entry_safe(core, tmp, &bus->cores, list) {
+ if (!core->dev_registered)
+ continue;
list_del(&core->list);
- if (core->dev_registered)
- device_unregister(&core->dev);
+ device_unregister(&core->dev);
}
if (bus->hosttype == BCMA_HOSTTYPE_SOC)
platform_device_unregister(bus->drv_cc.watchdog);
+
+ /* Now noone uses internally-handled cores, we can free them */
+ list_for_each_entry_safe(core, tmp, &bus->cores, list) {
+ list_del(&core->list);
+ kfree(core);
+ }
}
int bcma_bus_register(struct bcma_bus *bus)
@@ -369,10 +388,6 @@ int bcma_bus_register(struct bcma_bus *b
int err;
struct bcma_device *core;
- mutex_lock(&bcma_buses_mutex);
- bus->num = bcma_bus_next_num++;
- mutex_unlock(&bcma_buses_mutex);
-
/* Scan for devices (cores) */
err = bcma_bus_scan(bus);
if (err) {
@@ -387,6 +402,13 @@ int bcma_bus_register(struct bcma_bus *b
bcma_core_chipcommon_early_init(&bus->drv_cc);
}
+ /* Early init PCIE core */
+ core = bcma_find_core(bus, BCMA_CORE_PCIE);
+ if (core) {
+ bus->drv_pci[0].core = core;
+ bcma_core_pci_early_init(&bus->drv_pci[0]);
+ }
+
/* Cores providing flash access go before SPROM init */
list_for_each_entry(core, &bus->cores, list) {
if (bcma_is_core_needed_early(core->id.id))
@@ -459,7 +481,6 @@ int bcma_bus_register(struct bcma_bus *b
void bcma_bus_unregister(struct bcma_bus *bus)
{
- struct bcma_device *cores[3];
int err;
err = bcma_gpio_unregister(&bus->drv_cc);
@@ -470,46 +491,23 @@ void bcma_bus_unregister(struct bcma_bus
bcma_core_chipcommon_b_free(&bus->drv_cc_b);
- cores[0] = bcma_find_core(bus, BCMA_CORE_MIPS_74K);
- cores[1] = bcma_find_core(bus, BCMA_CORE_PCIE);
- cores[2] = bcma_find_core(bus, BCMA_CORE_4706_MAC_GBIT_COMMON);
-
bcma_unregister_cores(bus);
-
- kfree(cores[2]);
- kfree(cores[1]);
- kfree(cores[0]);
}
-int __init bcma_bus_early_register(struct bcma_bus *bus,
- struct bcma_device *core_cc,
- struct bcma_device *core_mips)
+/*
+ * This is a special version of bus registration function designed for SoCs.
+ * It scans bus and performs basic initialization of main cores only.
+ * Please note it requires memory allocation, however it won't try to sleep.
+ */
+int __init bcma_bus_early_register(struct bcma_bus *bus)
{
int err;
struct bcma_device *core;
- struct bcma_device_id match;
-
- match.manuf = BCMA_MANUF_BCM;
- match.id = bcma_cc_core_id(bus);
- match.class = BCMA_CL_SIM;
- match.rev = BCMA_ANY_REV;
- /* Scan for chip common core */
- err = bcma_bus_scan_early(bus, &match, core_cc);
- if (err) {
- bcma_err(bus, "Failed to scan for common core: %d\n", err);
- return -1;
- }
-
- match.manuf = BCMA_MANUF_MIPS;
- match.id = BCMA_CORE_MIPS_74K;
- match.class = BCMA_CL_SIM;
- match.rev = BCMA_ANY_REV;
-
- /* Scan for mips core */
- err = bcma_bus_scan_early(bus, &match, core_mips);
+ /* Scan for devices (cores) */
+ err = bcma_bus_scan(bus);
if (err) {
- bcma_err(bus, "Failed to scan for mips core: %d\n", err);
+ bcma_err(bus, "Failed to scan bus: %d\n", err);
return -1;
}
--- a/drivers/bcma/scan.c
+++ b/drivers/bcma/scan.c
@@ -435,15 +435,12 @@ static int bcma_get_next_core(struct bcm
return 0;
}
-void bcma_init_bus(struct bcma_bus *bus)
+void bcma_detect_chip(struct bcma_bus *bus)
{
s32 tmp;
struct bcma_chipinfo *chipinfo = &(bus->chipinfo);
char chip_id[8];
- INIT_LIST_HEAD(&bus->cores);
- bus->nr_cores = 0;
-
bcma_scan_switch_core(bus, BCMA_ADDR_BASE);
tmp = bcma_scan_read32(bus, 0, BCMA_CC_ID);
@@ -464,6 +461,10 @@ int bcma_bus_scan(struct bcma_bus *bus)
int err, core_num = 0;
+ /* Skip if bus was already scanned (e.g. during early register) */
+ if (bus->nr_cores)
+ return 0;
+
erombase = bcma_scan_read32(bus, 0, BCMA_CC_EROM);
if (bus->hosttype == BCMA_HOSTTYPE_SOC) {
eromptr = ioremap_nocache(erombase, BCMA_CORE_SIZE);
@@ -519,64 +520,6 @@ int bcma_bus_scan(struct bcma_bus *bus)
out:
if (bus->hosttype == BCMA_HOSTTYPE_SOC)
iounmap(eromptr);
-
- return err;
-}
-
-int __init bcma_bus_scan_early(struct bcma_bus *bus,
- struct bcma_device_id *match,
- struct bcma_device *core)
-{
- u32 erombase;
- u32 __iomem *eromptr, *eromend;
-
- int err = -ENODEV;
- int core_num = 0;
-
- erombase = bcma_scan_read32(bus, 0, BCMA_CC_EROM);
- if (bus->hosttype == BCMA_HOSTTYPE_SOC) {
- eromptr = ioremap_nocache(erombase, BCMA_CORE_SIZE);
- if (!eromptr)
- return -ENOMEM;
- } else {
- eromptr = bus->mmio;
- }
-
- eromend = eromptr + BCMA_CORE_SIZE / sizeof(u32);
-
- bcma_scan_switch_core(bus, erombase);
-
- while (eromptr < eromend) {
- memset(core, 0, sizeof(*core));
- INIT_LIST_HEAD(&core->list);
- core->bus = bus;
-
- err = bcma_get_next_core(bus, &eromptr, match, core_num, core);
- if (err == -ENODEV) {
- core_num++;
- continue;
- } else if (err == -ENXIO)
- continue;
- else if (err == -ESPIPE)
- break;
- else if (err < 0)
- goto out;
-
- core->core_index = core_num++;
- bus->nr_cores++;
- bcma_info(bus, "Core %d found: %s (manuf 0x%03X, id 0x%03X, rev 0x%02X, class 0x%X)\n",
- core->core_index, bcma_device_name(&core->id),
- core->id.manuf, core->id.id, core->id.rev,
- core->id.class);
-
- list_add_tail(&core->list, &bus->cores);
- err = 0;
- break;
- }
-
-out:
- if (bus->hosttype == BCMA_HOSTTYPE_SOC)
- iounmap(eromptr);
return err;
}
--- a/drivers/bcma/sprom.c
+++ b/drivers/bcma/sprom.c
@@ -579,7 +579,8 @@ int bcma_sprom_get(struct bcma_bus *bus)
u16 offset = BCMA_CC_SPROM;
u16 *sprom;
size_t sprom_sizes[] = { SSB_SPROMSIZE_WORDS_R4,
- SSB_SPROMSIZE_WORDS_R10, };
+ SSB_SPROMSIZE_WORDS_R10,
+ SSB_SPROMSIZE_WORDS_R11, };
int i, err = 0;
if (!bus->drv_cc.core)
--- a/include/linux/bcma/bcma.h
+++ b/include/linux/bcma/bcma.h
@@ -319,6 +319,7 @@ struct bcma_bus {
const struct bcma_host_ops *ops;
enum bcma_hosttype hosttype;
+ bool host_is_pcie2; /* Used for BCMA_HOSTTYPE_PCI only */
union {
/* Pointer to the PCI bus (only for BCMA_HOSTTYPE_PCI) */
struct pci_dev *host_pci;
--- a/include/linux/bcma/bcma_driver_pci.h
+++ b/include/linux/bcma/bcma_driver_pci.h
@@ -223,6 +223,7 @@ struct bcma_drv_pci_host {
struct bcma_drv_pci {
struct bcma_device *core;
+ u8 early_setup_done:1;
u8 setup_done:1;
u8 hostmode:1;
@@ -237,6 +238,7 @@ struct bcma_drv_pci {
#define pcicore_write16(pc, offset, val) bcma_write16((pc)->core, offset, val)
#define pcicore_write32(pc, offset, val) bcma_write32((pc)->core, offset, val)
+extern void bcma_core_pci_early_init(struct bcma_drv_pci *pc);
extern void bcma_core_pci_init(struct bcma_drv_pci *pc);
extern int bcma_core_pci_irq_ctl(struct bcma_drv_pci *pc,
struct bcma_device *core, bool enable);
--- a/include/linux/bcma/bcma_regs.h
+++ b/include/linux/bcma/bcma_regs.h
@@ -64,6 +64,8 @@
#define BCMA_PCI_GPIO_XTAL 0x40 /* PCI config space GPIO 14 for Xtal powerup */
#define BCMA_PCI_GPIO_PLL 0x80 /* PCI config space GPIO 15 for PLL powerdown */
+#define BCMA_PCIE2_BAR0_WIN2 0x70
+
/* SiliconBackplane Address Map.
* All regions may not exist on all chips.
*/
--- a/include/linux/bcma/bcma_soc.h
+++ b/include/linux/bcma/bcma_soc.h
@@ -5,8 +5,6 @@
struct bcma_soc {
struct bcma_bus bus;
- struct bcma_device core_cc;
- struct bcma_device core_mips;
};
int __init bcma_host_soc_register(struct bcma_soc *soc);
--- a/include/linux/ssb/ssb_regs.h
+++ b/include/linux/ssb/ssb_regs.h
@@ -173,6 +173,7 @@
#define SSB_SPROMSIZE_BYTES_R123 (SSB_SPROMSIZE_WORDS_R123 * sizeof(u16))
#define SSB_SPROMSIZE_BYTES_R4 (SSB_SPROMSIZE_WORDS_R4 * sizeof(u16))
#define SSB_SPROMSIZE_WORDS_R10 230
+#define SSB_SPROMSIZE_WORDS_R11 234
#define SSB_SPROM_BASE1 0x1000
#define SSB_SPROM_BASE31 0x0800
#define SSB_SPROM_REVISION 0x007E

680
target/linux/generic/pending-3.18/027-bcma-from-4.1.patch
View File

@ -1,680 +0,0 @@
--- a/drivers/bcma/bcma_private.h
+++ b/drivers/bcma/bcma_private.h
@@ -24,6 +24,7 @@ bool bcma_wait_value(struct bcma_device
int timeout);
void bcma_prepare_core(struct bcma_bus *bus, struct bcma_device *core);
void bcma_init_bus(struct bcma_bus *bus);
+void bcma_unregister_cores(struct bcma_bus *bus);
int bcma_bus_register(struct bcma_bus *bus);
void bcma_bus_unregister(struct bcma_bus *bus);
int __init bcma_bus_early_register(struct bcma_bus *bus);
@@ -40,6 +41,9 @@ int bcma_bus_scan(struct bcma_bus *bus);
int bcma_sprom_get(struct bcma_bus *bus);
/* driver_chipcommon.c */
+void bcma_core_chipcommon_early_init(struct bcma_drv_cc *cc);
+void bcma_core_chipcommon_init(struct bcma_drv_cc *cc);
+void bcma_chipco_bcm4331_ext_pa_lines_ctl(struct bcma_drv_cc *cc, bool enable);
#ifdef CONFIG_BCMA_DRIVER_MIPS
void bcma_chipco_serial_init(struct bcma_drv_cc *cc);
extern struct platform_device bcma_pflash_dev;
@@ -50,6 +54,8 @@ int bcma_core_chipcommon_b_init(struct b
void bcma_core_chipcommon_b_free(struct bcma_drv_cc_b *ccb);
/* driver_chipcommon_pmu.c */
+void bcma_pmu_early_init(struct bcma_drv_cc *cc);
+void bcma_pmu_init(struct bcma_drv_cc *cc);
u32 bcma_pmu_get_alp_clock(struct bcma_drv_cc *cc);
u32 bcma_pmu_get_cpu_clock(struct bcma_drv_cc *cc);
@@ -98,7 +104,35 @@ static inline void __exit bcma_host_soc_
#endif /* CONFIG_BCMA_HOST_SOC && CONFIG_OF */
/* driver_pci.c */
+#ifdef CONFIG_BCMA_DRIVER_PCI
u32 bcma_pcie_read(struct bcma_drv_pci *pc, u32 address);
+void bcma_core_pci_early_init(struct bcma_drv_pci *pc);
+void bcma_core_pci_init(struct bcma_drv_pci *pc);
+void bcma_core_pci_up(struct bcma_drv_pci *pc);
+void bcma_core_pci_down(struct bcma_drv_pci *pc);
+#else
+static inline void bcma_core_pci_early_init(struct bcma_drv_pci *pc)
+{
+ WARN_ON(pc->core->bus->hosttype == BCMA_HOSTTYPE_PCI);
+}
+static inline void bcma_core_pci_init(struct bcma_drv_pci *pc)
+{
+ /* Initialization is required for PCI hosted bus */
+ WARN_ON(pc->core->bus->hosttype == BCMA_HOSTTYPE_PCI);
+}
+#endif
+
+/* driver_pcie2.c */
+#ifdef CONFIG_BCMA_DRIVER_PCI
+void bcma_core_pcie2_init(struct bcma_drv_pcie2 *pcie2);
+void bcma_core_pcie2_up(struct bcma_drv_pcie2 *pcie2);
+#else
+static inline void bcma_core_pcie2_init(struct bcma_drv_pcie2 *pcie2)
+{
+ /* Initialization is required for PCI hosted bus */
+ WARN_ON(pcie2->core->bus->hosttype == BCMA_HOSTTYPE_PCI);
+}
+#endif
extern int bcma_chipco_watchdog_register(struct bcma_drv_cc *cc);
@@ -115,6 +149,39 @@ static inline void bcma_core_pci_hostmod
}
#endif /* CONFIG_BCMA_DRIVER_PCI_HOSTMODE */
+/**************************************************
+ * driver_mips.c
+ **************************************************/
+
+#ifdef CONFIG_BCMA_DRIVER_MIPS
+unsigned int bcma_core_mips_irq(struct bcma_device *dev);
+void bcma_core_mips_early_init(struct bcma_drv_mips *mcore);
+void bcma_core_mips_init(struct bcma_drv_mips *mcore);
+#else
+static inline unsigned int bcma_core_mips_irq(struct bcma_device *dev)
+{
+ return 0;
+}
+static inline void bcma_core_mips_early_init(struct bcma_drv_mips *mcore)
+{
+}
+static inline void bcma_core_mips_init(struct bcma_drv_mips *mcore)
+{
+}
+#endif
+
+/**************************************************
+ * driver_gmac_cmn.c
+ **************************************************/
+
+#ifdef CONFIG_BCMA_DRIVER_GMAC_CMN
+void bcma_core_gmac_cmn_init(struct bcma_drv_gmac_cmn *gc);
+#else
+static inline void bcma_core_gmac_cmn_init(struct bcma_drv_gmac_cmn *gc)
+{
+}
+#endif
+
#ifdef CONFIG_BCMA_DRIVER_GPIO
/* driver_gpio.c */
int bcma_gpio_init(struct bcma_drv_cc *cc);
--- a/drivers/bcma/driver_gpio.c
+++ b/drivers/bcma/driver_gpio.c
@@ -17,6 +17,8 @@
#include "bcma_private.h"
+#define BCMA_GPIO_MAX_PINS 32
+
static inline struct bcma_drv_cc *bcma_gpio_get_cc(struct gpio_chip *chip)
{
return container_of(chip, struct bcma_drv_cc, gpio);
@@ -76,7 +78,7 @@ static void bcma_gpio_free(struct gpio_c
bcma_chipco_gpio_pullup(cc, 1 << gpio, 0);
}
-#if IS_BUILTIN(CONFIG_BCM47XX)
+#if IS_BUILTIN(CONFIG_BCM47XX) || IS_BUILTIN(CONFIG_ARCH_BCM_5301X)
static int bcma_gpio_to_irq(struct gpio_chip *chip, unsigned gpio)
{
struct bcma_drv_cc *cc = bcma_gpio_get_cc(chip);
@@ -204,6 +206,7 @@ static void bcma_gpio_irq_domain_exit(st
int bcma_gpio_init(struct bcma_drv_cc *cc)
{
+ struct bcma_bus *bus = cc->core->bus;
struct gpio_chip *chip = &cc->gpio;
int err;
@@ -215,14 +218,14 @@ int bcma_gpio_init(struct bcma_drv_cc *c
chip->set = bcma_gpio_set_value;
chip->direction_input = bcma_gpio_direction_input;
chip->direction_output = bcma_gpio_direction_output;
-#if IS_BUILTIN(CONFIG_BCM47XX)
+#if IS_BUILTIN(CONFIG_BCM47XX) || IS_BUILTIN(CONFIG_ARCH_BCM_5301X)
chip->to_irq = bcma_gpio_to_irq;
#endif
#if IS_BUILTIN(CONFIG_OF)
if (cc->core->bus->hosttype == BCMA_HOSTTYPE_SOC)
chip->of_node = cc->core->dev.of_node;
#endif
- switch (cc->core->bus->chipinfo.id) {
+ switch (bus->chipinfo.id) {
case BCMA_CHIP_ID_BCM5357:
case BCMA_CHIP_ID_BCM53572:
chip->ngpio = 32;
@@ -231,13 +234,17 @@ int bcma_gpio_init(struct bcma_drv_cc *c
chip->ngpio = 16;
}
- /* There is just one SoC in one device and its GPIO addresses should be
- * deterministic to address them more easily. The other buses could get
- * a random base number. */
- if (cc->core->bus->hosttype == BCMA_HOSTTYPE_SOC)
- chip->base = 0;
- else
- chip->base = -1;
+ /*
+ * On MIPS we register GPIO devices (LEDs, buttons) using absolute GPIO
+ * pin numbers. We don't have Device Tree there and we can't really use
+ * relative (per chip) numbers.
+ * So let's use predictable base for BCM47XX and "random" for all other.
+ */
+#if IS_BUILTIN(CONFIG_BCM47XX)
+ chip->base = bus->num * BCMA_GPIO_MAX_PINS;
+#else
+ chip->base = -1;
+#endif
err = bcma_gpio_irq_domain_init(cc);
if (err)
--- a/drivers/bcma/driver_pci.c
+++ b/drivers/bcma/driver_pci.c
@@ -282,39 +282,6 @@ void bcma_core_pci_power_save(struct bcm
}
EXPORT_SYMBOL_GPL(bcma_core_pci_power_save);
-int bcma_core_pci_irq_ctl(struct bcma_drv_pci *pc, struct bcma_device *core,
- bool enable)
-{
- struct pci_dev *pdev;
- u32 coremask, tmp;
- int err = 0;
-
- if (!pc || core->bus->hosttype != BCMA_HOSTTYPE_PCI) {
- /* This bcma device is not on a PCI host-bus. So the IRQs are
- * not routed through the PCI core.
- * So we must not enable routing through the PCI core. */
- goto out;
- }
-
- pdev = pc->core->bus->host_pci;
-
- err = pci_read_config_dword(pdev, BCMA_PCI_IRQMASK, &tmp);
- if (err)
- goto out;
-
- coremask = BIT(core->core_index) << 8;
- if (enable)
- tmp |= coremask;
- else
- tmp &= ~coremask;
-
- err = pci_write_config_dword(pdev, BCMA_PCI_IRQMASK, tmp);
-
-out:
- return err;
-}
-EXPORT_SYMBOL_GPL(bcma_core_pci_irq_ctl);
-
static void bcma_core_pci_extend_L1timer(struct bcma_drv_pci *pc, bool extend)
{
u32 w;
@@ -328,28 +295,12 @@ static void bcma_core_pci_extend_L1timer
bcma_pcie_read(pc, BCMA_CORE_PCI_DLLP_PMTHRESHREG);
}
-void bcma_core_pci_up(struct bcma_bus *bus)
+void bcma_core_pci_up(struct bcma_drv_pci *pc)
{
- struct bcma_drv_pci *pc;
-
- if (bus->hosttype != BCMA_HOSTTYPE_PCI)
- return;
-
- pc = &bus->drv_pci[0];
-
bcma_core_pci_extend_L1timer(pc, true);
}
-EXPORT_SYMBOL_GPL(bcma_core_pci_up);
-void bcma_core_pci_down(struct bcma_bus *bus)
+void bcma_core_pci_down(struct bcma_drv_pci *pc)
{
- struct bcma_drv_pci *pc;
-
- if (bus->hosttype != BCMA_HOSTTYPE_PCI)
- return;
-
- pc = &bus->drv_pci[0];
-
bcma_core_pci_extend_L1timer(pc, false);
}
-EXPORT_SYMBOL_GPL(bcma_core_pci_down);
--- a/drivers/bcma/driver_pci_host.c
+++ b/drivers/bcma/driver_pci_host.c
@@ -11,6 +11,7 @@
#include "bcma_private.h"
#include <linux/pci.h>
+#include <linux/slab.h>
#include <linux/export.h>
#include <linux/bcma/bcma.h>
#include <asm/paccess.h>
--- a/drivers/bcma/driver_pcie2.c
+++ b/drivers/bcma/driver_pcie2.c
@@ -10,6 +10,7 @@
#include "bcma_private.h"
#include <linux/bcma/bcma.h>
+#include <linux/pci.h>
/**************************************************
* R/W ops.
@@ -156,14 +157,23 @@ static void pciedev_reg_pm_clk_period(st
void bcma_core_pcie2_init(struct bcma_drv_pcie2 *pcie2)
{
- struct bcma_chipinfo *ci = &pcie2->core->bus->chipinfo;
+ struct bcma_bus *bus = pcie2->core->bus;
+ struct bcma_chipinfo *ci = &bus->chipinfo;
u32 tmp;
tmp = pcie2_read32(pcie2, BCMA_CORE_PCIE2_SPROM(54));
if ((tmp & 0xe) >> 1 == 2)
bcma_core_pcie2_cfg_write(pcie2, 0x4e0, 0x17);
- /* TODO: Do we need pcie_reqsize? */
+ switch (bus->chipinfo.id) {
+ case BCMA_CHIP_ID_BCM4360:
+ case BCMA_CHIP_ID_BCM4352:
+ pcie2->reqsize = 1024;
+ break;
+ default:
+ pcie2->reqsize = 128;
+ break;
+ }
if (ci->id == BCMA_CHIP_ID_BCM4360 && ci->rev > 3)
bcma_core_pcie2_war_delay_perst_enab(pcie2, true);
@@ -173,3 +183,18 @@ void bcma_core_pcie2_init(struct bcma_dr
pciedev_crwlpciegen2_180(pcie2);
pciedev_crwlpciegen2_182(pcie2);
}
+
+/**************************************************
+ * Runtime ops.
+ **************************************************/
+
+void bcma_core_pcie2_up(struct bcma_drv_pcie2 *pcie2)
+{
+ struct bcma_bus *bus = pcie2->core->bus;
+ struct pci_dev *dev = bus->host_pci;
+ int err;
+
+ err = pcie_set_readrq(dev, pcie2->reqsize);
+ if (err)
+ bcma_err(bus, "Error setting PCI_EXP_DEVCTL_READRQ: %d\n", err);
+}
--- a/drivers/bcma/host_pci.c
+++ b/drivers/bcma/host_pci.c
@@ -213,16 +213,26 @@ static int bcma_host_pci_probe(struct pc
/* Initialize struct, detect chip */
bcma_init_bus(bus);
+ /* Scan bus to find out generation of PCIe core */
+ err = bcma_bus_scan(bus);
+ if (err)
+ goto err_pci_unmap_mmio;
+
+ if (bcma_find_core(bus, BCMA_CORE_PCIE2))
+ bus->host_is_pcie2 = true;
+
/* Register */
err = bcma_bus_register(bus);
if (err)
- goto err_pci_unmap_mmio;
+ goto err_unregister_cores;
pci_set_drvdata(dev, bus);
out:
return err;
+err_unregister_cores:
+ bcma_unregister_cores(bus);
err_pci_unmap_mmio:
pci_iounmap(dev, bus->mmio);
err_pci_release_regions:
@@ -283,9 +293,12 @@ static const struct pci_device_id bcma_p
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4357) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4358) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4359) },
+ { PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4360) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4365) },
+ { PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43a0) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43a9) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43aa) },
+ { PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43b1) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4727) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 43227) }, /* 0xa8db, BCM43217 (sic!) */
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 43228) }, /* 0xa8dc */
@@ -310,3 +323,65 @@ void __exit bcma_host_pci_exit(void)
{
pci_unregister_driver(&bcma_pci_bridge_driver);
}
+
+/**************************************************
+ * Runtime ops for drivers.
+ **************************************************/
+
+/* See also pcicore_up */
+void bcma_host_pci_up(struct bcma_bus *bus)
+{
+ if (bus->hosttype != BCMA_HOSTTYPE_PCI)
+ return;
+
+ if (bus->host_is_pcie2)
+ bcma_core_pcie2_up(&bus->drv_pcie2);
+ else
+ bcma_core_pci_up(&bus->drv_pci[0]);
+}
+EXPORT_SYMBOL_GPL(bcma_host_pci_up);
+
+/* See also pcicore_down */
+void bcma_host_pci_down(struct bcma_bus *bus)
+{
+ if (bus->hosttype != BCMA_HOSTTYPE_PCI)
+ return;
+
+ if (!bus->host_is_pcie2)
+ bcma_core_pci_down(&bus->drv_pci[0]);
+}
+EXPORT_SYMBOL_GPL(bcma_host_pci_down);
+
+/* See also si_pci_setup */
+int bcma_host_pci_irq_ctl(struct bcma_bus *bus, struct bcma_device *core,
+ bool enable)
+{
+ struct pci_dev *pdev;
+ u32 coremask, tmp;
+ int err = 0;
+
+ if (bus->hosttype != BCMA_HOSTTYPE_PCI) {
+ /* This bcma device is not on a PCI host-bus. So the IRQs are
+ * not routed through the PCI core.
+ * So we must not enable routing through the PCI core. */
+ goto out;
+ }
+
+ pdev = bus->host_pci;
+
+ err = pci_read_config_dword(pdev, BCMA_PCI_IRQMASK, &tmp);
+ if (err)
+ goto out;
+
+ coremask = BIT(core->core_index) << 8;
+ if (enable)
+ tmp |= coremask;
+ else
+ tmp &= ~coremask;
+
+ err = pci_write_config_dword(pdev, BCMA_PCI_IRQMASK, tmp);
+
+out:
+ return err;
+}
+EXPORT_SYMBOL_GPL(bcma_host_pci_irq_ctl);
--- a/drivers/bcma/main.c
+++ b/drivers/bcma/main.c
@@ -363,7 +363,7 @@ static int bcma_register_devices(struct
return 0;
}
-static void bcma_unregister_cores(struct bcma_bus *bus)
+void bcma_unregister_cores(struct bcma_bus *bus)
{
struct bcma_device *core, *tmp;
--- a/drivers/net/wireless/b43/main.c
+++ b/drivers/net/wireless/b43/main.c
@@ -4770,7 +4770,7 @@ static void b43_wireless_core_exit(struc
switch (dev->dev->bus_type) {
#ifdef CONFIG_B43_BCMA
case B43_BUS_BCMA:
- bcma_core_pci_down(dev->dev->bdev->bus);
+ bcma_host_pci_down(dev->dev->bdev->bus);
break;
#endif
#ifdef CONFIG_B43_SSB
@@ -4817,9 +4817,9 @@ static int b43_wireless_core_init(struct
switch (dev->dev->bus_type) {
#ifdef CONFIG_B43_BCMA
case B43_BUS_BCMA:
- bcma_core_pci_irq_ctl(&dev->dev->bdev->bus->drv_pci[0],
+ bcma_host_pci_irq_ctl(dev->dev->bdev->bus,
dev->dev->bdev, true);
- bcma_core_pci_up(dev->dev->bdev->bus);
+ bcma_host_pci_up(dev->dev->bdev->bus);
break;
#endif
#ifdef CONFIG_B43_SSB
--- a/drivers/net/wireless/brcm80211/brcmsmac/main.c
+++ b/drivers/net/wireless/brcm80211/brcmsmac/main.c
@@ -4669,7 +4669,7 @@ static int brcms_b_attach(struct brcms_c
brcms_c_coredisable(wlc_hw);
/* Match driver "down" state */
- bcma_core_pci_down(wlc_hw->d11core->bus);
+ bcma_host_pci_down(wlc_hw->d11core->bus);
/* turn off pll and xtal to match driver "down" state */
brcms_b_xtal(wlc_hw, OFF);
@@ -4960,7 +4960,7 @@ static int brcms_b_up_prep(struct brcms_
* Configure pci/pcmcia here instead of in brcms_c_attach()
* to allow mfg hotswap: down, hotswap (chip power cycle), up.
*/
- bcma_core_pci_irq_ctl(&wlc_hw->d11core->bus->drv_pci[0], wlc_hw->d11core,
+ bcma_host_pci_irq_ctl(wlc_hw->d11core->bus, wlc_hw->d11core,
true);
/*
@@ -4970,12 +4970,12 @@ static int brcms_b_up_prep(struct brcms_
*/
if (brcms_b_radio_read_hwdisabled(wlc_hw)) {
/* put SB PCI in down state again */
- bcma_core_pci_down(wlc_hw->d11core->bus);
+ bcma_host_pci_down(wlc_hw->d11core->bus);
brcms_b_xtal(wlc_hw, OFF);
return -ENOMEDIUM;
}
- bcma_core_pci_up(wlc_hw->d11core->bus);
+ bcma_host_pci_up(wlc_hw->d11core->bus);
/* reset the d11 core */
brcms_b_corereset(wlc_hw, BRCMS_USE_COREFLAGS);
@@ -5172,7 +5172,7 @@ static int brcms_b_down_finish(struct br
/* turn off primary xtal and pll */
if (!wlc_hw->noreset) {
- bcma_core_pci_down(wlc_hw->d11core->bus);
+ bcma_host_pci_down(wlc_hw->d11core->bus);
brcms_b_xtal(wlc_hw, OFF);
}
}
--- a/include/linux/bcma/bcma.h
+++ b/include/linux/bcma/bcma.h
@@ -435,6 +435,27 @@ static inline struct bcma_device *bcma_f
return bcma_find_core_unit(bus, coreid, 0);
}
+#ifdef CONFIG_BCMA_HOST_PCI
+extern void bcma_host_pci_up(struct bcma_bus *bus);
+extern void bcma_host_pci_down(struct bcma_bus *bus);
+extern int bcma_host_pci_irq_ctl(struct bcma_bus *bus,
+ struct bcma_device *core, bool enable);
+#else
+static inline void bcma_host_pci_up(struct bcma_bus *bus)
+{
+}
+static inline void bcma_host_pci_down(struct bcma_bus *bus)
+{
+}
+static inline int bcma_host_pci_irq_ctl(struct bcma_bus *bus,
+ struct bcma_device *core, bool enable)
+{
+ if (bus->hosttype == BCMA_HOSTTYPE_PCI)
+ return -ENOTSUPP;
+ return 0;
+}
+#endif
+
extern bool bcma_core_is_enabled(struct bcma_device *core);
extern void bcma_core_disable(struct bcma_device *core, u32 flags);
extern int bcma_core_enable(struct bcma_device *core, u32 flags);
--- a/include/linux/bcma/bcma_driver_pci.h
+++ b/include/linux/bcma/bcma_driver_pci.h
@@ -238,13 +238,13 @@ struct bcma_drv_pci {
#define pcicore_write16(pc, offset, val) bcma_write16((pc)->core, offset, val)
#define pcicore_write32(pc, offset, val) bcma_write32((pc)->core, offset, val)
-extern void bcma_core_pci_early_init(struct bcma_drv_pci *pc);
-extern void bcma_core_pci_init(struct bcma_drv_pci *pc);
-extern int bcma_core_pci_irq_ctl(struct bcma_drv_pci *pc,
- struct bcma_device *core, bool enable);
-extern void bcma_core_pci_up(struct bcma_bus *bus);
-extern void bcma_core_pci_down(struct bcma_bus *bus);
+#ifdef CONFIG_BCMA_DRIVER_PCI
extern void bcma_core_pci_power_save(struct bcma_bus *bus, bool up);
+#else
+static inline void bcma_core_pci_power_save(struct bcma_bus *bus, bool up)
+{
+}
+#endif
extern int bcma_core_pci_pcibios_map_irq(const struct pci_dev *dev);
extern int bcma_core_pci_plat_dev_init(struct pci_dev *dev);
--- a/include/linux/bcma/bcma_driver_pcie2.h
+++ b/include/linux/bcma/bcma_driver_pcie2.h
@@ -143,6 +143,8 @@
struct bcma_drv_pcie2 {
struct bcma_device *core;
+
+ u16 reqsize;
};
#define pcie2_read16(pcie2, offset) bcma_read16((pcie2)->core, offset)
@@ -153,6 +155,4 @@ struct bcma_drv_pcie2 {
#define pcie2_set32(pcie2, offset, set) bcma_set32((pcie2)->core, offset, set)
#define pcie2_mask32(pcie2, offset, mask) bcma_mask32((pcie2)->core, offset, mask)
-void bcma_core_pcie2_init(struct bcma_drv_pcie2 *pcie2);
-
#endif /* LINUX_BCMA_DRIVER_PCIE2_H_ */
--- a/drivers/bcma/Kconfig
+++ b/drivers/bcma/Kconfig
@@ -26,6 +26,7 @@ config BCMA_HOST_PCI_POSSIBLE
config BCMA_HOST_PCI
bool "Support for BCMA on PCI-host bus"
depends on BCMA_HOST_PCI_POSSIBLE
+ select BCMA_DRIVER_PCI
default y
config BCMA_DRIVER_PCI_HOSTMODE
@@ -44,6 +45,22 @@ config BCMA_HOST_SOC
If unsure, say N
+config BCMA_DRIVER_PCI
+ bool "BCMA Broadcom PCI core driver"
+ depends on BCMA && PCI
+ default y
+ help
+ BCMA bus may have many versions of PCIe core. This driver
+ supports:
+ 1) PCIe core working in clientmode
+ 2) PCIe Gen 2 clientmode core
+
+ In general PCIe (Gen 2) clientmode core is required on PCIe
+ hosted buses. It's responsible for initialization and basic
+ hardware management.
+ This driver is also prerequisite for a hostmode PCIe core
+ support.
+
config BCMA_DRIVER_MIPS
bool "BCMA Broadcom MIPS core driver"
depends on BCMA && MIPS
--- a/drivers/bcma/Makefile
+++ b/drivers/bcma/Makefile
@@ -3,8 +3,8 @@ bcma-y += driver_chipcommon.o driver
bcma-y += driver_chipcommon_b.o
bcma-$(CONFIG_BCMA_SFLASH) += driver_chipcommon_sflash.o
bcma-$(CONFIG_BCMA_NFLASH) += driver_chipcommon_nflash.o
-bcma-y += driver_pci.o
-bcma-y += driver_pcie2.o
+bcma-$(CONFIG_BCMA_DRIVER_PCI) += driver_pci.o
+bcma-$(CONFIG_BCMA_DRIVER_PCI) += driver_pcie2.o
bcma-$(CONFIG_BCMA_DRIVER_PCI_HOSTMODE) += driver_pci_host.o
bcma-$(CONFIG_BCMA_DRIVER_MIPS) += driver_mips.o
bcma-$(CONFIG_BCMA_DRIVER_GMAC_CMN) += driver_gmac_cmn.o
--- a/include/linux/bcma/bcma_driver_chipcommon.h
+++ b/include/linux/bcma/bcma_driver_chipcommon.h
@@ -663,14 +663,6 @@ struct bcma_drv_cc_b {
#define bcma_cc_maskset32(cc, offset, mask, set) \
bcma_cc_write32(cc, offset, (bcma_cc_read32(cc, offset) & (mask)) | (set))
-extern void bcma_core_chipcommon_init(struct bcma_drv_cc *cc);
-extern void bcma_core_chipcommon_early_init(struct bcma_drv_cc *cc);
-
-extern void bcma_chipco_suspend(struct bcma_drv_cc *cc);
-extern void bcma_chipco_resume(struct bcma_drv_cc *cc);
-
-void bcma_chipco_bcm4331_ext_pa_lines_ctl(struct bcma_drv_cc *cc, bool enable);
-
extern u32 bcma_chipco_watchdog_timer_set(struct bcma_drv_cc *cc, u32 ticks);
extern u32 bcma_chipco_get_alp_clock(struct bcma_drv_cc *cc);
@@ -690,9 +682,6 @@ u32 bcma_chipco_gpio_pullup(struct bcma_
u32 bcma_chipco_gpio_pulldown(struct bcma_drv_cc *cc, u32 mask, u32 value);
/* PMU support */
-extern void bcma_pmu_init(struct bcma_drv_cc *cc);
-extern void bcma_pmu_early_init(struct bcma_drv_cc *cc);
-
extern void bcma_chipco_pll_write(struct bcma_drv_cc *cc, u32 offset,
u32 value);
extern void bcma_chipco_pll_maskset(struct bcma_drv_cc *cc, u32 offset,
--- a/include/linux/bcma/bcma_driver_gmac_cmn.h
+++ b/include/linux/bcma/bcma_driver_gmac_cmn.h
@@ -91,10 +91,4 @@ struct bcma_drv_gmac_cmn {
#define gmac_cmn_write16(gc, offset, val) bcma_write16((gc)->core, offset, val)
#define gmac_cmn_write32(gc, offset, val) bcma_write32((gc)->core, offset, val)
-#ifdef CONFIG_BCMA_DRIVER_GMAC_CMN
-extern void bcma_core_gmac_cmn_init(struct bcma_drv_gmac_cmn *gc);
-#else
-static inline void bcma_core_gmac_cmn_init(struct bcma_drv_gmac_cmn *gc) { }
-#endif
-
#endif /* LINUX_BCMA_DRIVER_GMAC_CMN_H_ */
--- a/include/linux/bcma/bcma_driver_mips.h
+++ b/include/linux/bcma/bcma_driver_mips.h
@@ -39,21 +39,6 @@ struct bcma_drv_mips {
u8 early_setup_done:1;
};
-#ifdef CONFIG_BCMA_DRIVER_MIPS
-extern void bcma_core_mips_init(struct bcma_drv_mips *mcore);
-extern void bcma_core_mips_early_init(struct bcma_drv_mips *mcore);
-
-extern unsigned int bcma_core_mips_irq(struct bcma_device *dev);
-#else
-static inline void bcma_core_mips_init(struct bcma_drv_mips *mcore) { }
-static inline void bcma_core_mips_early_init(struct bcma_drv_mips *mcore) { }
-
-static inline unsigned int bcma_core_mips_irq(struct bcma_device *dev)
-{
- return 0;
-}
-#endif
-
extern u32 bcma_cpu_clock(struct bcma_drv_mips *mcore);
#endif /* LINUX_BCMA_DRIVER_MIPS_H_ */

86
target/linux/generic/pending-3.18/028-bcma-from-4.2.patch
View File

@ -1,86 +0,0 @@
--- a/drivers/bcma/driver_gpio.c
+++ b/drivers/bcma/driver_gpio.c
@@ -226,6 +226,7 @@ int bcma_gpio_init(struct bcma_drv_cc *c
chip->of_node = cc->core->dev.of_node;
#endif
switch (bus->chipinfo.id) {
+ case BCMA_CHIP_ID_BCM4707:
case BCMA_CHIP_ID_BCM5357:
case BCMA_CHIP_ID_BCM53572:
chip->ngpio = 32;
@@ -235,16 +236,17 @@ int bcma_gpio_init(struct bcma_drv_cc *c
}
/*
- * On MIPS we register GPIO devices (LEDs, buttons) using absolute GPIO
- * pin numbers. We don't have Device Tree there and we can't really use
- * relative (per chip) numbers.
- * So let's use predictable base for BCM47XX and "random" for all other.
+ * Register SoC GPIO devices with absolute GPIO pin base.
+ * On MIPS, we don't have Device Tree and we can't use relative (per chip)
+ * GPIO numbers.
+ * On some ARM devices, user space may want to access some system GPIO
+ * pins directly, which is easier to do with a predictable GPIO base.
*/
-#if IS_BUILTIN(CONFIG_BCM47XX)
- chip->base = bus->num * BCMA_GPIO_MAX_PINS;
-#else
- chip->base = -1;
-#endif
+ if (IS_BUILTIN(CONFIG_BCM47XX) ||
+ cc->core->bus->hosttype == BCMA_HOSTTYPE_SOC)
+ chip->base = bus->num * BCMA_GPIO_MAX_PINS;
+ else
+ chip->base = -1;
err = bcma_gpio_irq_domain_init(cc);
if (err)
--- a/drivers/bcma/Kconfig
+++ b/drivers/bcma/Kconfig
@@ -29,12 +29,6 @@ config BCMA_HOST_PCI
select BCMA_DRIVER_PCI
default y
-config BCMA_DRIVER_PCI_HOSTMODE
- bool "Driver for PCI core working in hostmode"
- depends on BCMA && MIPS && BCMA_HOST_PCI
- help
- PCI core hostmode operation (external PCI bus).
-
config BCMA_HOST_SOC
bool "Support for BCMA in a SoC"
depends on BCMA
@@ -61,6 +55,12 @@ config BCMA_DRIVER_PCI
This driver is also prerequisite for a hostmode PCIe core
support.
+config BCMA_DRIVER_PCI_HOSTMODE
+ bool "Driver for PCI core working in hostmode"
+ depends on BCMA && MIPS && BCMA_DRIVER_PCI
+ help
+ PCI core hostmode operation (external PCI bus).
+
config BCMA_DRIVER_MIPS
bool "BCMA Broadcom MIPS core driver"
depends on BCMA && MIPS
--- a/include/linux/bcma/bcma_driver_pci.h
+++ b/include/linux/bcma/bcma_driver_pci.h
@@ -246,7 +246,18 @@ static inline void bcma_core_pci_power_s
}
#endif
+#ifdef CONFIG_BCMA_DRIVER_PCI_HOSTMODE
extern int bcma_core_pci_pcibios_map_irq(const struct pci_dev *dev);
extern int bcma_core_pci_plat_dev_init(struct pci_dev *dev);
+#else
+static inline int bcma_core_pci_pcibios_map_irq(const struct pci_dev *dev)
+{
+ return -ENOTSUPP;
+}
+static inline int bcma_core_pci_plat_dev_init(struct pci_dev *dev)
+{
+ return -ENOTSUPP;
+}
+#endif
#endif /* LINUX_BCMA_DRIVER_PCI_H_ */

26
target/linux/generic/pending-3.18/029-bcma-from-4.4.patch
View File

@ -1,26 +0,0 @@
commit 55acca90da52b85299c033354e51ddaa7b73e019
Author: Hante Meuleman <meuleman@broadcom.com>
Date: Fri Sep 18 22:08:17 2015 +0200
brcmfmac: Add support for the BCM4365 and BCM4366 PCIE devices.
This patch adds support for the BCM4365 and BCM4366 11ac Wave2
PCIE devices.
Reviewed-by: Arend Van Spriel <arend@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
--- a/include/linux/bcma/bcma.h
+++ b/include/linux/bcma/bcma.h
@@ -151,6 +151,8 @@ struct bcma_host_ops {
#define BCMA_CORE_PCIE2 0x83C /* PCI Express Gen2 */
#define BCMA_CORE_USB30_DEV 0x83D
#define BCMA_CORE_ARM_CR4 0x83E
+#define BCMA_CORE_ARM_CA7 0x847
+#define BCMA_CORE_SYS_MEM 0x849
#define BCMA_CORE_DEFAULT 0xFFF
#define BCMA_MAX_NR_CORES 16

52
target/linux/generic/pending-3.18/030-backport_bcm47xx_nvram.patch
View File

@ -1,52 +0,0 @@
--- /dev/null
+++ b/include/linux/bcm47xx_nvram.h
@@ -0,0 +1,49 @@
+/*
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ */
+
+#ifndef __BCM47XX_NVRAM_H
+#define __BCM47XX_NVRAM_H
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/vmalloc.h>
+
+#ifdef CONFIG_BCM47XX_NVRAM
+int bcm47xx_nvram_init_from_mem(u32 base, u32 lim);
+int bcm47xx_nvram_getenv(const char *name, char *val, size_t val_len);
+int bcm47xx_nvram_gpio_pin(const char *name);
+char *bcm47xx_nvram_get_contents(size_t *val_len);
+static inline void bcm47xx_nvram_release_contents(char *nvram)
+{
+ vfree(nvram);
+};
+#else
+static inline int bcm47xx_nvram_init_from_mem(u32 base, u32 lim)
+{
+ return -ENOTSUPP;
+};
+static inline int bcm47xx_nvram_getenv(const char *name, char *val,
+ size_t val_len)
+{
+ return -ENOTSUPP;
+};
+static inline int bcm47xx_nvram_gpio_pin(const char *name)
+{
+ return -ENOTSUPP;
+};
+
+static inline char *bcm47xx_nvram_get_contents(size_t *val_len)
+{
+ return NULL;
+};
+
+static inline void bcm47xx_nvram_release_contents(char *nvram)
+{
+};
+#endif
+
+#endif /* __BCM47XX_NVRAM_H */

21
target/linux/generic/pending-3.18/030-nl80211-Allow-set-network-namespace-by-fd.patch
View File

@ -1,21 +0,0 @@
From: Vadim Kochan <vadim4j@gmail.com>
Date: Mon, 12 Jan 2015 16:34:05 +0200
Subject: [PATCH] nl80211: Allow set network namespace by fd
Added new NL80211_ATTR_NETNS_FD which allows to
set namespace via nl80211 by fd.
Signed-off-by: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
---
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -361,6 +361,7 @@ struct net *get_net_ns_by_fd(int fd)
return ERR_PTR(-EINVAL);
}
#endif
+EXPORT_SYMBOL_GPL(get_net_ns_by_fd);
struct net *get_net_ns_by_pid(pid_t pid)
{

49
target/linux/generic/pending-3.18/031-bcma-from-4.5.patch
View File

@ -1,49 +0,0 @@
--- a/drivers/bcma/main.c
+++ b/drivers/bcma/main.c
@@ -637,11 +637,36 @@ static int bcma_device_uevent(struct dev
core->id.rev, core->id.class);
}
-static int __init bcma_modinit(void)
+static unsigned int bcma_bus_registered;
+
+/*
+ * If built-in, bus has to be registered early, before any driver calls
+ * bcma_driver_register.
+ * Otherwise registering driver would trigger BUG in driver_register.
+ */
+static int __init bcma_init_bus_register(void)
{
int err;
+ if (bcma_bus_registered)
+ return 0;
+
err = bus_register(&bcma_bus_type);
+ if (!err)
+ bcma_bus_registered = 1;
+
+ return err;
+}
+#ifndef MODULE
+fs_initcall(bcma_init_bus_register);
+#endif
+
+/* Main initialization has to be done with SPI/mtd/NAND/SPROM available */
+static int __init bcma_modinit(void)
+{
+ int err;
+
+ err = bcma_init_bus_register();
if (err)
return err;
@@ -660,7 +685,7 @@ static int __init bcma_modinit(void)
return err;
}
-fs_initcall(bcma_modinit);
+module_init(bcma_modinit);
static void __exit bcma_modexit(void)
{

716
target/linux/generic/pending-3.18/032-bcma-from-4.6.patch
View File

@ -1,716 +0,0 @@
--- a/drivers/bcma/driver_chipcommon.c
+++ b/drivers/bcma/driver_chipcommon.c
@@ -15,6 +15,8 @@
#include <linux/platform_device.h>
#include <linux/bcma/bcma.h>
+static void bcma_chipco_serial_init(struct bcma_drv_cc *cc);
+
static inline u32 bcma_cc_write32_masked(struct bcma_drv_cc *cc, u16 offset,
u32 mask, u32 value)
{
@@ -113,8 +115,37 @@ int bcma_chipco_watchdog_register(struct
return 0;
}
+static void bcma_core_chipcommon_flash_detect(struct bcma_drv_cc *cc)
+{
+ struct bcma_bus *bus = cc->core->bus;
+
+ switch (cc->capabilities & BCMA_CC_CAP_FLASHT) {
+ case BCMA_CC_FLASHT_STSER:
+ case BCMA_CC_FLASHT_ATSER:
+ bcma_debug(bus, "Found serial flash\n");
+ bcma_sflash_init(cc);
+ break;
+ case BCMA_CC_FLASHT_PARA:
+ bcma_debug(bus, "Found parallel flash\n");
+ bcma_pflash_init(cc);
+ break;
+ default:
+ bcma_err(bus, "Flash type not supported\n");
+ }
+
+ if (cc->core->id.rev == 38 ||
+ bus->chipinfo.id == BCMA_CHIP_ID_BCM4706) {
+ if (cc->capabilities & BCMA_CC_CAP_NFLASH) {
+ bcma_debug(bus, "Found NAND flash\n");
+ bcma_nflash_init(cc);
+ }
+ }
+}
+
void bcma_core_chipcommon_early_init(struct bcma_drv_cc *cc)
{
+ struct bcma_bus *bus = cc->core->bus;
+
if (cc->early_setup_done)
return;
@@ -129,6 +160,12 @@ void bcma_core_chipcommon_early_init(str
if (cc->capabilities & BCMA_CC_CAP_PMU)
bcma_pmu_early_init(cc);
+ if (IS_BUILTIN(CONFIG_BCM47XX) && bus->hosttype == BCMA_HOSTTYPE_SOC)
+ bcma_chipco_serial_init(cc);
+
+ if (bus->hosttype == BCMA_HOSTTYPE_SOC)
+ bcma_core_chipcommon_flash_detect(cc);
+
cc->early_setup_done = true;
}
@@ -185,11 +222,12 @@ u32 bcma_chipco_watchdog_timer_set(struc
ticks = 2;
else if (ticks > maxt)
ticks = maxt;
- bcma_cc_write32(cc, BCMA_CC_PMU_WATCHDOG, ticks);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_WATCHDOG, ticks);
} else {
struct bcma_bus *bus = cc->core->bus;
if (bus->chipinfo.id != BCMA_CHIP_ID_BCM4707 &&
+ bus->chipinfo.id != BCMA_CHIP_ID_BCM47094 &&
bus->chipinfo.id != BCMA_CHIP_ID_BCM53018)
bcma_core_set_clockmode(cc->core,
ticks ? BCMA_CLKMODE_FAST : BCMA_CLKMODE_DYNAMIC);
@@ -314,9 +352,9 @@ u32 bcma_chipco_gpio_pulldown(struct bcm
return res;
}
-#ifdef CONFIG_BCMA_DRIVER_MIPS
-void bcma_chipco_serial_init(struct bcma_drv_cc *cc)
+static void bcma_chipco_serial_init(struct bcma_drv_cc *cc)
{
+#if IS_BUILTIN(CONFIG_BCM47XX)
unsigned int irq;
u32 baud_base;
u32 i;
@@ -358,5 +396,5 @@ void bcma_chipco_serial_init(struct bcma
ports[i].baud_base = baud_base;
ports[i].reg_shift = 0;
}
+#endif /* CONFIG_BCM47XX */
}
-#endif /* CONFIG_BCMA_DRIVER_MIPS */
--- a/drivers/bcma/driver_chipcommon_pmu.c
+++ b/drivers/bcma/driver_chipcommon_pmu.c
@@ -15,44 +15,44 @@
u32 bcma_chipco_pll_read(struct bcma_drv_cc *cc, u32 offset)
{
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR, offset);
- bcma_cc_read32(cc, BCMA_CC_PLLCTL_ADDR);
- return bcma_cc_read32(cc, BCMA_CC_PLLCTL_DATA);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR, offset);
+ bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_ADDR);
+ return bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_DATA);
}
EXPORT_SYMBOL_GPL(bcma_chipco_pll_read);
void bcma_chipco_pll_write(struct bcma_drv_cc *cc, u32 offset, u32 value)
{
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR, offset);
- bcma_cc_read32(cc, BCMA_CC_PLLCTL_ADDR);
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_DATA, value);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR, offset);
+ bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_ADDR);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_DATA, value);
}
EXPORT_SYMBOL_GPL(bcma_chipco_pll_write);
void bcma_chipco_pll_maskset(struct bcma_drv_cc *cc, u32 offset, u32 mask,
u32 set)
{
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR, offset);
- bcma_cc_read32(cc, BCMA_CC_PLLCTL_ADDR);
- bcma_cc_maskset32(cc, BCMA_CC_PLLCTL_DATA, mask, set);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR, offset);
+ bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_ADDR);
+ bcma_pmu_maskset32(cc, BCMA_CC_PMU_PLLCTL_DATA, mask, set);
}
EXPORT_SYMBOL_GPL(bcma_chipco_pll_maskset);
void bcma_chipco_chipctl_maskset(struct bcma_drv_cc *cc,
u32 offset, u32 mask, u32 set)
{
- bcma_cc_write32(cc, BCMA_CC_CHIPCTL_ADDR, offset);
- bcma_cc_read32(cc, BCMA_CC_CHIPCTL_ADDR);
- bcma_cc_maskset32(cc, BCMA_CC_CHIPCTL_DATA, mask, set);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_CHIPCTL_ADDR, offset);
+ bcma_pmu_read32(cc, BCMA_CC_PMU_CHIPCTL_ADDR);
+ bcma_pmu_maskset32(cc, BCMA_CC_PMU_CHIPCTL_DATA, mask, set);
}
EXPORT_SYMBOL_GPL(bcma_chipco_chipctl_maskset);
void bcma_chipco_regctl_maskset(struct bcma_drv_cc *cc, u32 offset, u32 mask,
u32 set)
{
- bcma_cc_write32(cc, BCMA_CC_REGCTL_ADDR, offset);
- bcma_cc_read32(cc, BCMA_CC_REGCTL_ADDR);
- bcma_cc_maskset32(cc, BCMA_CC_REGCTL_DATA, mask, set);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_REGCTL_ADDR, offset);
+ bcma_pmu_read32(cc, BCMA_CC_PMU_REGCTL_ADDR);
+ bcma_pmu_maskset32(cc, BCMA_CC_PMU_REGCTL_DATA, mask, set);
}
EXPORT_SYMBOL_GPL(bcma_chipco_regctl_maskset);
@@ -60,18 +60,18 @@ static u32 bcma_pmu_xtalfreq(struct bcma
{
u32 ilp_ctl, alp_hz;
- if (!(bcma_cc_read32(cc, BCMA_CC_PMU_STAT) &
+ if (!(bcma_pmu_read32(cc, BCMA_CC_PMU_STAT) &
BCMA_CC_PMU_STAT_EXT_LPO_AVAIL))
return 0;
- bcma_cc_write32(cc, BCMA_CC_PMU_XTAL_FREQ,
- BIT(BCMA_CC_PMU_XTAL_FREQ_MEASURE_SHIFT));
+ bcma_pmu_write32(cc, BCMA_CC_PMU_XTAL_FREQ,
+ BIT(BCMA_CC_PMU_XTAL_FREQ_MEASURE_SHIFT));
usleep_range(1000, 2000);
- ilp_ctl = bcma_cc_read32(cc, BCMA_CC_PMU_XTAL_FREQ);
+ ilp_ctl = bcma_pmu_read32(cc, BCMA_CC_PMU_XTAL_FREQ);
ilp_ctl &= BCMA_CC_PMU_XTAL_FREQ_ILPCTL_MASK;
- bcma_cc_write32(cc, BCMA_CC_PMU_XTAL_FREQ, 0);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_XTAL_FREQ, 0);
alp_hz = ilp_ctl * 32768 / 4;
return (alp_hz + 50000) / 100000 * 100;
@@ -127,8 +127,8 @@ static void bcma_pmu2_pll_init0(struct b
mask = (u32)~(BCMA_RES_4314_HT_AVAIL |
BCMA_RES_4314_MACPHY_CLK_AVAIL);
- bcma_cc_mask32(cc, BCMA_CC_PMU_MINRES_MSK, mask);
- bcma_cc_mask32(cc, BCMA_CC_PMU_MAXRES_MSK, mask);
+ bcma_pmu_mask32(cc, BCMA_CC_PMU_MINRES_MSK, mask);
+ bcma_pmu_mask32(cc, BCMA_CC_PMU_MAXRES_MSK, mask);
bcma_wait_value(cc->core, BCMA_CLKCTLST,
BCMA_CLKCTLST_HAVEHT, 0, 20000);
break;
@@ -140,7 +140,7 @@ static void bcma_pmu2_pll_init0(struct b
/* Flush */
if (cc->pmu.rev >= 2)
- bcma_cc_set32(cc, BCMA_CC_PMU_CTL, BCMA_CC_PMU_CTL_PLL_UPD);
+ bcma_pmu_set32(cc, BCMA_CC_PMU_CTL, BCMA_CC_PMU_CTL_PLL_UPD);
/* TODO: Do we need to update OTP? */
}
@@ -195,9 +195,9 @@ static void bcma_pmu_resources_init(stru
/* Set the resource masks. */
if (min_msk)
- bcma_cc_write32(cc, BCMA_CC_PMU_MINRES_MSK, min_msk);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_MINRES_MSK, min_msk);
if (max_msk)
- bcma_cc_write32(cc, BCMA_CC_PMU_MAXRES_MSK, max_msk);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_MAXRES_MSK, max_msk);
/*
* Add some delay; allow resources to come up and settle.
@@ -269,23 +269,33 @@ static void bcma_pmu_workarounds(struct
void bcma_pmu_early_init(struct bcma_drv_cc *cc)
{
+ struct bcma_bus *bus = cc->core->bus;
u32 pmucap;
- pmucap = bcma_cc_read32(cc, BCMA_CC_PMU_CAP);
+ if (cc->core->id.rev >= 35 &&
+ cc->capabilities_ext & BCMA_CC_CAP_EXT_AOB_PRESENT) {
+ cc->pmu.core = bcma_find_core(bus, BCMA_CORE_PMU);
+ if (!cc->pmu.core)
+ bcma_warn(bus, "Couldn't find expected PMU core");
+ }
+ if (!cc->pmu.core)
+ cc->pmu.core = cc->core;
+
+ pmucap = bcma_pmu_read32(cc, BCMA_CC_PMU_CAP);
cc->pmu.rev = (pmucap & BCMA_CC_PMU_CAP_REVISION);
- bcma_debug(cc->core->bus, "Found rev %u PMU (capabilities 0x%08X)\n",
- cc->pmu.rev, pmucap);
+ bcma_debug(bus, "Found rev %u PMU (capabilities 0x%08X)\n", cc->pmu.rev,
+ pmucap);
}
void bcma_pmu_init(struct bcma_drv_cc *cc)
{
if (cc->pmu.rev == 1)
- bcma_cc_mask32(cc, BCMA_CC_PMU_CTL,
- ~BCMA_CC_PMU_CTL_NOILPONW);
+ bcma_pmu_mask32(cc, BCMA_CC_PMU_CTL,
+ ~BCMA_CC_PMU_CTL_NOILPONW);
else
- bcma_cc_set32(cc, BCMA_CC_PMU_CTL,
- BCMA_CC_PMU_CTL_NOILPONW);
+ bcma_pmu_set32(cc, BCMA_CC_PMU_CTL,
+ BCMA_CC_PMU_CTL_NOILPONW);
bcma_pmu_pll_init(cc);
bcma_pmu_resources_init(cc);
@@ -472,8 +482,8 @@ u32 bcma_pmu_get_cpu_clock(struct bcma_d
static void bcma_pmu_spuravoid_pll_write(struct bcma_drv_cc *cc, u32 offset,
u32 value)
{
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR, offset);
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_DATA, value);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR, offset);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_DATA, value);
}
void bcma_pmu_spuravoid_pllupdate(struct bcma_drv_cc *cc, int spuravoid)
@@ -497,20 +507,20 @@ void bcma_pmu_spuravoid_pllupdate(struct
bus->chipinfo.id == BCMA_CHIP_ID_BCM53572) ? 6 : 0;
/* RMW only the P1 divider */
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR,
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR,
BCMA_CC_PMU_PLL_CTL0 + phypll_offset);
- tmp = bcma_cc_read32(cc, BCMA_CC_PLLCTL_DATA);
+ tmp = bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_DATA);
tmp &= (~(BCMA_CC_PMU1_PLL0_PC0_P1DIV_MASK));
tmp |= (bcm5357_bcm43236_p1div[spuravoid] << BCMA_CC_PMU1_PLL0_PC0_P1DIV_SHIFT);
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_DATA, tmp);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_DATA, tmp);
/* RMW only the int feedback divider */
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_ADDR,
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_ADDR,
BCMA_CC_PMU_PLL_CTL2 + phypll_offset);
- tmp = bcma_cc_read32(cc, BCMA_CC_PLLCTL_DATA);
+ tmp = bcma_pmu_read32(cc, BCMA_CC_PMU_PLLCTL_DATA);
tmp &= ~(BCMA_CC_PMU1_PLL0_PC2_NDIV_INT_MASK);
tmp |= (bcm5357_bcm43236_ndiv[spuravoid]) << BCMA_CC_PMU1_PLL0_PC2_NDIV_INT_SHIFT;
- bcma_cc_write32(cc, BCMA_CC_PLLCTL_DATA, tmp);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_PLLCTL_DATA, tmp);
tmp = BCMA_CC_PMU_CTL_PLL_UPD;
break;
@@ -646,7 +656,7 @@ void bcma_pmu_spuravoid_pllupdate(struct
break;
}
- tmp |= bcma_cc_read32(cc, BCMA_CC_PMU_CTL);
- bcma_cc_write32(cc, BCMA_CC_PMU_CTL, tmp);
+ tmp |= bcma_pmu_read32(cc, BCMA_CC_PMU_CTL);
+ bcma_pmu_write32(cc, BCMA_CC_PMU_CTL, tmp);
}
EXPORT_SYMBOL_GPL(bcma_pmu_spuravoid_pllupdate);
--- a/drivers/bcma/driver_chipcommon_sflash.c
+++ b/drivers/bcma/driver_chipcommon_sflash.c
@@ -38,6 +38,7 @@ static const struct bcma_sflash_tbl_e bc
{ "M25P32", 0x15, 0x10000, 64, },
{ "M25P64", 0x16, 0x10000, 128, },
{ "M25FL128", 0x17, 0x10000, 256, },
+ { "MX25L25635F", 0x18, 0x10000, 512, },
{ NULL },
};
--- a/drivers/bcma/scan.c
+++ b/drivers/bcma/scan.c
@@ -98,6 +98,9 @@ static const struct bcma_device_id_name
{ BCMA_CORE_SHIM, "SHIM" },
{ BCMA_CORE_PCIE2, "PCIe Gen2" },
{ BCMA_CORE_ARM_CR4, "ARM CR4" },
+ { BCMA_CORE_GCI, "GCI" },
+ { BCMA_CORE_CMEM, "CNDS DDR2/3 memory controller" },
+ { BCMA_CORE_ARM_CA7, "ARM CA7" },
{ BCMA_CORE_DEFAULT, "Default" },
};
@@ -315,6 +318,8 @@ static int bcma_get_next_core(struct bcm
switch (core->id.id) {
case BCMA_CORE_4706_MAC_GBIT_COMMON:
case BCMA_CORE_NS_CHIPCOMMON_B:
+ case BCMA_CORE_PMU:
+ case BCMA_CORE_GCI:
/* Not used yet: case BCMA_CORE_OOB_ROUTER: */
break;
default:
--- a/drivers/net/wireless/b43/main.c
+++ b/drivers/net/wireless/b43/main.c
@@ -1215,10 +1215,10 @@ void b43_wireless_core_phy_pll_reset(str
case B43_BUS_BCMA:
bcma_cc = &dev->dev->bdev->bus->drv_cc;
- bcma_cc_write32(bcma_cc, BCMA_CC_CHIPCTL_ADDR, 0);
- bcma_cc_mask32(bcma_cc, BCMA_CC_CHIPCTL_DATA, ~0x4);
- bcma_cc_set32(bcma_cc, BCMA_CC_CHIPCTL_DATA, 0x4);
- bcma_cc_mask32(bcma_cc, BCMA_CC_CHIPCTL_DATA, ~0x4);
+ bcma_cc_write32(bcma_cc, BCMA_CC_PMU_CHIPCTL_ADDR, 0);
+ bcma_cc_mask32(bcma_cc, BCMA_CC_PMU_CHIPCTL_DATA, ~0x4);
+ bcma_cc_set32(bcma_cc, BCMA_CC_PMU_CHIPCTL_DATA, 0x4);
+ bcma_cc_mask32(bcma_cc, BCMA_CC_PMU_CHIPCTL_DATA, ~0x4);
break;
#endif
#ifdef CONFIG_B43_SSB
--- a/include/linux/bcma/bcma.h
+++ b/include/linux/bcma/bcma.h
@@ -151,6 +151,8 @@ struct bcma_host_ops {
#define BCMA_CORE_PCIE2 0x83C /* PCI Express Gen2 */
#define BCMA_CORE_USB30_DEV 0x83D
#define BCMA_CORE_ARM_CR4 0x83E
+#define BCMA_CORE_GCI 0x840
+#define BCMA_CORE_CMEM 0x846 /* CNDS DDR2/3 memory controller */
#define BCMA_CORE_ARM_CA7 0x847
#define BCMA_CORE_SYS_MEM 0x849
#define BCMA_CORE_DEFAULT 0xFFF
@@ -200,6 +202,7 @@ struct bcma_host_ops {
#define BCMA_PKG_ID_BCM4707 1
#define BCMA_PKG_ID_BCM4708 2
#define BCMA_PKG_ID_BCM4709 0
+#define BCMA_CHIP_ID_BCM47094 53030
#define BCMA_CHIP_ID_BCM53018 53018
/* Board types (on PCI usually equals to the subsystem dev id) */
--- a/include/linux/bcma/bcma_driver_chipcommon.h
+++ b/include/linux/bcma/bcma_driver_chipcommon.h
@@ -217,6 +217,11 @@
#define BCMA_CC_CLKDIV_JTAG_SHIFT 8
#define BCMA_CC_CLKDIV_UART 0x000000FF
#define BCMA_CC_CAP_EXT 0x00AC /* Capabilities */
+#define BCMA_CC_CAP_EXT_SECI_PRESENT 0x00000001
+#define BCMA_CC_CAP_EXT_GSIO_PRESENT 0x00000002
+#define BCMA_CC_CAP_EXT_GCI_PRESENT 0x00000004
+#define BCMA_CC_CAP_EXT_SECI_PUART_PRESENT 0x00000008 /* UART present */
+#define BCMA_CC_CAP_EXT_AOB_PRESENT 0x00000040
#define BCMA_CC_PLLONDELAY 0x00B0 /* Rev >= 4 only */
#define BCMA_CC_FREFSELDELAY 0x00B4 /* Rev >= 4 only */
#define BCMA_CC_SLOWCLKCTL 0x00B8 /* 6 <= Rev <= 9 only */
@@ -351,12 +356,12 @@
#define BCMA_CC_PMU_RES_REQTS 0x0640 /* PMU res req timer sel */
#define BCMA_CC_PMU_RES_REQT 0x0644 /* PMU res req timer */
#define BCMA_CC_PMU_RES_REQM 0x0648 /* PMU res req mask */
-#define BCMA_CC_CHIPCTL_ADDR 0x0650
-#define BCMA_CC_CHIPCTL_DATA 0x0654
-#define BCMA_CC_REGCTL_ADDR 0x0658
-#define BCMA_CC_REGCTL_DATA 0x065C
-#define BCMA_CC_PLLCTL_ADDR 0x0660
-#define BCMA_CC_PLLCTL_DATA 0x0664
+#define BCMA_CC_PMU_CHIPCTL_ADDR 0x0650
+#define BCMA_CC_PMU_CHIPCTL_DATA 0x0654
+#define BCMA_CC_PMU_REGCTL_ADDR 0x0658
+#define BCMA_CC_PMU_REGCTL_DATA 0x065C
+#define BCMA_CC_PMU_PLLCTL_ADDR 0x0660
+#define BCMA_CC_PMU_PLLCTL_DATA 0x0664
#define BCMA_CC_PMU_STRAPOPT 0x0668 /* (corerev >= 28) */
#define BCMA_CC_PMU_XTAL_FREQ 0x066C /* (pmurev >= 10) */
#define BCMA_CC_PMU_XTAL_FREQ_ILPCTL_MASK 0x00001FFF
@@ -566,17 +571,16 @@
* Check availability with ((struct bcma_chipcommon)->capabilities & BCMA_CC_CAP_PMU)
*/
struct bcma_chipcommon_pmu {
+ struct bcma_device *core; /* Can be separated core or just ChipCommon one */
u8 rev; /* PMU revision */
u32 crystalfreq; /* The active crystal frequency (in kHz) */
};
-#ifdef CONFIG_BCMA_DRIVER_MIPS
+#ifdef CONFIG_BCMA_PFLASH
struct bcma_pflash {
bool present;
- u8 buswidth;
- u32 window;
- u32 window_size;
};
+#endif
#ifdef CONFIG_BCMA_SFLASH
struct bcma_sflash {
@@ -602,6 +606,7 @@ struct bcma_nflash {
};
#endif
+#ifdef CONFIG_BCMA_DRIVER_MIPS
struct bcma_serial_port {
void *regs;
unsigned long clockspeed;
@@ -621,8 +626,9 @@ struct bcma_drv_cc {
/* Fast Powerup Delay constant */
u16 fast_pwrup_delay;
struct bcma_chipcommon_pmu pmu;
-#ifdef CONFIG_BCMA_DRIVER_MIPS
+#ifdef CONFIG_BCMA_PFLASH
struct bcma_pflash pflash;
+#endif
#ifdef CONFIG_BCMA_SFLASH
struct bcma_sflash sflash;
#endif
@@ -630,6 +636,7 @@ struct bcma_drv_cc {
struct bcma_nflash nflash;
#endif
+#ifdef CONFIG_BCMA_DRIVER_MIPS
int nr_serial_ports;
struct bcma_serial_port serial_ports[4];
#endif /* CONFIG_BCMA_DRIVER_MIPS */
@@ -663,6 +670,19 @@ struct bcma_drv_cc_b {
#define bcma_cc_maskset32(cc, offset, mask, set) \
bcma_cc_write32(cc, offset, (bcma_cc_read32(cc, offset) & (mask)) | (set))
+/* PMU registers access */
+#define bcma_pmu_read32(cc, offset) \
+ bcma_read32((cc)->pmu.core, offset)
+#define bcma_pmu_write32(cc, offset, val) \
+ bcma_write32((cc)->pmu.core, offset, val)
+
+#define bcma_pmu_mask32(cc, offset, mask) \
+ bcma_pmu_write32(cc, offset, bcma_pmu_read32(cc, offset) & (mask))
+#define bcma_pmu_set32(cc, offset, set) \
+ bcma_pmu_write32(cc, offset, bcma_pmu_read32(cc, offset) | (set))
+#define bcma_pmu_maskset32(cc, offset, mask, set) \
+ bcma_pmu_write32(cc, offset, (bcma_pmu_read32(cc, offset) & (mask)) | (set))
+
extern u32 bcma_chipco_watchdog_timer_set(struct bcma_drv_cc *cc, u32 ticks);
extern u32 bcma_chipco_get_alp_clock(struct bcma_drv_cc *cc);
--- a/drivers/bcma/bcma_private.h
+++ b/drivers/bcma/bcma_private.h
@@ -44,10 +44,6 @@ int bcma_sprom_get(struct bcma_bus *bus)
void bcma_core_chipcommon_early_init(struct bcma_drv_cc *cc);
void bcma_core_chipcommon_init(struct bcma_drv_cc *cc);
void bcma_chipco_bcm4331_ext_pa_lines_ctl(struct bcma_drv_cc *cc, bool enable);
-#ifdef CONFIG_BCMA_DRIVER_MIPS
-void bcma_chipco_serial_init(struct bcma_drv_cc *cc);
-extern struct platform_device bcma_pflash_dev;
-#endif /* CONFIG_BCMA_DRIVER_MIPS */
/* driver_chipcommon_b.c */
int bcma_core_chipcommon_b_init(struct bcma_drv_cc_b *ccb);
@@ -59,6 +55,21 @@ void bcma_pmu_init(struct bcma_drv_cc *c
u32 bcma_pmu_get_alp_clock(struct bcma_drv_cc *cc);
u32 bcma_pmu_get_cpu_clock(struct bcma_drv_cc *cc);
+/**************************************************
+ * driver_chipcommon_sflash.c
+ **************************************************/
+
+#ifdef CONFIG_BCMA_PFLASH
+extern struct platform_device bcma_pflash_dev;
+int bcma_pflash_init(struct bcma_drv_cc *cc);
+#else
+static inline int bcma_pflash_init(struct bcma_drv_cc *cc)
+{
+ bcma_err(cc->core->bus, "Parallel flash not supported\n");
+ return 0;
+}
+#endif /* CONFIG_BCMA_PFLASH */
+
#ifdef CONFIG_BCMA_SFLASH
/* driver_chipcommon_sflash.c */
int bcma_sflash_init(struct bcma_drv_cc *cc);
--- a/drivers/bcma/driver_gpio.c
+++ b/drivers/bcma/driver_gpio.c
@@ -229,6 +229,7 @@ int bcma_gpio_init(struct bcma_drv_cc *c
case BCMA_CHIP_ID_BCM4707:
case BCMA_CHIP_ID_BCM5357:
case BCMA_CHIP_ID_BCM53572:
+ case BCMA_CHIP_ID_BCM47094:
chip->ngpio = 32;
break;
default:
--- a/drivers/bcma/driver_mips.c
+++ b/drivers/bcma/driver_mips.c
@@ -14,8 +14,6 @@
#include <linux/bcma/bcma.h>
-#include <linux/mtd/physmap.h>
-#include <linux/platform_device.h>
#include <linux/serial.h>
#include <linux/serial_core.h>
#include <linux/serial_reg.h>
@@ -29,26 +27,6 @@ enum bcma_boot_dev {
BCMA_BOOT_DEV_NAND,
};
-static const char * const part_probes[] = { "bcm47xxpart", NULL };
-
-static struct physmap_flash_data bcma_pflash_data = {
- .part_probe_types = part_probes,
-};
-
-static struct resource bcma_pflash_resource = {
- .name = "bcma_pflash",
- .flags = IORESOURCE_MEM,
-};
-
-struct platform_device bcma_pflash_dev = {
- .name = "physmap-flash",
- .dev = {
- .platform_data = &bcma_pflash_data,
- },
- .resource = &bcma_pflash_resource,
- .num_resources = 1,
-};
-
/* The 47162a0 hangs when reading MIPS DMP registers registers */
static inline bool bcma_core_mips_bcm47162a0_quirk(struct bcma_device *dev)
{
@@ -269,48 +247,11 @@ static enum bcma_boot_dev bcma_boot_dev(
return BCMA_BOOT_DEV_SERIAL;
}
-static void bcma_core_mips_flash_detect(struct bcma_drv_mips *mcore)
+static void bcma_core_mips_nvram_init(struct bcma_drv_mips *mcore)
{
struct bcma_bus *bus = mcore->core->bus;
- struct bcma_drv_cc *cc = &bus->drv_cc;
- struct bcma_pflash *pflash = &cc->pflash;
enum bcma_boot_dev boot_dev;
- switch (cc->capabilities & BCMA_CC_CAP_FLASHT) {
- case BCMA_CC_FLASHT_STSER:
- case BCMA_CC_FLASHT_ATSER:
- bcma_debug(bus, "Found serial flash\n");
- bcma_sflash_init(cc);
- break;
- case BCMA_CC_FLASHT_PARA:
- bcma_debug(bus, "Found parallel flash\n");
- pflash->present = true;
- pflash->window = BCMA_SOC_FLASH2;
- pflash->window_size = BCMA_SOC_FLASH2_SZ;
-
- if ((bcma_read32(cc->core, BCMA_CC_FLASH_CFG) &
- BCMA_CC_FLASH_CFG_DS) == 0)
- pflash->buswidth = 1;
- else
- pflash->buswidth = 2;
-
- bcma_pflash_data.width = pflash->buswidth;
- bcma_pflash_resource.start = pflash->window;
- bcma_pflash_resource.end = pflash->window + pflash->window_size;
-
- break;
- default:
- bcma_err(bus, "Flash type not supported\n");
- }
-
- if (cc->core->id.rev == 38 ||
- bus->chipinfo.id == BCMA_CHIP_ID_BCM4706) {
- if (cc->capabilities & BCMA_CC_CAP_NFLASH) {
- bcma_debug(bus, "Found NAND flash\n");
- bcma_nflash_init(cc);
- }
- }
-
/* Determine flash type this SoC boots from */
boot_dev = bcma_boot_dev(bus);
switch (boot_dev) {
@@ -328,13 +269,10 @@ static void bcma_core_mips_flash_detect(
void bcma_core_mips_early_init(struct bcma_drv_mips *mcore)
{
- struct bcma_bus *bus = mcore->core->bus;
-
if (mcore->early_setup_done)
return;
- bcma_chipco_serial_init(&bus->drv_cc);
- bcma_core_mips_flash_detect(mcore);
+ bcma_core_mips_nvram_init(mcore);
mcore->early_setup_done = true;
}
--- a/drivers/bcma/host_pci.c
+++ b/drivers/bcma/host_pci.c
@@ -294,7 +294,7 @@ static const struct pci_device_id bcma_p
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4358) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4359) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4360) },
- { PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4365) },
+ { PCI_DEVICE_SUB(PCI_VENDOR_ID_BROADCOM, 0x4365, PCI_VENDOR_ID_DELL, 0x0016) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43a0) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43a9) },
{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x43aa) },
--- a/drivers/bcma/Kconfig
+++ b/drivers/bcma/Kconfig
@@ -70,6 +70,11 @@ config BCMA_DRIVER_MIPS
If unsure, say N
+config BCMA_PFLASH
+ bool
+ depends on BCMA_DRIVER_MIPS
+ default y
+
config BCMA_SFLASH
bool
depends on BCMA_DRIVER_MIPS
--- a/drivers/bcma/Makefile
+++ b/drivers/bcma/Makefile
@@ -1,6 +1,7 @@
bcma-y += main.o scan.o core.o sprom.o
bcma-y += driver_chipcommon.o driver_chipcommon_pmu.o
bcma-y += driver_chipcommon_b.o
+bcma-$(CONFIG_BCMA_PFLASH) += driver_chipcommon_pflash.o
bcma-$(CONFIG_BCMA_SFLASH) += driver_chipcommon_sflash.o
bcma-$(CONFIG_BCMA_NFLASH) += driver_chipcommon_nflash.o
bcma-$(CONFIG_BCMA_DRIVER_PCI) += driver_pci.o
--- /dev/null
+++ b/drivers/bcma/driver_chipcommon_pflash.c
@@ -0,0 +1,49 @@
+/*
+ * Broadcom specific AMBA
+ * ChipCommon parallel flash
+ *
+ * Licensed under the GNU/GPL. See COPYING for details.
+ */
+
+#include "bcma_private.h"
+
+#include <linux/bcma/bcma.h>
+#include <linux/mtd/physmap.h>
+#include <linux/platform_device.h>
+
+static const char * const part_probes[] = { "bcm47xxpart", NULL };
+
+static struct physmap_flash_data bcma_pflash_data = {
+ .part_probe_types = part_probes,
+};
+
+static struct resource bcma_pflash_resource = {
+ .name = "bcma_pflash",
+ .flags = IORESOURCE_MEM,
+};
+
+struct platform_device bcma_pflash_dev = {
+ .name = "physmap-flash",
+ .dev = {
+ .platform_data = &bcma_pflash_data,
+ },
+ .resource = &bcma_pflash_resource,
+ .num_resources = 1,
+};
+
+int bcma_pflash_init(struct bcma_drv_cc *cc)
+{
+ struct bcma_pflash *pflash = &cc->pflash;
+
+ pflash->present = true;
+
+ if (!(bcma_read32(cc->core, BCMA_CC_FLASH_CFG) & BCMA_CC_FLASH_CFG_DS))
+ bcma_pflash_data.width = 1;
+ else
+ bcma_pflash_data.width = 2;
+
+ bcma_pflash_resource.start = BCMA_SOC_FLASH2;
+ bcma_pflash_resource.end = BCMA_SOC_FLASH2 + BCMA_SOC_FLASH2_SZ;
+
+ return 0;
+}
--- a/drivers/bcma/main.c
+++ b/drivers/bcma/main.c
@@ -325,7 +325,7 @@ static int bcma_register_devices(struct
bcma_register_core(bus, core);
}
-#ifdef CONFIG_BCMA_DRIVER_MIPS
+#ifdef CONFIG_BCMA_PFLASH
if (bus->drv_cc.pflash.present) {
err = platform_device_register(&bcma_pflash_dev);
if (err)

50
target/linux/generic/pending-3.18/040-mtd-bcm47xxpart-backports-from-3.19.patch
View File

@ -1,50 +0,0 @@
--- a/drivers/mtd/bcm47xxpart.c
+++ b/drivers/mtd/bcm47xxpart.c
@@ -15,8 +15,12 @@
#include <linux/mtd/mtd.h>
#include <linux/mtd/partitions.h>
-/* 10 parts were found on sflash on Netgear WNDR4500 */
-#define BCM47XXPART_MAX_PARTS 12
+/*
+ * NAND flash on Netgear R6250 was verified to contain 15 partitions.
+ * This will result in allocating too big array for some old devices, but the
+ * memory will be freed soon anyway (see mtd_device_parse_register).
+ */
+#define BCM47XXPART_MAX_PARTS 20
/*
* Amount of bytes we read when analyzing each block of flash memory.
@@ -168,18 +172,26 @@ static int bcm47xxpart_parse(struct mtd_
i++;
}
- bcm47xxpart_add_part(&parts[curr_part++], "linux",
- offset + trx->offset[i], 0);
- i++;
+ if (trx->offset[i]) {
+ bcm47xxpart_add_part(&parts[curr_part++],
+ "linux",
+ offset + trx->offset[i],
+ 0);
+ i++;
+ }
/*
* Pure rootfs size is known and can be calculated as:
* trx->length - trx->offset[i]. We don't fill it as
* we want to have jffs2 (overlay) in the same mtd.
*/
- bcm47xxpart_add_part(&parts[curr_part++], "rootfs",
- offset + trx->offset[i], 0);
- i++;
+ if (trx->offset[i]) {
+ bcm47xxpart_add_part(&parts[curr_part++],
+ "rootfs",
+ offset + trx->offset[i],
+ 0);
+ i++;
+ }
last_trx_part = curr_part - 1;

95
target/linux/generic/pending-3.18/041-mtd-bcm47xxpart-backports-from-3.20.patch
View File

@ -1,95 +0,0 @@
--- a/drivers/mtd/bcm47xxpart.c
+++ b/drivers/mtd/bcm47xxpart.c
@@ -15,6 +15,8 @@
#include <linux/mtd/mtd.h>
#include <linux/mtd/partitions.h>
+#include <uapi/linux/magic.h>
+
/*
* NAND flash on Netgear R6250 was verified to contain 15 partitions.
* This will result in allocating too big array for some old devices, but the
@@ -39,7 +41,8 @@
#define ML_MAGIC1 0x39685a42
#define ML_MAGIC2 0x26594131
#define TRX_MAGIC 0x30524448
-#define SQSH_MAGIC 0x71736873 /* shsq */
+#define SHSQ_MAGIC 0x71736873 /* shsq (weird ZTE H218N endianness) */
+#define UBI_EC_MAGIC 0x23494255 /* UBI# */
struct trx_header {
uint32_t magic;
@@ -50,7 +53,7 @@ struct trx_header {
uint32_t offset[3];
} __packed;
-static void bcm47xxpart_add_part(struct mtd_partition *part, char *name,
+static void bcm47xxpart_add_part(struct mtd_partition *part, const char *name,
u64 offset, uint32_t mask_flags)
{
part->name = name;
@@ -58,6 +61,26 @@ static void bcm47xxpart_add_part(struct
part->mask_flags = mask_flags;
}
+static const char *bcm47xxpart_trx_data_part_name(struct mtd_info *master,
+ size_t offset)
+{
+ uint32_t buf;
+ size_t bytes_read;
+
+ if (mtd_read(master, offset, sizeof(buf), &bytes_read,
+ (uint8_t *)&buf) < 0) {
+ pr_err("mtd_read error while parsing (offset: 0x%X)!\n",
+ offset);
+ goto out_default;
+ }
+
+ if (buf == UBI_EC_MAGIC)
+ return "ubi";
+
+out_default:
+ return "rootfs";
+}
+
static int bcm47xxpart_parse(struct mtd_info *master,
struct mtd_partition **pparts,
struct mtd_part_parser_data *data)
@@ -73,8 +96,12 @@ static int bcm47xxpart_parse(struct mtd_
int last_trx_part = -1;
int possible_nvram_sizes[] = { 0x8000, 0xF000, 0x10000, };
- if (blocksize <= 0x10000)
- blocksize = 0x10000;
+ /*
+ * Some really old flashes (like AT45DB*) had smaller erasesize-s, but
+ * partitions were aligned to at least 0x1000 anyway.
+ */
+ if (blocksize < 0x1000)
+ blocksize = 0x1000;
/* Alloc */
parts = kzalloc(sizeof(struct mtd_partition) * BCM47XXPART_MAX_PARTS,
@@ -186,8 +213,11 @@ static int bcm47xxpart_parse(struct mtd_
* we want to have jffs2 (overlay) in the same mtd.
*/
if (trx->offset[i]) {
+ const char *name;
+
+ name = bcm47xxpart_trx_data_part_name(master, offset + trx->offset[i]);
bcm47xxpart_add_part(&parts[curr_part++],
- "rootfs",
+ name,
offset + trx->offset[i],
0);
i++;
@@ -203,7 +233,8 @@ static int bcm47xxpart_parse(struct mtd_
}
/* Squashfs on devices not using TRX */
- if (buf[0x000 / 4] == SQSH_MAGIC) {
+ if (le32_to_cpu(buf[0x000 / 4]) == SQUASHFS_MAGIC ||
+ buf[0x000 / 4] == SHSQ_MAGIC) {
bcm47xxpart_add_part(&parts[curr_part++], "rootfs",
offset, 0);
continue;

15
target/linux/generic/pending-3.18/043-mtd_GD25Q128B_support_backport_from_3.19.patch
View File

@ -1,15 +0,0 @@
mtd: spi-nor: support for (GigaDevice) GD25Q128B
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
--- a/drivers/mtd/spi-nor/spi-nor.c
+++ b/drivers/mtd/spi-nor/spi-nor.c
@@ -510,6 +510,7 @@ static const struct spi_device_id spi_no
/* GigaDevice */
{ "gd25q32", INFO(0xc84016, 0, 64 * 1024, 64, SECT_4K) },
{ "gd25q64", INFO(0xc84017, 0, 64 * 1024, 128, SECT_4K) },
+ { "gd25q128", INFO(0xc84018, 0, 64 * 1024, 256, SECT_4K) },
/* Intel/Numonyx -- xxxs33b */
{ "160s33b", INFO(0x898911, 0, 64 * 1024, 32, 0) },

39
target/linux/generic/pending-3.18/044-backport-m25p80-jedec-probe.patch
View File

@ -1,39 +0,0 @@
--- a/drivers/mtd/devices/m25p80.c
+++ b/drivers/mtd/devices/m25p80.c
@@ -310,11 +310,21 @@ static const struct spi_device_id m25p_i
};
MODULE_DEVICE_TABLE(spi, m25p_ids);
+static const struct of_device_id m25p_of_table[] = {
+ /*
+ * Generic compatibility for SPI NOR that can be identified by the
+ * JEDEC READ ID opcode (0x9F). Use this, if possible.
+ */
+ { .compatible = "jedec,spi-nor" },
+ {}
+};
+MODULE_DEVICE_TABLE(of, m25p_of_table);
static struct spi_driver m25p80_driver = {
.driver = {
.name = "m25p80",
.owner = THIS_MODULE,
+ .of_match_table = m25p_of_table,
},
.id_table = m25p_ids,
.probe = m25p_probe,
--- a/drivers/mtd/spi-nor/spi-nor.c
+++ b/drivers/mtd/spi-nor/spi-nor.c
@@ -934,8 +934,11 @@ int spi_nor_scan(struct spi_nor *nor, co
if (ret)
return ret;
- id = spi_nor_match_id(name);
+ if (name)
+ id = spi_nor_match_id(name);
if (!id)
+ id = nor->read_id(nor);
+ if (IS_ERR_OR_NULL(id))
return -ENOENT;
info = (void *)id->driver_data;

509
target/linux/generic/pending-3.18/050-backport_netfilter_rtcache.patch
View File

@ -1,509 +0,0 @@
Subject: netfilter: conntrack: cache route for forwarded connections
... to avoid per-packet FIB lookup if possible.
The cached dst is re-used provided the input interface
is the same as that of the previous packet in the same direction.
If not, the cached dst is invalidated.
For ipv6 we also need to store sernum, else dst_check doesn't work,
pointed out by Eric Dumazet.
This should speed up forwarding when conntrack is already in use
anyway, especially when using reverse path filtering -- active RPF
enforces two FIB lookups for each packet.
Before the routing cache removal this didn't matter since RPF was performed
only when route cache didn't yield a result; but without route cache it
comes at higher price.
Julian Anastasov suggested to add NETDEV_UNREGISTER handler to
avoid holding on to dsts of 'frozen' conntracks.
Signed-off-by: Florian Westphal <fw@strlen.de>
--- a/include/net/netfilter/nf_conntrack_extend.h
+++ b/include/net/netfilter/nf_conntrack_extend.h
@@ -30,6 +30,9 @@ enum nf_ct_ext_id {
#if IS_ENABLED(CONFIG_NETFILTER_SYNPROXY)
NF_CT_EXT_SYNPROXY,
#endif
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_RTCACHE)
+ NF_CT_EXT_RTCACHE,
+#endif
NF_CT_EXT_NUM,
};
@@ -43,6 +46,7 @@ enum nf_ct_ext_id {
#define NF_CT_EXT_TIMEOUT_TYPE struct nf_conn_timeout
#define NF_CT_EXT_LABELS_TYPE struct nf_conn_labels
#define NF_CT_EXT_SYNPROXY_TYPE struct nf_conn_synproxy
+#define NF_CT_EXT_RTCACHE_TYPE struct nf_conn_rtcache
/* Extensions: optional stuff which isn't permanently in struct. */
struct nf_ct_ext {
--- /dev/null
+++ b/include/net/netfilter/nf_conntrack_rtcache.h
@@ -0,0 +1,34 @@
+#include <linux/gfp.h>
+#include <net/netfilter/nf_conntrack.h>
+#include <net/netfilter/nf_conntrack_extend.h>
+
+struct dst_entry;
+
+struct nf_conn_dst_cache {
+ struct dst_entry *dst;
+ int iif;
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV6)
+ u32 cookie;
+#endif
+
+};
+
+struct nf_conn_rtcache {
+ struct nf_conn_dst_cache cached_dst[IP_CT_DIR_MAX];
+};
+
+static inline
+struct nf_conn_rtcache *nf_ct_rtcache_find(const struct nf_conn *ct)
+{
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_RTCACHE)
+ return nf_ct_ext_find(ct, NF_CT_EXT_RTCACHE);
+#else
+ return NULL;
+#endif
+}
+
+static inline int nf_conn_rtcache_iif_get(const struct nf_conn_rtcache *rtc,
+ enum ip_conntrack_dir dir)
+{
+ return rtc->cached_dst[dir].iif;
+}
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -106,6 +106,18 @@ config NF_CONNTRACK_EVENTS
If unsure, say `N'.
+config NF_CONNTRACK_RTCACHE
+ tristate "Cache route entries in conntrack objects"
+ depends on NETFILTER_ADVANCED
+ depends on NF_CONNTRACK
+ help
+ If this option is enabled, the connection tracking code will
+ cache routing information for each connection that is being
+ forwarded, at a cost of 32 bytes per conntrack object.
+
+ To compile it as a module, choose M here. If unsure, say N.
+ The module will be called nf_conntrack_rtcache.
+
config NF_CONNTRACK_TIMEOUT
bool 'Connection tracking timeout'
depends on NETFILTER_ADVANCED
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -18,6 +18,9 @@ obj-$(CONFIG_NETFILTER_NETLINK_LOG) += n
# connection tracking
obj-$(CONFIG_NF_CONNTRACK) += nf_conntrack.o
+# optional conntrack route cache extension
+obj-$(CONFIG_NF_CONNTRACK_RTCACHE) += nf_conntrack_rtcache.o
+
# SCTP protocol connection tracking
obj-$(CONFIG_NF_CT_PROTO_DCCP) += nf_conntrack_proto_dccp.o
obj-$(CONFIG_NF_CT_PROTO_GRE) += nf_conntrack_proto_gre.o
--- /dev/null
+++ b/net/netfilter/nf_conntrack_rtcache.c
@@ -0,0 +1,391 @@
+/* route cache for netfilter.
+ *
+ * (C) 2014 Red Hat GmbH
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/types.h>
+#include <linux/netfilter.h>
+#include <linux/skbuff.h>
+#include <linux/stddef.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/export.h>
+#include <linux/module.h>
+
+#include <net/dst.h>
+
+#include <net/netfilter/nf_conntrack.h>
+#include <net/netfilter/nf_conntrack_core.h>
+#include <net/netfilter/nf_conntrack_extend.h>
+#include <net/netfilter/nf_conntrack_rtcache.h>
+
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV6)
+#include <net/ip6_fib.h>
+#endif
+
+static void __nf_conn_rtcache_destroy(struct nf_conn_rtcache *rtc,
+ enum ip_conntrack_dir dir)
+{
+ struct dst_entry *dst = rtc->cached_dst[dir].dst;
+
+ dst_release(dst);
+}
+
+static void nf_conn_rtcache_destroy(struct nf_conn *ct)
+{
+ struct nf_conn_rtcache *rtc = nf_ct_rtcache_find(ct);
+
+ if (!rtc)
+ return;
+
+ __nf_conn_rtcache_destroy(rtc, IP_CT_DIR_ORIGINAL);
+ __nf_conn_rtcache_destroy(rtc, IP_CT_DIR_REPLY);
+}
+
+static void nf_ct_rtcache_ext_add(struct nf_conn *ct)
+{
+ struct nf_conn_rtcache *rtc;
+
+ rtc = nf_ct_ext_add(ct, NF_CT_EXT_RTCACHE, GFP_ATOMIC);
+ if (rtc) {
+ rtc->cached_dst[IP_CT_DIR_ORIGINAL].iif = -1;
+ rtc->cached_dst[IP_CT_DIR_ORIGINAL].dst = NULL;
+ rtc->cached_dst[IP_CT_DIR_REPLY].iif = -1;
+ rtc->cached_dst[IP_CT_DIR_REPLY].dst = NULL;
+ }
+}
+
+static struct nf_conn_rtcache *nf_ct_rtcache_find_usable(struct nf_conn *ct)
+{
+ if (nf_ct_is_untracked(ct))
+ return NULL;
+ return nf_ct_rtcache_find(ct);
+}
+
+static struct dst_entry *
+nf_conn_rtcache_dst_get(const struct nf_conn_rtcache *rtc,
+ enum ip_conntrack_dir dir)
+{
+ return rtc->cached_dst[dir].dst;
+}
+
+static u32 nf_rtcache_get_cookie(int pf, const struct dst_entry *dst)
+{
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV6)
+ if (pf == NFPROTO_IPV6) {
+ const struct rt6_info *rt = (const struct rt6_info *)dst;
+
+ if (rt->rt6i_node)
+ return (u32)rt->rt6i_node->fn_sernum;
+ }
+#endif
+ return 0;
+}
+
+static void nf_conn_rtcache_dst_set(int pf,
+ struct nf_conn_rtcache *rtc,
+ struct dst_entry *dst,
+ enum ip_conntrack_dir dir, int iif)
+{
+ if (rtc->cached_dst[dir].iif != iif)
+ rtc->cached_dst[dir].iif = iif;
+
+ if (rtc->cached_dst[dir].dst != dst) {
+ struct dst_entry *old;
+
+ dst_hold(dst);
+
+ old = xchg(&rtc->cached_dst[dir].dst, dst);
+ dst_release(old);
+
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV6)
+ if (pf == NFPROTO_IPV6)
+ rtc->cached_dst[dir].cookie =
+ nf_rtcache_get_cookie(pf, dst);
+#endif
+ }
+}
+
+static void nf_conn_rtcache_dst_obsolete(struct nf_conn_rtcache *rtc,
+ enum ip_conntrack_dir dir)
+{
+ struct dst_entry *old;
+
+ pr_debug("Invalidate iif %d for dir %d on cache %p\n",
+ rtc->cached_dst[dir].iif, dir, rtc);
+
+ old = xchg(&rtc->cached_dst[dir].dst, NULL);
+ dst_release(old);
+ rtc->cached_dst[dir].iif = -1;
+}
+
+static unsigned int nf_rtcache_in(const struct nf_hook_ops *ops,
+ struct sk_buff *skb,
+ const struct net_device *in,
+ const struct net_device *out,
+ int (*okfn)(struct sk_buff *))
+{
+ struct nf_conn_rtcache *rtc;
+ enum ip_conntrack_info ctinfo;
+ enum ip_conntrack_dir dir;
+ struct dst_entry *dst;
+ struct nf_conn *ct;
+ int iif;
+ u32 cookie;
+
+ if (skb_dst(skb) || skb->sk)
+ return NF_ACCEPT;
+
+ ct = nf_ct_get(skb, &ctinfo);
+ if (!ct)
+ return NF_ACCEPT;
+
+ rtc = nf_ct_rtcache_find_usable(ct);
+ if (!rtc)
+ return NF_ACCEPT;
+
+ /* if iif changes, don't use cache and let ip stack
+ * do route lookup.
+ *
+ * If rp_filter is enabled it might toss skb, so
+ * we don't want to avoid these checks.
+ */
+ dir = CTINFO2DIR(ctinfo);
+ iif = nf_conn_rtcache_iif_get(rtc, dir);
+ if (in->ifindex != iif) {
+ pr_debug("ct %p, iif %d, cached iif %d, skip cached entry\n",
+ ct, iif, in->ifindex);
+ return NF_ACCEPT;
+ }
+ dst = nf_conn_rtcache_dst_get(rtc, dir);
+ if (dst == NULL)
+ return NF_ACCEPT;
+
+ cookie = nf_rtcache_get_cookie(ops->pf, dst);
+
+ dst = dst_check(dst, cookie);
+ pr_debug("obtained dst %p for skb %p, cookie %d\n", dst, skb, cookie);
+ if (likely(dst))
+ skb_dst_set_noref_force(skb, dst);
+ else
+ nf_conn_rtcache_dst_obsolete(rtc, dir);
+
+ return NF_ACCEPT;
+}
+
+static unsigned int nf_rtcache_forward(const struct nf_hook_ops *ops,
+ struct sk_buff *skb,
+ const struct net_device *in,
+ const struct net_device *out,
+ int (*okfn)(struct sk_buff *))
+{
+ struct nf_conn_rtcache *rtc;
+ enum ip_conntrack_info ctinfo;
+ enum ip_conntrack_dir dir;
+ struct nf_conn *ct;
+ struct dst_entry *dst = skb_dst(skb);
+ int iif;
+
+ ct = nf_ct_get(skb, &ctinfo);
+ if (!ct)
+ return NF_ACCEPT;
+
+ if (dst && dst_xfrm(dst))
+ return NF_ACCEPT;
+
+ if (!nf_ct_is_confirmed(ct)) {
+ if (WARN_ON(nf_ct_rtcache_find(ct)))
+ return NF_ACCEPT;
+ nf_ct_rtcache_ext_add(ct);
+ return NF_ACCEPT;
+ }
+
+ rtc = nf_ct_rtcache_find_usable(ct);
+ if (!rtc)
+ return NF_ACCEPT;
+
+ dir = CTINFO2DIR(ctinfo);
+ iif = nf_conn_rtcache_iif_get(rtc, dir);
+ pr_debug("ct %p, skb %p, dir %d, iif %d, cached iif %d\n",
+ ct, skb, dir, iif, in->ifindex);
+ if (likely(in->ifindex == iif))
+ return NF_ACCEPT;
+
+ nf_conn_rtcache_dst_set(ops->pf, rtc, skb_dst(skb), dir, in->ifindex);
+ return NF_ACCEPT;
+}
+
+static int nf_rtcache_dst_remove(struct nf_conn *ct, void *data)
+{
+ struct nf_conn_rtcache *rtc = nf_ct_rtcache_find(ct);
+ struct net_device *dev = data;
+
+ if (!rtc)
+ return 0;
+
+ if (dev->ifindex == rtc->cached_dst[IP_CT_DIR_ORIGINAL].iif ||
+ dev->ifindex == rtc->cached_dst[IP_CT_DIR_REPLY].iif) {
+ nf_conn_rtcache_dst_obsolete(rtc, IP_CT_DIR_ORIGINAL);
+ nf_conn_rtcache_dst_obsolete(rtc, IP_CT_DIR_REPLY);
+ }
+
+ return 0;
+}
+
+static int nf_rtcache_netdev_event(struct notifier_block *this,
+ unsigned long event, void *ptr)
+{
+ struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+ struct net *net = dev_net(dev);
+
+ if (event == NETDEV_DOWN)
+ nf_ct_iterate_cleanup(net, nf_rtcache_dst_remove, dev, 0, 0);
+
+ return NOTIFY_DONE;
+}
+
+static struct notifier_block nf_rtcache_notifier = {
+ .notifier_call = nf_rtcache_netdev_event,
+};
+
+static struct nf_hook_ops rtcache_ops[] = {
+ {
+ .hook = nf_rtcache_in,
+ .owner = THIS_MODULE,
+ .pf = NFPROTO_IPV4,
+ .hooknum = NF_INET_PRE_ROUTING,
+ .priority = NF_IP_PRI_LAST,
+ },
+ {
+ .hook = nf_rtcache_forward,
+ .owner = THIS_MODULE,
+ .pf = NFPROTO_IPV4,
+ .hooknum = NF_INET_FORWARD,
+ .priority = NF_IP_PRI_LAST,
+ },
+#if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV6)
+ {
+ .hook = nf_rtcache_in,
+ .owner = THIS_MODULE,
+ .pf = NFPROTO_IPV6,
+ .hooknum = NF_INET_PRE_ROUTING,
+ .priority = NF_IP_PRI_LAST,
+ },
+ {
+ .hook = nf_rtcache_forward,
+ .owner = THIS_MODULE,
+ .pf = NFPROTO_IPV6,
+ .hooknum = NF_INET_FORWARD,
+ .priority = NF_IP_PRI_LAST,
+ },
+#endif
+};
+
+static struct nf_ct_ext_type rtcache_extend __read_mostly = {
+ .len = sizeof(struct nf_conn_rtcache),
+ .align = __alignof__(struct nf_conn_rtcache),
+ .id = NF_CT_EXT_RTCACHE,
+ .destroy = nf_conn_rtcache_destroy,
+};
+
+static int __init nf_conntrack_rtcache_init(void)
+{
+ int ret = nf_ct_extend_register(&rtcache_extend);
+
+ if (ret < 0) {
+ pr_err("nf_conntrack_rtcache: Unable to register extension\n");
+ return ret;
+ }
+
+ ret = nf_register_hooks(rtcache_ops, ARRAY_SIZE(rtcache_ops));
+ if (ret < 0) {
+ nf_ct_extend_unregister(&rtcache_extend);
+ return ret;
+ }
+
+ ret = register_netdevice_notifier(&nf_rtcache_notifier);
+ if (ret) {
+ nf_unregister_hooks(rtcache_ops, ARRAY_SIZE(rtcache_ops));
+ nf_ct_extend_unregister(&rtcache_extend);
+ }
+
+ return ret;
+}
+
+static int nf_rtcache_ext_remove(struct nf_conn *ct, void *data)
+{
+ struct nf_conn_rtcache *rtc = nf_ct_rtcache_find(ct);
+
+ return rtc != NULL;
+}
+
+static bool __exit nf_conntrack_rtcache_wait_for_dying(struct net *net)
+{
+ bool wait = false;
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ struct nf_conntrack_tuple_hash *h;
+ struct hlist_nulls_node *n;
+ struct nf_conn *ct;
+ struct ct_pcpu *pcpu = per_cpu_ptr(net->ct.pcpu_lists, cpu);
+
+ rcu_read_lock();
+ spin_lock_bh(&pcpu->lock);
+
+ hlist_nulls_for_each_entry(h, n, &pcpu->dying, hnnode) {
+ ct = nf_ct_tuplehash_to_ctrack(h);
+ if (nf_ct_rtcache_find(ct) != NULL) {
+ wait = true;
+ break;
+ }
+ }
+ spin_unlock_bh(&pcpu->lock);
+ rcu_read_unlock();
+ }
+
+ return wait;
+}
+
+static void __exit nf_conntrack_rtcache_fini(void)
+{
+ struct net *net;
+ int count = 0;
+
+ /* remove hooks so no new connections get rtcache extension */
+ nf_unregister_hooks(rtcache_ops, ARRAY_SIZE(rtcache_ops));
+
+ synchronize_net();
+
+ unregister_netdevice_notifier(&nf_rtcache_notifier);
+
+ rtnl_lock();
+
+ /* zap all conntracks with rtcache extension */
+ for_each_net(net)
+ nf_ct_iterate_cleanup(net, nf_rtcache_ext_remove, NULL, 0, 0);
+
+ for_each_net(net) {
+ /* .. and make sure they're gone from dying list, too */
+ while (nf_conntrack_rtcache_wait_for_dying(net)) {
+ msleep(200);
+ WARN_ONCE(++count > 25, "Waiting for all rtcache conntracks to go away\n");
+ }
+ }
+
+ rtnl_unlock();
+ synchronize_net();
+ nf_ct_extend_unregister(&rtcache_extend);
+}
+module_init(nf_conntrack_rtcache_init);
+module_exit(nf_conntrack_rtcache_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Florian Westphal <fw@strlen.de>");
+MODULE_DESCRIPTION("Conntrack route cache extension");

99
target/linux/generic/pending-3.18/051-02-bridge-allow-setting-hash_max-multicast_router-if-in.patch
View File

@ -1,99 +0,0 @@
From 6ae4ae8e512bd229f806c22f8a2cd751e4f987c2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Linus=20L=C3=BCssing?= <linus.luessing@c0d3.blue>
Date: Sat, 23 May 2015 03:12:34 +0200
Subject: [PATCH] bridge: allow setting hash_max + multicast_router if
interface is down
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Network managers like netifd (used in OpenWRT for instance) try to
configure interface options after creation but before setting the
interface up.
Unfortunately the sysfs / bridge currently only allows to configure the
hash_max and multicast_router options when the bridge interface is up.
But since br_multicast_init() doesn't start any timers and only sets
default values and initializes timers it should be save to reconfigure
the default values after that, before things actually get active after
the bridge is set up.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
net/bridge/br_multicast.c | 24 +++---------------------
1 file changed, 3 insertions(+), 21 deletions(-)
--- a/net/bridge/br_multicast.c
+++ b/net/bridge/br_multicast.c
@@ -1948,11 +1948,9 @@ out:
int br_multicast_set_router(struct net_bridge *br, unsigned long val)
{
- int err = -ENOENT;
+ int err = -EINVAL;
spin_lock_bh(&br->multicast_lock);
- if (!netif_running(br->dev))
- goto unlock;
switch (val) {
case 0:
@@ -1963,13 +1961,8 @@ int br_multicast_set_router(struct net_b
br->multicast_router = val;
err = 0;
break;
-
- default:
- err = -EINVAL;
- break;
}
-unlock:
spin_unlock_bh(&br->multicast_lock);
return err;
@@ -1978,11 +1971,9 @@ unlock:
int br_multicast_set_port_router(struct net_bridge_port *p, unsigned long val)
{
struct net_bridge *br = p->br;
- int err = -ENOENT;
+ int err = -EINVAL;
spin_lock(&br->multicast_lock);
- if (!netif_running(br->dev) || p->state == BR_STATE_DISABLED)
- goto unlock;
switch (val) {
case 0:
@@ -2004,13 +1995,8 @@ int br_multicast_set_port_router(struct
br_multicast_add_router(br, p);
break;
-
- default:
- err = -EINVAL;
- break;
}
-unlock:
spin_unlock(&br->multicast_lock);
return err;
@@ -2115,15 +2101,11 @@ unlock:
int br_multicast_set_hash_max(struct net_bridge *br, unsigned long val)
{
- int err = -ENOENT;
+ int err = -EINVAL;
u32 old;
struct net_bridge_mdb_htable *mdb;
spin_lock_bh(&br->multicast_lock);
- if (!netif_running(br->dev))
- goto unlock;
-
- err = -EINVAL;
if (!is_power_of_2(val))
goto unlock;

22
target/linux/generic/pending-3.18/060-mips_decompressor_memmove.patch
View File

@ -1,22 +0,0 @@
--- a/arch/mips/boot/compressed/string.c
+++ b/arch/mips/boot/compressed/string.c
@@ -26,3 +26,19 @@ void *memset(void *s, int c, size_t n)
ss[i] = c;
return s;
}
+
+void *memmove(void *__dest, __const void *__src, size_t count)
+{
+ unsigned char *d = __dest;
+ const unsigned char *s = __src;
+
+ if (__dest == __src)
+ return __dest;
+
+ if (__dest < __src)
+ return memcpy(__dest, __src, count);
+
+ while (count--)
+ d[count] = s[count];
+ return __dest;
+}

44
target/linux/generic/pending-3.18/070-bgmac-register-napi-before-the-device.patch
View File

@ -1,44 +0,0 @@
From 6216642f200258708e47170ff14ba8ecb486f4f0 Mon Sep 17 00:00:00 2001
From: Hauke Mehrtens <hauke@hauke-m.de>
Date: Sun, 18 Jan 2015 19:49:58 +0100
Subject: [PATCH] bgmac: register napi before the device
napi should get registered before the netdev and not after.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/ethernet/broadcom/bgmac.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -1521,6 +1521,8 @@ static int bgmac_probe(struct bcma_devic
if (core->bus->sprom.boardflags_lo & BGMAC_BFL_ENETADM)
bgmac_warn(bgmac, "Support for ADMtek ethernet switch not implemented\n");
+ netif_napi_add(net_dev, &bgmac->napi, bgmac_poll, BGMAC_WEIGHT);
+
err = bgmac_mii_register(bgmac);
if (err) {
bgmac_err(bgmac, "Cannot register MDIO\n");
@@ -1535,8 +1537,6 @@ static int bgmac_probe(struct bcma_devic
netif_carrier_off(net_dev);
- netif_napi_add(net_dev, &bgmac->napi, bgmac_poll, BGMAC_WEIGHT);
-
return 0;
err_mii_unregister:
@@ -1555,9 +1555,9 @@ static void bgmac_remove(struct bcma_dev
{
struct bgmac *bgmac = bcma_get_drvdata(core);
- netif_napi_del(&bgmac->napi);
unregister_netdev(bgmac->net_dev);
bgmac_mii_unregister(bgmac);
+ netif_napi_del(&bgmac->napi);
bgmac_dma_free(bgmac);
bcma_set_drvdata(core, NULL);
free_netdev(bgmac->net_dev);

30
target/linux/generic/pending-3.18/071-bgmac-activate-irqs-only-if-there-is-nothing-to-poll.patch
View File

@ -1,30 +0,0 @@
From 43f159c60a99318b1ef7d1d7c16c4dfdd06bfd90 Mon Sep 17 00:00:00 2001
From: Hauke Mehrtens <hauke@hauke-m.de>
Date: Sun, 18 Jan 2015 19:49:59 +0100
Subject: [PATCH] bgmac: activate irqs only if there is nothing to poll
IRQs should only get activated when there is nothing to poll in the
queue any more and to after every poll.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/ethernet/broadcom/bgmac.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -1167,10 +1167,10 @@ static int bgmac_poll(struct napi_struct
bgmac->int_status = 0;
}
- if (handled < weight)
+ if (handled < weight) {
napi_complete(napi);
-
- bgmac_chip_intrs_on(bgmac);
+ bgmac_chip_intrs_on(bgmac);
+ }
return handled;
}

50
target/linux/generic/pending-3.18/073-bgmac-Clean-warning-messages.patch
View File

@ -1,50 +0,0 @@
From 8edfe3b6fad28da191c8fa15e4e0d8f7335a0091 Mon Sep 17 00:00:00 2001
From: Peter Senna Tschudin <peter.senna@gmail.com>
Date: Sat, 7 Mar 2015 12:10:26 +0100
Subject: [PATCH] bgmac: Clean warning messages
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
On my test environment the throughput of a file transfer drops
from 4.4MBps to 116KBps due the number of repeated warning
messages. This patch removes the warning messages as DMA works
correctly with addresses using 0xC0000000 bits.
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/ethernet/broadcom/bgmac.c | 7 -------
1 file changed, 7 deletions(-)
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -302,9 +302,6 @@ static int bgmac_dma_rx_skb_for_slot(str
slot->skb = skb;
slot->dma_addr = dma_addr;
- if (slot->dma_addr & 0xC0000000)
- bgmac_warn(bgmac, "DMA address using 0xC0000000 bit(s), it may need translation trick\n");
-
return 0;
}
@@ -505,8 +502,6 @@ static int bgmac_dma_alloc(struct bgmac
ring->mmio_base);
goto err_dma_free;
}
- if (ring->dma_base & 0xC0000000)
- bgmac_warn(bgmac, "DMA address using 0xC0000000 bit(s), it may need translation trick\n");
ring->unaligned = bgmac_dma_unaligned(bgmac, ring,
BGMAC_DMA_RING_TX);
@@ -536,8 +531,6 @@ static int bgmac_dma_alloc(struct bgmac
err = -ENOMEM;
goto err_dma_free;
}
- if (ring->dma_base & 0xC0000000)
- bgmac_warn(bgmac, "DMA address using 0xC0000000 bit(s), it may need translation trick\n");
ring->unaligned = bgmac_dma_unaligned(bgmac, ring,
BGMAC_DMA_RING_RX);

76
target/linux/generic/pending-3.18/074-bgmac-register-fixed-PHY-for-ARM-BCM470X-BCM5301X-ch.patch
View File

@ -1,76 +0,0 @@
From c25b23b8a387e7d31f7a74af8e37b61e9e6ebb21 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <zajec5@gmail.com>
Date: Fri, 20 Mar 2015 23:14:31 +0100
Subject: [PATCH] bgmac: register fixed PHY for ARM BCM470X / BCM5301X chipsets
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
On ARM SoCs with bgmac Ethernet hardware we don't have any normal PHY.
There is always a switch attached but it's not even controlled over MDIO
like in case of MIPS devices.
We need a fixed PHY to be able to send/receive packets from the switch.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/ethernet/broadcom/bgmac.c | 34 ++++++++++++++++++++++++++++++++++
1 file changed, 34 insertions(+)
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -14,6 +14,7 @@
#include <linux/etherdevice.h>
#include <linux/mii.h>
#include <linux/phy.h>
+#include <linux/phy_fixed.h>
#include <linux/interrupt.h>
#include <linux/dma-mapping.h>
#include <bcm47xx_nvram.h>
@@ -1330,13 +1331,46 @@ static void bgmac_adjust_link(struct net
}
}
+static int bgmac_fixed_phy_register(struct bgmac *bgmac)
+{
+ struct fixed_phy_status fphy_status = {
+ .link = 1,
+ .speed = SPEED_1000,
+ .duplex = DUPLEX_FULL,
+ };
+ struct phy_device *phy_dev;
+ int err;
+
+ phy_dev = fixed_phy_register(PHY_POLL, &fphy_status, NULL);
+ if (!phy_dev || IS_ERR(phy_dev)) {
+ bgmac_err(bgmac, "Failed to register fixed PHY device\n");
+ return -ENODEV;
+ }
+
+ err = phy_connect_direct(bgmac->net_dev, phy_dev, bgmac_adjust_link,
+ PHY_INTERFACE_MODE_MII);
+ if (err) {
+ bgmac_err(bgmac, "Connecting PHY failed\n");
+ return err;
+ }
+
+ bgmac->phy_dev = phy_dev;
+
+ return err;
+}
+
static int bgmac_mii_register(struct bgmac *bgmac)
{
+ struct bcma_chipinfo *ci = &bgmac->core->bus->chipinfo;
struct mii_bus *mii_bus;
struct phy_device *phy_dev;
char bus_id[MII_BUS_ID_SIZE + 3];
int i, err = 0;
+ if (ci->id == BCMA_CHIP_ID_BCM4707 ||
+ ci->id == BCMA_CHIP_ID_BCM53018)
+ return bgmac_fixed_phy_register(bgmac);
+
mii_bus = mdiobus_alloc();
if (!mii_bus)
return -ENOMEM;

28
target/linux/generic/pending-3.18/075-bgmac-allow-enabling-on-ARCH_BCM_5301X.patch
View File

@ -1,28 +0,0 @@
From fc300dc3733fdc328e6e10c7b8379b60c26cd648 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <zajec5@gmail.com>
Date: Fri, 20 Mar 2015 23:14:32 +0100
Subject: [PATCH] bgmac: allow enabling on ARCH_BCM_5301X
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Home routers based on ARM SoCs like BCM4708 also have bcma bus with core
supported by bgmac.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/ethernet/broadcom/Kconfig | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/drivers/net/ethernet/broadcom/Kconfig
+++ b/drivers/net/ethernet/broadcom/Kconfig
@@ -143,7 +143,7 @@ config BNX2X_SRIOV
config BGMAC
tristate "BCMA bus GBit core support"
- depends on BCMA_HOST_SOC && HAS_DMA && BCM47XX
+ depends on BCMA_HOST_SOC && HAS_DMA && (BCM47XX || ARCH_BCM_5301X)
select PHYLIB
---help---
This driver supports GBit MAC and BCM4706 GBit MAC cores on BCMA bus.

30
target/linux/generic/pending-3.18/076-net-phy-export-fixed_phy_register.patch
View File

@ -1,30 +0,0 @@
From 37e9a6904520b525b542ecd67201164d06fdb95a Mon Sep 17 00:00:00 2001
From: Mark Salter <msalter@redhat.com>
Date: Thu, 11 Dec 2014 23:03:26 -0500
Subject: [PATCH] net: phy: export fixed_phy_register()
When building the bcmgenet driver as module, I get:
ERROR: "fixed_phy_register" [drivers/net/ethernet/broadcom/genet/genet.ko] undefined!
commit b0ba512e225d72 ("net: bcmgenet: enable driver to work without device
tree") which added a call to fixed_phy_register. But fixed_phy_register
needs to be exported if used from a module.
Signed-off-by: Mark Salter <msalter@redhat.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
drivers/net/phy/fixed.c | 1 +
1 file changed, 1 insertion(+)
--- a/drivers/net/phy/fixed.c
+++ b/drivers/net/phy/fixed.c
@@ -274,6 +274,7 @@ struct phy_device *fixed_phy_register(un
return phy;
}
+EXPORT_SYMBOL_GPL(fixed_phy_register);
static int __init fixed_mdio_bus_init(void)
{

24
target/linux/generic/pending-3.18/077-01-bgmac-fix-descriptor-frame-start-end-definitions.patch
View File

@ -1,24 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Mon, 23 Mar 2015 02:40:06 +0100
Subject: [PATCH] bgmac: fix descriptor frame start/end definitions
The start-of-frame and end-of-frame bits were accidentally swapped.
In the current code it does not make any difference, since they are
always used together.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -345,8 +345,8 @@
#define BGMAC_DESC_CTL0_EOT 0x10000000 /* End of ring */
#define BGMAC_DESC_CTL0_IOC 0x20000000 /* IRQ on complete */
-#define BGMAC_DESC_CTL0_SOF 0x40000000 /* Start of frame */
-#define BGMAC_DESC_CTL0_EOF 0x80000000 /* End of frame */
+#define BGMAC_DESC_CTL0_EOF 0x40000000 /* End of frame */
+#define BGMAC_DESC_CTL0_SOF 0x80000000 /* Start of frame */
#define BGMAC_DESC_CTL1_LEN 0x00001FFF
#define BGMAC_PHY_NOREGS 0x1E

189
target/linux/generic/pending-3.18/077-02-bgmac-implement-GRO-and-use-build_skb.patch
View File

@ -1,189 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Mon, 23 Mar 2015 02:41:25 +0100
Subject: [PATCH] bgmac: implement GRO and use build_skb
This improves performance for routing and local rx
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -276,31 +276,31 @@ static int bgmac_dma_rx_skb_for_slot(str
struct bgmac_slot_info *slot)
{
struct device *dma_dev = bgmac->core->dma_dev;
- struct sk_buff *skb;
dma_addr_t dma_addr;
struct bgmac_rx_header *rx;
+ void *buf;
/* Alloc skb */
- skb = netdev_alloc_skb(bgmac->net_dev, BGMAC_RX_BUF_SIZE);
- if (!skb)
+ buf = netdev_alloc_frag(BGMAC_RX_ALLOC_SIZE);
+ if (!buf)
return -ENOMEM;
/* Poison - if everything goes fine, hardware will overwrite it */
- rx = (struct bgmac_rx_header *)skb->data;
+ rx = buf;
rx->len = cpu_to_le16(0xdead);
rx->flags = cpu_to_le16(0xbeef);
/* Map skb for the DMA */
- dma_addr = dma_map_single(dma_dev, skb->data,
- BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
+ dma_addr = dma_map_single(dma_dev, buf, BGMAC_RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
if (dma_mapping_error(dma_dev, dma_addr)) {
bgmac_err(bgmac, "DMA mapping error\n");
- dev_kfree_skb(skb);
+ put_page(virt_to_head_page(buf));
return -ENOMEM;
}
/* Update the slot */
- slot->skb = skb;
+ slot->buf = buf;
slot->dma_addr = dma_addr;
return 0;
@@ -343,8 +343,9 @@ static int bgmac_dma_rx_read(struct bgma
while (ring->start != ring->end) {
struct device *dma_dev = bgmac->core->dma_dev;
struct bgmac_slot_info *slot = &ring->slots[ring->start];
- struct sk_buff *skb = slot->skb;
- struct bgmac_rx_header *rx;
+ struct bgmac_rx_header *rx = slot->buf;
+ struct sk_buff *skb;
+ void *buf = slot->buf;
u16 len, flags;
/* Unmap buffer to make it accessible to the CPU */
@@ -352,7 +353,6 @@ static int bgmac_dma_rx_read(struct bgma
BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
/* Get info from the header */
- rx = (struct bgmac_rx_header *)skb->data;
len = le16_to_cpu(rx->len);
flags = le16_to_cpu(rx->flags);
@@ -393,12 +393,13 @@ static int bgmac_dma_rx_read(struct bgma
dma_unmap_single(dma_dev, old_dma_addr,
BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
+ skb = build_skb(buf, BGMAC_RX_ALLOC_SIZE);
skb_put(skb, BGMAC_RX_FRAME_OFFSET + len);
skb_pull(skb, BGMAC_RX_FRAME_OFFSET);
skb_checksum_none_assert(skb);
skb->protocol = eth_type_trans(skb, bgmac->net_dev);
- netif_receive_skb(skb);
+ napi_gro_receive(&bgmac->napi, skb);
handled++;
} while (0);
@@ -434,12 +435,11 @@ static bool bgmac_dma_unaligned(struct b
return false;
}
-static void bgmac_dma_ring_free(struct bgmac *bgmac,
- struct bgmac_dma_ring *ring)
+static void bgmac_dma_tx_ring_free(struct bgmac *bgmac,
+ struct bgmac_dma_ring *ring)
{
struct device *dma_dev = bgmac->core->dma_dev;
struct bgmac_slot_info *slot;
- int size;
int i;
for (i = 0; i < ring->num_slots; i++) {
@@ -451,23 +451,55 @@ static void bgmac_dma_ring_free(struct b
dev_kfree_skb(slot->skb);
}
}
+}
+
+static void bgmac_dma_rx_ring_free(struct bgmac *bgmac,
+ struct bgmac_dma_ring *ring)
+{
+ struct device *dma_dev = bgmac->core->dma_dev;
+ struct bgmac_slot_info *slot;
+ int i;
+
+ for (i = 0; i < ring->num_slots; i++) {
+ slot = &ring->slots[i];
+ if (!slot->buf)
+ continue;
- if (ring->cpu_base) {
- /* Free ring of descriptors */
- size = ring->num_slots * sizeof(struct bgmac_dma_desc);
- dma_free_coherent(dma_dev, size, ring->cpu_base,
- ring->dma_base);
+ if (slot->dma_addr)
+ dma_unmap_single(dma_dev, slot->dma_addr,
+ BGMAC_RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
+ put_page(virt_to_head_page(slot->buf));
}
}
+static void bgmac_dma_ring_desc_free(struct bgmac *bgmac,
+ struct bgmac_dma_ring *ring)
+{
+ struct device *dma_dev = bgmac->core->dma_dev;
+ int size;
+
+ if (!ring->cpu_base)
+ return;
+
+ /* Free ring of descriptors */
+ size = ring->num_slots * sizeof(struct bgmac_dma_desc);
+ dma_free_coherent(dma_dev, size, ring->cpu_base,
+ ring->dma_base);
+}
+
static void bgmac_dma_free(struct bgmac *bgmac)
{
int i;
- for (i = 0; i < BGMAC_MAX_TX_RINGS; i++)
- bgmac_dma_ring_free(bgmac, &bgmac->tx_ring[i]);
- for (i = 0; i < BGMAC_MAX_RX_RINGS; i++)
- bgmac_dma_ring_free(bgmac, &bgmac->rx_ring[i]);
+ for (i = 0; i < BGMAC_MAX_TX_RINGS; i++) {
+ bgmac_dma_tx_ring_free(bgmac, &bgmac->tx_ring[i]);
+ bgmac_dma_ring_desc_free(bgmac, &bgmac->tx_ring[i]);
+ }
+ for (i = 0; i < BGMAC_MAX_RX_RINGS; i++) {
+ bgmac_dma_rx_ring_free(bgmac, &bgmac->rx_ring[i]);
+ bgmac_dma_ring_desc_free(bgmac, &bgmac->rx_ring[i]);
+ }
}
static int bgmac_dma_alloc(struct bgmac *bgmac)
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -362,6 +362,8 @@
#define BGMAC_RX_FRAME_OFFSET 30 /* There are 2 unused bytes between header and real data */
#define BGMAC_RX_MAX_FRAME_SIZE 1536 /* Copied from b44/tg3 */
#define BGMAC_RX_BUF_SIZE (BGMAC_RX_FRAME_OFFSET + BGMAC_RX_MAX_FRAME_SIZE)
+#define BGMAC_RX_ALLOC_SIZE (SKB_DATA_ALIGN(BGMAC_RX_BUF_SIZE) + \
+ SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
#define BGMAC_BFL_ENETROBO 0x0010 /* has ephy roboswitch spi */
#define BGMAC_BFL_ENETADM 0x0080 /* has ADMtek switch */
@@ -383,7 +385,10 @@
#define ETHER_MAX_LEN 1518
struct bgmac_slot_info {
- struct sk_buff *skb;
+ union {
+ struct sk_buff *skb;
+ void *buf;
+ };
dma_addr_t dma_addr;
};

267
target/linux/generic/pending-3.18/077-03-bgmac-implement-scatter-gather-support.patch
View File

@ -1,267 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Mon, 23 Mar 2015 02:42:26 +0100
Subject: [PATCH] bgmac: implement scatter/gather support
Always use software checksumming, since the hardware does not have any
checksum offload support.
This significantly improves local TCP tx performance.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -115,53 +115,91 @@ static void bgmac_dma_tx_enable(struct b
bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_TX_CTL, ctl);
}
+static void
+bgmac_dma_tx_add_buf(struct bgmac *bgmac, struct bgmac_dma_ring *ring,
+ int i, int len, u32 ctl0)
+{
+ struct bgmac_slot_info *slot;
+ struct bgmac_dma_desc *dma_desc;
+ u32 ctl1;
+
+ if (i == ring->num_slots - 1)
+ ctl0 |= BGMAC_DESC_CTL0_EOT;
+
+ ctl1 = len & BGMAC_DESC_CTL1_LEN;
+
+ slot = &ring->slots[i];
+ dma_desc = &ring->cpu_base[i];
+ dma_desc->addr_low = cpu_to_le32(lower_32_bits(slot->dma_addr));
+ dma_desc->addr_high = cpu_to_le32(upper_32_bits(slot->dma_addr));
+ dma_desc->ctl0 = cpu_to_le32(ctl0);
+ dma_desc->ctl1 = cpu_to_le32(ctl1);
+}
+
static netdev_tx_t bgmac_dma_tx_add(struct bgmac *bgmac,
struct bgmac_dma_ring *ring,
struct sk_buff *skb)
{
struct device *dma_dev = bgmac->core->dma_dev;
struct net_device *net_dev = bgmac->net_dev;
- struct bgmac_dma_desc *dma_desc;
- struct bgmac_slot_info *slot;
- u32 ctl0, ctl1;
+ struct bgmac_slot_info *slot = &ring->slots[ring->end];
int free_slots;
+ int nr_frags;
+ u32 flags;
+ int index = ring->end;
+ int i;
if (skb->len > BGMAC_DESC_CTL1_LEN) {
bgmac_err(bgmac, "Too long skb (%d)\n", skb->len);
- goto err_stop_drop;
+ goto err_drop;
}
+ if (skb->ip_summed == CHECKSUM_PARTIAL)
+ skb_checksum_help(skb);
+
+ nr_frags = skb_shinfo(skb)->nr_frags;
+
if (ring->start <= ring->end)
free_slots = ring->start - ring->end + BGMAC_TX_RING_SLOTS;
else
free_slots = ring->start - ring->end;
- if (free_slots == 1) {
+
+ if (free_slots <= nr_frags + 1) {
bgmac_err(bgmac, "TX ring is full, queue should be stopped!\n");
netif_stop_queue(net_dev);
return NETDEV_TX_BUSY;
}
- slot = &ring->slots[ring->end];
- slot->skb = skb;
- slot->dma_addr = dma_map_single(dma_dev, skb->data, skb->len,
+ slot->dma_addr = dma_map_single(dma_dev, skb->data, skb_headlen(skb),
DMA_TO_DEVICE);
- if (dma_mapping_error(dma_dev, slot->dma_addr)) {
- bgmac_err(bgmac, "Mapping error of skb on ring 0x%X\n",
- ring->mmio_base);
- goto err_stop_drop;
- }
+ if (unlikely(dma_mapping_error(dma_dev, slot->dma_addr)))
+ goto err_dma_head;
- ctl0 = BGMAC_DESC_CTL0_IOC | BGMAC_DESC_CTL0_SOF | BGMAC_DESC_CTL0_EOF;
- if (ring->end == ring->num_slots - 1)
- ctl0 |= BGMAC_DESC_CTL0_EOT;
- ctl1 = skb->len & BGMAC_DESC_CTL1_LEN;
+ flags = BGMAC_DESC_CTL0_SOF;
+ if (!nr_frags)
+ flags |= BGMAC_DESC_CTL0_EOF | BGMAC_DESC_CTL0_IOC;
+
+ bgmac_dma_tx_add_buf(bgmac, ring, index, skb_headlen(skb), flags);
+ flags = 0;
+
+ for (i = 0; i < nr_frags; i++) {
+ struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i];
+ int len = skb_frag_size(frag);
+
+ index = (index + 1) % BGMAC_TX_RING_SLOTS;
+ slot = &ring->slots[index];
+ slot->dma_addr = skb_frag_dma_map(dma_dev, frag, 0,
+ len, DMA_TO_DEVICE);
+ if (unlikely(dma_mapping_error(dma_dev, slot->dma_addr)))
+ goto err_dma;
- dma_desc = ring->cpu_base;
- dma_desc += ring->end;
- dma_desc->addr_low = cpu_to_le32(lower_32_bits(slot->dma_addr));
- dma_desc->addr_high = cpu_to_le32(upper_32_bits(slot->dma_addr));
- dma_desc->ctl0 = cpu_to_le32(ctl0);
- dma_desc->ctl1 = cpu_to_le32(ctl1);
+ if (i == nr_frags - 1)
+ flags |= BGMAC_DESC_CTL0_EOF | BGMAC_DESC_CTL0_IOC;
+
+ bgmac_dma_tx_add_buf(bgmac, ring, index, len, flags);
+ }
+
+ slot->skb = skb;
netdev_sent_queue(net_dev, skb->len);
@@ -170,20 +208,35 @@ static netdev_tx_t bgmac_dma_tx_add(stru
/* Increase ring->end to point empty slot. We tell hardware the first
* slot it should *not* read.
*/
- if (++ring->end >= BGMAC_TX_RING_SLOTS)
- ring->end = 0;
+ ring->end = (index + 1) % BGMAC_TX_RING_SLOTS;
bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_TX_INDEX,
ring->index_base +
ring->end * sizeof(struct bgmac_dma_desc));
- /* Always keep one slot free to allow detecting bugged calls. */
- if (--free_slots == 1)
+ free_slots -= nr_frags + 1;
+ if (free_slots < 8)
netif_stop_queue(net_dev);
return NETDEV_TX_OK;
-err_stop_drop:
- netif_stop_queue(net_dev);
+err_dma:
+ dma_unmap_single(dma_dev, slot->dma_addr, skb_headlen(skb),
+ DMA_TO_DEVICE);
+
+ while (i > 0) {
+ int index = (ring->end + i) % BGMAC_TX_RING_SLOTS;
+ struct bgmac_slot_info *slot = &ring->slots[index];
+ u32 ctl1 = le32_to_cpu(ring->cpu_base[index].ctl1);
+ int len = ctl1 & BGMAC_DESC_CTL1_LEN;
+
+ dma_unmap_page(dma_dev, slot->dma_addr, len, DMA_TO_DEVICE);
+ }
+
+err_dma_head:
+ bgmac_err(bgmac, "Mapping error of skb on ring 0x%X\n",
+ ring->mmio_base);
+
+err_drop:
dev_kfree_skb(skb);
return NETDEV_TX_OK;
}
@@ -205,32 +258,45 @@ static void bgmac_dma_tx_free(struct bgm
while (ring->start != empty_slot) {
struct bgmac_slot_info *slot = &ring->slots[ring->start];
+ u32 ctl1 = le32_to_cpu(ring->cpu_base[ring->start].ctl1);
+ int len = ctl1 & BGMAC_DESC_CTL1_LEN;
- if (slot->skb) {
+ if (!slot->dma_addr) {
+ bgmac_err(bgmac, "Hardware reported transmission for empty TX ring slot %d! End of ring: %d\n",
+ ring->start, ring->end);
+ goto next;
+ }
+
+ if (ctl1 & BGMAC_DESC_CTL0_SOF)
/* Unmap no longer used buffer */
- dma_unmap_single(dma_dev, slot->dma_addr,
- slot->skb->len, DMA_TO_DEVICE);
- slot->dma_addr = 0;
+ dma_unmap_single(dma_dev, slot->dma_addr, len,
+ DMA_TO_DEVICE);
+ else
+ dma_unmap_page(dma_dev, slot->dma_addr, len,
+ DMA_TO_DEVICE);
+ if (slot->skb) {
bytes_compl += slot->skb->len;
pkts_compl++;
/* Free memory! :) */
dev_kfree_skb(slot->skb);
slot->skb = NULL;
- } else {
- bgmac_err(bgmac, "Hardware reported transmission for empty TX ring slot %d! End of ring: %d\n",
- ring->start, ring->end);
}
+next:
+ slot->dma_addr = 0;
if (++ring->start >= BGMAC_TX_RING_SLOTS)
ring->start = 0;
freed = true;
}
+ if (!pkts_compl)
+ return;
+
netdev_completed_queue(bgmac->net_dev, pkts_compl, bytes_compl);
- if (freed && netif_queue_stopped(bgmac->net_dev))
+ if (netif_queue_stopped(bgmac->net_dev))
netif_wake_queue(bgmac->net_dev);
}
@@ -439,17 +505,25 @@ static void bgmac_dma_tx_ring_free(struc
struct bgmac_dma_ring *ring)
{
struct device *dma_dev = bgmac->core->dma_dev;
+ struct bgmac_dma_desc *dma_desc = ring->cpu_base;
struct bgmac_slot_info *slot;
int i;
for (i = 0; i < ring->num_slots; i++) {
+ int len = dma_desc[i].ctl1 & BGMAC_DESC_CTL1_LEN;
+
slot = &ring->slots[i];
- if (slot->skb) {
- if (slot->dma_addr)
- dma_unmap_single(dma_dev, slot->dma_addr,
- slot->skb->len, DMA_TO_DEVICE);
- dev_kfree_skb(slot->skb);
- }
+ dev_kfree_skb(slot->skb);
+
+ if (!slot->dma_addr)
+ continue;
+
+ if (slot->skb)
+ dma_unmap_single(dma_dev, slot->dma_addr,
+ len, DMA_TO_DEVICE);
+ else
+ dma_unmap_page(dma_dev, slot->dma_addr,
+ len, DMA_TO_DEVICE);
}
}
@@ -1588,6 +1662,10 @@ static int bgmac_probe(struct bcma_devic
goto err_dma_free;
}
+ net_dev->features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM;
+ net_dev->hw_features = net_dev->features;
+ net_dev->vlan_features = net_dev->features;
+
err = register_netdev(bgmac->net_dev);
if (err) {
bgmac_err(bgmac, "Cannot register net device\n");

125
target/linux/generic/pending-3.18/077-04-bgmac-simplify-tx-ring-index-handling.patch
View File

@ -1,125 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 09:58:56 +0200
Subject: [PATCH] bgmac: simplify tx ring index handling
Keep incrementing ring->start and ring->end instead of pointing it to
the actual ring slot entry. This simplifies the calculation of the
number of free slots.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -142,11 +142,10 @@ static netdev_tx_t bgmac_dma_tx_add(stru
{
struct device *dma_dev = bgmac->core->dma_dev;
struct net_device *net_dev = bgmac->net_dev;
- struct bgmac_slot_info *slot = &ring->slots[ring->end];
- int free_slots;
+ int index = ring->end % BGMAC_TX_RING_SLOTS;
+ struct bgmac_slot_info *slot = &ring->slots[index];
int nr_frags;
u32 flags;
- int index = ring->end;
int i;
if (skb->len > BGMAC_DESC_CTL1_LEN) {
@@ -159,12 +158,10 @@ static netdev_tx_t bgmac_dma_tx_add(stru
nr_frags = skb_shinfo(skb)->nr_frags;
- if (ring->start <= ring->end)
- free_slots = ring->start - ring->end + BGMAC_TX_RING_SLOTS;
- else
- free_slots = ring->start - ring->end;
-
- if (free_slots <= nr_frags + 1) {
+ /* ring->end - ring->start will return the number of valid slots,
+ * even when ring->end overflows
+ */
+ if (ring->end - ring->start + nr_frags + 1 >= BGMAC_TX_RING_SLOTS) {
bgmac_err(bgmac, "TX ring is full, queue should be stopped!\n");
netif_stop_queue(net_dev);
return NETDEV_TX_BUSY;
@@ -200,7 +197,7 @@ static netdev_tx_t bgmac_dma_tx_add(stru
}
slot->skb = skb;
-
+ ring->end += nr_frags + 1;
netdev_sent_queue(net_dev, skb->len);
wmb();
@@ -208,13 +205,12 @@ static netdev_tx_t bgmac_dma_tx_add(stru
/* Increase ring->end to point empty slot. We tell hardware the first
* slot it should *not* read.
*/
- ring->end = (index + 1) % BGMAC_TX_RING_SLOTS;
bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_TX_INDEX,
ring->index_base +
- ring->end * sizeof(struct bgmac_dma_desc));
+ (ring->end % BGMAC_TX_RING_SLOTS) *
+ sizeof(struct bgmac_dma_desc));
- free_slots -= nr_frags + 1;
- if (free_slots < 8)
+ if (ring->end - ring->start >= BGMAC_TX_RING_SLOTS - 8)
netif_stop_queue(net_dev);
return NETDEV_TX_OK;
@@ -256,17 +252,17 @@ static void bgmac_dma_tx_free(struct bgm
empty_slot &= BGMAC_DMA_TX_STATDPTR;
empty_slot /= sizeof(struct bgmac_dma_desc);
- while (ring->start != empty_slot) {
- struct bgmac_slot_info *slot = &ring->slots[ring->start];
- u32 ctl1 = le32_to_cpu(ring->cpu_base[ring->start].ctl1);
- int len = ctl1 & BGMAC_DESC_CTL1_LEN;
+ while (ring->start != ring->end) {
+ int slot_idx = ring->start % BGMAC_TX_RING_SLOTS;
+ struct bgmac_slot_info *slot = &ring->slots[slot_idx];
+ u32 ctl1;
+ int len;
- if (!slot->dma_addr) {
- bgmac_err(bgmac, "Hardware reported transmission for empty TX ring slot %d! End of ring: %d\n",
- ring->start, ring->end);
- goto next;
- }
+ if (slot_idx == empty_slot)
+ break;
+ ctl1 = le32_to_cpu(ring->cpu_base[slot_idx].ctl1);
+ len = ctl1 & BGMAC_DESC_CTL1_LEN;
if (ctl1 & BGMAC_DESC_CTL0_SOF)
/* Unmap no longer used buffer */
dma_unmap_single(dma_dev, slot->dma_addr, len,
@@ -284,10 +280,8 @@ static void bgmac_dma_tx_free(struct bgm
slot->skb = NULL;
}
-next:
slot->dma_addr = 0;
- if (++ring->start >= BGMAC_TX_RING_SLOTS)
- ring->start = 0;
+ ring->start++;
freed = true;
}
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -414,10 +414,10 @@ enum bgmac_dma_ring_type {
* empty.
*/
struct bgmac_dma_ring {
- u16 num_slots;
- u16 start;
- u16 end;
+ u32 start;
+ u32 end;
+ u16 num_slots;
u16 mmio_base;
struct bgmac_dma_desc *cpu_base;
dma_addr_t dma_base;

87
target/linux/generic/pending-3.18/077-05-bgmac-leave-interrupts-disabled-as-long-as-there-is-.patch
View File

@ -1,87 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 10:08:04 +0200
Subject: [PATCH] bgmac: leave interrupts disabled as long as there is work
to do
Always poll rx and tx during NAPI poll instead of relying on the status
of the first interrupt. This prevents bgmac_poll from leaving unfinished
work around until the next IRQ.
In my tests this makes bridging/routing throughput under heavy load more
stable and ensures that no new IRQs arrive as long as bgmac_poll uses up
the entire budget.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -1109,8 +1109,6 @@ static void bgmac_chip_reset(struct bgma
bgmac_phy_init(bgmac);
netdev_reset_queue(bgmac->net_dev);
-
- bgmac->int_status = 0;
}
static void bgmac_chip_intrs_on(struct bgmac *bgmac)
@@ -1225,14 +1223,13 @@ static irqreturn_t bgmac_interrupt(int i
if (!int_status)
return IRQ_NONE;
- /* Ack */
- bgmac_write(bgmac, BGMAC_INT_STATUS, int_status);
+ int_status &= ~(BGMAC_IS_TX0 | BGMAC_IS_RX);
+ if (int_status)
+ bgmac_err(bgmac, "Unknown IRQs: 0x%08X\n", int_status);
/* Disable new interrupts until handling existing ones */
bgmac_chip_intrs_off(bgmac);
- bgmac->int_status = int_status;
-
napi_schedule(&bgmac->napi);
return IRQ_HANDLED;
@@ -1241,25 +1238,17 @@ static irqreturn_t bgmac_interrupt(int i
static int bgmac_poll(struct napi_struct *napi, int weight)
{
struct bgmac *bgmac = container_of(napi, struct bgmac, napi);
- struct bgmac_dma_ring *ring;
int handled = 0;
- if (bgmac->int_status & BGMAC_IS_TX0) {
- ring = &bgmac->tx_ring[0];
- bgmac_dma_tx_free(bgmac, ring);
- bgmac->int_status &= ~BGMAC_IS_TX0;
- }
+ /* Ack */
+ bgmac_write(bgmac, BGMAC_INT_STATUS, ~0);
- if (bgmac->int_status & BGMAC_IS_RX) {
- ring = &bgmac->rx_ring[0];
- handled += bgmac_dma_rx_read(bgmac, ring, weight);
- bgmac->int_status &= ~BGMAC_IS_RX;
- }
+ bgmac_dma_tx_free(bgmac, &bgmac->tx_ring[0]);
+ handled += bgmac_dma_rx_read(bgmac, &bgmac->rx_ring[0], weight);
- if (bgmac->int_status) {
- bgmac_err(bgmac, "Unknown IRQs: 0x%08X\n", bgmac->int_status);
- bgmac->int_status = 0;
- }
+ /* Poll again if more events arrived in the meantime */
+ if (bgmac_read(bgmac, BGMAC_INT_STATUS) & (BGMAC_IS_TX0 | BGMAC_IS_RX))
+ return handled;
if (handled < weight) {
napi_complete(napi);
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -452,7 +452,6 @@ struct bgmac {
/* Int */
u32 int_mask;
- u32 int_status;
/* Current MAC state */
int mac_speed;

66
target/linux/generic/pending-3.18/077-06-bgmac-set-received-skb-headroom-to-NET_SKB_PAD.patch
View File

@ -1,66 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 10:13:28 +0200
Subject: [PATCH] bgmac: set received skb headroom to NET_SKB_PAD
A packet buffer offset of 30 bytes is inefficient, because the first 2
bytes end up in a different cacheline.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -346,13 +346,13 @@ static int bgmac_dma_rx_skb_for_slot(str
return -ENOMEM;
/* Poison - if everything goes fine, hardware will overwrite it */
- rx = buf;
+ rx = buf + BGMAC_RX_BUF_OFFSET;
rx->len = cpu_to_le16(0xdead);
rx->flags = cpu_to_le16(0xbeef);
/* Map skb for the DMA */
- dma_addr = dma_map_single(dma_dev, buf, BGMAC_RX_BUF_SIZE,
- DMA_FROM_DEVICE);
+ dma_addr = dma_map_single(dma_dev, buf + BGMAC_RX_BUF_OFFSET,
+ BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
if (dma_mapping_error(dma_dev, dma_addr)) {
bgmac_err(bgmac, "DMA mapping error\n");
put_page(virt_to_head_page(buf));
@@ -403,7 +403,7 @@ static int bgmac_dma_rx_read(struct bgma
while (ring->start != ring->end) {
struct device *dma_dev = bgmac->core->dma_dev;
struct bgmac_slot_info *slot = &ring->slots[ring->start];
- struct bgmac_rx_header *rx = slot->buf;
+ struct bgmac_rx_header *rx = slot->buf + BGMAC_RX_BUF_OFFSET;
struct sk_buff *skb;
void *buf = slot->buf;
u16 len, flags;
@@ -454,8 +454,10 @@ static int bgmac_dma_rx_read(struct bgma
BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
skb = build_skb(buf, BGMAC_RX_ALLOC_SIZE);
- skb_put(skb, BGMAC_RX_FRAME_OFFSET + len);
- skb_pull(skb, BGMAC_RX_FRAME_OFFSET);
+ skb_put(skb, BGMAC_RX_FRAME_OFFSET +
+ BGMAC_RX_BUF_OFFSET + len);
+ skb_pull(skb, BGMAC_RX_FRAME_OFFSET +
+ BGMAC_RX_BUF_OFFSET);
skb_checksum_none_assert(skb);
skb->protocol = eth_type_trans(skb, bgmac->net_dev);
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -360,9 +360,11 @@
#define BGMAC_RX_HEADER_LEN 28 /* Last 24 bytes are unused. Well... */
#define BGMAC_RX_FRAME_OFFSET 30 /* There are 2 unused bytes between header and real data */
+#define BGMAC_RX_BUF_OFFSET (NET_SKB_PAD + NET_IP_ALIGN - \
+ BGMAC_RX_FRAME_OFFSET)
#define BGMAC_RX_MAX_FRAME_SIZE 1536 /* Copied from b44/tg3 */
#define BGMAC_RX_BUF_SIZE (BGMAC_RX_FRAME_OFFSET + BGMAC_RX_MAX_FRAME_SIZE)
-#define BGMAC_RX_ALLOC_SIZE (SKB_DATA_ALIGN(BGMAC_RX_BUF_SIZE) + \
+#define BGMAC_RX_ALLOC_SIZE (SKB_DATA_ALIGN(BGMAC_RX_BUF_SIZE + BGMAC_RX_BUF_OFFSET) + \
SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
#define BGMAC_BFL_ENETROBO 0x0010 /* has ephy roboswitch spi */

130
target/linux/generic/pending-3.18/077-07-bgmac-simplify-rx-DMA-error-handling.patch
View File

@ -1,130 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 22:23:07 +0200
Subject: [PATCH] bgmac: simplify/optimize rx DMA error handling
Allocate a new buffer before processing the completed one. If allocation
fails, reuse the old buffer.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -386,6 +386,19 @@ static void bgmac_dma_rx_setup_desc(stru
dma_desc->ctl1 = cpu_to_le32(ctl1);
}
+static void bgmac_dma_rx_poison_buf(struct device *dma_dev,
+ struct bgmac_slot_info *slot)
+{
+ struct bgmac_rx_header *rx = slot->buf + BGMAC_RX_BUF_OFFSET;
+
+ dma_sync_single_for_cpu(dma_dev, slot->dma_addr, BGMAC_RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
+ rx->len = cpu_to_le16(0xdead);
+ rx->flags = cpu_to_le16(0xbeef);
+ dma_sync_single_for_device(dma_dev, slot->dma_addr, BGMAC_RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
+}
+
static int bgmac_dma_rx_read(struct bgmac *bgmac, struct bgmac_dma_ring *ring,
int weight)
{
@@ -406,53 +419,35 @@ static int bgmac_dma_rx_read(struct bgma
struct bgmac_rx_header *rx = slot->buf + BGMAC_RX_BUF_OFFSET;
struct sk_buff *skb;
void *buf = slot->buf;
+ dma_addr_t dma_addr = slot->dma_addr;
u16 len, flags;
- /* Unmap buffer to make it accessible to the CPU */
- dma_sync_single_for_cpu(dma_dev, slot->dma_addr,
- BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
-
- /* Get info from the header */
- len = le16_to_cpu(rx->len);
- flags = le16_to_cpu(rx->flags);
-
do {
- dma_addr_t old_dma_addr = slot->dma_addr;
- int err;
+ /* Prepare new skb as replacement */
+ if (bgmac_dma_rx_skb_for_slot(bgmac, slot)) {
+ bgmac_dma_rx_poison_buf(dma_dev, slot);
+ break;
+ }
+
+ /* Unmap buffer to make it accessible to the CPU */
+ dma_unmap_single(dma_dev, dma_addr,
+ BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
+
+ /* Get info from the header */
+ len = le16_to_cpu(rx->len);
+ flags = le16_to_cpu(rx->flags);
/* Check for poison and drop or pass the packet */
if (len == 0xdead && flags == 0xbeef) {
bgmac_err(bgmac, "Found poisoned packet at slot %d, DMA issue!\n",
ring->start);
- dma_sync_single_for_device(dma_dev,
- slot->dma_addr,
- BGMAC_RX_BUF_SIZE,
- DMA_FROM_DEVICE);
+ put_page(virt_to_head_page(buf));
break;
}
/* Omit CRC. */
len -= ETH_FCS_LEN;
- /* Prepare new skb as replacement */
- err = bgmac_dma_rx_skb_for_slot(bgmac, slot);
- if (err) {
- /* Poison the old skb */
- rx->len = cpu_to_le16(0xdead);
- rx->flags = cpu_to_le16(0xbeef);
-
- dma_sync_single_for_device(dma_dev,
- slot->dma_addr,
- BGMAC_RX_BUF_SIZE,
- DMA_FROM_DEVICE);
- break;
- }
- bgmac_dma_rx_setup_desc(bgmac, ring, ring->start);
-
- /* Unmap old skb, we'll pass it to the netfif */
- dma_unmap_single(dma_dev, old_dma_addr,
- BGMAC_RX_BUF_SIZE, DMA_FROM_DEVICE);
-
skb = build_skb(buf, BGMAC_RX_ALLOC_SIZE);
skb_put(skb, BGMAC_RX_FRAME_OFFSET +
BGMAC_RX_BUF_OFFSET + len);
@@ -465,6 +460,8 @@ static int bgmac_dma_rx_read(struct bgma
handled++;
} while (0);
+ bgmac_dma_rx_setup_desc(bgmac, ring, ring->start);
+
if (++ring->start >= BGMAC_RX_RING_SLOTS)
ring->start = 0;
@@ -532,14 +529,14 @@ static void bgmac_dma_rx_ring_free(struc
for (i = 0; i < ring->num_slots; i++) {
slot = &ring->slots[i];
- if (!slot->buf)
+ if (!slot->dma_addr)
continue;
- if (slot->dma_addr)
- dma_unmap_single(dma_dev, slot->dma_addr,
- BGMAC_RX_BUF_SIZE,
- DMA_FROM_DEVICE);
+ dma_unmap_single(dma_dev, slot->dma_addr,
+ BGMAC_RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
put_page(virt_to_head_page(slot->buf));
+ slot->dma_addr = 0;
}
}

27
target/linux/generic/pending-3.18/077-08-bgmac-add-check-for-oversized-packets.patch
View File

@ -1,27 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 22:28:20 +0200
Subject: [PATCH] bgmac: add check for oversized packets
In very rare cases, the MAC can catch an internal buffer that is bigger
than it's supposed to be. Instead of crashing the kernel, simply pass
the buffer back to the hardware
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -445,6 +445,13 @@ static int bgmac_dma_rx_read(struct bgma
break;
}
+ if (len > BGMAC_RX_ALLOC_SIZE) {
+ bgmac_err(bgmac, "Found oversized packet at slot %d, DMA issue!\n",
+ ring->start);
+ put_page(virt_to_head_page(buf));
+ break;
+ }
+
/* Omit CRC. */
len -= ETH_FCS_LEN;

23
target/linux/generic/pending-3.18/077-09-bgmac-increase-rx-ring-size-from-511-to-512.patch
View File

@ -1,23 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 22:36:16 +0200
Subject: [PATCH] bgmac: increase rx ring size from 511 to 512
Limiting it to 511 looks like a failed attempt at leaving one descriptor
empty to allow the hardware to stop processing a buffer that has not
been prepared yet. However, this doesn't work because this affects the
total ring size as well
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -356,7 +356,7 @@
#define BGMAC_MAX_RX_RINGS 1
#define BGMAC_TX_RING_SLOTS 128
-#define BGMAC_RX_RING_SLOTS 512 - 1 /* Why -1? Well, Broadcom does that... */
+#define BGMAC_RX_RING_SLOTS 512
#define BGMAC_RX_HEADER_LEN 28 /* Last 24 bytes are unused. Well... */
#define BGMAC_RX_FRAME_OFFSET 30 /* There are 2 unused bytes between header and real data */

184
target/linux/generic/pending-3.18/077-10-bgmac-simplify-dma-init-cleanup.patch
View File

@ -1,184 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 23:19:32 +0200
Subject: [PATCH] bgmac: simplify dma init/cleanup
Instead of allocating buffers at device init time and initializing
descriptors at device open, do both at the same time (during open).
Free all buffers when closing the device.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -562,18 +562,26 @@ static void bgmac_dma_ring_desc_free(str
ring->dma_base);
}
-static void bgmac_dma_free(struct bgmac *bgmac)
+static void bgmac_dma_cleanup(struct bgmac *bgmac)
{
int i;
- for (i = 0; i < BGMAC_MAX_TX_RINGS; i++) {
+ for (i = 0; i < BGMAC_MAX_TX_RINGS; i++)
bgmac_dma_tx_ring_free(bgmac, &bgmac->tx_ring[i]);
- bgmac_dma_ring_desc_free(bgmac, &bgmac->tx_ring[i]);
- }
- for (i = 0; i < BGMAC_MAX_RX_RINGS; i++) {
+
+ for (i = 0; i < BGMAC_MAX_RX_RINGS; i++)
bgmac_dma_rx_ring_free(bgmac, &bgmac->rx_ring[i]);
+}
+
+static void bgmac_dma_free(struct bgmac *bgmac)
+{
+ int i;
+
+ for (i = 0; i < BGMAC_MAX_TX_RINGS; i++)
+ bgmac_dma_ring_desc_free(bgmac, &bgmac->tx_ring[i]);
+
+ for (i = 0; i < BGMAC_MAX_RX_RINGS; i++)
bgmac_dma_ring_desc_free(bgmac, &bgmac->rx_ring[i]);
- }
}
static int bgmac_dma_alloc(struct bgmac *bgmac)
@@ -621,8 +629,6 @@ static int bgmac_dma_alloc(struct bgmac
}
for (i = 0; i < BGMAC_MAX_RX_RINGS; i++) {
- int j;
-
ring = &bgmac->rx_ring[i];
ring->num_slots = BGMAC_RX_RING_SLOTS;
ring->mmio_base = ring_base[i];
@@ -645,15 +651,6 @@ static int bgmac_dma_alloc(struct bgmac
ring->index_base = lower_32_bits(ring->dma_base);
else
ring->index_base = 0;
-
- /* Alloc RX slots */
- for (j = 0; j < ring->num_slots; j++) {
- err = bgmac_dma_rx_skb_for_slot(bgmac, &ring->slots[j]);
- if (err) {
- bgmac_err(bgmac, "Can't allocate skb for slot in RX ring\n");
- goto err_dma_free;
- }
- }
}
return 0;
@@ -663,10 +660,10 @@ err_dma_free:
return -ENOMEM;
}
-static void bgmac_dma_init(struct bgmac *bgmac)
+static int bgmac_dma_init(struct bgmac *bgmac)
{
struct bgmac_dma_ring *ring;
- int i;
+ int i, err;
for (i = 0; i < BGMAC_MAX_TX_RINGS; i++) {
ring = &bgmac->tx_ring[i];
@@ -698,8 +695,13 @@ static void bgmac_dma_init(struct bgmac
if (ring->unaligned)
bgmac_dma_rx_enable(bgmac, ring);
- for (j = 0; j < ring->num_slots; j++)
+ for (j = 0; j < ring->num_slots; j++) {
+ err = bgmac_dma_rx_skb_for_slot(bgmac, &ring->slots[j]);
+ if (err)
+ goto error;
+
bgmac_dma_rx_setup_desc(bgmac, ring, j);
+ }
bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_RX_INDEX,
ring->index_base +
@@ -708,6 +710,12 @@ static void bgmac_dma_init(struct bgmac
ring->start = 0;
ring->end = 0;
}
+
+ return 0;
+
+error:
+ bgmac_dma_cleanup(bgmac);
+ return err;
}
/**************************************************
@@ -1183,11 +1191,8 @@ static void bgmac_enable(struct bgmac *b
}
/* http://bcm-v4.sipsolutions.net/mac-gbit/gmac/chipinit */
-static void bgmac_chip_init(struct bgmac *bgmac, bool full_init)
+static void bgmac_chip_init(struct bgmac *bgmac)
{
- struct bgmac_dma_ring *ring;
- int i;
-
/* 1 interrupt per received frame */
bgmac_write(bgmac, BGMAC_INT_RECV_LAZY, 1 << BGMAC_IRL_FC_SHIFT);
@@ -1205,16 +1210,7 @@ static void bgmac_chip_init(struct bgmac
bgmac_write(bgmac, BGMAC_RXMAX_LENGTH, 32 + ETHER_MAX_LEN);
- if (full_init) {
- bgmac_dma_init(bgmac);
- if (1) /* FIXME: is there any case we don't want IRQs? */
- bgmac_chip_intrs_on(bgmac);
- } else {
- for (i = 0; i < BGMAC_MAX_RX_RINGS; i++) {
- ring = &bgmac->rx_ring[i];
- bgmac_dma_rx_enable(bgmac, ring);
- }
- }
+ bgmac_chip_intrs_on(bgmac);
bgmac_enable(bgmac);
}
@@ -1274,23 +1270,27 @@ static int bgmac_open(struct net_device
int err = 0;
bgmac_chip_reset(bgmac);
+
+ err = bgmac_dma_init(bgmac);
+ if (err)
+ return err;
+
/* Specs say about reclaiming rings here, but we do that in DMA init */
- bgmac_chip_init(bgmac, true);
+ bgmac_chip_init(bgmac);
err = request_irq(bgmac->core->irq, bgmac_interrupt, IRQF_SHARED,
KBUILD_MODNAME, net_dev);
if (err < 0) {
bgmac_err(bgmac, "IRQ request error: %d!\n", err);
- goto err_out;
+ bgmac_dma_cleanup(bgmac);
+ return err;
}
napi_enable(&bgmac->napi);
phy_start(bgmac->phy_dev);
netif_carrier_on(net_dev);
-
-err_out:
- return err;
+ return 0;
}
static int bgmac_stop(struct net_device *net_dev)
@@ -1306,6 +1306,7 @@ static int bgmac_stop(struct net_device
free_irq(bgmac->core->irq, net_dev);
bgmac_chip_reset(bgmac);
+ bgmac_dma_cleanup(bgmac);
return 0;
}

88
target/linux/generic/pending-3.18/077-11-bgmac-fix-DMA-rx-corruption.patch
View File

@ -1,88 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 11:59:47 +0200
Subject: [PATCH] bgmac: fix DMA rx corruption
The driver needs to inform the hardware about the first invalid (not yet
filled) rx slot, by writing its DMA descriptor pointer offset to the
BGMAC_DMA_RX_INDEX register.
This register was set to a value exceeding the rx ring size, effectively
allowing the hardware constant access to the full ring, regardless of
which slots are initialized.
To fix this issue, always mark the last filled rx slot as invalid.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -366,6 +366,16 @@ static int bgmac_dma_rx_skb_for_slot(str
return 0;
}
+static void bgmac_dma_rx_update_index(struct bgmac *bgmac,
+ struct bgmac_dma_ring *ring)
+{
+ wmb();
+
+ bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_RX_INDEX,
+ ring->index_base +
+ ring->end * sizeof(struct bgmac_dma_desc));
+}
+
static void bgmac_dma_rx_setup_desc(struct bgmac *bgmac,
struct bgmac_dma_ring *ring, int desc_idx)
{
@@ -384,6 +394,8 @@ static void bgmac_dma_rx_setup_desc(stru
dma_desc->addr_high = cpu_to_le32(upper_32_bits(ring->slots[desc_idx].dma_addr));
dma_desc->ctl0 = cpu_to_le32(ctl0);
dma_desc->ctl1 = cpu_to_le32(ctl1);
+
+ ring->end = desc_idx;
}
static void bgmac_dma_rx_poison_buf(struct device *dma_dev,
@@ -411,9 +423,7 @@ static int bgmac_dma_rx_read(struct bgma
end_slot &= BGMAC_DMA_RX_STATDPTR;
end_slot /= sizeof(struct bgmac_dma_desc);
- ring->end = end_slot;
-
- while (ring->start != ring->end) {
+ while (ring->start != end_slot) {
struct device *dma_dev = bgmac->core->dma_dev;
struct bgmac_slot_info *slot = &ring->slots[ring->start];
struct bgmac_rx_header *rx = slot->buf + BGMAC_RX_BUF_OFFSET;
@@ -476,6 +486,8 @@ static int bgmac_dma_rx_read(struct bgma
break;
}
+ bgmac_dma_rx_update_index(bgmac, ring);
+
return handled;
}
@@ -695,6 +707,8 @@ static int bgmac_dma_init(struct bgmac *
if (ring->unaligned)
bgmac_dma_rx_enable(bgmac, ring);
+ ring->start = 0;
+ ring->end = 0;
for (j = 0; j < ring->num_slots; j++) {
err = bgmac_dma_rx_skb_for_slot(bgmac, &ring->slots[j]);
if (err)
@@ -703,12 +717,7 @@ static int bgmac_dma_init(struct bgmac *
bgmac_dma_rx_setup_desc(bgmac, ring, j);
}
- bgmac_write(bgmac, ring->mmio_base + BGMAC_DMA_RX_INDEX,
- ring->index_base +
- ring->num_slots * sizeof(struct bgmac_dma_desc));
-
- ring->start = 0;
- ring->end = 0;
+ bgmac_dma_rx_update_index(bgmac, ring);
}
return 0;

132
target/linux/generic/pending-3.18/077-12-bgmac-drop-ring-num_slots.patch
View File

@ -1,132 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sun, 12 Apr 2015 23:28:38 +0200
Subject: [PATCH] bgmac: drop ring->num_slots
The ring size is always known at compile time, so make the code a bit
more efficient
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -123,7 +123,7 @@ bgmac_dma_tx_add_buf(struct bgmac *bgmac
struct bgmac_dma_desc *dma_desc;
u32 ctl1;
- if (i == ring->num_slots - 1)
+ if (i == BGMAC_TX_RING_SLOTS - 1)
ctl0 |= BGMAC_DESC_CTL0_EOT;
ctl1 = len & BGMAC_DESC_CTL1_LEN;
@@ -382,7 +382,7 @@ static void bgmac_dma_rx_setup_desc(stru
struct bgmac_dma_desc *dma_desc = ring->cpu_base + desc_idx;
u32 ctl0 = 0, ctl1 = 0;
- if (desc_idx == ring->num_slots - 1)
+ if (desc_idx == BGMAC_RX_RING_SLOTS - 1)
ctl0 |= BGMAC_DESC_CTL0_EOT;
ctl1 |= BGMAC_RX_BUF_SIZE & BGMAC_DESC_CTL1_LEN;
/* Is there any BGMAC device that requires extension? */
@@ -521,7 +521,7 @@ static void bgmac_dma_tx_ring_free(struc
struct bgmac_slot_info *slot;
int i;
- for (i = 0; i < ring->num_slots; i++) {
+ for (i = 0; i < BGMAC_TX_RING_SLOTS; i++) {
int len = dma_desc[i].ctl1 & BGMAC_DESC_CTL1_LEN;
slot = &ring->slots[i];
@@ -546,7 +546,7 @@ static void bgmac_dma_rx_ring_free(struc
struct bgmac_slot_info *slot;
int i;
- for (i = 0; i < ring->num_slots; i++) {
+ for (i = 0; i < BGMAC_RX_RING_SLOTS; i++) {
slot = &ring->slots[i];
if (!slot->dma_addr)
continue;
@@ -560,7 +560,8 @@ static void bgmac_dma_rx_ring_free(struc
}
static void bgmac_dma_ring_desc_free(struct bgmac *bgmac,
- struct bgmac_dma_ring *ring)
+ struct bgmac_dma_ring *ring,
+ int num_slots)
{
struct device *dma_dev = bgmac->core->dma_dev;
int size;
@@ -569,7 +570,7 @@ static void bgmac_dma_ring_desc_free(str
return;
/* Free ring of descriptors */
- size = ring->num_slots * sizeof(struct bgmac_dma_desc);
+ size = num_slots * sizeof(struct bgmac_dma_desc);
dma_free_coherent(dma_dev, size, ring->cpu_base,
ring->dma_base);
}
@@ -590,10 +591,12 @@ static void bgmac_dma_free(struct bgmac
int i;
for (i = 0; i < BGMAC_MAX_TX_RINGS; i++)
- bgmac_dma_ring_desc_free(bgmac, &bgmac->tx_ring[i]);
+ bgmac_dma_ring_desc_free(bgmac, &bgmac->tx_ring[i],
+ BGMAC_TX_RING_SLOTS);
for (i = 0; i < BGMAC_MAX_RX_RINGS; i++)
- bgmac_dma_ring_desc_free(bgmac, &bgmac->rx_ring[i]);
+ bgmac_dma_ring_desc_free(bgmac, &bgmac->rx_ring[i],
+ BGMAC_RX_RING_SLOTS);
}
static int bgmac_dma_alloc(struct bgmac *bgmac)
@@ -616,11 +619,10 @@ static int bgmac_dma_alloc(struct bgmac
for (i = 0; i < BGMAC_MAX_TX_RINGS; i++) {
ring = &bgmac->tx_ring[i];
- ring->num_slots = BGMAC_TX_RING_SLOTS;
ring->mmio_base = ring_base[i];
/* Alloc ring of descriptors */
- size = ring->num_slots * sizeof(struct bgmac_dma_desc);
+ size = BGMAC_TX_RING_SLOTS * sizeof(struct bgmac_dma_desc);
ring->cpu_base = dma_zalloc_coherent(dma_dev, size,
&ring->dma_base,
GFP_KERNEL);
@@ -642,11 +644,10 @@ static int bgmac_dma_alloc(struct bgmac
for (i = 0; i < BGMAC_MAX_RX_RINGS; i++) {
ring = &bgmac->rx_ring[i];
- ring->num_slots = BGMAC_RX_RING_SLOTS;
ring->mmio_base = ring_base[i];
/* Alloc ring of descriptors */
- size = ring->num_slots * sizeof(struct bgmac_dma_desc);
+ size = BGMAC_RX_RING_SLOTS * sizeof(struct bgmac_dma_desc);
ring->cpu_base = dma_zalloc_coherent(dma_dev, size,
&ring->dma_base,
GFP_KERNEL);
@@ -709,7 +710,7 @@ static int bgmac_dma_init(struct bgmac *
ring->start = 0;
ring->end = 0;
- for (j = 0; j < ring->num_slots; j++) {
+ for (j = 0; j < BGMAC_RX_RING_SLOTS; j++) {
err = bgmac_dma_rx_skb_for_slot(bgmac, &ring->slots[j]);
if (err)
goto error;
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -419,11 +419,10 @@ struct bgmac_dma_ring {
u32 start;
u32 end;
- u16 num_slots;
- u16 mmio_base;
struct bgmac_dma_desc *cpu_base;
dma_addr_t dma_base;
u32 index_base; /* Used for unaligned rings only, otherwise 0 */
+ u16 mmio_base;
bool unaligned;
struct bgmac_slot_info slots[BGMAC_RX_RING_SLOTS];

31
target/linux/generic/pending-3.18/078-bgmac-reset-enable-Ethernet-core-before-using-it.patch
View File

@ -1,31 +0,0 @@
From b4dfd8e92956b396d3438212bc9a0be6267b8b34 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <zajec5@gmail.com>
Date: Tue, 12 Apr 2016 13:30:45 +0200
Subject: [PATCH] bgmac: reset & enable Ethernet core before using it
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
This fixes Ethernet on D-Link DIR-885L with BCM47094 SoC. Felix reported
similar fix was needed for his BCM4709 device (Buffalo WXR-1900DHP?).
I tested this for regressions on BCM4706, BCM4708A0 and BCM47081A0.
Cc: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ethernet/broadcom/bgmac.c
+++ b/drivers/net/ethernet/broadcom/bgmac.c
@@ -1569,6 +1569,11 @@ static int bgmac_probe(struct bcma_devic
*/
bcma_core_enable(core, 0);
+ /* This (reset &) enable is not preset in specs or reference driver but
+ * Broadcom does it in arch PCI code when enabling fake PCI device.
+ */
+ bcma_core_enable(core, 0);
+
/* Allocation and references */
net_dev = alloc_etherdev(sizeof(*bgmac));
if (!net_dev)

34
target/linux/generic/pending-3.18/079-bgmac-fix-MAC-soft-reset-bit-for-corerev-4.patch
View File

@ -1,34 +0,0 @@
From c02bc350f9dbce7d637c394a6e1c4d29dc5b28b2 Mon Sep 17 00:00:00 2001
From: Felix Fietkau <nbd@nbd.name>
Date: Tue, 12 Apr 2016 18:27:29 +0200
Subject: [PATCH] bgmac: fix MAC soft-reset bit for corerev > 4
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Only core revisions older than 4 use BGMAC_CMDCFG_SR_REV0. This mainly
fixes support for BCM4708A0KF SoCs with Ethernet core rev 5 (it means
only some devices as most of BCM4708A0KF-s got core rev 4).
This was tested for regressions on BCM47094 which doesn't seem to care
which bit gets used.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ethernet/broadcom/bgmac.h
+++ b/drivers/net/ethernet/broadcom/bgmac.h
@@ -198,9 +198,9 @@
#define BGMAC_CMDCFG_TAI 0x00000200
#define BGMAC_CMDCFG_HD 0x00000400 /* Set if in half duplex mode */
#define BGMAC_CMDCFG_HD_SHIFT 10
-#define BGMAC_CMDCFG_SR_REV0 0x00000800 /* Set to reset mode, for other revs */
-#define BGMAC_CMDCFG_SR_REV4 0x00002000 /* Set to reset mode, only for core rev 4 */
-#define BGMAC_CMDCFG_SR(rev) ((rev == 4) ? BGMAC_CMDCFG_SR_REV4 : BGMAC_CMDCFG_SR_REV0)
+#define BGMAC_CMDCFG_SR_REV0 0x00000800 /* Set to reset mode, for core rev 0-3 */
+#define BGMAC_CMDCFG_SR_REV4 0x00002000 /* Set to reset mode, for core rev >= 4 */
+#define BGMAC_CMDCFG_SR(rev) ((rev >= 4) ? BGMAC_CMDCFG_SR_REV4 : BGMAC_CMDCFG_SR_REV0)
#define BGMAC_CMDCFG_ML 0x00008000 /* Set to activate mac loopback mode */
#define BGMAC_CMDCFG_AE 0x00400000
#define BGMAC_CMDCFG_CFE 0x00800000

46
target/linux/generic/pending-3.18/080-00-fib_trie-Fix-proc-net-fib_trie-when-CONFIG_IP_MULTIP.patch
View File

@ -1,46 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Tue, 2 Dec 2014 10:58:21 -0800
Subject: [PATCH] fib_trie: Fix /proc/net/fib_trie when
CONFIG_IP_MULTIPLE_TABLES is not defined
In recent testing I had disabled CONFIG_IP_MULTIPLE_TABLES and as a result
when I ran "cat /proc/net/fib_trie" the main trie was displayed multiple
times. I found that the problem line of code was in the function
fib_trie_seq_next. Specifically the line below caused the indexes to go in
the opposite direction of our traversal:
h = tb->tb_id & (FIB_TABLE_HASHSZ - 1);
This issue was that the RT tables are defined such that RT_TABLE_LOCAL is ID
255, while it is located at TABLE_LOCAL_INDEX of 0, and RT_TABLE_MAIN is 254
with a TABLE_MAIN_INDEX of 1. This means that the above line will return 1
for the local table and 0 for main. The result is that fib_trie_seq_next
will return NULL at the end of the local table, fib_trie_seq_start will
return the start of the main table, and then fib_trie_seq_next will loop on
main forever as h will always return 0.
The fix for this is to reverse the ordering of the two tables. It has the
advantage of making it so that the tables now print in the same order
regardless of if multiple tables are enabled or not. In order to make the
definition consistent with the multiple tables case I simply masked the to
RT_TABLE_XXX values by (FIB_TABLE_HASHSZ - 1). This way the two table
layouts should always stay consistent.
Fixes: 93456b6 ("[IPV4]: Unify access to the routing tables")
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -201,8 +201,8 @@ void fib_free_table(struct fib_table *tb
#ifndef CONFIG_IP_MULTIPLE_TABLES
-#define TABLE_LOCAL_INDEX 0
-#define TABLE_MAIN_INDEX 1
+#define TABLE_LOCAL_INDEX (RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1))
+#define TABLE_MAIN_INDEX (RT_TABLE_MAIN & (FIB_TABLE_HASHSZ - 1))
static inline struct fib_table *fib_get_table(struct net *net, u32 id)
{

72
target/linux/generic/pending-3.18/080-01-fib_trie-Fix-trie-balancing-issue-if-new-node-pushes.patch
View File

@ -1,72 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 10 Dec 2014 21:49:22 -0800
Subject: [PATCH] fib_trie: Fix trie balancing issue if new node pushes down
existing node
This patch addresses an issue with the level compression of the fib_trie.
Specifically in the case of adding a new leaf that triggers a new node to
be added that takes the place of the old node. The result is a trie where
the 1 child tnode is on one side and one leaf is on the other which gives
you a very deep trie. Below is the script I used to generate a trie on
dummy0 with a 10.X.X.X family of addresses.
ip link add type dummy
ipval=184549374
bit=2
for i in `seq 1 23`
do
ifconfig dummy0:$bit $ipval/8
ipval=`expr $ipval - $bit`
bit=`expr $bit \* 2`
done
cat /proc/net/fib_triestat
Running the script before the patch:
Local:
Aver depth: 10.82
Max depth: 23
Leaves: 29
Prefixes: 30
Internal nodes: 27
1: 26 2: 1
Pointers: 56
Null ptrs: 1
Total size: 5 kB
After applying the patch and repeating:
Local:
Aver depth: 4.72
Max depth: 9
Leaves: 29
Prefixes: 30
Internal nodes: 12
1: 3 2: 2 3: 7
Pointers: 70
Null ptrs: 30
Total size: 4 kB
What this fix does is start the rebalance at the newly created tnode
instead of at the parent tnode. This way if there is a gap between the
parent and the new node it doesn't prevent the new tnode from being
coalesced with any pre-existing nodes that may have been pushed into one
of the new nodes child branches.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1143,8 +1143,9 @@ static struct list_head *fib_insert_node
put_child(tp, cindex, (struct rt_trie_node *)tn);
} else {
rcu_assign_pointer(t->trie, (struct rt_trie_node *)tn);
- tp = tn;
}
+
+ tp = tn;
}
if (tp && tp->pos + tp->bits > 32)

200
target/linux/generic/pending-3.18/080-02-fib_trie-Update-usage-stats-to-be-percpu-instead-of-.patch
View File

@ -1,200 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:55:29 -0800
Subject: [PATCH] fib_trie: Update usage stats to be percpu instead of
global variables
The trie usage stats were currently being shared by all threads that were
calling fib_table_lookup. As a result when multiple threads were
performing lookups simultaneously the trie would begin to cache bounce
between those threads.
In order to prevent this I have updated the usage stats to use a set of
percpu variables. By doing this we should be able to avoid the cache
bouncing and still make use of these stats.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -67,7 +67,7 @@ static int __net_init fib4_rules_init(st
return 0;
fail:
- kfree(local_table);
+ fib_free_table(local_table);
return -ENOMEM;
}
#else
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -153,7 +153,7 @@ struct trie_stat {
struct trie {
struct rt_trie_node __rcu *trie;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- struct trie_use_stats stats;
+ struct trie_use_stats __percpu *stats;
#endif
};
@@ -631,7 +631,7 @@ static struct rt_trie_node *resize(struc
if (IS_ERR(tn)) {
tn = old_tn;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.resize_node_skipped++;
+ this_cpu_inc(t->stats->resize_node_skipped);
#endif
break;
}
@@ -658,7 +658,7 @@ static struct rt_trie_node *resize(struc
if (IS_ERR(tn)) {
tn = old_tn;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.resize_node_skipped++;
+ this_cpu_inc(t->stats->resize_node_skipped);
#endif
break;
}
@@ -1357,7 +1357,7 @@ static int check_leaf(struct fib_table *
err = fib_props[fa->fa_type].error;
if (err) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.semantic_match_passed++;
+ this_cpu_inc(t->stats->semantic_match_passed);
#endif
return err;
}
@@ -1372,7 +1372,7 @@ static int check_leaf(struct fib_table *
continue;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.semantic_match_passed++;
+ this_cpu_inc(t->stats->semantic_match_passed);
#endif
res->prefixlen = li->plen;
res->nh_sel = nhsel;
@@ -1388,7 +1388,7 @@ static int check_leaf(struct fib_table *
}
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.semantic_match_miss++;
+ this_cpu_inc(t->stats->semantic_match_miss);
#endif
}
@@ -1399,6 +1399,9 @@ int fib_table_lookup(struct fib_table *t
struct fib_result *res, int fib_flags)
{
struct trie *t = (struct trie *) tb->tb_data;
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ struct trie_use_stats __percpu *stats = t->stats;
+#endif
int ret;
struct rt_trie_node *n;
struct tnode *pn;
@@ -1417,7 +1420,7 @@ int fib_table_lookup(struct fib_table *t
goto failed;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.gets++;
+ this_cpu_inc(stats->gets);
#endif
/* Just a leaf? */
@@ -1441,7 +1444,7 @@ int fib_table_lookup(struct fib_table *t
if (n == NULL) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.null_node_hit++;
+ this_cpu_inc(stats->null_node_hit);
#endif
goto backtrace;
}
@@ -1576,7 +1579,7 @@ backtrace:
chopped_off = 0;
#ifdef CONFIG_IP_FIB_TRIE_STATS
- t->stats.backtrack++;
+ this_cpu_inc(stats->backtrack);
#endif
goto backtrace;
}
@@ -1830,6 +1833,11 @@ int fib_table_flush(struct fib_table *tb
void fib_free_table(struct fib_table *tb)
{
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ struct trie *t = (struct trie *)tb->tb_data;
+
+ free_percpu(t->stats);
+#endif /* CONFIG_IP_FIB_TRIE_STATS */
kfree(tb);
}
@@ -1973,7 +1981,14 @@ struct fib_table *fib_trie_table(u32 id)
tb->tb_num_default = 0;
t = (struct trie *) tb->tb_data;
- memset(t, 0, sizeof(*t));
+ RCU_INIT_POINTER(t->trie, NULL);
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ t->stats = alloc_percpu(struct trie_use_stats);
+ if (!t->stats) {
+ kfree(tb);
+ tb = NULL;
+ }
+#endif
return tb;
}
@@ -2139,18 +2154,31 @@ static void trie_show_stats(struct seq_f
#ifdef CONFIG_IP_FIB_TRIE_STATS
static void trie_show_usage(struct seq_file *seq,
- const struct trie_use_stats *stats)
+ const struct trie_use_stats __percpu *stats)
{
+ struct trie_use_stats s = { 0 };
+ int cpu;
+
+ /* loop through all of the CPUs and gather up the stats */
+ for_each_possible_cpu(cpu) {
+ const struct trie_use_stats *pcpu = per_cpu_ptr(stats, cpu);
+
+ s.gets += pcpu->gets;
+ s.backtrack += pcpu->backtrack;
+ s.semantic_match_passed += pcpu->semantic_match_passed;
+ s.semantic_match_miss += pcpu->semantic_match_miss;
+ s.null_node_hit += pcpu->null_node_hit;
+ s.resize_node_skipped += pcpu->resize_node_skipped;
+ }
+
seq_printf(seq, "\nCounters:\n---------\n");
- seq_printf(seq, "gets = %u\n", stats->gets);
- seq_printf(seq, "backtracks = %u\n", stats->backtrack);
+ seq_printf(seq, "gets = %u\n", s.gets);
+ seq_printf(seq, "backtracks = %u\n", s.backtrack);
seq_printf(seq, "semantic match passed = %u\n",
- stats->semantic_match_passed);
- seq_printf(seq, "semantic match miss = %u\n",
- stats->semantic_match_miss);
- seq_printf(seq, "null node hit= %u\n", stats->null_node_hit);
- seq_printf(seq, "skipped node resize = %u\n\n",
- stats->resize_node_skipped);
+ s.semantic_match_passed);
+ seq_printf(seq, "semantic match miss = %u\n", s.semantic_match_miss);
+ seq_printf(seq, "null node hit= %u\n", s.null_node_hit);
+ seq_printf(seq, "skipped node resize = %u\n\n", s.resize_node_skipped);
}
#endif /* CONFIG_IP_FIB_TRIE_STATS */
@@ -2191,7 +2219,7 @@ static int fib_triestat_seq_show(struct
trie_collect_stats(t, &stat);
trie_show_stats(seq, &stat);
#ifdef CONFIG_IP_FIB_TRIE_STATS
- trie_show_usage(seq, &t->stats);
+ trie_show_usage(seq, t->stats);
#endif
}
}

421
target/linux/generic/pending-3.18/080-03-fib_trie-Make-leaf-and-tnode-more-uniform.patch
View File

@ -1,421 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:55:35 -0800
Subject: [PATCH] fib_trie: Make leaf and tnode more uniform
This change makes some fundamental changes to the way leaves and tnodes are
constructed. The big differences are:
1. Leaves now populate pos and bits indicating their full key size.
2. Trie nodes now mask out their lower bits to be consistent with the leaf
3. Both structures have been reordered so that rt_trie_node now consisists
of a much larger region including the pos, bits, and rcu portions of
the tnode structure.
On 32b systems this will result in the leaf being 4B larger as the pos and
bits values were added to a hole created by the key as it was only 4B in
length.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -87,24 +87,38 @@
typedef unsigned int t_key;
-#define T_TNODE 0
-#define T_LEAF 1
-#define NODE_TYPE_MASK 0x1UL
-#define NODE_TYPE(node) ((node)->parent & NODE_TYPE_MASK)
+#define IS_TNODE(n) ((n)->bits)
+#define IS_LEAF(n) (!(n)->bits)
-#define IS_TNODE(n) (!(n->parent & T_LEAF))
-#define IS_LEAF(n) (n->parent & T_LEAF)
+struct tnode {
+ t_key key;
+ unsigned char bits; /* 2log(KEYLENGTH) bits needed */
+ unsigned char pos; /* 2log(KEYLENGTH) bits needed */
+ struct tnode __rcu *parent;
+ union {
+ struct rcu_head rcu;
+ struct tnode *tnode_free;
+ };
+ unsigned int full_children; /* KEYLENGTH bits needed */
+ unsigned int empty_children; /* KEYLENGTH bits needed */
+ struct rt_trie_node __rcu *child[0];
+};
struct rt_trie_node {
- unsigned long parent;
t_key key;
+ unsigned char bits;
+ unsigned char pos;
+ struct tnode __rcu *parent;
+ struct rcu_head rcu;
};
struct leaf {
- unsigned long parent;
t_key key;
- struct hlist_head list;
+ unsigned char bits;
+ unsigned char pos;
+ struct tnode __rcu *parent;
struct rcu_head rcu;
+ struct hlist_head list;
};
struct leaf_info {
@@ -115,20 +129,6 @@ struct leaf_info {
struct rcu_head rcu;
};
-struct tnode {
- unsigned long parent;
- t_key key;
- unsigned char pos; /* 2log(KEYLENGTH) bits needed */
- unsigned char bits; /* 2log(KEYLENGTH) bits needed */
- unsigned int full_children; /* KEYLENGTH bits needed */
- unsigned int empty_children; /* KEYLENGTH bits needed */
- union {
- struct rcu_head rcu;
- struct tnode *tnode_free;
- };
- struct rt_trie_node __rcu *child[0];
-};
-
#ifdef CONFIG_IP_FIB_TRIE_STATS
struct trie_use_stats {
unsigned int gets;
@@ -176,38 +176,27 @@ static const int sync_pages = 128;
static struct kmem_cache *fn_alias_kmem __read_mostly;
static struct kmem_cache *trie_leaf_kmem __read_mostly;
-/*
- * caller must hold RTNL
- */
-static inline struct tnode *node_parent(const struct rt_trie_node *node)
-{
- unsigned long parent;
+/* caller must hold RTNL */
+#define node_parent(n) rtnl_dereference((n)->parent)
- parent = rcu_dereference_index_check(node->parent, lockdep_rtnl_is_held());
+/* caller must hold RCU read lock or RTNL */
+#define node_parent_rcu(n) rcu_dereference_rtnl((n)->parent)
- return (struct tnode *)(parent & ~NODE_TYPE_MASK);
-}
-
-/*
- * caller must hold RCU read lock or RTNL
- */
-static inline struct tnode *node_parent_rcu(const struct rt_trie_node *node)
+/* wrapper for rcu_assign_pointer */
+static inline void node_set_parent(struct rt_trie_node *node, struct tnode *ptr)
{
- unsigned long parent;
-
- parent = rcu_dereference_index_check(node->parent, rcu_read_lock_held() ||
- lockdep_rtnl_is_held());
-
- return (struct tnode *)(parent & ~NODE_TYPE_MASK);
+ if (node)
+ rcu_assign_pointer(node->parent, ptr);
}
-/* Same as rcu_assign_pointer
- * but that macro() assumes that value is a pointer.
+#define NODE_INIT_PARENT(n, p) RCU_INIT_POINTER((n)->parent, p)
+
+/* This provides us with the number of children in this node, in the case of a
+ * leaf this will return 0 meaning none of the children are accessible.
*/
-static inline void node_set_parent(struct rt_trie_node *node, struct tnode *ptr)
+static inline int tnode_child_length(const struct tnode *tn)
{
- smp_wmb();
- node->parent = (unsigned long)ptr | NODE_TYPE(node);
+ return (1ul << tn->bits) & ~(1ul);
}
/*
@@ -215,7 +204,7 @@ static inline void node_set_parent(struc
*/
static inline struct rt_trie_node *tnode_get_child(const struct tnode *tn, unsigned int i)
{
- BUG_ON(i >= 1U << tn->bits);
+ BUG_ON(i >= tnode_child_length(tn));
return rtnl_dereference(tn->child[i]);
}
@@ -225,16 +214,11 @@ static inline struct rt_trie_node *tnode
*/
static inline struct rt_trie_node *tnode_get_child_rcu(const struct tnode *tn, unsigned int i)
{
- BUG_ON(i >= 1U << tn->bits);
+ BUG_ON(i >= tnode_child_length(tn));
return rcu_dereference_rtnl(tn->child[i]);
}
-static inline int tnode_child_length(const struct tnode *tn)
-{
- return 1 << tn->bits;
-}
-
static inline t_key mask_pfx(t_key k, unsigned int l)
{
return (l == 0) ? 0 : k >> (KEYLENGTH-l) << (KEYLENGTH-l);
@@ -336,11 +320,6 @@ static inline int tkey_mismatch(t_key a,
*/
-static inline void check_tnode(const struct tnode *tn)
-{
- WARN_ON(tn && tn->pos+tn->bits > 32);
-}
-
static const int halve_threshold = 25;
static const int inflate_threshold = 50;
static const int halve_threshold_root = 15;
@@ -426,11 +405,20 @@ static void tnode_free_flush(void)
}
}
-static struct leaf *leaf_new(void)
+static struct leaf *leaf_new(t_key key)
{
struct leaf *l = kmem_cache_alloc(trie_leaf_kmem, GFP_KERNEL);
if (l) {
- l->parent = T_LEAF;
+ l->parent = NULL;
+ /* set key and pos to reflect full key value
+ * any trailing zeros in the key should be ignored
+ * as the nodes are searched
+ */
+ l->key = key;
+ l->pos = KEYLENGTH;
+ /* set bits to 0 indicating we are not a tnode */
+ l->bits = 0;
+
INIT_HLIST_HEAD(&l->list);
}
return l;
@@ -451,12 +439,16 @@ static struct tnode *tnode_new(t_key key
{
size_t sz = sizeof(struct tnode) + (sizeof(struct rt_trie_node *) << bits);
struct tnode *tn = tnode_alloc(sz);
+ unsigned int shift = pos + bits;
+
+ /* verify bits and pos their msb bits clear and values are valid */
+ BUG_ON(!bits || (shift > KEYLENGTH));
if (tn) {
- tn->parent = T_TNODE;
+ tn->parent = NULL;
tn->pos = pos;
tn->bits = bits;
- tn->key = key;
+ tn->key = mask_pfx(key, pos);
tn->full_children = 0;
tn->empty_children = 1<<bits;
}
@@ -473,10 +465,7 @@ static struct tnode *tnode_new(t_key key
static inline int tnode_full(const struct tnode *tn, const struct rt_trie_node *n)
{
- if (n == NULL || IS_LEAF(n))
- return 0;
-
- return ((struct tnode *) n)->pos == tn->pos + tn->bits;
+ return n && IS_TNODE(n) && (n->pos == (tn->pos + tn->bits));
}
static inline void put_child(struct tnode *tn, int i,
@@ -514,8 +503,7 @@ static void tnode_put_child_reorg(struct
else if (!wasfull && isfull)
tn->full_children++;
- if (n)
- node_set_parent(n, tn);
+ node_set_parent(n, tn);
rcu_assign_pointer(tn->child[i], n);
}
@@ -523,7 +511,7 @@ static void tnode_put_child_reorg(struct
#define MAX_WORK 10
static struct rt_trie_node *resize(struct trie *t, struct tnode *tn)
{
- int i;
+ struct rt_trie_node *n = NULL;
struct tnode *old_tn;
int inflate_threshold_use;
int halve_threshold_use;
@@ -536,12 +524,11 @@ static struct rt_trie_node *resize(struc
tn, inflate_threshold, halve_threshold);
/* No children */
- if (tn->empty_children == tnode_child_length(tn)) {
- tnode_free_safe(tn);
- return NULL;
- }
+ if (tn->empty_children > (tnode_child_length(tn) - 1))
+ goto no_children;
+
/* One child */
- if (tn->empty_children == tnode_child_length(tn) - 1)
+ if (tn->empty_children == (tnode_child_length(tn) - 1))
goto one_child;
/*
* Double as long as the resulting node has a number of
@@ -607,11 +594,9 @@ static struct rt_trie_node *resize(struc
*
*/
- check_tnode(tn);
-
/* Keep root node larger */
- if (!node_parent((struct rt_trie_node *)tn)) {
+ if (!node_parent(tn)) {
inflate_threshold_use = inflate_threshold_root;
halve_threshold_use = halve_threshold_root;
} else {
@@ -637,8 +622,6 @@ static struct rt_trie_node *resize(struc
}
}
- check_tnode(tn);
-
/* Return if at least one inflate is run */
if (max_work != MAX_WORK)
return (struct rt_trie_node *) tn;
@@ -666,21 +649,16 @@ static struct rt_trie_node *resize(struc
/* Only one child remains */
- if (tn->empty_children == tnode_child_length(tn) - 1) {
+ if (tn->empty_children == (tnode_child_length(tn) - 1)) {
+ unsigned long i;
one_child:
- for (i = 0; i < tnode_child_length(tn); i++) {
- struct rt_trie_node *n;
-
- n = rtnl_dereference(tn->child[i]);
- if (!n)
- continue;
-
- /* compress one level */
-
- node_set_parent(n, NULL);
- tnode_free_safe(tn);
- return n;
- }
+ for (i = tnode_child_length(tn); !n && i;)
+ n = tnode_get_child(tn, --i);
+no_children:
+ /* compress one level */
+ node_set_parent(n, NULL);
+ tnode_free_safe(tn);
+ return n;
}
return (struct rt_trie_node *) tn;
}
@@ -760,8 +738,7 @@ static struct tnode *inflate(struct trie
/* A leaf or an internal node with skipped bits */
- if (IS_LEAF(node) || ((struct tnode *) node)->pos >
- tn->pos + tn->bits - 1) {
+ if (IS_LEAF(node) || (node->pos > (tn->pos + tn->bits - 1))) {
put_child(tn,
tkey_extract_bits(node->key, oldtnode->pos, oldtnode->bits + 1),
node);
@@ -958,11 +935,9 @@ fib_find_node(struct trie *t, u32 key)
pos = 0;
n = rcu_dereference_rtnl(t->trie);
- while (n != NULL && NODE_TYPE(n) == T_TNODE) {
+ while (n && IS_TNODE(n)) {
tn = (struct tnode *) n;
- check_tnode(tn);
-
if (tkey_sub_equals(tn->key, pos, tn->pos-pos, key)) {
pos = tn->pos + tn->bits;
n = tnode_get_child_rcu(tn,
@@ -988,7 +963,7 @@ static void trie_rebalance(struct trie *
key = tn->key;
- while (tn != NULL && (tp = node_parent((struct rt_trie_node *)tn)) != NULL) {
+ while (tn != NULL && (tp = node_parent(tn)) != NULL) {
cindex = tkey_extract_bits(key, tp->pos, tp->bits);
wasfull = tnode_full(tp, tnode_get_child(tp, cindex));
tn = (struct tnode *)resize(t, tn);
@@ -996,7 +971,7 @@ static void trie_rebalance(struct trie *
tnode_put_child_reorg(tp, cindex,
(struct rt_trie_node *)tn, wasfull);
- tp = node_parent((struct rt_trie_node *) tn);
+ tp = node_parent(tn);
if (!tp)
rcu_assign_pointer(t->trie, (struct rt_trie_node *)tn);
@@ -1048,11 +1023,9 @@ static struct list_head *fib_insert_node
* If it doesn't, we need to replace it with a T_TNODE.
*/
- while (n != NULL && NODE_TYPE(n) == T_TNODE) {
+ while (n && IS_TNODE(n)) {
tn = (struct tnode *) n;
- check_tnode(tn);
-
if (tkey_sub_equals(tn->key, pos, tn->pos-pos, key)) {
tp = tn;
pos = tn->pos + tn->bits;
@@ -1087,12 +1060,11 @@ static struct list_head *fib_insert_node
insert_leaf_info(&l->list, li);
goto done;
}
- l = leaf_new();
+ l = leaf_new(key);
if (!l)
return NULL;
- l->key = key;
li = leaf_info_new(plen);
if (!li) {
@@ -1569,7 +1541,7 @@ backtrace:
if (chopped_off <= pn->bits) {
cindex &= ~(1 << (chopped_off-1));
} else {
- struct tnode *parent = node_parent_rcu((struct rt_trie_node *) pn);
+ struct tnode *parent = node_parent_rcu(pn);
if (!parent)
goto failed;
@@ -1597,7 +1569,7 @@ EXPORT_SYMBOL_GPL(fib_table_lookup);
*/
static void trie_leaf_remove(struct trie *t, struct leaf *l)
{
- struct tnode *tp = node_parent((struct rt_trie_node *) l);
+ struct tnode *tp = node_parent(l);
pr_debug("entering trie_leaf_remove(%p)\n", l);
@@ -2374,7 +2346,7 @@ static int fib_trie_seq_show(struct seq_
if (IS_TNODE(n)) {
struct tnode *tn = (struct tnode *) n;
- __be32 prf = htonl(mask_pfx(tn->key, tn->pos));
+ __be32 prf = htonl(tn->key);
seq_indent(seq, iter->depth-1);
seq_printf(seq, " +-- %pI4/%d %d %d %d\n",

209
target/linux/generic/pending-3.18/080-04-fib_trie-Merge-tnode_free-and-leaf_free-into-node_fr.patch
View File

@ -1,209 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:55:41 -0800
Subject: [PATCH] fib_trie: Merge tnode_free and leaf_free into node_free
Both the leaf and the tnode had an rcu_head in them, but they had them in
slightly different places. Since we now have them in the same spot and
know that any node with bits == 0 is a leaf and the rest are either vmalloc
or kmalloc tnodes depending on the value of bits it makes it easy to combine
the functions and reduce overhead.
In addition I have taken advantage of the rcu_head pointer to go ahead and
put together a simple linked list instead of using the tnode pointer as
this way we can merge either type of structure for freeing.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -95,15 +95,17 @@ struct tnode {
unsigned char bits; /* 2log(KEYLENGTH) bits needed */
unsigned char pos; /* 2log(KEYLENGTH) bits needed */
struct tnode __rcu *parent;
- union {
- struct rcu_head rcu;
- struct tnode *tnode_free;
- };
+ struct rcu_head rcu;
+ /* everything above this comment must be the same as rt_trie_node */
unsigned int full_children; /* KEYLENGTH bits needed */
unsigned int empty_children; /* KEYLENGTH bits needed */
struct rt_trie_node __rcu *child[0];
};
+/* This struct represents the shared bits between tnode and leaf. If any
+ * ordering is changed here is must also be updated in tnode and leaf as
+ * well.
+ */
struct rt_trie_node {
t_key key;
unsigned char bits;
@@ -118,6 +120,7 @@ struct leaf {
unsigned char pos;
struct tnode __rcu *parent;
struct rcu_head rcu;
+ /* everything above this comment must be the same as rt_trie_node */
struct hlist_head list;
};
@@ -163,7 +166,7 @@ static struct rt_trie_node *resize(struc
static struct tnode *inflate(struct trie *t, struct tnode *tn);
static struct tnode *halve(struct trie *t, struct tnode *tn);
/* tnodes to free after resize(); protected by RTNL */
-static struct tnode *tnode_free_head;
+static struct callback_head *tnode_free_head;
static size_t tnode_free_size;
/*
@@ -336,17 +339,23 @@ static inline void alias_free_mem_rcu(st
call_rcu(&fa->rcu, __alias_free_mem);
}
-static void __leaf_free_rcu(struct rcu_head *head)
-{
- struct leaf *l = container_of(head, struct leaf, rcu);
- kmem_cache_free(trie_leaf_kmem, l);
-}
+#define TNODE_KMALLOC_MAX \
+ ilog2((PAGE_SIZE - sizeof(struct tnode)) / sizeof(struct rt_trie_node *))
-static inline void free_leaf(struct leaf *l)
+static void __node_free_rcu(struct rcu_head *head)
{
- call_rcu(&l->rcu, __leaf_free_rcu);
+ struct rt_trie_node *n = container_of(head, struct rt_trie_node, rcu);
+
+ if (IS_LEAF(n))
+ kmem_cache_free(trie_leaf_kmem, n);
+ else if (n->bits <= TNODE_KMALLOC_MAX)
+ kfree(n);
+ else
+ vfree(n);
}
+#define node_free(n) call_rcu(&n->rcu, __node_free_rcu)
+
static inline void free_leaf_info(struct leaf_info *leaf)
{
kfree_rcu(leaf, rcu);
@@ -360,43 +369,24 @@ static struct tnode *tnode_alloc(size_t
return vzalloc(size);
}
-static void __tnode_free_rcu(struct rcu_head *head)
-{
- struct tnode *tn = container_of(head, struct tnode, rcu);
- size_t size = sizeof(struct tnode) +
- (sizeof(struct rt_trie_node *) << tn->bits);
-
- if (size <= PAGE_SIZE)
- kfree(tn);
- else
- vfree(tn);
-}
-
-static inline void tnode_free(struct tnode *tn)
-{
- if (IS_LEAF(tn))
- free_leaf((struct leaf *) tn);
- else
- call_rcu(&tn->rcu, __tnode_free_rcu);
-}
-
static void tnode_free_safe(struct tnode *tn)
{
BUG_ON(IS_LEAF(tn));
- tn->tnode_free = tnode_free_head;
- tnode_free_head = tn;
- tnode_free_size += sizeof(struct tnode) +
- (sizeof(struct rt_trie_node *) << tn->bits);
+ tn->rcu.next = tnode_free_head;
+ tnode_free_head = &tn->rcu;
}
static void tnode_free_flush(void)
{
- struct tnode *tn;
+ struct callback_head *head;
+
+ while ((head = tnode_free_head)) {
+ struct tnode *tn = container_of(head, struct tnode, rcu);
+
+ tnode_free_head = head->next;
+ tnode_free_size += offsetof(struct tnode, child[1 << tn->bits]);
- while ((tn = tnode_free_head)) {
- tnode_free_head = tn->tnode_free;
- tn->tnode_free = NULL;
- tnode_free(tn);
+ node_free(tn);
}
if (tnode_free_size >= PAGE_SIZE * sync_pages) {
@@ -437,7 +427,7 @@ static struct leaf_info *leaf_info_new(i
static struct tnode *tnode_new(t_key key, int pos, int bits)
{
- size_t sz = sizeof(struct tnode) + (sizeof(struct rt_trie_node *) << bits);
+ size_t sz = offsetof(struct tnode, child[1 << bits]);
struct tnode *tn = tnode_alloc(sz);
unsigned int shift = pos + bits;
@@ -666,15 +656,15 @@ no_children:
static void tnode_clean_free(struct tnode *tn)
{
+ struct rt_trie_node *tofree;
int i;
- struct tnode *tofree;
for (i = 0; i < tnode_child_length(tn); i++) {
- tofree = (struct tnode *)rtnl_dereference(tn->child[i]);
+ tofree = rtnl_dereference(tn->child[i]);
if (tofree)
- tnode_free(tofree);
+ node_free(tofree);
}
- tnode_free(tn);
+ node_free(tn);
}
static struct tnode *inflate(struct trie *t, struct tnode *tn)
@@ -717,7 +707,7 @@ static struct tnode *inflate(struct trie
inode->bits - 1);
if (!right) {
- tnode_free(left);
+ node_free(left);
goto nomem;
}
@@ -1068,7 +1058,7 @@ static struct list_head *fib_insert_node
li = leaf_info_new(plen);
if (!li) {
- free_leaf(l);
+ node_free(l);
return NULL;
}
@@ -1100,7 +1090,7 @@ static struct list_head *fib_insert_node
if (!tn) {
free_leaf_info(li);
- free_leaf(l);
+ node_free(l);
return NULL;
}
@@ -1580,7 +1570,7 @@ static void trie_leaf_remove(struct trie
} else
RCU_INIT_POINTER(t->trie, NULL);
- free_leaf(l);
+ node_free(l);
}
/*

928
target/linux/generic/pending-3.18/080-05-fib_trie-Merge-leaf-into-tnode.patch
View File

@ -1,928 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:55:47 -0800
Subject: [PATCH] fib_trie: Merge leaf into tnode
This change makes it so that leaf and tnode are the same struct. As a
result there is no need for rt_trie_node anymore since everyting can be
merged into tnode.
On 32b systems this results in the leaf being 4 bytes larger, however I
don't know if that is really an issue as this and an eariler patch that
added bits & pos have increased the size from 20 to 28. If I am not
mistaken slub/slab allocate on power of 2 sizes so 20 was likely being
rounded up to 32 anyway.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -96,32 +96,16 @@ struct tnode {
unsigned char pos; /* 2log(KEYLENGTH) bits needed */
struct tnode __rcu *parent;
struct rcu_head rcu;
- /* everything above this comment must be the same as rt_trie_node */
- unsigned int full_children; /* KEYLENGTH bits needed */
- unsigned int empty_children; /* KEYLENGTH bits needed */
- struct rt_trie_node __rcu *child[0];
-};
-
-/* This struct represents the shared bits between tnode and leaf. If any
- * ordering is changed here is must also be updated in tnode and leaf as
- * well.
- */
-struct rt_trie_node {
- t_key key;
- unsigned char bits;
- unsigned char pos;
- struct tnode __rcu *parent;
- struct rcu_head rcu;
-};
-
-struct leaf {
- t_key key;
- unsigned char bits;
- unsigned char pos;
- struct tnode __rcu *parent;
- struct rcu_head rcu;
- /* everything above this comment must be the same as rt_trie_node */
- struct hlist_head list;
+ union {
+ /* The fields in this struct are valid if bits > 0 (TNODE) */
+ struct {
+ unsigned int full_children; /* KEYLENGTH bits needed */
+ unsigned int empty_children; /* KEYLENGTH bits needed */
+ struct tnode __rcu *child[0];
+ };
+ /* This list pointer if valid if bits == 0 (LEAF) */
+ struct hlist_head list;
+ };
};
struct leaf_info {
@@ -154,15 +138,15 @@ struct trie_stat {
};
struct trie {
- struct rt_trie_node __rcu *trie;
+ struct tnode __rcu *trie;
#ifdef CONFIG_IP_FIB_TRIE_STATS
struct trie_use_stats __percpu *stats;
#endif
};
-static void tnode_put_child_reorg(struct tnode *tn, int i, struct rt_trie_node *n,
+static void tnode_put_child_reorg(struct tnode *tn, int i, struct tnode *n,
int wasfull);
-static struct rt_trie_node *resize(struct trie *t, struct tnode *tn);
+static struct tnode *resize(struct trie *t, struct tnode *tn);
static struct tnode *inflate(struct trie *t, struct tnode *tn);
static struct tnode *halve(struct trie *t, struct tnode *tn);
/* tnodes to free after resize(); protected by RTNL */
@@ -186,10 +170,10 @@ static struct kmem_cache *trie_leaf_kmem
#define node_parent_rcu(n) rcu_dereference_rtnl((n)->parent)
/* wrapper for rcu_assign_pointer */
-static inline void node_set_parent(struct rt_trie_node *node, struct tnode *ptr)
+static inline void node_set_parent(struct tnode *n, struct tnode *tp)
{
- if (node)
- rcu_assign_pointer(node->parent, ptr);
+ if (n)
+ rcu_assign_pointer(n->parent, tp);
}
#define NODE_INIT_PARENT(n, p) RCU_INIT_POINTER((n)->parent, p)
@@ -205,7 +189,7 @@ static inline int tnode_child_length(con
/*
* caller must hold RTNL
*/
-static inline struct rt_trie_node *tnode_get_child(const struct tnode *tn, unsigned int i)
+static inline struct tnode *tnode_get_child(const struct tnode *tn, unsigned int i)
{
BUG_ON(i >= tnode_child_length(tn));
@@ -215,7 +199,7 @@ static inline struct rt_trie_node *tnode
/*
* caller must hold RCU read lock or RTNL
*/
-static inline struct rt_trie_node *tnode_get_child_rcu(const struct tnode *tn, unsigned int i)
+static inline struct tnode *tnode_get_child_rcu(const struct tnode *tn, unsigned int i)
{
BUG_ON(i >= tnode_child_length(tn));
@@ -340,11 +324,11 @@ static inline void alias_free_mem_rcu(st
}
#define TNODE_KMALLOC_MAX \
- ilog2((PAGE_SIZE - sizeof(struct tnode)) / sizeof(struct rt_trie_node *))
+ ilog2((PAGE_SIZE - sizeof(struct tnode)) / sizeof(struct tnode *))
static void __node_free_rcu(struct rcu_head *head)
{
- struct rt_trie_node *n = container_of(head, struct rt_trie_node, rcu);
+ struct tnode *n = container_of(head, struct tnode, rcu);
if (IS_LEAF(n))
kmem_cache_free(trie_leaf_kmem, n);
@@ -395,9 +379,9 @@ static void tnode_free_flush(void)
}
}
-static struct leaf *leaf_new(t_key key)
+static struct tnode *leaf_new(t_key key)
{
- struct leaf *l = kmem_cache_alloc(trie_leaf_kmem, GFP_KERNEL);
+ struct tnode *l = kmem_cache_alloc(trie_leaf_kmem, GFP_KERNEL);
if (l) {
l->parent = NULL;
/* set key and pos to reflect full key value
@@ -444,7 +428,7 @@ static struct tnode *tnode_new(t_key key
}
pr_debug("AT %p s=%zu %zu\n", tn, sizeof(struct tnode),
- sizeof(struct rt_trie_node *) << bits);
+ sizeof(struct tnode *) << bits);
return tn;
}
@@ -453,13 +437,13 @@ static struct tnode *tnode_new(t_key key
* and no bits are skipped. See discussion in dyntree paper p. 6
*/
-static inline int tnode_full(const struct tnode *tn, const struct rt_trie_node *n)
+static inline int tnode_full(const struct tnode *tn, const struct tnode *n)
{
return n && IS_TNODE(n) && (n->pos == (tn->pos + tn->bits));
}
static inline void put_child(struct tnode *tn, int i,
- struct rt_trie_node *n)
+ struct tnode *n)
{
tnode_put_child_reorg(tn, i, n, -1);
}
@@ -469,10 +453,10 @@ static inline void put_child(struct tnod
* Update the value of full_children and empty_children.
*/
-static void tnode_put_child_reorg(struct tnode *tn, int i, struct rt_trie_node *n,
+static void tnode_put_child_reorg(struct tnode *tn, int i, struct tnode *n,
int wasfull)
{
- struct rt_trie_node *chi = rtnl_dereference(tn->child[i]);
+ struct tnode *chi = rtnl_dereference(tn->child[i]);
int isfull;
BUG_ON(i >= 1<<tn->bits);
@@ -499,10 +483,9 @@ static void tnode_put_child_reorg(struct
}
#define MAX_WORK 10
-static struct rt_trie_node *resize(struct trie *t, struct tnode *tn)
+static struct tnode *resize(struct trie *t, struct tnode *tn)
{
- struct rt_trie_node *n = NULL;
- struct tnode *old_tn;
+ struct tnode *old_tn, *n = NULL;
int inflate_threshold_use;
int halve_threshold_use;
int max_work;
@@ -614,7 +597,7 @@ static struct rt_trie_node *resize(struc
/* Return if at least one inflate is run */
if (max_work != MAX_WORK)
- return (struct rt_trie_node *) tn;
+ return tn;
/*
* Halve as long as the number of empty children in this
@@ -650,13 +633,13 @@ no_children:
tnode_free_safe(tn);
return n;
}
- return (struct rt_trie_node *) tn;
+ return tn;
}
static void tnode_clean_free(struct tnode *tn)
{
- struct rt_trie_node *tofree;
+ struct tnode *tofree;
int i;
for (i = 0; i < tnode_child_length(tn); i++) {
@@ -667,10 +650,10 @@ static void tnode_clean_free(struct tnod
node_free(tn);
}
-static struct tnode *inflate(struct trie *t, struct tnode *tn)
+static struct tnode *inflate(struct trie *t, struct tnode *oldtnode)
{
- struct tnode *oldtnode = tn;
- int olen = tnode_child_length(tn);
+ int olen = tnode_child_length(oldtnode);
+ struct tnode *tn;
int i;
pr_debug("In inflate\n");
@@ -690,11 +673,8 @@ static struct tnode *inflate(struct trie
for (i = 0; i < olen; i++) {
struct tnode *inode;
- inode = (struct tnode *) tnode_get_child(oldtnode, i);
- if (inode &&
- IS_TNODE(inode) &&
- inode->pos == oldtnode->pos + oldtnode->bits &&
- inode->bits > 1) {
+ inode = tnode_get_child(oldtnode, i);
+ if (tnode_full(oldtnode, inode) && inode->bits > 1) {
struct tnode *left, *right;
t_key m = ~0U << (KEYLENGTH - 1) >> inode->pos;
@@ -711,33 +691,29 @@ static struct tnode *inflate(struct trie
goto nomem;
}
- put_child(tn, 2*i, (struct rt_trie_node *) left);
- put_child(tn, 2*i+1, (struct rt_trie_node *) right);
+ put_child(tn, 2*i, left);
+ put_child(tn, 2*i+1, right);
}
}
for (i = 0; i < olen; i++) {
- struct tnode *inode;
- struct rt_trie_node *node = tnode_get_child(oldtnode, i);
+ struct tnode *inode = tnode_get_child(oldtnode, i);
struct tnode *left, *right;
int size, j;
/* An empty child */
- if (node == NULL)
+ if (inode == NULL)
continue;
/* A leaf or an internal node with skipped bits */
-
- if (IS_LEAF(node) || (node->pos > (tn->pos + tn->bits - 1))) {
+ if (!tnode_full(oldtnode, inode)) {
put_child(tn,
- tkey_extract_bits(node->key, oldtnode->pos, oldtnode->bits + 1),
- node);
+ tkey_extract_bits(inode->key, tn->pos, tn->bits),
+ inode);
continue;
}
/* An internal node with two children */
- inode = (struct tnode *) node;
-
if (inode->bits == 1) {
put_child(tn, 2*i, rtnl_dereference(inode->child[0]));
put_child(tn, 2*i+1, rtnl_dereference(inode->child[1]));
@@ -769,12 +745,12 @@ static struct tnode *inflate(struct trie
* bit to zero.
*/
- left = (struct tnode *) tnode_get_child(tn, 2*i);
+ left = tnode_get_child(tn, 2*i);
put_child(tn, 2*i, NULL);
BUG_ON(!left);
- right = (struct tnode *) tnode_get_child(tn, 2*i+1);
+ right = tnode_get_child(tn, 2*i+1);
put_child(tn, 2*i+1, NULL);
BUG_ON(!right);
@@ -796,12 +772,11 @@ nomem:
return ERR_PTR(-ENOMEM);
}
-static struct tnode *halve(struct trie *t, struct tnode *tn)
+static struct tnode *halve(struct trie *t, struct tnode *oldtnode)
{
- struct tnode *oldtnode = tn;
- struct rt_trie_node *left, *right;
+ int olen = tnode_child_length(oldtnode);
+ struct tnode *tn, *left, *right;
int i;
- int olen = tnode_child_length(tn);
pr_debug("In halve\n");
@@ -830,7 +805,7 @@ static struct tnode *halve(struct trie *
if (!newn)
goto nomem;
- put_child(tn, i/2, (struct rt_trie_node *)newn);
+ put_child(tn, i/2, newn);
}
}
@@ -855,7 +830,7 @@ static struct tnode *halve(struct trie *
}
/* Two nonempty children */
- newBinNode = (struct tnode *) tnode_get_child(tn, i/2);
+ newBinNode = tnode_get_child(tn, i/2);
put_child(tn, i/2, NULL);
put_child(newBinNode, 0, left);
put_child(newBinNode, 1, right);
@@ -871,7 +846,7 @@ nomem:
/* readside must use rcu_read_lock currently dump routines
via get_fa_head and dump */
-static struct leaf_info *find_leaf_info(struct leaf *l, int plen)
+static struct leaf_info *find_leaf_info(struct tnode *l, int plen)
{
struct hlist_head *head = &l->list;
struct leaf_info *li;
@@ -883,7 +858,7 @@ static struct leaf_info *find_leaf_info(
return NULL;
}
-static inline struct list_head *get_fa_head(struct leaf *l, int plen)
+static inline struct list_head *get_fa_head(struct tnode *l, int plen)
{
struct leaf_info *li = find_leaf_info(l, plen);
@@ -915,32 +890,25 @@ static void insert_leaf_info(struct hlis
/* rcu_read_lock needs to be hold by caller from readside */
-static struct leaf *
-fib_find_node(struct trie *t, u32 key)
+static struct tnode *fib_find_node(struct trie *t, u32 key)
{
- int pos;
- struct tnode *tn;
- struct rt_trie_node *n;
-
- pos = 0;
- n = rcu_dereference_rtnl(t->trie);
+ struct tnode *n = rcu_dereference_rtnl(t->trie);
+ int pos = 0;
while (n && IS_TNODE(n)) {
- tn = (struct tnode *) n;
-
- if (tkey_sub_equals(tn->key, pos, tn->pos-pos, key)) {
- pos = tn->pos + tn->bits;
- n = tnode_get_child_rcu(tn,
+ if (tkey_sub_equals(n->key, pos, n->pos-pos, key)) {
+ pos = n->pos + n->bits;
+ n = tnode_get_child_rcu(n,
tkey_extract_bits(key,
- tn->pos,
- tn->bits));
+ n->pos,
+ n->bits));
} else
break;
}
/* Case we have found a leaf. Compare prefixes */
if (n != NULL && IS_LEAF(n) && tkey_equals(key, n->key))
- return (struct leaf *)n;
+ return n;
return NULL;
}
@@ -956,14 +924,13 @@ static void trie_rebalance(struct trie *
while (tn != NULL && (tp = node_parent(tn)) != NULL) {
cindex = tkey_extract_bits(key, tp->pos, tp->bits);
wasfull = tnode_full(tp, tnode_get_child(tp, cindex));
- tn = (struct tnode *)resize(t, tn);
+ tn = resize(t, tn);
- tnode_put_child_reorg(tp, cindex,
- (struct rt_trie_node *)tn, wasfull);
+ tnode_put_child_reorg(tp, cindex, tn, wasfull);
tp = node_parent(tn);
if (!tp)
- rcu_assign_pointer(t->trie, (struct rt_trie_node *)tn);
+ rcu_assign_pointer(t->trie, tn);
tnode_free_flush();
if (!tp)
@@ -973,9 +940,9 @@ static void trie_rebalance(struct trie *
/* Handle last (top) tnode */
if (IS_TNODE(tn))
- tn = (struct tnode *)resize(t, tn);
+ tn = resize(t, tn);
- rcu_assign_pointer(t->trie, (struct rt_trie_node *)tn);
+ rcu_assign_pointer(t->trie, tn);
tnode_free_flush();
}
@@ -985,8 +952,8 @@ static struct list_head *fib_insert_node
{
int pos, newpos;
struct tnode *tp = NULL, *tn = NULL;
- struct rt_trie_node *n;
- struct leaf *l;
+ struct tnode *n;
+ struct tnode *l;
int missbit;
struct list_head *fa_head = NULL;
struct leaf_info *li;
@@ -1014,17 +981,15 @@ static struct list_head *fib_insert_node
*/
while (n && IS_TNODE(n)) {
- tn = (struct tnode *) n;
-
- if (tkey_sub_equals(tn->key, pos, tn->pos-pos, key)) {
- tp = tn;
- pos = tn->pos + tn->bits;
- n = tnode_get_child(tn,
+ if (tkey_sub_equals(n->key, pos, n->pos-pos, key)) {
+ tp = n;
+ pos = n->pos + n->bits;
+ n = tnode_get_child(n,
tkey_extract_bits(key,
- tn->pos,
- tn->bits));
+ n->pos,
+ n->bits));
- BUG_ON(n && node_parent(n) != tn);
+ BUG_ON(n && node_parent(n) != tp);
} else
break;
}
@@ -1040,14 +1005,13 @@ static struct list_head *fib_insert_node
/* Case 1: n is a leaf. Compare prefixes */
if (n != NULL && IS_LEAF(n) && tkey_equals(key, n->key)) {
- l = (struct leaf *) n;
li = leaf_info_new(plen);
if (!li)
return NULL;
fa_head = &li->falh;
- insert_leaf_info(&l->list, li);
+ insert_leaf_info(&n->list, li);
goto done;
}
l = leaf_new(key);
@@ -1068,10 +1032,10 @@ static struct list_head *fib_insert_node
if (t->trie && n == NULL) {
/* Case 2: n is NULL, and will just insert a new leaf */
- node_set_parent((struct rt_trie_node *)l, tp);
+ node_set_parent(l, tp);
cindex = tkey_extract_bits(key, tp->pos, tp->bits);
- put_child(tp, cindex, (struct rt_trie_node *)l);
+ put_child(tp, cindex, l);
} else {
/* Case 3: n is a LEAF or a TNODE and the key doesn't match. */
/*
@@ -1094,17 +1058,17 @@ static struct list_head *fib_insert_node
return NULL;
}
- node_set_parent((struct rt_trie_node *)tn, tp);
+ node_set_parent(tn, tp);
missbit = tkey_extract_bits(key, newpos, 1);
- put_child(tn, missbit, (struct rt_trie_node *)l);
+ put_child(tn, missbit, l);
put_child(tn, 1-missbit, n);
if (tp) {
cindex = tkey_extract_bits(key, tp->pos, tp->bits);
- put_child(tp, cindex, (struct rt_trie_node *)tn);
+ put_child(tp, cindex, tn);
} else {
- rcu_assign_pointer(t->trie, (struct rt_trie_node *)tn);
+ rcu_assign_pointer(t->trie, tn);
}
tp = tn;
@@ -1134,7 +1098,7 @@ int fib_table_insert(struct fib_table *t
u8 tos = cfg->fc_tos;
u32 key, mask;
int err;
- struct leaf *l;
+ struct tnode *l;
if (plen > 32)
return -EINVAL;
@@ -1292,7 +1256,7 @@ err:
}
/* should be called with rcu_read_lock */
-static int check_leaf(struct fib_table *tb, struct trie *t, struct leaf *l,
+static int check_leaf(struct fib_table *tb, struct trie *t, struct tnode *l,
t_key key, const struct flowi4 *flp,
struct fib_result *res, int fib_flags)
{
@@ -1365,7 +1329,7 @@ int fib_table_lookup(struct fib_table *t
struct trie_use_stats __percpu *stats = t->stats;
#endif
int ret;
- struct rt_trie_node *n;
+ struct tnode *n;
struct tnode *pn;
unsigned int pos, bits;
t_key key = ntohl(flp->daddr);
@@ -1387,11 +1351,11 @@ int fib_table_lookup(struct fib_table *t
/* Just a leaf? */
if (IS_LEAF(n)) {
- ret = check_leaf(tb, t, (struct leaf *)n, key, flp, res, fib_flags);
+ ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
goto found;
}
- pn = (struct tnode *) n;
+ pn = n;
chopped_off = 0;
while (pn) {
@@ -1412,13 +1376,13 @@ int fib_table_lookup(struct fib_table *t
}
if (IS_LEAF(n)) {
- ret = check_leaf(tb, t, (struct leaf *)n, key, flp, res, fib_flags);
+ ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
if (ret > 0)
goto backtrace;
goto found;
}
- cn = (struct tnode *)n;
+ cn = n;
/*
* It's a tnode, and we can do some extra checks here if we
@@ -1506,7 +1470,7 @@ int fib_table_lookup(struct fib_table *t
current_prefix_length = mp;
}
- pn = (struct tnode *)n; /* Descend */
+ pn = n; /* Descend */
chopped_off = 0;
continue;
@@ -1557,7 +1521,7 @@ EXPORT_SYMBOL_GPL(fib_table_lookup);
/*
* Remove the leaf and return parent.
*/
-static void trie_leaf_remove(struct trie *t, struct leaf *l)
+static void trie_leaf_remove(struct trie *t, struct tnode *l)
{
struct tnode *tp = node_parent(l);
@@ -1584,7 +1548,7 @@ int fib_table_delete(struct fib_table *t
u8 tos = cfg->fc_tos;
struct fib_alias *fa, *fa_to_delete;
struct list_head *fa_head;
- struct leaf *l;
+ struct tnode *l;
struct leaf_info *li;
if (plen > 32)
@@ -1682,7 +1646,7 @@ static int trie_flush_list(struct list_h
return found;
}
-static int trie_flush_leaf(struct leaf *l)
+static int trie_flush_leaf(struct tnode *l)
{
int found = 0;
struct hlist_head *lih = &l->list;
@@ -1704,7 +1668,7 @@ static int trie_flush_leaf(struct leaf *
* Scan for the next right leaf starting at node p->child[idx]
* Since we have back pointer, no recursion necessary.
*/
-static struct leaf *leaf_walk_rcu(struct tnode *p, struct rt_trie_node *c)
+static struct tnode *leaf_walk_rcu(struct tnode *p, struct tnode *c)
{
do {
t_key idx;
@@ -1720,47 +1684,46 @@ static struct leaf *leaf_walk_rcu(struct
continue;
if (IS_LEAF(c))
- return (struct leaf *) c;
+ return c;
/* Rescan start scanning in new node */
- p = (struct tnode *) c;
+ p = c;
idx = 0;
}
/* Node empty, walk back up to parent */
- c = (struct rt_trie_node *) p;
+ c = p;
} while ((p = node_parent_rcu(c)) != NULL);
return NULL; /* Root of trie */
}
-static struct leaf *trie_firstleaf(struct trie *t)
+static struct tnode *trie_firstleaf(struct trie *t)
{
- struct tnode *n = (struct tnode *)rcu_dereference_rtnl(t->trie);
+ struct tnode *n = rcu_dereference_rtnl(t->trie);
if (!n)
return NULL;
if (IS_LEAF(n)) /* trie is just a leaf */
- return (struct leaf *) n;
+ return n;
return leaf_walk_rcu(n, NULL);
}
-static struct leaf *trie_nextleaf(struct leaf *l)
+static struct tnode *trie_nextleaf(struct tnode *l)
{
- struct rt_trie_node *c = (struct rt_trie_node *) l;
- struct tnode *p = node_parent_rcu(c);
+ struct tnode *p = node_parent_rcu(l);
if (!p)
return NULL; /* trie with just one leaf */
- return leaf_walk_rcu(p, c);
+ return leaf_walk_rcu(p, l);
}
-static struct leaf *trie_leafindex(struct trie *t, int index)
+static struct tnode *trie_leafindex(struct trie *t, int index)
{
- struct leaf *l = trie_firstleaf(t);
+ struct tnode *l = trie_firstleaf(t);
while (l && index-- > 0)
l = trie_nextleaf(l);
@@ -1775,7 +1738,7 @@ static struct leaf *trie_leafindex(struc
int fib_table_flush(struct fib_table *tb)
{
struct trie *t = (struct trie *) tb->tb_data;
- struct leaf *l, *ll = NULL;
+ struct tnode *l, *ll = NULL;
int found = 0;
for (l = trie_firstleaf(t); l; l = trie_nextleaf(l)) {
@@ -1840,7 +1803,7 @@ static int fn_trie_dump_fa(t_key key, in
return skb->len;
}
-static int fn_trie_dump_leaf(struct leaf *l, struct fib_table *tb,
+static int fn_trie_dump_leaf(struct tnode *l, struct fib_table *tb,
struct sk_buff *skb, struct netlink_callback *cb)
{
struct leaf_info *li;
@@ -1876,7 +1839,7 @@ static int fn_trie_dump_leaf(struct leaf
int fib_table_dump(struct fib_table *tb, struct sk_buff *skb,
struct netlink_callback *cb)
{
- struct leaf *l;
+ struct tnode *l;
struct trie *t = (struct trie *) tb->tb_data;
t_key key = cb->args[2];
int count = cb->args[3];
@@ -1922,7 +1885,7 @@ void __init fib_trie_init(void)
0, SLAB_PANIC, NULL);
trie_leaf_kmem = kmem_cache_create("ip_fib_trie",
- max(sizeof(struct leaf),
+ max(sizeof(struct tnode),
sizeof(struct leaf_info)),
0, SLAB_PANIC, NULL);
}
@@ -1965,7 +1928,7 @@ struct fib_trie_iter {
unsigned int depth;
};
-static struct rt_trie_node *fib_trie_get_next(struct fib_trie_iter *iter)
+static struct tnode *fib_trie_get_next(struct fib_trie_iter *iter)
{
struct tnode *tn = iter->tnode;
unsigned int cindex = iter->index;
@@ -1979,7 +1942,7 @@ static struct rt_trie_node *fib_trie_get
iter->tnode, iter->index, iter->depth);
rescan:
while (cindex < (1<<tn->bits)) {
- struct rt_trie_node *n = tnode_get_child_rcu(tn, cindex);
+ struct tnode *n = tnode_get_child_rcu(tn, cindex);
if (n) {
if (IS_LEAF(n)) {
@@ -1987,7 +1950,7 @@ rescan:
iter->index = cindex + 1;
} else {
/* push down one level */
- iter->tnode = (struct tnode *) n;
+ iter->tnode = n;
iter->index = 0;
++iter->depth;
}
@@ -1998,7 +1961,7 @@ rescan:
}
/* Current node exhausted, pop back up */
- p = node_parent_rcu((struct rt_trie_node *)tn);
+ p = node_parent_rcu(tn);
if (p) {
cindex = tkey_extract_bits(tn->key, p->pos, p->bits)+1;
tn = p;
@@ -2010,10 +1973,10 @@ rescan:
return NULL;
}
-static struct rt_trie_node *fib_trie_get_first(struct fib_trie_iter *iter,
+static struct tnode *fib_trie_get_first(struct fib_trie_iter *iter,
struct trie *t)
{
- struct rt_trie_node *n;
+ struct tnode *n;
if (!t)
return NULL;
@@ -2023,7 +1986,7 @@ static struct rt_trie_node *fib_trie_get
return NULL;
if (IS_TNODE(n)) {
- iter->tnode = (struct tnode *) n;
+ iter->tnode = n;
iter->index = 0;
iter->depth = 1;
} else {
@@ -2037,7 +2000,7 @@ static struct rt_trie_node *fib_trie_get
static void trie_collect_stats(struct trie *t, struct trie_stat *s)
{
- struct rt_trie_node *n;
+ struct tnode *n;
struct fib_trie_iter iter;
memset(s, 0, sizeof(*s));
@@ -2045,7 +2008,6 @@ static void trie_collect_stats(struct tr
rcu_read_lock();
for (n = fib_trie_get_first(&iter, t); n; n = fib_trie_get_next(&iter)) {
if (IS_LEAF(n)) {
- struct leaf *l = (struct leaf *)n;
struct leaf_info *li;
s->leaves++;
@@ -2053,18 +2015,17 @@ static void trie_collect_stats(struct tr
if (iter.depth > s->maxdepth)
s->maxdepth = iter.depth;
- hlist_for_each_entry_rcu(li, &l->list, hlist)
+ hlist_for_each_entry_rcu(li, &n->list, hlist)
++s->prefixes;
} else {
- const struct tnode *tn = (const struct tnode *) n;
int i;
s->tnodes++;
- if (tn->bits < MAX_STAT_DEPTH)
- s->nodesizes[tn->bits]++;
+ if (n->bits < MAX_STAT_DEPTH)
+ s->nodesizes[n->bits]++;
- for (i = 0; i < (1<<tn->bits); i++)
- if (!tn->child[i])
+ for (i = 0; i < tnode_child_length(n); i++)
+ if (!rcu_access_pointer(n->child[i]))
s->nullpointers++;
}
}
@@ -2088,7 +2049,7 @@ static void trie_show_stats(struct seq_f
seq_printf(seq, "\tMax depth: %u\n", stat->maxdepth);
seq_printf(seq, "\tLeaves: %u\n", stat->leaves);
- bytes = sizeof(struct leaf) * stat->leaves;
+ bytes = sizeof(struct tnode) * stat->leaves;
seq_printf(seq, "\tPrefixes: %u\n", stat->prefixes);
bytes += sizeof(struct leaf_info) * stat->prefixes;
@@ -2109,7 +2070,7 @@ static void trie_show_stats(struct seq_f
seq_putc(seq, '\n');
seq_printf(seq, "\tPointers: %u\n", pointers);
- bytes += sizeof(struct rt_trie_node *) * pointers;
+ bytes += sizeof(struct tnode *) * pointers;
seq_printf(seq, "Null ptrs: %u\n", stat->nullpointers);
seq_printf(seq, "Total size: %u kB\n", (bytes + 1023) / 1024);
}
@@ -2163,7 +2124,7 @@ static int fib_triestat_seq_show(struct
seq_printf(seq,
"Basic info: size of leaf:"
" %Zd bytes, size of tnode: %Zd bytes.\n",
- sizeof(struct leaf), sizeof(struct tnode));
+ sizeof(struct tnode), sizeof(struct tnode));
for (h = 0; h < FIB_TABLE_HASHSZ; h++) {
struct hlist_head *head = &net->ipv4.fib_table_hash[h];
@@ -2202,7 +2163,7 @@ static const struct file_operations fib_
.release = single_release_net,
};
-static struct rt_trie_node *fib_trie_get_idx(struct seq_file *seq, loff_t pos)
+static struct tnode *fib_trie_get_idx(struct seq_file *seq, loff_t pos)
{
struct fib_trie_iter *iter = seq->private;
struct net *net = seq_file_net(seq);
@@ -2214,7 +2175,7 @@ static struct rt_trie_node *fib_trie_get
struct fib_table *tb;
hlist_for_each_entry_rcu(tb, head, tb_hlist) {
- struct rt_trie_node *n;
+ struct tnode *n;
for (n = fib_trie_get_first(iter,
(struct trie *) tb->tb_data);
@@ -2243,7 +2204,7 @@ static void *fib_trie_seq_next(struct se
struct fib_table *tb = iter->tb;
struct hlist_node *tb_node;
unsigned int h;
- struct rt_trie_node *n;
+ struct tnode *n;
++*pos;
/* next node in same table */
@@ -2329,29 +2290,26 @@ static inline const char *rtn_type(char
static int fib_trie_seq_show(struct seq_file *seq, void *v)
{
const struct fib_trie_iter *iter = seq->private;
- struct rt_trie_node *n = v;
+ struct tnode *n = v;
if (!node_parent_rcu(n))
fib_table_print(seq, iter->tb);
if (IS_TNODE(n)) {
- struct tnode *tn = (struct tnode *) n;
- __be32 prf = htonl(tn->key);
+ __be32 prf = htonl(n->key);
- seq_indent(seq, iter->depth-1);
+ seq_indent(seq, iter->depth - 1);
seq_printf(seq, " +-- %pI4/%d %d %d %d\n",
- &prf, tn->pos, tn->bits, tn->full_children,
- tn->empty_children);
-
+ &prf, n->pos, n->bits, n->full_children,
+ n->empty_children);
} else {
- struct leaf *l = (struct leaf *) n;
struct leaf_info *li;
- __be32 val = htonl(l->key);
+ __be32 val = htonl(n->key);
seq_indent(seq, iter->depth);
seq_printf(seq, " |-- %pI4\n", &val);
- hlist_for_each_entry_rcu(li, &l->list, hlist) {
+ hlist_for_each_entry_rcu(li, &n->list, hlist) {
struct fib_alias *fa;
list_for_each_entry_rcu(fa, &li->falh, fa_list) {
@@ -2401,9 +2359,9 @@ struct fib_route_iter {
t_key key;
};
-static struct leaf *fib_route_get_idx(struct fib_route_iter *iter, loff_t pos)
+static struct tnode *fib_route_get_idx(struct fib_route_iter *iter, loff_t pos)
{
- struct leaf *l = NULL;
+ struct tnode *l = NULL;
struct trie *t = iter->main_trie;
/* use cache location of last found key */
@@ -2448,7 +2406,7 @@ static void *fib_route_seq_start(struct
static void *fib_route_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct fib_route_iter *iter = seq->private;
- struct leaf *l = v;
+ struct tnode *l = v;
++*pos;
if (v == SEQ_START_TOKEN) {
@@ -2494,7 +2452,7 @@ static unsigned int fib_flag_trans(int t
*/
static int fib_route_seq_show(struct seq_file *seq, void *v)
{
- struct leaf *l = v;
+ struct tnode *l = v;
struct leaf_info *li;
if (v == SEQ_START_TOKEN) {

343
target/linux/generic/pending-3.18/080-06-fib_trie-Optimize-fib_table_lookup-to-avoid-wasting-.patch
View File

@ -1,343 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:55:54 -0800
Subject: [PATCH] fib_trie: Optimize fib_table_lookup to avoid wasting
time on loops/variables
This patch is meant to reduce the complexity of fib_table_lookup by reducing
the number of variables to the bare minimum while still keeping the same if
not improved functionality versus the original.
Most of this change was started off by the desire to rid the function of
chopped_off and current_prefix_length as they actually added very little to
the function since they only applied when computing the cindex. I was able
to replace them mostly with just a check for the prefix match. As long as
the prefix between the key and the node being tested was the same we know
we can search the tnode fully versus just testing cindex 0.
The second portion of the change ended up being a massive reordering.
Originally the calls to check_leaf were up near the start of the loop, and
the backtracing and descending into lower levels of tnodes was later. This
didn't make much sense as the structure of the tree means the leaves are
always the last thing to be tested. As such I reordered things so that we
instead have a loop that will delve into the tree and only exit when we
have either found a leaf or we have exhausted the tree. The advantage of
rearranging things like this is that we can fully inline check_leaf since
there is now only one reference to it in the function.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -90,6 +90,9 @@ typedef unsigned int t_key;
#define IS_TNODE(n) ((n)->bits)
#define IS_LEAF(n) (!(n)->bits)
+#define get_shift(_kv) (KEYLENGTH - (_kv)->pos - (_kv)->bits)
+#define get_index(_key, _kv) (((_key) ^ (_kv)->key) >> get_shift(_kv))
+
struct tnode {
t_key key;
unsigned char bits; /* 2log(KEYLENGTH) bits needed */
@@ -1281,7 +1284,7 @@ static int check_leaf(struct fib_table *
continue;
fib_alias_accessed(fa);
err = fib_props[fa->fa_type].error;
- if (err) {
+ if (unlikely(err < 0)) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(t->stats->semantic_match_passed);
#endif
@@ -1303,7 +1306,7 @@ static int check_leaf(struct fib_table *
res->prefixlen = li->plen;
res->nh_sel = nhsel;
res->type = fa->fa_type;
- res->scope = fa->fa_info->fib_scope;
+ res->scope = fi->fib_scope;
res->fi = fi;
res->table = tb;
res->fa_head = &li->falh;
@@ -1321,23 +1324,24 @@ static int check_leaf(struct fib_table *
return 1;
}
+static inline t_key prefix_mismatch(t_key key, struct tnode *n)
+{
+ t_key prefix = n->key;
+
+ return (key ^ prefix) & (prefix | -prefix);
+}
+
int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
struct fib_result *res, int fib_flags)
{
- struct trie *t = (struct trie *) tb->tb_data;
+ struct trie *t = (struct trie *)tb->tb_data;
#ifdef CONFIG_IP_FIB_TRIE_STATS
struct trie_use_stats __percpu *stats = t->stats;
#endif
- int ret;
- struct tnode *n;
- struct tnode *pn;
- unsigned int pos, bits;
- t_key key = ntohl(flp->daddr);
- unsigned int chopped_off;
- t_key cindex = 0;
- unsigned int current_prefix_length = KEYLENGTH;
- struct tnode *cn;
- t_key pref_mismatch;
+ const t_key key = ntohl(flp->daddr);
+ struct tnode *n, *pn;
+ t_key cindex;
+ int ret = 1;
rcu_read_lock();
@@ -1349,170 +1353,102 @@ int fib_table_lookup(struct fib_table *t
this_cpu_inc(stats->gets);
#endif
- /* Just a leaf? */
- if (IS_LEAF(n)) {
- ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
- goto found;
- }
-
pn = n;
- chopped_off = 0;
-
- while (pn) {
- pos = pn->pos;
- bits = pn->bits;
+ cindex = 0;
- if (!chopped_off)
- cindex = tkey_extract_bits(mask_pfx(key, current_prefix_length),
- pos, bits);
-
- n = tnode_get_child_rcu(pn, cindex);
-
- if (n == NULL) {
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(stats->null_node_hit);
-#endif
- goto backtrace;
- }
+ /* Step 1: Travel to the longest prefix match in the trie */
+ for (;;) {
+ unsigned long index = get_index(key, n);
+
+ /* This bit of code is a bit tricky but it combines multiple
+ * checks into a single check. The prefix consists of the
+ * prefix plus zeros for the "bits" in the prefix. The index
+ * is the difference between the key and this value. From
+ * this we can actually derive several pieces of data.
+ * if !(index >> bits)
+ * we know the value is child index
+ * else
+ * we have a mismatch in skip bits and failed
+ */
+ if (index >> n->bits)
+ break;
- if (IS_LEAF(n)) {
- ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
- if (ret > 0)
- goto backtrace;
+ /* we have found a leaf. Prefixes have already been compared */
+ if (IS_LEAF(n))
goto found;
- }
-
- cn = n;
- /*
- * It's a tnode, and we can do some extra checks here if we
- * like, to avoid descending into a dead-end branch.
- * This tnode is in the parent's child array at index
- * key[p_pos..p_pos+p_bits] but potentially with some bits
- * chopped off, so in reality the index may be just a
- * subprefix, padded with zero at the end.
- * We can also take a look at any skipped bits in this
- * tnode - everything up to p_pos is supposed to be ok,
- * and the non-chopped bits of the index (se previous
- * paragraph) are also guaranteed ok, but the rest is
- * considered unknown.
- *
- * The skipped bits are key[pos+bits..cn->pos].
- */
-
- /* If current_prefix_length < pos+bits, we are already doing
- * actual prefix matching, which means everything from
- * pos+(bits-chopped_off) onward must be zero along some
- * branch of this subtree - otherwise there is *no* valid
- * prefix present. Here we can only check the skipped
- * bits. Remember, since we have already indexed into the
- * parent's child array, we know that the bits we chopped of
- * *are* zero.
+ /* only record pn and cindex if we are going to be chopping
+ * bits later. Otherwise we are just wasting cycles.
*/
-
- /* NOTA BENE: Checking only skipped bits
- for the new node here */
-
- if (current_prefix_length < pos+bits) {
- if (tkey_extract_bits(cn->key, current_prefix_length,
- cn->pos - current_prefix_length)
- || !(cn->child[0]))
- goto backtrace;
+ if (index) {
+ pn = n;
+ cindex = index;
}
- /*
- * If chopped_off=0, the index is fully validated and we
- * only need to look at the skipped bits for this, the new,
- * tnode. What we actually want to do is to find out if
- * these skipped bits match our key perfectly, or if we will
- * have to count on finding a matching prefix further down,
- * because if we do, we would like to have some way of
- * verifying the existence of such a prefix at this point.
- */
-
- /* The only thing we can do at this point is to verify that
- * any such matching prefix can indeed be a prefix to our
- * key, and if the bits in the node we are inspecting that
- * do not match our key are not ZERO, this cannot be true.
- * Thus, find out where there is a mismatch (before cn->pos)
- * and verify that all the mismatching bits are zero in the
- * new tnode's key.
- */
+ n = rcu_dereference(n->child[index]);
+ if (unlikely(!n))
+ goto backtrace;
+ }
- /*
- * Note: We aren't very concerned about the piece of
- * the key that precede pn->pos+pn->bits, since these
- * have already been checked. The bits after cn->pos
- * aren't checked since these are by definition
- * "unknown" at this point. Thus, what we want to see
- * is if we are about to enter the "prefix matching"
- * state, and in that case verify that the skipped
- * bits that will prevail throughout this subtree are
- * zero, as they have to be if we are to find a
- * matching prefix.
+ /* Step 2: Sort out leaves and begin backtracing for longest prefix */
+ for (;;) {
+ /* record the pointer where our next node pointer is stored */
+ struct tnode __rcu **cptr = n->child;
+
+ /* This test verifies that none of the bits that differ
+ * between the key and the prefix exist in the region of
+ * the lsb and higher in the prefix.
*/
+ if (unlikely(prefix_mismatch(key, n)))
+ goto backtrace;
- pref_mismatch = mask_pfx(cn->key ^ key, cn->pos);
+ /* exit out and process leaf */
+ if (unlikely(IS_LEAF(n)))
+ break;
- /*
- * In short: If skipped bits in this node do not match
- * the search key, enter the "prefix matching"
- * state.directly.
+ /* Don't bother recording parent info. Since we are in
+ * prefix match mode we will have to come back to wherever
+ * we started this traversal anyway
*/
- if (pref_mismatch) {
- /* fls(x) = __fls(x) + 1 */
- int mp = KEYLENGTH - __fls(pref_mismatch) - 1;
-
- if (tkey_extract_bits(cn->key, mp, cn->pos - mp) != 0)
- goto backtrace;
-
- if (current_prefix_length >= cn->pos)
- current_prefix_length = mp;
- }
-
- pn = n; /* Descend */
- chopped_off = 0;
- continue;
+ while ((n = rcu_dereference(*cptr)) == NULL) {
backtrace:
- chopped_off++;
-
- /* As zero don't change the child key (cindex) */
- while ((chopped_off <= pn->bits)
- && !(cindex & (1<<(chopped_off-1))))
- chopped_off++;
-
- /* Decrease current_... with bits chopped off */
- if (current_prefix_length > pn->pos + pn->bits - chopped_off)
- current_prefix_length = pn->pos + pn->bits
- - chopped_off;
-
- /*
- * Either we do the actual chop off according or if we have
- * chopped off all bits in this tnode walk up to our parent.
- */
-
- if (chopped_off <= pn->bits) {
- cindex &= ~(1 << (chopped_off-1));
- } else {
- struct tnode *parent = node_parent_rcu(pn);
- if (!parent)
- goto failed;
-
- /* Get Child's index */
- cindex = tkey_extract_bits(pn->key, parent->pos, parent->bits);
- pn = parent;
- chopped_off = 0;
-
#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(stats->backtrack);
+ if (!n)
+ this_cpu_inc(stats->null_node_hit);
#endif
- goto backtrace;
+ /* If we are at cindex 0 there are no more bits for
+ * us to strip at this level so we must ascend back
+ * up one level to see if there are any more bits to
+ * be stripped there.
+ */
+ while (!cindex) {
+ t_key pkey = pn->key;
+
+ pn = node_parent_rcu(pn);
+ if (unlikely(!pn))
+ goto failed;
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(stats->backtrack);
+#endif
+ /* Get Child's index */
+ cindex = get_index(pkey, pn);
+ }
+
+ /* strip the least significant bit from the cindex */
+ cindex &= cindex - 1;
+
+ /* grab pointer for next child node */
+ cptr = &pn->child[cindex];
}
}
-failed:
- ret = 1;
+
found:
+ /* Step 3: Process the leaf, if that fails fall back to backtracing */
+ ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
+ if (unlikely(ret > 0))
+ goto backtrace;
+failed:
rcu_read_unlock();
return ret;
}

64
target/linux/generic/pending-3.18/080-07-fib_trie-Optimize-fib_find_node.patch
View File

@ -1,64 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:00 -0800
Subject: [PATCH] fib_trie: Optimize fib_find_node
This patch makes use of the same features I made use of for
fib_table_lookup to streamline fib_find_node. The resultant code should be
smaller and run faster than the original.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -892,28 +892,34 @@ static void insert_leaf_info(struct hlis
}
/* rcu_read_lock needs to be hold by caller from readside */
-
static struct tnode *fib_find_node(struct trie *t, u32 key)
{
struct tnode *n = rcu_dereference_rtnl(t->trie);
- int pos = 0;
- while (n && IS_TNODE(n)) {
- if (tkey_sub_equals(n->key, pos, n->pos-pos, key)) {
- pos = n->pos + n->bits;
- n = tnode_get_child_rcu(n,
- tkey_extract_bits(key,
- n->pos,
- n->bits));
- } else
+ while (n) {
+ unsigned long index = get_index(key, n);
+
+ /* This bit of code is a bit tricky but it combines multiple
+ * checks into a single check. The prefix consists of the
+ * prefix plus zeros for the bits in the cindex. The index
+ * is the difference between the key and this value. From
+ * this we can actually derive several pieces of data.
+ * if !(index >> bits)
+ * we know the value is cindex
+ * else
+ * we have a mismatch in skip bits and failed
+ */
+ if (index >> n->bits)
+ return NULL;
+
+ /* we have found a leaf. Prefixes have already been compared */
+ if (IS_LEAF(n))
break;
- }
- /* Case we have found a leaf. Compare prefixes */
- if (n != NULL && IS_LEAF(n) && tkey_equals(key, n->key))
- return n;
+ n = rcu_dereference_rtnl(n->child[index]);
+ }
- return NULL;
+ return n;
}
static void trie_rebalance(struct trie *t, struct tnode *tn)

276
target/linux/generic/pending-3.18/080-08-fib_trie-Optimize-fib_table_insert.patch
View File

@ -1,276 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:06 -0800
Subject: [PATCH] fib_trie: Optimize fib_table_insert
This patch updates the fib_table_insert function to take advantage of the
changes made to improve the performance of fib_table_lookup. As a result
the code should be smaller and run faster then the original.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -222,31 +222,6 @@ static inline t_key tkey_extract_bits(t_
return 0;
}
-static inline int tkey_equals(t_key a, t_key b)
-{
- return a == b;
-}
-
-static inline int tkey_sub_equals(t_key a, int offset, int bits, t_key b)
-{
- if (bits == 0 || offset >= KEYLENGTH)
- return 1;
- bits = bits > KEYLENGTH ? KEYLENGTH : bits;
- return ((a ^ b) << offset) >> (KEYLENGTH - bits) == 0;
-}
-
-static inline int tkey_mismatch(t_key a, int offset, t_key b)
-{
- t_key diff = a ^ b;
- int i = offset;
-
- if (!diff)
- return 0;
- while ((diff << i) >> (KEYLENGTH-1) == 0)
- i++;
- return i;
-}
-
/*
To understand this stuff, an understanding of keys and all their bits is
necessary. Every node in the trie has a key associated with it, but not
@@ -485,6 +460,15 @@ static void tnode_put_child_reorg(struct
rcu_assign_pointer(tn->child[i], n);
}
+static void put_child_root(struct tnode *tp, struct trie *t,
+ t_key key, struct tnode *n)
+{
+ if (tp)
+ put_child(tp, get_index(key, tp), n);
+ else
+ rcu_assign_pointer(t->trie, n);
+}
+
#define MAX_WORK 10
static struct tnode *resize(struct trie *t, struct tnode *tn)
{
@@ -959,138 +943,100 @@ static void trie_rebalance(struct trie *
static struct list_head *fib_insert_node(struct trie *t, u32 key, int plen)
{
- int pos, newpos;
- struct tnode *tp = NULL, *tn = NULL;
- struct tnode *n;
- struct tnode *l;
- int missbit;
struct list_head *fa_head = NULL;
+ struct tnode *l, *n, *tp = NULL;
struct leaf_info *li;
- t_key cindex;
- pos = 0;
+ li = leaf_info_new(plen);
+ if (!li)
+ return NULL;
+ fa_head = &li->falh;
+
n = rtnl_dereference(t->trie);
/* If we point to NULL, stop. Either the tree is empty and we should
* just put a new leaf in if, or we have reached an empty child slot,
* and we should just put our new leaf in that.
- * If we point to a T_TNODE, check if it matches our key. Note that
- * a T_TNODE might be skipping any number of bits - its 'pos' need
- * not be the parent's 'pos'+'bits'!
- *
- * If it does match the current key, get pos/bits from it, extract
- * the index from our key, push the T_TNODE and walk the tree.
- *
- * If it doesn't, we have to replace it with a new T_TNODE.
*
- * If we point to a T_LEAF, it might or might not have the same key
- * as we do. If it does, just change the value, update the T_LEAF's
- * value, and return it.
- * If it doesn't, we need to replace it with a T_TNODE.
+ * If we hit a node with a key that does't match then we should stop
+ * and create a new tnode to replace that node and insert ourselves
+ * and the other node into the new tnode.
*/
+ while (n) {
+ unsigned long index = get_index(key, n);
- while (n && IS_TNODE(n)) {
- if (tkey_sub_equals(n->key, pos, n->pos-pos, key)) {
- tp = n;
- pos = n->pos + n->bits;
- n = tnode_get_child(n,
- tkey_extract_bits(key,
- n->pos,
- n->bits));
-
- BUG_ON(n && node_parent(n) != tp);
- } else
+ /* This bit of code is a bit tricky but it combines multiple
+ * checks into a single check. The prefix consists of the
+ * prefix plus zeros for the "bits" in the prefix. The index
+ * is the difference between the key and this value. From
+ * this we can actually derive several pieces of data.
+ * if !(index >> bits)
+ * we know the value is child index
+ * else
+ * we have a mismatch in skip bits and failed
+ */
+ if (index >> n->bits)
break;
- }
-
- /*
- * n ----> NULL, LEAF or TNODE
- *
- * tp is n's (parent) ----> NULL or TNODE
- */
- BUG_ON(tp && IS_LEAF(tp));
-
- /* Case 1: n is a leaf. Compare prefixes */
-
- if (n != NULL && IS_LEAF(n) && tkey_equals(key, n->key)) {
- li = leaf_info_new(plen);
-
- if (!li)
- return NULL;
+ /* we have found a leaf. Prefixes have already been compared */
+ if (IS_LEAF(n)) {
+ /* Case 1: n is a leaf, and prefixes match*/
+ insert_leaf_info(&n->list, li);
+ return fa_head;
+ }
- fa_head = &li->falh;
- insert_leaf_info(&n->list, li);
- goto done;
+ tp = n;
+ n = rcu_dereference_rtnl(n->child[index]);
}
- l = leaf_new(key);
-
- if (!l)
- return NULL;
- li = leaf_info_new(plen);
-
- if (!li) {
- node_free(l);
+ l = leaf_new(key);
+ if (!l) {
+ free_leaf_info(li);
return NULL;
}
- fa_head = &li->falh;
insert_leaf_info(&l->list, li);
- if (t->trie && n == NULL) {
- /* Case 2: n is NULL, and will just insert a new leaf */
-
- node_set_parent(l, tp);
-
- cindex = tkey_extract_bits(key, tp->pos, tp->bits);
- put_child(tp, cindex, l);
- } else {
- /* Case 3: n is a LEAF or a TNODE and the key doesn't match. */
- /*
- * Add a new tnode here
- * first tnode need some special handling
- */
+ /* Case 2: n is a LEAF or a TNODE and the key doesn't match.
+ *
+ * Add a new tnode here
+ * first tnode need some special handling
+ * leaves us in position for handling as case 3
+ */
+ if (n) {
+ struct tnode *tn;
+ int newpos;
- if (n) {
- pos = tp ? tp->pos+tp->bits : 0;
- newpos = tkey_mismatch(key, pos, n->key);
- tn = tnode_new(n->key, newpos, 1);
- } else {
- newpos = 0;
- tn = tnode_new(key, newpos, 1); /* First tnode */
- }
+ newpos = KEYLENGTH - __fls(n->key ^ key) - 1;
+ tn = tnode_new(key, newpos, 1);
if (!tn) {
free_leaf_info(li);
node_free(l);
return NULL;
}
- node_set_parent(tn, tp);
-
- missbit = tkey_extract_bits(key, newpos, 1);
- put_child(tn, missbit, l);
- put_child(tn, 1-missbit, n);
-
- if (tp) {
- cindex = tkey_extract_bits(key, tp->pos, tp->bits);
- put_child(tp, cindex, tn);
- } else {
- rcu_assign_pointer(t->trie, tn);
- }
+ /* initialize routes out of node */
+ NODE_INIT_PARENT(tn, tp);
+ put_child(tn, get_index(key, tn) ^ 1, n);
+
+ /* start adding routes into the node */
+ put_child_root(tp, t, key, tn);
+ node_set_parent(n, tn);
+ /* parent now has a NULL spot where the leaf can go */
tp = tn;
}
- if (tp && tp->pos + tp->bits > 32)
- pr_warn("fib_trie tp=%p pos=%d, bits=%d, key=%0x plen=%d\n",
- tp, tp->pos, tp->bits, key, plen);
-
- /* Rebalance the trie */
+ /* Case 3: n is NULL, and will just insert a new leaf */
+ if (tp) {
+ NODE_INIT_PARENT(l, tp);
+ put_child(tp, get_index(key, tp), l);
+ trie_rebalance(t, tp);
+ } else {
+ rcu_assign_pointer(t->trie, l);
+ }
- trie_rebalance(t, tp);
-done:
return fa_head;
}
@@ -1470,11 +1416,11 @@ static void trie_leaf_remove(struct trie
pr_debug("entering trie_leaf_remove(%p)\n", l);
if (tp) {
- t_key cindex = tkey_extract_bits(l->key, tp->pos, tp->bits);
- put_child(tp, cindex, NULL);
+ put_child(tp, get_index(l->key, tp), NULL);
trie_rebalance(t, tp);
- } else
+ } else {
RCU_INIT_POINTER(t->trie, NULL);
+ }
node_free(l);
}

346
target/linux/generic/pending-3.18/080-09-fib_trie-Update-meaning-of-pos-to-represent-unchecke.patch
View File

@ -1,346 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:12 -0800
Subject: [PATCH] fib_trie: Update meaning of pos to represent unchecked
bits
This change moves the pos value to the other side of the "bits" field. By
doing this it actually simplifies a significant amount of code in the trie.
For example when halving a tree we know that the bit lost exists at
oldnode->pos, and if we inflate the tree the new bit being add is at
tn->pos. Previously to find those bits you would have to subtract pos and
bits from the keylength or start with a value of (1 << 31) and then shift
that.
There are a number of spots throughout the code that benefit from this. In
the case of the hot-path searches the main advantage is that we can drop 2
or more operations from the search path as we no longer need to compute the
value for the index to be shifted by and can instead just use the raw pos
value.
In addition the tkey_extract_bits is now defunct and can be replaced by
get_index since the two operations were doing the same thing, but now
get_index does it much more quickly as it is only an xor and shift versus a
pair of shifts and a subtraction.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -90,8 +90,7 @@ typedef unsigned int t_key;
#define IS_TNODE(n) ((n)->bits)
#define IS_LEAF(n) (!(n)->bits)
-#define get_shift(_kv) (KEYLENGTH - (_kv)->pos - (_kv)->bits)
-#define get_index(_key, _kv) (((_key) ^ (_kv)->key) >> get_shift(_kv))
+#define get_index(_key, _kv) (((_key) ^ (_kv)->key) >> (_kv)->pos)
struct tnode {
t_key key;
@@ -209,81 +208,64 @@ static inline struct tnode *tnode_get_ch
return rcu_dereference_rtnl(tn->child[i]);
}
-static inline t_key mask_pfx(t_key k, unsigned int l)
-{
- return (l == 0) ? 0 : k >> (KEYLENGTH-l) << (KEYLENGTH-l);
-}
-
-static inline t_key tkey_extract_bits(t_key a, unsigned int offset, unsigned int bits)
-{
- if (offset < KEYLENGTH)
- return ((t_key)(a << offset)) >> (KEYLENGTH - bits);
- else
- return 0;
-}
-
-/*
- To understand this stuff, an understanding of keys and all their bits is
- necessary. Every node in the trie has a key associated with it, but not
- all of the bits in that key are significant.
-
- Consider a node 'n' and its parent 'tp'.
-
- If n is a leaf, every bit in its key is significant. Its presence is
- necessitated by path compression, since during a tree traversal (when
- searching for a leaf - unless we are doing an insertion) we will completely
- ignore all skipped bits we encounter. Thus we need to verify, at the end of
- a potentially successful search, that we have indeed been walking the
- correct key path.
-
- Note that we can never "miss" the correct key in the tree if present by
- following the wrong path. Path compression ensures that segments of the key
- that are the same for all keys with a given prefix are skipped, but the
- skipped part *is* identical for each node in the subtrie below the skipped
- bit! trie_insert() in this implementation takes care of that - note the
- call to tkey_sub_equals() in trie_insert().
-
- if n is an internal node - a 'tnode' here, the various parts of its key
- have many different meanings.
-
- Example:
- _________________________________________________________________
- | i | i | i | i | i | i | i | N | N | N | S | S | S | S | S | C |
- -----------------------------------------------------------------
- 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
-
- _________________________________________________________________
- | C | C | C | u | u | u | u | u | u | u | u | u | u | u | u | u |
- -----------------------------------------------------------------
- 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
-
- tp->pos = 7
- tp->bits = 3
- n->pos = 15
- n->bits = 4
-
- First, let's just ignore the bits that come before the parent tp, that is
- the bits from 0 to (tp->pos-1). They are *known* but at this point we do
- not use them for anything.
-
- The bits from (tp->pos) to (tp->pos + tp->bits - 1) - "N", above - are the
- index into the parent's child array. That is, they will be used to find
- 'n' among tp's children.
-
- The bits from (tp->pos + tp->bits) to (n->pos - 1) - "S" - are skipped bits
- for the node n.
-
- All the bits we have seen so far are significant to the node n. The rest
- of the bits are really not needed or indeed known in n->key.
-
- The bits from (n->pos) to (n->pos + n->bits - 1) - "C" - are the index into
- n's child array, and will of course be different for each child.
-
-
- The rest of the bits, from (n->pos + n->bits) onward, are completely unknown
- at this point.
-
-*/
+/* To understand this stuff, an understanding of keys and all their bits is
+ * necessary. Every node in the trie has a key associated with it, but not
+ * all of the bits in that key are significant.
+ *
+ * Consider a node 'n' and its parent 'tp'.
+ *
+ * If n is a leaf, every bit in its key is significant. Its presence is
+ * necessitated by path compression, since during a tree traversal (when
+ * searching for a leaf - unless we are doing an insertion) we will completely
+ * ignore all skipped bits we encounter. Thus we need to verify, at the end of
+ * a potentially successful search, that we have indeed been walking the
+ * correct key path.
+ *
+ * Note that we can never "miss" the correct key in the tree if present by
+ * following the wrong path. Path compression ensures that segments of the key
+ * that are the same for all keys with a given prefix are skipped, but the
+ * skipped part *is* identical for each node in the subtrie below the skipped
+ * bit! trie_insert() in this implementation takes care of that.
+ *
+ * if n is an internal node - a 'tnode' here, the various parts of its key
+ * have many different meanings.
+ *
+ * Example:
+ * _________________________________________________________________
+ * | i | i | i | i | i | i | i | N | N | N | S | S | S | S | S | C |
+ * -----------------------------------------------------------------
+ * 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16
+ *
+ * _________________________________________________________________
+ * | C | C | C | u | u | u | u | u | u | u | u | u | u | u | u | u |
+ * -----------------------------------------------------------------
+ * 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
+ *
+ * tp->pos = 22
+ * tp->bits = 3
+ * n->pos = 13
+ * n->bits = 4
+ *
+ * First, let's just ignore the bits that come before the parent tp, that is
+ * the bits from (tp->pos + tp->bits) to 31. They are *known* but at this
+ * point we do not use them for anything.
+ *
+ * The bits from (tp->pos) to (tp->pos + tp->bits - 1) - "N", above - are the
+ * index into the parent's child array. That is, they will be used to find
+ * 'n' among tp's children.
+ *
+ * The bits from (n->pos + n->bits) to (tn->pos - 1) - "S" - are skipped bits
+ * for the node n.
+ *
+ * All the bits we have seen so far are significant to the node n. The rest
+ * of the bits are really not needed or indeed known in n->key.
+ *
+ * The bits from (n->pos) to (n->pos + n->bits - 1) - "C" - are the index into
+ * n's child array, and will of course be different for each child.
+ *
+ * The rest of the bits, from 0 to (n->pos + n->bits), are completely unknown
+ * at this point.
+ */
static const int halve_threshold = 25;
static const int inflate_threshold = 50;
@@ -367,7 +349,7 @@ static struct tnode *leaf_new(t_key key)
* as the nodes are searched
*/
l->key = key;
- l->pos = KEYLENGTH;
+ l->pos = 0;
/* set bits to 0 indicating we are not a tnode */
l->bits = 0;
@@ -400,7 +382,7 @@ static struct tnode *tnode_new(t_key key
tn->parent = NULL;
tn->pos = pos;
tn->bits = bits;
- tn->key = mask_pfx(key, pos);
+ tn->key = (shift < KEYLENGTH) ? (key >> shift) << shift : 0;
tn->full_children = 0;
tn->empty_children = 1<<bits;
}
@@ -410,14 +392,12 @@ static struct tnode *tnode_new(t_key key
return tn;
}
-/*
- * Check whether a tnode 'n' is "full", i.e. it is an internal node
+/* Check whether a tnode 'n' is "full", i.e. it is an internal node
* and no bits are skipped. See discussion in dyntree paper p. 6
*/
-
static inline int tnode_full(const struct tnode *tn, const struct tnode *n)
{
- return n && IS_TNODE(n) && (n->pos == (tn->pos + tn->bits));
+ return n && ((n->pos + n->bits) == tn->pos) && IS_TNODE(n);
}
static inline void put_child(struct tnode *tn, int i,
@@ -641,11 +621,12 @@ static struct tnode *inflate(struct trie
{
int olen = tnode_child_length(oldtnode);
struct tnode *tn;
+ t_key m;
int i;
pr_debug("In inflate\n");
- tn = tnode_new(oldtnode->key, oldtnode->pos, oldtnode->bits + 1);
+ tn = tnode_new(oldtnode->key, oldtnode->pos - 1, oldtnode->bits + 1);
if (!tn)
return ERR_PTR(-ENOMEM);
@@ -656,21 +637,18 @@ static struct tnode *inflate(struct trie
* fails. In case of failure we return the oldnode and inflate
* of tnode is ignored.
*/
+ for (i = 0, m = 1u << tn->pos; i < olen; i++) {
+ struct tnode *inode = tnode_get_child(oldtnode, i);
- for (i = 0; i < olen; i++) {
- struct tnode *inode;
-
- inode = tnode_get_child(oldtnode, i);
- if (tnode_full(oldtnode, inode) && inode->bits > 1) {
+ if (tnode_full(oldtnode, inode) && (inode->bits > 1)) {
struct tnode *left, *right;
- t_key m = ~0U << (KEYLENGTH - 1) >> inode->pos;
- left = tnode_new(inode->key&(~m), inode->pos + 1,
+ left = tnode_new(inode->key & ~m, inode->pos,
inode->bits - 1);
if (!left)
goto nomem;
- right = tnode_new(inode->key|m, inode->pos + 1,
+ right = tnode_new(inode->key | m, inode->pos,
inode->bits - 1);
if (!right) {
@@ -694,9 +672,7 @@ static struct tnode *inflate(struct trie
/* A leaf or an internal node with skipped bits */
if (!tnode_full(oldtnode, inode)) {
- put_child(tn,
- tkey_extract_bits(inode->key, tn->pos, tn->bits),
- inode);
+ put_child(tn, get_index(inode->key, tn), inode);
continue;
}
@@ -767,7 +743,7 @@ static struct tnode *halve(struct trie *
pr_debug("In halve\n");
- tn = tnode_new(oldtnode->key, oldtnode->pos, oldtnode->bits - 1);
+ tn = tnode_new(oldtnode->key, oldtnode->pos + 1, oldtnode->bits - 1);
if (!tn)
return ERR_PTR(-ENOMEM);
@@ -787,7 +763,7 @@ static struct tnode *halve(struct trie *
if (left && right) {
struct tnode *newn;
- newn = tnode_new(left->key, tn->pos + tn->bits, 1);
+ newn = tnode_new(left->key, oldtnode->pos, 1);
if (!newn)
goto nomem;
@@ -915,7 +891,7 @@ static void trie_rebalance(struct trie *
key = tn->key;
while (tn != NULL && (tp = node_parent(tn)) != NULL) {
- cindex = tkey_extract_bits(key, tp->pos, tp->bits);
+ cindex = get_index(key, tp);
wasfull = tnode_full(tp, tnode_get_child(tp, cindex));
tn = resize(t, tn);
@@ -1005,11 +981,8 @@ static struct list_head *fib_insert_node
*/
if (n) {
struct tnode *tn;
- int newpos;
-
- newpos = KEYLENGTH - __fls(n->key ^ key) - 1;
- tn = tnode_new(key, newpos, 1);
+ tn = tnode_new(key, __fls(key ^ n->key), 1);
if (!tn) {
free_leaf_info(li);
node_free(l);
@@ -1559,12 +1532,7 @@ static int trie_flush_leaf(struct tnode
static struct tnode *leaf_walk_rcu(struct tnode *p, struct tnode *c)
{
do {
- t_key idx;
-
- if (c)
- idx = tkey_extract_bits(c->key, p->pos, p->bits) + 1;
- else
- idx = 0;
+ t_key idx = c ? idx = get_index(c->key, p) + 1 : 0;
while (idx < 1u << p->bits) {
c = tnode_get_child_rcu(p, idx++);
@@ -1851,7 +1819,7 @@ rescan:
/* Current node exhausted, pop back up */
p = node_parent_rcu(tn);
if (p) {
- cindex = tkey_extract_bits(tn->key, p->pos, p->bits)+1;
+ cindex = get_index(tn->key, p) + 1;
tn = p;
--iter->depth;
goto rescan;
@@ -2186,10 +2154,10 @@ static int fib_trie_seq_show(struct seq_
if (IS_TNODE(n)) {
__be32 prf = htonl(n->key);
- seq_indent(seq, iter->depth - 1);
- seq_printf(seq, " +-- %pI4/%d %d %d %d\n",
- &prf, n->pos, n->bits, n->full_children,
- n->empty_children);
+ seq_indent(seq, iter->depth-1);
+ seq_printf(seq, " +-- %pI4/%zu %u %u %u\n",
+ &prf, KEYLENGTH - n->pos - n->bits, n->bits,
+ n->full_children, n->empty_children);
} else {
struct leaf_info *li;
__be32 val = htonl(n->key);

186
target/linux/generic/pending-3.18/080-10-fib_trie-Use-unsigned-long-for-anything-dealing-with.patch
View File

@ -1,186 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:18 -0800
Subject: [PATCH] fib_trie: Use unsigned long for anything dealing with a
shift by bits
This change makes it so that anything that can be shifted by, or compared
to a value shifted by bits is updated to be an unsigned long. This is
mostly a precaution against an insanely huge address space that somehow
starts coming close to the 2^32 root node size which would require
something like 1.5 billion addresses.
I chose unsigned long instead of unsigned long long since I do not believe
it is possible to allocate a 32 bit tnode on a 32 bit system as the memory
consumed would be 16GB + 28B which exceeds the addressible space for any
one process.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -146,8 +146,8 @@ struct trie {
#endif
};
-static void tnode_put_child_reorg(struct tnode *tn, int i, struct tnode *n,
- int wasfull);
+static void tnode_put_child_reorg(struct tnode *tn, unsigned long i,
+ struct tnode *n, int wasfull);
static struct tnode *resize(struct trie *t, struct tnode *tn);
static struct tnode *inflate(struct trie *t, struct tnode *tn);
static struct tnode *halve(struct trie *t, struct tnode *tn);
@@ -183,25 +183,23 @@ static inline void node_set_parent(struc
/* This provides us with the number of children in this node, in the case of a
* leaf this will return 0 meaning none of the children are accessible.
*/
-static inline int tnode_child_length(const struct tnode *tn)
+static inline unsigned long tnode_child_length(const struct tnode *tn)
{
return (1ul << tn->bits) & ~(1ul);
}
-/*
- * caller must hold RTNL
- */
-static inline struct tnode *tnode_get_child(const struct tnode *tn, unsigned int i)
+/* caller must hold RTNL */
+static inline struct tnode *tnode_get_child(const struct tnode *tn,
+ unsigned long i)
{
BUG_ON(i >= tnode_child_length(tn));
return rtnl_dereference(tn->child[i]);
}
-/*
- * caller must hold RCU read lock or RTNL
- */
-static inline struct tnode *tnode_get_child_rcu(const struct tnode *tn, unsigned int i)
+/* caller must hold RCU read lock or RTNL */
+static inline struct tnode *tnode_get_child_rcu(const struct tnode *tn,
+ unsigned long i)
{
BUG_ON(i >= tnode_child_length(tn));
@@ -400,7 +398,7 @@ static inline int tnode_full(const struc
return n && ((n->pos + n->bits) == tn->pos) && IS_TNODE(n);
}
-static inline void put_child(struct tnode *tn, int i,
+static inline void put_child(struct tnode *tn, unsigned long i,
struct tnode *n)
{
tnode_put_child_reorg(tn, i, n, -1);
@@ -411,13 +409,13 @@ static inline void put_child(struct tnod
* Update the value of full_children and empty_children.
*/
-static void tnode_put_child_reorg(struct tnode *tn, int i, struct tnode *n,
- int wasfull)
+static void tnode_put_child_reorg(struct tnode *tn, unsigned long i,
+ struct tnode *n, int wasfull)
{
struct tnode *chi = rtnl_dereference(tn->child[i]);
int isfull;
- BUG_ON(i >= 1<<tn->bits);
+ BUG_ON(i >= tnode_child_length(tn));
/* update emptyChildren */
if (n == NULL && chi != NULL)
@@ -607,10 +605,10 @@ no_children:
static void tnode_clean_free(struct tnode *tn)
{
struct tnode *tofree;
- int i;
+ unsigned long i;
for (i = 0; i < tnode_child_length(tn); i++) {
- tofree = rtnl_dereference(tn->child[i]);
+ tofree = tnode_get_child(tn, i);
if (tofree)
node_free(tofree);
}
@@ -619,10 +617,10 @@ static void tnode_clean_free(struct tnod
static struct tnode *inflate(struct trie *t, struct tnode *oldtnode)
{
- int olen = tnode_child_length(oldtnode);
+ unsigned long olen = tnode_child_length(oldtnode);
struct tnode *tn;
+ unsigned long i;
t_key m;
- int i;
pr_debug("In inflate\n");
@@ -664,7 +662,7 @@ static struct tnode *inflate(struct trie
for (i = 0; i < olen; i++) {
struct tnode *inode = tnode_get_child(oldtnode, i);
struct tnode *left, *right;
- int size, j;
+ unsigned long size, j;
/* An empty child */
if (inode == NULL)
@@ -737,7 +735,7 @@ nomem:
static struct tnode *halve(struct trie *t, struct tnode *oldtnode)
{
- int olen = tnode_child_length(oldtnode);
+ unsigned long olen = tnode_child_length(oldtnode);
struct tnode *tn, *left, *right;
int i;
@@ -1532,9 +1530,9 @@ static int trie_flush_leaf(struct tnode
static struct tnode *leaf_walk_rcu(struct tnode *p, struct tnode *c)
{
do {
- t_key idx = c ? idx = get_index(c->key, p) + 1 : 0;
+ unsigned long idx = c ? idx = get_index(c->key, p) + 1 : 0;
- while (idx < 1u << p->bits) {
+ while (idx < tnode_child_length(p)) {
c = tnode_get_child_rcu(p, idx++);
if (!c)
continue;
@@ -1786,8 +1784,8 @@ struct fib_trie_iter {
static struct tnode *fib_trie_get_next(struct fib_trie_iter *iter)
{
+ unsigned long cindex = iter->index;
struct tnode *tn = iter->tnode;
- unsigned int cindex = iter->index;
struct tnode *p;
/* A single entry routing table */
@@ -1797,7 +1795,7 @@ static struct tnode *fib_trie_get_next(s
pr_debug("get_next iter={node=%p index=%d depth=%d}\n",
iter->tnode, iter->index, iter->depth);
rescan:
- while (cindex < (1<<tn->bits)) {
+ while (cindex < tnode_child_length(tn)) {
struct tnode *n = tnode_get_child_rcu(tn, cindex);
if (n) {
@@ -1874,15 +1872,16 @@ static void trie_collect_stats(struct tr
hlist_for_each_entry_rcu(li, &n->list, hlist)
++s->prefixes;
} else {
- int i;
+ unsigned long i;
s->tnodes++;
if (n->bits < MAX_STAT_DEPTH)
s->nodesizes[n->bits]++;
- for (i = 0; i < tnode_child_length(n); i++)
+ for (i = 0; i < tnode_child_length(n); i++) {
if (!rcu_access_pointer(n->child[i]))
s->nullpointers++;
+ }
}
}
rcu_read_unlock();

403
target/linux/generic/pending-3.18/080-11-fib_trie-Push-rcu_read_lock-unlock-to-callers.patch
View File

@ -1,403 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:24 -0800
Subject: [PATCH] fib_trie: Push rcu_read_lock/unlock to callers
This change is to start cleaning up some of the rcu_read_lock/unlock
handling. I realized while reviewing the code there are several spots that
I don't believe are being handled correctly or are masking warnings by
locally calling rcu_read_lock/unlock instead of calling them at the correct
level.
A common example is a call to fib_get_table followed by fib_table_lookup.
The rcu_read_lock/unlock ought to wrap both but there are several spots where
they were not wrapped.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -222,16 +222,19 @@ static inline struct fib_table *fib_new_
static inline int fib_lookup(struct net *net, const struct flowi4 *flp,
struct fib_result *res)
{
- struct fib_table *table;
+ int err = -ENETUNREACH;
- table = fib_get_table(net, RT_TABLE_LOCAL);
- if (!fib_table_lookup(table, flp, res, FIB_LOOKUP_NOREF))
- return 0;
-
- table = fib_get_table(net, RT_TABLE_MAIN);
- if (!fib_table_lookup(table, flp, res, FIB_LOOKUP_NOREF))
- return 0;
- return -ENETUNREACH;
+ rcu_read_lock();
+
+ if (!fib_table_lookup(fib_get_table(net, RT_TABLE_LOCAL), flp, res,
+ FIB_LOOKUP_NOREF) ||
+ !fib_table_lookup(fib_get_table(net, RT_TABLE_MAIN), flp, res,
+ FIB_LOOKUP_NOREF))
+ err = 0;
+
+ rcu_read_unlock();
+
+ return err;
}
#else /* CONFIG_IP_MULTIPLE_TABLES */
@@ -247,20 +250,25 @@ static inline int fib_lookup(struct net
struct fib_result *res)
{
if (!net->ipv4.fib_has_custom_rules) {
+ int err = -ENETUNREACH;
+
+ rcu_read_lock();
+
res->tclassid = 0;
- if (net->ipv4.fib_local &&
- !fib_table_lookup(net->ipv4.fib_local, flp, res,
- FIB_LOOKUP_NOREF))
- return 0;
- if (net->ipv4.fib_main &&
- !fib_table_lookup(net->ipv4.fib_main, flp, res,
- FIB_LOOKUP_NOREF))
- return 0;
- if (net->ipv4.fib_default &&
- !fib_table_lookup(net->ipv4.fib_default, flp, res,
- FIB_LOOKUP_NOREF))
- return 0;
- return -ENETUNREACH;
+ if ((net->ipv4.fib_local &&
+ !fib_table_lookup(net->ipv4.fib_local, flp, res,
+ FIB_LOOKUP_NOREF)) ||
+ (net->ipv4.fib_main &&
+ !fib_table_lookup(net->ipv4.fib_main, flp, res,
+ FIB_LOOKUP_NOREF)) ||
+ (net->ipv4.fib_default &&
+ !fib_table_lookup(net->ipv4.fib_default, flp, res,
+ FIB_LOOKUP_NOREF)))
+ err = 0;
+
+ rcu_read_unlock();
+
+ return err;
}
return __fib_lookup(net, flp, res);
}
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -109,6 +109,7 @@ struct fib_table *fib_new_table(struct n
return tb;
}
+/* caller must hold either rtnl or rcu read lock */
struct fib_table *fib_get_table(struct net *net, u32 id)
{
struct fib_table *tb;
@@ -119,15 +120,11 @@ struct fib_table *fib_get_table(struct n
id = RT_TABLE_MAIN;
h = id & (FIB_TABLE_HASHSZ - 1);
- rcu_read_lock();
head = &net->ipv4.fib_table_hash[h];
hlist_for_each_entry_rcu(tb, head, tb_hlist) {
- if (tb->tb_id == id) {
- rcu_read_unlock();
+ if (tb->tb_id == id)
return tb;
- }
}
- rcu_read_unlock();
return NULL;
}
#endif /* CONFIG_IP_MULTIPLE_TABLES */
@@ -167,16 +164,18 @@ static inline unsigned int __inet_dev_ad
if (ipv4_is_multicast(addr))
return RTN_MULTICAST;
+ rcu_read_lock();
+
local_table = fib_get_table(net, RT_TABLE_LOCAL);
if (local_table) {
ret = RTN_UNICAST;
- rcu_read_lock();
if (!fib_table_lookup(local_table, &fl4, &res, FIB_LOOKUP_NOREF)) {
if (!dev || dev == res.fi->fib_dev)
ret = res.type;
}
- rcu_read_unlock();
}
+
+ rcu_read_unlock();
return ret;
}
@@ -923,7 +922,7 @@ no_promotions:
#undef BRD1_OK
}
-static void nl_fib_lookup(struct fib_result_nl *frn, struct fib_table *tb)
+static void nl_fib_lookup(struct net *net, struct fib_result_nl *frn)
{
struct fib_result res;
@@ -933,6 +932,11 @@ static void nl_fib_lookup(struct fib_res
.flowi4_tos = frn->fl_tos,
.flowi4_scope = frn->fl_scope,
};
+ struct fib_table *tb;
+
+ rcu_read_lock();
+
+ tb = fib_get_table(net, frn->tb_id_in);
frn->err = -ENOENT;
if (tb) {
@@ -949,6 +953,8 @@ static void nl_fib_lookup(struct fib_res
}
local_bh_enable();
}
+
+ rcu_read_unlock();
}
static void nl_fib_input(struct sk_buff *skb)
@@ -956,7 +962,6 @@ static void nl_fib_input(struct sk_buff
struct net *net;
struct fib_result_nl *frn;
struct nlmsghdr *nlh;
- struct fib_table *tb;
u32 portid;
net = sock_net(skb->sk);
@@ -972,9 +977,7 @@ static void nl_fib_input(struct sk_buff
nlh = nlmsg_hdr(skb);
frn = (struct fib_result_nl *) nlmsg_data(nlh);
- tb = fib_get_table(net, frn->tb_id_in);
-
- nl_fib_lookup(frn, tb);
+ nl_fib_lookup(net, frn);
portid = NETLINK_CB(skb).portid; /* netlink portid */
NETLINK_CB(skb).portid = 0; /* from kernel */
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c
@@ -81,27 +81,25 @@ static int fib4_rule_action(struct fib_r
break;
case FR_ACT_UNREACHABLE:
- err = -ENETUNREACH;
- goto errout;
+ return -ENETUNREACH;
case FR_ACT_PROHIBIT:
- err = -EACCES;
- goto errout;
+ return -EACCES;
case FR_ACT_BLACKHOLE:
default:
- err = -EINVAL;
- goto errout;
+ return -EINVAL;
}
+ rcu_read_lock();
+
tbl = fib_get_table(rule->fr_net, rule->table);
- if (!tbl)
- goto errout;
+ if (tbl)
+ err = fib_table_lookup(tbl, &flp->u.ip4,
+ (struct fib_result *)arg->result,
+ arg->flags);
- err = fib_table_lookup(tbl, &flp->u.ip4, (struct fib_result *) arg->result, arg->flags);
- if (err > 0)
- err = -EAGAIN;
-errout:
+ rcu_read_unlock();
return err;
}
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1181,72 +1181,6 @@ err:
return err;
}
-/* should be called with rcu_read_lock */
-static int check_leaf(struct fib_table *tb, struct trie *t, struct tnode *l,
- t_key key, const struct flowi4 *flp,
- struct fib_result *res, int fib_flags)
-{
- struct leaf_info *li;
- struct hlist_head *hhead = &l->list;
-
- hlist_for_each_entry_rcu(li, hhead, hlist) {
- struct fib_alias *fa;
-
- if (l->key != (key & li->mask_plen))
- continue;
-
- list_for_each_entry_rcu(fa, &li->falh, fa_list) {
- struct fib_info *fi = fa->fa_info;
- int nhsel, err;
-
- if (fa->fa_tos && fa->fa_tos != flp->flowi4_tos)
- continue;
- if (fi->fib_dead)
- continue;
- if (fa->fa_info->fib_scope < flp->flowi4_scope)
- continue;
- fib_alias_accessed(fa);
- err = fib_props[fa->fa_type].error;
- if (unlikely(err < 0)) {
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(t->stats->semantic_match_passed);
-#endif
- return err;
- }
- if (fi->fib_flags & RTNH_F_DEAD)
- continue;
- for (nhsel = 0; nhsel < fi->fib_nhs; nhsel++) {
- const struct fib_nh *nh = &fi->fib_nh[nhsel];
-
- if (nh->nh_flags & RTNH_F_DEAD)
- continue;
- if (flp->flowi4_oif && flp->flowi4_oif != nh->nh_oif)
- continue;
-
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(t->stats->semantic_match_passed);
-#endif
- res->prefixlen = li->plen;
- res->nh_sel = nhsel;
- res->type = fa->fa_type;
- res->scope = fi->fib_scope;
- res->fi = fi;
- res->table = tb;
- res->fa_head = &li->falh;
- if (!(fib_flags & FIB_LOOKUP_NOREF))
- atomic_inc(&fi->fib_clntref);
- return 0;
- }
- }
-
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(t->stats->semantic_match_miss);
-#endif
- }
-
- return 1;
-}
-
static inline t_key prefix_mismatch(t_key key, struct tnode *n)
{
t_key prefix = n->key;
@@ -1254,6 +1188,7 @@ static inline t_key prefix_mismatch(t_ke
return (key ^ prefix) & (prefix | -prefix);
}
+/* should be called with rcu_read_lock */
int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
struct fib_result *res, int fib_flags)
{
@@ -1263,14 +1198,12 @@ int fib_table_lookup(struct fib_table *t
#endif
const t_key key = ntohl(flp->daddr);
struct tnode *n, *pn;
+ struct leaf_info *li;
t_key cindex;
- int ret = 1;
-
- rcu_read_lock();
n = rcu_dereference(t->trie);
if (!n)
- goto failed;
+ return -EAGAIN;
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(stats->gets);
@@ -1350,7 +1283,7 @@ backtrace:
pn = node_parent_rcu(pn);
if (unlikely(!pn))
- goto failed;
+ return -EAGAIN;
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(stats->backtrack);
#endif
@@ -1368,12 +1301,62 @@ backtrace:
found:
/* Step 3: Process the leaf, if that fails fall back to backtracing */
- ret = check_leaf(tb, t, n, key, flp, res, fib_flags);
- if (unlikely(ret > 0))
- goto backtrace;
-failed:
- rcu_read_unlock();
- return ret;
+ hlist_for_each_entry_rcu(li, &n->list, hlist) {
+ struct fib_alias *fa;
+
+ if ((key ^ n->key) & li->mask_plen)
+ continue;
+
+ list_for_each_entry_rcu(fa, &li->falh, fa_list) {
+ struct fib_info *fi = fa->fa_info;
+ int nhsel, err;
+
+ if (fa->fa_tos && fa->fa_tos != flp->flowi4_tos)
+ continue;
+ if (fi->fib_dead)
+ continue;
+ if (fa->fa_info->fib_scope < flp->flowi4_scope)
+ continue;
+ fib_alias_accessed(fa);
+ err = fib_props[fa->fa_type].error;
+ if (unlikely(err < 0)) {
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(stats->semantic_match_passed);
+#endif
+ return err;
+ }
+ if (fi->fib_flags & RTNH_F_DEAD)
+ continue;
+ for (nhsel = 0; nhsel < fi->fib_nhs; nhsel++) {
+ const struct fib_nh *nh = &fi->fib_nh[nhsel];
+
+ if (nh->nh_flags & RTNH_F_DEAD)
+ continue;
+ if (flp->flowi4_oif && flp->flowi4_oif != nh->nh_oif)
+ continue;
+
+ if (!(fib_flags & FIB_LOOKUP_NOREF))
+ atomic_inc(&fi->fib_clntref);
+
+ res->prefixlen = li->plen;
+ res->nh_sel = nhsel;
+ res->type = fa->fa_type;
+ res->scope = fi->fib_scope;
+ res->fi = fi;
+ res->table = tb;
+ res->fa_head = &li->falh;
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(stats->semantic_match_passed);
+#endif
+ return err;
+ }
+ }
+
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(stats->semantic_match_miss);
+#endif
+ }
+ goto backtrace;
}
EXPORT_SYMBOL_GPL(fib_table_lookup);

345
target/linux/generic/pending-3.18/080-12-fib_trie-Move-resize-to-after-inflate-halve.patch
View File

@ -1,345 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:31 -0800
Subject: [PATCH] fib_trie: Move resize to after inflate/halve
This change consists of a cut/paste of resize to behind inflate and halve
so that I could remove the two function prototypes.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -149,8 +149,6 @@ struct trie {
static void tnode_put_child_reorg(struct tnode *tn, unsigned long i,
struct tnode *n, int wasfull);
static struct tnode *resize(struct trie *t, struct tnode *tn);
-static struct tnode *inflate(struct trie *t, struct tnode *tn);
-static struct tnode *halve(struct trie *t, struct tnode *tn);
/* tnodes to free after resize(); protected by RTNL */
static struct callback_head *tnode_free_head;
static size_t tnode_free_size;
@@ -447,161 +445,6 @@ static void put_child_root(struct tnode
rcu_assign_pointer(t->trie, n);
}
-#define MAX_WORK 10
-static struct tnode *resize(struct trie *t, struct tnode *tn)
-{
- struct tnode *old_tn, *n = NULL;
- int inflate_threshold_use;
- int halve_threshold_use;
- int max_work;
-
- if (!tn)
- return NULL;
-
- pr_debug("In tnode_resize %p inflate_threshold=%d threshold=%d\n",
- tn, inflate_threshold, halve_threshold);
-
- /* No children */
- if (tn->empty_children > (tnode_child_length(tn) - 1))
- goto no_children;
-
- /* One child */
- if (tn->empty_children == (tnode_child_length(tn) - 1))
- goto one_child;
- /*
- * Double as long as the resulting node has a number of
- * nonempty nodes that are above the threshold.
- */
-
- /*
- * From "Implementing a dynamic compressed trie" by Stefan Nilsson of
- * the Helsinki University of Technology and Matti Tikkanen of Nokia
- * Telecommunications, page 6:
- * "A node is doubled if the ratio of non-empty children to all
- * children in the *doubled* node is at least 'high'."
- *
- * 'high' in this instance is the variable 'inflate_threshold'. It
- * is expressed as a percentage, so we multiply it with
- * tnode_child_length() and instead of multiplying by 2 (since the
- * child array will be doubled by inflate()) and multiplying
- * the left-hand side by 100 (to handle the percentage thing) we
- * multiply the left-hand side by 50.
- *
- * The left-hand side may look a bit weird: tnode_child_length(tn)
- * - tn->empty_children is of course the number of non-null children
- * in the current node. tn->full_children is the number of "full"
- * children, that is non-null tnodes with a skip value of 0.
- * All of those will be doubled in the resulting inflated tnode, so
- * we just count them one extra time here.
- *
- * A clearer way to write this would be:
- *
- * to_be_doubled = tn->full_children;
- * not_to_be_doubled = tnode_child_length(tn) - tn->empty_children -
- * tn->full_children;
- *
- * new_child_length = tnode_child_length(tn) * 2;
- *
- * new_fill_factor = 100 * (not_to_be_doubled + 2*to_be_doubled) /
- * new_child_length;
- * if (new_fill_factor >= inflate_threshold)
- *
- * ...and so on, tho it would mess up the while () loop.
- *
- * anyway,
- * 100 * (not_to_be_doubled + 2*to_be_doubled) / new_child_length >=
- * inflate_threshold
- *
- * avoid a division:
- * 100 * (not_to_be_doubled + 2*to_be_doubled) >=
- * inflate_threshold * new_child_length
- *
- * expand not_to_be_doubled and to_be_doubled, and shorten:
- * 100 * (tnode_child_length(tn) - tn->empty_children +
- * tn->full_children) >= inflate_threshold * new_child_length
- *
- * expand new_child_length:
- * 100 * (tnode_child_length(tn) - tn->empty_children +
- * tn->full_children) >=
- * inflate_threshold * tnode_child_length(tn) * 2
- *
- * shorten again:
- * 50 * (tn->full_children + tnode_child_length(tn) -
- * tn->empty_children) >= inflate_threshold *
- * tnode_child_length(tn)
- *
- */
-
- /* Keep root node larger */
-
- if (!node_parent(tn)) {
- inflate_threshold_use = inflate_threshold_root;
- halve_threshold_use = halve_threshold_root;
- } else {
- inflate_threshold_use = inflate_threshold;
- halve_threshold_use = halve_threshold;
- }
-
- max_work = MAX_WORK;
- while ((tn->full_children > 0 && max_work-- &&
- 50 * (tn->full_children + tnode_child_length(tn)
- - tn->empty_children)
- >= inflate_threshold_use * tnode_child_length(tn))) {
-
- old_tn = tn;
- tn = inflate(t, tn);
-
- if (IS_ERR(tn)) {
- tn = old_tn;
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(t->stats->resize_node_skipped);
-#endif
- break;
- }
- }
-
- /* Return if at least one inflate is run */
- if (max_work != MAX_WORK)
- return tn;
-
- /*
- * Halve as long as the number of empty children in this
- * node is above threshold.
- */
-
- max_work = MAX_WORK;
- while (tn->bits > 1 && max_work-- &&
- 100 * (tnode_child_length(tn) - tn->empty_children) <
- halve_threshold_use * tnode_child_length(tn)) {
-
- old_tn = tn;
- tn = halve(t, tn);
- if (IS_ERR(tn)) {
- tn = old_tn;
-#ifdef CONFIG_IP_FIB_TRIE_STATS
- this_cpu_inc(t->stats->resize_node_skipped);
-#endif
- break;
- }
- }
-
-
- /* Only one child remains */
- if (tn->empty_children == (tnode_child_length(tn) - 1)) {
- unsigned long i;
-one_child:
- for (i = tnode_child_length(tn); !n && i;)
- n = tnode_get_child(tn, --i);
-no_children:
- /* compress one level */
- node_set_parent(n, NULL);
- tnode_free_safe(tn);
- return n;
- }
- return tn;
-}
-
-
static void tnode_clean_free(struct tnode *tn)
{
struct tnode *tofree;
@@ -804,6 +647,160 @@ nomem:
return ERR_PTR(-ENOMEM);
}
+#define MAX_WORK 10
+static struct tnode *resize(struct trie *t, struct tnode *tn)
+{
+ struct tnode *old_tn, *n = NULL;
+ int inflate_threshold_use;
+ int halve_threshold_use;
+ int max_work;
+
+ if (!tn)
+ return NULL;
+
+ pr_debug("In tnode_resize %p inflate_threshold=%d threshold=%d\n",
+ tn, inflate_threshold, halve_threshold);
+
+ /* No children */
+ if (tn->empty_children > (tnode_child_length(tn) - 1))
+ goto no_children;
+
+ /* One child */
+ if (tn->empty_children == (tnode_child_length(tn) - 1))
+ goto one_child;
+ /*
+ * Double as long as the resulting node has a number of
+ * nonempty nodes that are above the threshold.
+ */
+
+ /*
+ * From "Implementing a dynamic compressed trie" by Stefan Nilsson of
+ * the Helsinki University of Technology and Matti Tikkanen of Nokia
+ * Telecommunications, page 6:
+ * "A node is doubled if the ratio of non-empty children to all
+ * children in the *doubled* node is at least 'high'."
+ *
+ * 'high' in this instance is the variable 'inflate_threshold'. It
+ * is expressed as a percentage, so we multiply it with
+ * tnode_child_length() and instead of multiplying by 2 (since the
+ * child array will be doubled by inflate()) and multiplying
+ * the left-hand side by 100 (to handle the percentage thing) we
+ * multiply the left-hand side by 50.
+ *
+ * The left-hand side may look a bit weird: tnode_child_length(tn)
+ * - tn->empty_children is of course the number of non-null children
+ * in the current node. tn->full_children is the number of "full"
+ * children, that is non-null tnodes with a skip value of 0.
+ * All of those will be doubled in the resulting inflated tnode, so
+ * we just count them one extra time here.
+ *
+ * A clearer way to write this would be:
+ *
+ * to_be_doubled = tn->full_children;
+ * not_to_be_doubled = tnode_child_length(tn) - tn->empty_children -
+ * tn->full_children;
+ *
+ * new_child_length = tnode_child_length(tn) * 2;
+ *
+ * new_fill_factor = 100 * (not_to_be_doubled + 2*to_be_doubled) /
+ * new_child_length;
+ * if (new_fill_factor >= inflate_threshold)
+ *
+ * ...and so on, tho it would mess up the while () loop.
+ *
+ * anyway,
+ * 100 * (not_to_be_doubled + 2*to_be_doubled) / new_child_length >=
+ * inflate_threshold
+ *
+ * avoid a division:
+ * 100 * (not_to_be_doubled + 2*to_be_doubled) >=
+ * inflate_threshold * new_child_length
+ *
+ * expand not_to_be_doubled and to_be_doubled, and shorten:
+ * 100 * (tnode_child_length(tn) - tn->empty_children +
+ * tn->full_children) >= inflate_threshold * new_child_length
+ *
+ * expand new_child_length:
+ * 100 * (tnode_child_length(tn) - tn->empty_children +
+ * tn->full_children) >=
+ * inflate_threshold * tnode_child_length(tn) * 2
+ *
+ * shorten again:
+ * 50 * (tn->full_children + tnode_child_length(tn) -
+ * tn->empty_children) >= inflate_threshold *
+ * tnode_child_length(tn)
+ *
+ */
+
+ /* Keep root node larger */
+
+ if (!node_parent(tn)) {
+ inflate_threshold_use = inflate_threshold_root;
+ halve_threshold_use = halve_threshold_root;
+ } else {
+ inflate_threshold_use = inflate_threshold;
+ halve_threshold_use = halve_threshold;
+ }
+
+ max_work = MAX_WORK;
+ while ((tn->full_children > 0 && max_work-- &&
+ 50 * (tn->full_children + tnode_child_length(tn)
+ - tn->empty_children)
+ >= inflate_threshold_use * tnode_child_length(tn))) {
+
+ old_tn = tn;
+ tn = inflate(t, tn);
+
+ if (IS_ERR(tn)) {
+ tn = old_tn;
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(t->stats->resize_node_skipped);
+#endif
+ break;
+ }
+ }
+
+ /* Return if at least one inflate is run */
+ if (max_work != MAX_WORK)
+ return tn;
+
+ /*
+ * Halve as long as the number of empty children in this
+ * node is above threshold.
+ */
+
+ max_work = MAX_WORK;
+ while (tn->bits > 1 && max_work-- &&
+ 100 * (tnode_child_length(tn) - tn->empty_children) <
+ halve_threshold_use * tnode_child_length(tn)) {
+
+ old_tn = tn;
+ tn = halve(t, tn);
+ if (IS_ERR(tn)) {
+ tn = old_tn;
+#ifdef CONFIG_IP_FIB_TRIE_STATS
+ this_cpu_inc(t->stats->resize_node_skipped);
+#endif
+ break;
+ }
+ }
+
+
+ /* Only one child remains */
+ if (tn->empty_children == (tnode_child_length(tn) - 1)) {
+ unsigned long i;
+one_child:
+ for (i = tnode_child_length(tn); !n && i;)
+ n = tnode_get_child(tn, --i);
+no_children:
+ /* compress one level */
+ node_set_parent(n, NULL);
+ tnode_free_safe(tn);
+ return n;
+ }
+ return tn;
+}
+
/* readside must use rcu_read_lock currently dump routines
via get_fa_head and dump */

250
target/linux/generic/pending-3.18/080-13-fib_trie-Add-functions-should_inflate-and-should_hal.patch
View File

@ -1,250 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:37 -0800
Subject: [PATCH] fib_trie: Add functions should_inflate and should_halve
This change pulls the logic for if we should inflate/halve the nodes out
into separate functions. It also addresses what I believe is a bug where 1
full node is all that is needed to keep a node from ever being halved.
Simple script to reproduce the issue:
modprobe dummy; ifconfig dummy0 up
for i in `seq 0 255`; do ifconfig dummy0:$i 10.0.${i}.1/24 up; done
ifconfig dummy0:256 10.0.255.33/16 up
for i in `seq 0 254`; do ifconfig dummy0:$i down; done
Results from /proc/net/fib_triestat
Before:
Local:
Aver depth: 3.00
Max depth: 4
Leaves: 17
Prefixes: 18
Internal nodes: 11
1: 8 2: 2 10: 1
Pointers: 1048
Null ptrs: 1021
Total size: 11 kB
After:
Local:
Aver depth: 3.41
Max depth: 5
Leaves: 17
Prefixes: 18
Internal nodes: 12
1: 8 2: 3 3: 1
Pointers: 36
Null ptrs: 8
Total size: 3 kB
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -647,12 +647,94 @@ nomem:
return ERR_PTR(-ENOMEM);
}
+/* From "Implementing a dynamic compressed trie" by Stefan Nilsson of
+ * the Helsinki University of Technology and Matti Tikkanen of Nokia
+ * Telecommunications, page 6:
+ * "A node is doubled if the ratio of non-empty children to all
+ * children in the *doubled* node is at least 'high'."
+ *
+ * 'high' in this instance is the variable 'inflate_threshold'. It
+ * is expressed as a percentage, so we multiply it with
+ * tnode_child_length() and instead of multiplying by 2 (since the
+ * child array will be doubled by inflate()) and multiplying
+ * the left-hand side by 100 (to handle the percentage thing) we
+ * multiply the left-hand side by 50.
+ *
+ * The left-hand side may look a bit weird: tnode_child_length(tn)
+ * - tn->empty_children is of course the number of non-null children
+ * in the current node. tn->full_children is the number of "full"
+ * children, that is non-null tnodes with a skip value of 0.
+ * All of those will be doubled in the resulting inflated tnode, so
+ * we just count them one extra time here.
+ *
+ * A clearer way to write this would be:
+ *
+ * to_be_doubled = tn->full_children;
+ * not_to_be_doubled = tnode_child_length(tn) - tn->empty_children -
+ * tn->full_children;
+ *
+ * new_child_length = tnode_child_length(tn) * 2;
+ *
+ * new_fill_factor = 100 * (not_to_be_doubled + 2*to_be_doubled) /
+ * new_child_length;
+ * if (new_fill_factor >= inflate_threshold)
+ *
+ * ...and so on, tho it would mess up the while () loop.
+ *
+ * anyway,
+ * 100 * (not_to_be_doubled + 2*to_be_doubled) / new_child_length >=
+ * inflate_threshold
+ *
+ * avoid a division:
+ * 100 * (not_to_be_doubled + 2*to_be_doubled) >=
+ * inflate_threshold * new_child_length
+ *
+ * expand not_to_be_doubled and to_be_doubled, and shorten:
+ * 100 * (tnode_child_length(tn) - tn->empty_children +
+ * tn->full_children) >= inflate_threshold * new_child_length
+ *
+ * expand new_child_length:
+ * 100 * (tnode_child_length(tn) - tn->empty_children +
+ * tn->full_children) >=
+ * inflate_threshold * tnode_child_length(tn) * 2
+ *
+ * shorten again:
+ * 50 * (tn->full_children + tnode_child_length(tn) -
+ * tn->empty_children) >= inflate_threshold *
+ * tnode_child_length(tn)
+ *
+ */
+static bool should_inflate(const struct tnode *tn)
+{
+ unsigned long used = tnode_child_length(tn);
+ unsigned long threshold = used;
+
+ /* Keep root node larger */
+ threshold *= node_parent(tn) ? inflate_threshold :
+ inflate_threshold_root;
+ used += tn->full_children;
+ used -= tn->empty_children;
+
+ return tn->pos && ((50 * used) >= threshold);
+}
+
+static bool should_halve(const struct tnode *tn)
+{
+ unsigned long used = tnode_child_length(tn);
+ unsigned long threshold = used;
+
+ /* Keep root node larger */
+ threshold *= node_parent(tn) ? halve_threshold :
+ halve_threshold_root;
+ used -= tn->empty_children;
+
+ return (tn->bits > 1) && ((100 * used) < threshold);
+}
+
#define MAX_WORK 10
static struct tnode *resize(struct trie *t, struct tnode *tn)
{
struct tnode *old_tn, *n = NULL;
- int inflate_threshold_use;
- int halve_threshold_use;
int max_work;
if (!tn)
@@ -668,86 +750,12 @@ static struct tnode *resize(struct trie
/* One child */
if (tn->empty_children == (tnode_child_length(tn) - 1))
goto one_child;
- /*
- * Double as long as the resulting node has a number of
- * nonempty nodes that are above the threshold.
- */
- /*
- * From "Implementing a dynamic compressed trie" by Stefan Nilsson of
- * the Helsinki University of Technology and Matti Tikkanen of Nokia
- * Telecommunications, page 6:
- * "A node is doubled if the ratio of non-empty children to all
- * children in the *doubled* node is at least 'high'."
- *
- * 'high' in this instance is the variable 'inflate_threshold'. It
- * is expressed as a percentage, so we multiply it with
- * tnode_child_length() and instead of multiplying by 2 (since the
- * child array will be doubled by inflate()) and multiplying
- * the left-hand side by 100 (to handle the percentage thing) we
- * multiply the left-hand side by 50.
- *
- * The left-hand side may look a bit weird: tnode_child_length(tn)
- * - tn->empty_children is of course the number of non-null children
- * in the current node. tn->full_children is the number of "full"
- * children, that is non-null tnodes with a skip value of 0.
- * All of those will be doubled in the resulting inflated tnode, so
- * we just count them one extra time here.
- *
- * A clearer way to write this would be:
- *
- * to_be_doubled = tn->full_children;
- * not_to_be_doubled = tnode_child_length(tn) - tn->empty_children -
- * tn->full_children;
- *
- * new_child_length = tnode_child_length(tn) * 2;
- *
- * new_fill_factor = 100 * (not_to_be_doubled + 2*to_be_doubled) /
- * new_child_length;
- * if (new_fill_factor >= inflate_threshold)
- *
- * ...and so on, tho it would mess up the while () loop.
- *
- * anyway,
- * 100 * (not_to_be_doubled + 2*to_be_doubled) / new_child_length >=
- * inflate_threshold
- *
- * avoid a division:
- * 100 * (not_to_be_doubled + 2*to_be_doubled) >=
- * inflate_threshold * new_child_length
- *
- * expand not_to_be_doubled and to_be_doubled, and shorten:
- * 100 * (tnode_child_length(tn) - tn->empty_children +
- * tn->full_children) >= inflate_threshold * new_child_length
- *
- * expand new_child_length:
- * 100 * (tnode_child_length(tn) - tn->empty_children +
- * tn->full_children) >=
- * inflate_threshold * tnode_child_length(tn) * 2
- *
- * shorten again:
- * 50 * (tn->full_children + tnode_child_length(tn) -
- * tn->empty_children) >= inflate_threshold *
- * tnode_child_length(tn)
- *
+ /* Double as long as the resulting node has a number of
+ * nonempty nodes that are above the threshold.
*/
-
- /* Keep root node larger */
-
- if (!node_parent(tn)) {
- inflate_threshold_use = inflate_threshold_root;
- halve_threshold_use = halve_threshold_root;
- } else {
- inflate_threshold_use = inflate_threshold;
- halve_threshold_use = halve_threshold;
- }
-
max_work = MAX_WORK;
- while ((tn->full_children > 0 && max_work-- &&
- 50 * (tn->full_children + tnode_child_length(tn)
- - tn->empty_children)
- >= inflate_threshold_use * tnode_child_length(tn))) {
-
+ while (should_inflate(tn) && max_work--) {
old_tn = tn;
tn = inflate(t, tn);
@@ -764,16 +772,11 @@ static struct tnode *resize(struct trie
if (max_work != MAX_WORK)
return tn;
- /*
- * Halve as long as the number of empty children in this
+ /* Halve as long as the number of empty children in this
* node is above threshold.
*/
-
max_work = MAX_WORK;
- while (tn->bits > 1 && max_work-- &&
- 100 * (tnode_child_length(tn) - tn->empty_children) <
- halve_threshold_use * tnode_child_length(tn)) {
-
+ while (should_halve(tn) && max_work--) {
old_tn = tn;
tn = halve(t, tn);
if (IS_ERR(tn)) {

336
target/linux/generic/pending-3.18/080-14-fib_trie-Push-assignment-of-child-to-parent-down-int.patch
View File

@ -1,336 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:43 -0800
Subject: [PATCH] fib_trie: Push assignment of child to parent down into
inflate/halve
This change makes it so that the assignment of the tnode to the parent is
handled directly within whatever function is currently handling the node be
it inflate, halve, or resize. By doing this we can avoid some of the need
to set NULL pointers in the tree while we are resizing the subnodes.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -146,9 +146,7 @@ struct trie {
#endif
};
-static void tnode_put_child_reorg(struct tnode *tn, unsigned long i,
- struct tnode *n, int wasfull);
-static struct tnode *resize(struct trie *t, struct tnode *tn);
+static void resize(struct trie *t, struct tnode *tn);
/* tnodes to free after resize(); protected by RTNL */
static struct callback_head *tnode_free_head;
static size_t tnode_free_size;
@@ -396,22 +394,13 @@ static inline int tnode_full(const struc
return n && ((n->pos + n->bits) == tn->pos) && IS_TNODE(n);
}
-static inline void put_child(struct tnode *tn, unsigned long i,
- struct tnode *n)
-{
- tnode_put_child_reorg(tn, i, n, -1);
-}
-
- /*
- * Add a child at position i overwriting the old value.
- * Update the value of full_children and empty_children.
- */
-
-static void tnode_put_child_reorg(struct tnode *tn, unsigned long i,
- struct tnode *n, int wasfull)
+/* Add a child at position i overwriting the old value.
+ * Update the value of full_children and empty_children.
+ */
+static void put_child(struct tnode *tn, unsigned long i, struct tnode *n)
{
struct tnode *chi = rtnl_dereference(tn->child[i]);
- int isfull;
+ int isfull, wasfull;
BUG_ON(i >= tnode_child_length(tn));
@@ -422,10 +411,9 @@ static void tnode_put_child_reorg(struct
tn->empty_children--;
/* update fullChildren */
- if (wasfull == -1)
- wasfull = tnode_full(tn, chi);
-
+ wasfull = tnode_full(tn, chi);
isfull = tnode_full(tn, n);
+
if (wasfull && !isfull)
tn->full_children--;
else if (!wasfull && isfull)
@@ -458,9 +446,10 @@ static void tnode_clean_free(struct tnod
node_free(tn);
}
-static struct tnode *inflate(struct trie *t, struct tnode *oldtnode)
+static int inflate(struct trie *t, struct tnode *oldtnode)
{
unsigned long olen = tnode_child_length(oldtnode);
+ struct tnode *tp = node_parent(oldtnode);
struct tnode *tn;
unsigned long i;
t_key m;
@@ -468,9 +457,8 @@ static struct tnode *inflate(struct trie
pr_debug("In inflate\n");
tn = tnode_new(oldtnode->key, oldtnode->pos - 1, oldtnode->bits + 1);
-
if (!tn)
- return ERR_PTR(-ENOMEM);
+ return -ENOMEM;
/*
* Preallocate and store tnodes before the actual work so we
@@ -564,30 +552,36 @@ static struct tnode *inflate(struct trie
put_child(left, j, rtnl_dereference(inode->child[j]));
put_child(right, j, rtnl_dereference(inode->child[j + size]));
}
- put_child(tn, 2*i, resize(t, left));
- put_child(tn, 2*i+1, resize(t, right));
+
+ put_child(tn, 2 * i, left);
+ put_child(tn, 2 * i + 1, right);
tnode_free_safe(inode);
+
+ resize(t, left);
+ resize(t, right);
}
+
+ put_child_root(tp, t, tn->key, tn);
tnode_free_safe(oldtnode);
- return tn;
+ return 0;
nomem:
tnode_clean_free(tn);
- return ERR_PTR(-ENOMEM);
+ return -ENOMEM;
}
-static struct tnode *halve(struct trie *t, struct tnode *oldtnode)
+static int halve(struct trie *t, struct tnode *oldtnode)
{
unsigned long olen = tnode_child_length(oldtnode);
+ struct tnode *tp = node_parent(oldtnode);
struct tnode *tn, *left, *right;
int i;
pr_debug("In halve\n");
tn = tnode_new(oldtnode->key, oldtnode->pos + 1, oldtnode->bits - 1);
-
if (!tn)
- return ERR_PTR(-ENOMEM);
+ return -ENOMEM;
/*
* Preallocate and store tnodes before the actual work so we
@@ -606,8 +600,10 @@ static struct tnode *halve(struct trie *
newn = tnode_new(left->key, oldtnode->pos, 1);
- if (!newn)
- goto nomem;
+ if (!newn) {
+ tnode_clean_free(tn);
+ return -ENOMEM;
+ }
put_child(tn, i/2, newn);
}
@@ -635,16 +631,18 @@ static struct tnode *halve(struct trie *
/* Two nonempty children */
newBinNode = tnode_get_child(tn, i/2);
- put_child(tn, i/2, NULL);
put_child(newBinNode, 0, left);
put_child(newBinNode, 1, right);
- put_child(tn, i/2, resize(t, newBinNode));
+
+ put_child(tn, i / 2, newBinNode);
+
+ resize(t, newBinNode);
}
+
+ put_child_root(tp, t, tn->key, tn);
tnode_free_safe(oldtnode);
- return tn;
-nomem:
- tnode_clean_free(tn);
- return ERR_PTR(-ENOMEM);
+
+ return 0;
}
/* From "Implementing a dynamic compressed trie" by Stefan Nilsson of
@@ -704,45 +702,48 @@ nomem:
* tnode_child_length(tn)
*
*/
-static bool should_inflate(const struct tnode *tn)
+static bool should_inflate(const struct tnode *tp, const struct tnode *tn)
{
unsigned long used = tnode_child_length(tn);
unsigned long threshold = used;
/* Keep root node larger */
- threshold *= node_parent(tn) ? inflate_threshold :
- inflate_threshold_root;
+ threshold *= tp ? inflate_threshold : inflate_threshold_root;
used += tn->full_children;
used -= tn->empty_children;
return tn->pos && ((50 * used) >= threshold);
}
-static bool should_halve(const struct tnode *tn)
+static bool should_halve(const struct tnode *tp, const struct tnode *tn)
{
unsigned long used = tnode_child_length(tn);
unsigned long threshold = used;
/* Keep root node larger */
- threshold *= node_parent(tn) ? halve_threshold :
- halve_threshold_root;
+ threshold *= tp ? halve_threshold : halve_threshold_root;
used -= tn->empty_children;
return (tn->bits > 1) && ((100 * used) < threshold);
}
#define MAX_WORK 10
-static struct tnode *resize(struct trie *t, struct tnode *tn)
+static void resize(struct trie *t, struct tnode *tn)
{
- struct tnode *old_tn, *n = NULL;
+ struct tnode *tp = node_parent(tn), *n = NULL;
+ struct tnode __rcu **cptr;
int max_work;
- if (!tn)
- return NULL;
-
pr_debug("In tnode_resize %p inflate_threshold=%d threshold=%d\n",
tn, inflate_threshold, halve_threshold);
+ /* track the tnode via the pointer from the parent instead of
+ * doing it ourselves. This way we can let RCU fully do its
+ * thing without us interfering
+ */
+ cptr = tp ? &tp->child[get_index(tn->key, tp)] : &t->trie;
+ BUG_ON(tn != rtnl_dereference(*cptr));
+
/* No children */
if (tn->empty_children > (tnode_child_length(tn) - 1))
goto no_children;
@@ -755,39 +756,35 @@ static struct tnode *resize(struct trie
* nonempty nodes that are above the threshold.
*/
max_work = MAX_WORK;
- while (should_inflate(tn) && max_work--) {
- old_tn = tn;
- tn = inflate(t, tn);
-
- if (IS_ERR(tn)) {
- tn = old_tn;
+ while (should_inflate(tp, tn) && max_work--) {
+ if (inflate(t, tn)) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(t->stats->resize_node_skipped);
#endif
break;
}
+
+ tn = rtnl_dereference(*cptr);
}
/* Return if at least one inflate is run */
if (max_work != MAX_WORK)
- return tn;
+ return;
/* Halve as long as the number of empty children in this
* node is above threshold.
*/
max_work = MAX_WORK;
- while (should_halve(tn) && max_work--) {
- old_tn = tn;
- tn = halve(t, tn);
- if (IS_ERR(tn)) {
- tn = old_tn;
+ while (should_halve(tp, tn) && max_work--) {
+ if (halve(t, tn)) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(t->stats->resize_node_skipped);
#endif
break;
}
- }
+ tn = rtnl_dereference(*cptr);
+ }
/* Only one child remains */
if (tn->empty_children == (tnode_child_length(tn) - 1)) {
@@ -797,11 +794,12 @@ one_child:
n = tnode_get_child(tn, --i);
no_children:
/* compress one level */
- node_set_parent(n, NULL);
+ put_child_root(tp, t, tn->key, n);
+ node_set_parent(n, tp);
+
+ /* drop dead node */
tnode_free_safe(tn);
- return n;
}
- return tn;
}
/* readside must use rcu_read_lock currently dump routines
@@ -882,34 +880,19 @@ static struct tnode *fib_find_node(struc
static void trie_rebalance(struct trie *t, struct tnode *tn)
{
- int wasfull;
- t_key cindex, key;
struct tnode *tp;
- key = tn->key;
-
- while (tn != NULL && (tp = node_parent(tn)) != NULL) {
- cindex = get_index(key, tp);
- wasfull = tnode_full(tp, tnode_get_child(tp, cindex));
- tn = resize(t, tn);
-
- tnode_put_child_reorg(tp, cindex, tn, wasfull);
-
- tp = node_parent(tn);
- if (!tp)
- rcu_assign_pointer(t->trie, tn);
+ while ((tp = node_parent(tn)) != NULL) {
+ resize(t, tn);
tnode_free_flush();
- if (!tp)
- break;
tn = tp;
}
/* Handle last (top) tnode */
if (IS_TNODE(tn))
- tn = resize(t, tn);
+ resize(t, tn);
- rcu_assign_pointer(t->trie, tn);
tnode_free_flush();
}

237
target/linux/generic/pending-3.18/080-15-fib_trie-Push-tnode-flushing-down-to-inflate-halve.patch
View File

@ -1,237 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:49 -0800
Subject: [PATCH] fib_trie: Push tnode flushing down to inflate/halve
This change pushes the tnode freeing down into the inflate and halve
functions. It makes more sense here as we have a better grasp of what is
going on and when a given cluster of nodes is ready to be freed.
I believe this may address a bug in the freeing logic as well. For some
reason if the freelist got to a certain size we would call
synchronize_rcu(). I'm assuming that what they meant to do is call
synchronize_rcu() after they had handed off that much memory via
call_rcu(). As such that is what I have updated the behavior to be.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -147,8 +147,6 @@ struct trie {
};
static void resize(struct trie *t, struct tnode *tn);
-/* tnodes to free after resize(); protected by RTNL */
-static struct callback_head *tnode_free_head;
static size_t tnode_free_size;
/*
@@ -307,32 +305,6 @@ static struct tnode *tnode_alloc(size_t
return vzalloc(size);
}
-static void tnode_free_safe(struct tnode *tn)
-{
- BUG_ON(IS_LEAF(tn));
- tn->rcu.next = tnode_free_head;
- tnode_free_head = &tn->rcu;
-}
-
-static void tnode_free_flush(void)
-{
- struct callback_head *head;
-
- while ((head = tnode_free_head)) {
- struct tnode *tn = container_of(head, struct tnode, rcu);
-
- tnode_free_head = head->next;
- tnode_free_size += offsetof(struct tnode, child[1 << tn->bits]);
-
- node_free(tn);
- }
-
- if (tnode_free_size >= PAGE_SIZE * sync_pages) {
- tnode_free_size = 0;
- synchronize_rcu();
- }
-}
-
static struct tnode *leaf_new(t_key key)
{
struct tnode *l = kmem_cache_alloc(trie_leaf_kmem, GFP_KERNEL);
@@ -433,17 +405,33 @@ static void put_child_root(struct tnode
rcu_assign_pointer(t->trie, n);
}
-static void tnode_clean_free(struct tnode *tn)
+static inline void tnode_free_init(struct tnode *tn)
{
- struct tnode *tofree;
- unsigned long i;
+ tn->rcu.next = NULL;
+}
+
+static inline void tnode_free_append(struct tnode *tn, struct tnode *n)
+{
+ n->rcu.next = tn->rcu.next;
+ tn->rcu.next = &n->rcu;
+}
- for (i = 0; i < tnode_child_length(tn); i++) {
- tofree = tnode_get_child(tn, i);
- if (tofree)
- node_free(tofree);
+static void tnode_free(struct tnode *tn)
+{
+ struct callback_head *head = &tn->rcu;
+
+ while (head) {
+ head = head->next;
+ tnode_free_size += offsetof(struct tnode, child[1 << tn->bits]);
+ node_free(tn);
+
+ tn = container_of(head, struct tnode, rcu);
+ }
+
+ if (tnode_free_size >= PAGE_SIZE * sync_pages) {
+ tnode_free_size = 0;
+ synchronize_rcu();
}
- node_free(tn);
}
static int inflate(struct trie *t, struct tnode *oldtnode)
@@ -476,20 +464,23 @@ static int inflate(struct trie *t, struc
inode->bits - 1);
if (!left)
goto nomem;
+ tnode_free_append(tn, left);
right = tnode_new(inode->key | m, inode->pos,
inode->bits - 1);
- if (!right) {
- node_free(left);
+ if (!right)
goto nomem;
- }
+ tnode_free_append(tn, right);
put_child(tn, 2*i, left);
put_child(tn, 2*i+1, right);
}
}
+ /* prepare oldtnode to be freed */
+ tnode_free_init(oldtnode);
+
for (i = 0; i < olen; i++) {
struct tnode *inode = tnode_get_child(oldtnode, i);
struct tnode *left, *right;
@@ -505,12 +496,13 @@ static int inflate(struct trie *t, struc
continue;
}
+ /* drop the node in the old tnode free list */
+ tnode_free_append(oldtnode, inode);
+
/* An internal node with two children */
if (inode->bits == 1) {
put_child(tn, 2*i, rtnl_dereference(inode->child[0]));
put_child(tn, 2*i+1, rtnl_dereference(inode->child[1]));
-
- tnode_free_safe(inode);
continue;
}
@@ -556,17 +548,19 @@ static int inflate(struct trie *t, struc
put_child(tn, 2 * i, left);
put_child(tn, 2 * i + 1, right);
- tnode_free_safe(inode);
-
+ /* resize child nodes */
resize(t, left);
resize(t, right);
}
put_child_root(tp, t, tn->key, tn);
- tnode_free_safe(oldtnode);
+
+ /* we completed without error, prepare to free old node */
+ tnode_free(oldtnode);
return 0;
nomem:
- tnode_clean_free(tn);
+ /* all pointers should be clean so we are done */
+ tnode_free(tn);
return -ENOMEM;
}
@@ -599,17 +593,20 @@ static int halve(struct trie *t, struct
struct tnode *newn;
newn = tnode_new(left->key, oldtnode->pos, 1);
-
if (!newn) {
- tnode_clean_free(tn);
+ tnode_free(tn);
return -ENOMEM;
}
+ tnode_free_append(tn, newn);
put_child(tn, i/2, newn);
}
}
+ /* prepare oldtnode to be freed */
+ tnode_free_init(oldtnode);
+
for (i = 0; i < olen; i += 2) {
struct tnode *newBinNode;
@@ -636,11 +633,14 @@ static int halve(struct trie *t, struct
put_child(tn, i / 2, newBinNode);
+ /* resize child node */
resize(t, newBinNode);
}
put_child_root(tp, t, tn->key, tn);
- tnode_free_safe(oldtnode);
+
+ /* all pointers should be clean so we are done */
+ tnode_free(oldtnode);
return 0;
}
@@ -798,7 +798,8 @@ no_children:
node_set_parent(n, tp);
/* drop dead node */
- tnode_free_safe(tn);
+ tnode_free_init(tn);
+ tnode_free(tn);
}
}
@@ -884,16 +885,12 @@ static void trie_rebalance(struct trie *
while ((tp = node_parent(tn)) != NULL) {
resize(t, tn);
-
- tnode_free_flush();
tn = tp;
}
/* Handle last (top) tnode */
if (IS_TNODE(tn))
resize(t, tn);
-
- tnode_free_flush();
}
/* only used from updater-side */

345
target/linux/generic/pending-3.18/080-16-fib_trie-inflate-halve-nodes-in-a-more-RCU-friendly-.patch
View File

@ -1,345 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:56:55 -0800
Subject: [PATCH] fib_trie: inflate/halve nodes in a more RCU friendly
way
This change pulls the node_set_parent functionality out of put_child_reorg
and instead leaves that to the function to take care of as well. By doing
this we can fully construct the new cluster of tnodes and all of the
pointers out of it before we start routing pointers into it.
I am suspecting this will likely fix some concurency issues though I don't
have a good test to show as such.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -391,8 +391,6 @@ static void put_child(struct tnode *tn,
else if (!wasfull && isfull)
tn->full_children++;
- node_set_parent(n, tn);
-
rcu_assign_pointer(tn->child[i], n);
}
@@ -436,10 +434,8 @@ static void tnode_free(struct tnode *tn)
static int inflate(struct trie *t, struct tnode *oldtnode)
{
- unsigned long olen = tnode_child_length(oldtnode);
- struct tnode *tp = node_parent(oldtnode);
- struct tnode *tn;
- unsigned long i;
+ struct tnode *inode, *node0, *node1, *tn, *tp;
+ unsigned long i, j, k;
t_key m;
pr_debug("In inflate\n");
@@ -448,43 +444,13 @@ static int inflate(struct trie *t, struc
if (!tn)
return -ENOMEM;
- /*
- * Preallocate and store tnodes before the actual work so we
- * don't get into an inconsistent state if memory allocation
- * fails. In case of failure we return the oldnode and inflate
- * of tnode is ignored.
+ /* Assemble all of the pointers in our cluster, in this case that
+ * represents all of the pointers out of our allocated nodes that
+ * point to existing tnodes and the links between our allocated
+ * nodes.
*/
- for (i = 0, m = 1u << tn->pos; i < olen; i++) {
- struct tnode *inode = tnode_get_child(oldtnode, i);
-
- if (tnode_full(oldtnode, inode) && (inode->bits > 1)) {
- struct tnode *left, *right;
-
- left = tnode_new(inode->key & ~m, inode->pos,
- inode->bits - 1);
- if (!left)
- goto nomem;
- tnode_free_append(tn, left);
-
- right = tnode_new(inode->key | m, inode->pos,
- inode->bits - 1);
-
- if (!right)
- goto nomem;
- tnode_free_append(tn, right);
-
- put_child(tn, 2*i, left);
- put_child(tn, 2*i+1, right);
- }
- }
-
- /* prepare oldtnode to be freed */
- tnode_free_init(oldtnode);
-
- for (i = 0; i < olen; i++) {
- struct tnode *inode = tnode_get_child(oldtnode, i);
- struct tnode *left, *right;
- unsigned long size, j;
+ for (i = tnode_child_length(oldtnode), m = 1u << tn->pos; i;) {
+ inode = tnode_get_child(oldtnode, --i);
/* An empty child */
if (inode == NULL)
@@ -496,65 +462,99 @@ static int inflate(struct trie *t, struc
continue;
}
- /* drop the node in the old tnode free list */
- tnode_free_append(oldtnode, inode);
-
/* An internal node with two children */
if (inode->bits == 1) {
- put_child(tn, 2*i, rtnl_dereference(inode->child[0]));
- put_child(tn, 2*i+1, rtnl_dereference(inode->child[1]));
+ put_child(tn, 2 * i + 1, tnode_get_child(inode, 1));
+ put_child(tn, 2 * i, tnode_get_child(inode, 0));
continue;
}
- /* An internal node with more than two children */
-
/* We will replace this node 'inode' with two new
- * ones, 'left' and 'right', each with half of the
+ * ones, 'node0' and 'node1', each with half of the
* original children. The two new nodes will have
* a position one bit further down the key and this
* means that the "significant" part of their keys
* (see the discussion near the top of this file)
* will differ by one bit, which will be "0" in
- * left's key and "1" in right's key. Since we are
+ * node0's key and "1" in node1's key. Since we are
* moving the key position by one step, the bit that
* we are moving away from - the bit at position
- * (inode->pos) - is the one that will differ between
- * left and right. So... we synthesize that bit in the
- * two new keys.
- * The mask 'm' below will be a single "one" bit at
- * the position (inode->pos)
+ * (tn->pos) - is the one that will differ between
+ * node0 and node1. So... we synthesize that bit in the
+ * two new keys.
*/
+ node1 = tnode_new(inode->key | m, inode->pos, inode->bits - 1);
+ if (!node1)
+ goto nomem;
+ tnode_free_append(tn, node1);
+
+ node0 = tnode_new(inode->key & ~m, inode->pos, inode->bits - 1);
+ if (!node0)
+ goto nomem;
+ tnode_free_append(tn, node0);
+
+ /* populate child pointers in new nodes */
+ for (k = tnode_child_length(inode), j = k / 2; j;) {
+ put_child(node1, --j, tnode_get_child(inode, --k));
+ put_child(node0, j, tnode_get_child(inode, j));
+ put_child(node1, --j, tnode_get_child(inode, --k));
+ put_child(node0, j, tnode_get_child(inode, j));
+ }
+
+ /* link new nodes to parent */
+ NODE_INIT_PARENT(node1, tn);
+ NODE_INIT_PARENT(node0, tn);
+
+ /* link parent to nodes */
+ put_child(tn, 2 * i + 1, node1);
+ put_child(tn, 2 * i, node0);
+ }
+
+ /* setup the parent pointer into and out of this node */
+ tp = node_parent(oldtnode);
+ NODE_INIT_PARENT(tn, tp);
+ put_child_root(tp, t, tn->key, tn);
- /* Use the old key, but set the new significant
- * bit to zero.
- */
+ /* prepare oldtnode to be freed */
+ tnode_free_init(oldtnode);
- left = tnode_get_child(tn, 2*i);
- put_child(tn, 2*i, NULL);
+ /* update all child nodes parent pointers to route to us */
+ for (i = tnode_child_length(oldtnode); i;) {
+ inode = tnode_get_child(oldtnode, --i);
- BUG_ON(!left);
+ /* A leaf or an internal node with skipped bits */
+ if (!tnode_full(oldtnode, inode)) {
+ node_set_parent(inode, tn);
+ continue;
+ }
- right = tnode_get_child(tn, 2*i+1);
- put_child(tn, 2*i+1, NULL);
+ /* drop the node in the old tnode free list */
+ tnode_free_append(oldtnode, inode);
- BUG_ON(!right);
+ /* fetch new nodes */
+ node1 = tnode_get_child(tn, 2 * i + 1);
+ node0 = tnode_get_child(tn, 2 * i);
- size = tnode_child_length(left);
- for (j = 0; j < size; j++) {
- put_child(left, j, rtnl_dereference(inode->child[j]));
- put_child(right, j, rtnl_dereference(inode->child[j + size]));
+ /* bits == 1 then node0 and node1 represent inode's children */
+ if (inode->bits == 1) {
+ node_set_parent(node1, tn);
+ node_set_parent(node0, tn);
+ continue;
}
- put_child(tn, 2 * i, left);
- put_child(tn, 2 * i + 1, right);
+ /* update parent pointers in child node's children */
+ for (k = tnode_child_length(inode), j = k / 2; j;) {
+ node_set_parent(tnode_get_child(inode, --k), node1);
+ node_set_parent(tnode_get_child(inode, --j), node0);
+ node_set_parent(tnode_get_child(inode, --k), node1);
+ node_set_parent(tnode_get_child(inode, --j), node0);
+ }
/* resize child nodes */
- resize(t, left);
- resize(t, right);
+ resize(t, node1);
+ resize(t, node0);
}
- put_child_root(tp, t, tn->key, tn);
-
/* we completed without error, prepare to free old node */
tnode_free(oldtnode);
return 0;
@@ -566,10 +566,8 @@ nomem:
static int halve(struct trie *t, struct tnode *oldtnode)
{
- unsigned long olen = tnode_child_length(oldtnode);
- struct tnode *tp = node_parent(oldtnode);
- struct tnode *tn, *left, *right;
- int i;
+ struct tnode *tn, *tp, *inode, *node0, *node1;
+ unsigned long i;
pr_debug("In halve\n");
@@ -577,68 +575,64 @@ static int halve(struct trie *t, struct
if (!tn)
return -ENOMEM;
- /*
- * Preallocate and store tnodes before the actual work so we
- * don't get into an inconsistent state if memory allocation
- * fails. In case of failure we return the oldnode and halve
- * of tnode is ignored.
+ /* Assemble all of the pointers in our cluster, in this case that
+ * represents all of the pointers out of our allocated nodes that
+ * point to existing tnodes and the links between our allocated
+ * nodes.
*/
+ for (i = tnode_child_length(oldtnode); i;) {
+ node1 = tnode_get_child(oldtnode, --i);
+ node0 = tnode_get_child(oldtnode, --i);
- for (i = 0; i < olen; i += 2) {
- left = tnode_get_child(oldtnode, i);
- right = tnode_get_child(oldtnode, i+1);
+ /* At least one of the children is empty */
+ if (!node1 || !node0) {
+ put_child(tn, i / 2, node1 ? : node0);
+ continue;
+ }
/* Two nonempty children */
- if (left && right) {
- struct tnode *newn;
-
- newn = tnode_new(left->key, oldtnode->pos, 1);
- if (!newn) {
- tnode_free(tn);
- return -ENOMEM;
- }
- tnode_free_append(tn, newn);
-
- put_child(tn, i/2, newn);
+ inode = tnode_new(node0->key, oldtnode->pos, 1);
+ if (!inode) {
+ tnode_free(tn);
+ return -ENOMEM;
}
+ tnode_free_append(tn, inode);
+ /* initialize pointers out of node */
+ put_child(inode, 1, node1);
+ put_child(inode, 0, node0);
+ NODE_INIT_PARENT(inode, tn);
+
+ /* link parent to node */
+ put_child(tn, i / 2, inode);
}
+ /* setup the parent pointer out of and back into this node */
+ tp = node_parent(oldtnode);
+ NODE_INIT_PARENT(tn, tp);
+ put_child_root(tp, t, tn->key, tn);
+
/* prepare oldtnode to be freed */
tnode_free_init(oldtnode);
- for (i = 0; i < olen; i += 2) {
- struct tnode *newBinNode;
-
- left = tnode_get_child(oldtnode, i);
- right = tnode_get_child(oldtnode, i+1);
-
- /* At least one of the children is empty */
- if (left == NULL) {
- if (right == NULL) /* Both are empty */
- continue;
- put_child(tn, i/2, right);
- continue;
- }
-
- if (right == NULL) {
- put_child(tn, i/2, left);
+ /* update all of the child parent pointers */
+ for (i = tnode_child_length(tn); i;) {
+ inode = tnode_get_child(tn, --i);
+
+ /* only new tnodes will be considered "full" nodes */
+ if (!tnode_full(tn, inode)) {
+ node_set_parent(inode, tn);
continue;
}
/* Two nonempty children */
- newBinNode = tnode_get_child(tn, i/2);
- put_child(newBinNode, 0, left);
- put_child(newBinNode, 1, right);
-
- put_child(tn, i / 2, newBinNode);
+ node_set_parent(tnode_get_child(inode, 1), inode);
+ node_set_parent(tnode_get_child(inode, 0), inode);
/* resize child node */
- resize(t, newBinNode);
+ resize(t, inode);
}
- put_child_root(tp, t, tn->key, tn);
-
/* all pointers should be clean so we are done */
tnode_free(oldtnode);

95
target/linux/generic/pending-3.18/080-17-fib_trie-Remove-checks-for-index-tnode_child_length-.patch
View File

@ -1,95 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:57:02 -0800
Subject: [PATCH] fib_trie: Remove checks for index >= tnode_child_length
from tnode_get_child
For some reason the compiler doesn't seem to understand that when we are in
a loop that runs from tnode_child_length - 1 to 0 we don't expect the value
of tn->bits to change. As such every call to tnode_get_child was rerunning
tnode_chile_length which ended up consuming quite a bit of space in the
resultant assembly code.
I have gone though and verified that in all cases where tnode_get_child
is used we are either winding though a fixed loop from tnode_child_length -
1 to 0, or are in a fastpath case where we are verifying the value by
either checking for any remaining bits after shifting index by bits and
testing for leaf, or by using tnode_child_length.
size net/ipv4/fib_trie.o
Before:
text data bss dec hex filename
15506 376 8 15890 3e12 net/ipv4/fib_trie.o
After:
text data bss dec hex filename
14827 376 8 15211 3b6b net/ipv4/fib_trie.o
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -186,8 +186,6 @@ static inline unsigned long tnode_child_
static inline struct tnode *tnode_get_child(const struct tnode *tn,
unsigned long i)
{
- BUG_ON(i >= tnode_child_length(tn));
-
return rtnl_dereference(tn->child[i]);
}
@@ -195,8 +193,6 @@ static inline struct tnode *tnode_get_ch
static inline struct tnode *tnode_get_child_rcu(const struct tnode *tn,
unsigned long i)
{
- BUG_ON(i >= tnode_child_length(tn));
-
return rcu_dereference_rtnl(tn->child[i]);
}
@@ -371,7 +367,7 @@ static inline int tnode_full(const struc
*/
static void put_child(struct tnode *tn, unsigned long i, struct tnode *n)
{
- struct tnode *chi = rtnl_dereference(tn->child[i]);
+ struct tnode *chi = tnode_get_child(tn, i);
int isfull, wasfull;
BUG_ON(i >= tnode_child_length(tn));
@@ -867,7 +863,7 @@ static struct tnode *fib_find_node(struc
if (IS_LEAF(n))
break;
- n = rcu_dereference_rtnl(n->child[index]);
+ n = tnode_get_child_rcu(n, index);
}
return n;
@@ -934,7 +930,7 @@ static struct list_head *fib_insert_node
}
tp = n;
- n = rcu_dereference_rtnl(n->child[index]);
+ n = tnode_get_child_rcu(n, index);
}
l = leaf_new(key);
@@ -1215,7 +1211,7 @@ int fib_table_lookup(struct fib_table *t
cindex = index;
}
- n = rcu_dereference(n->child[index]);
+ n = tnode_get_child_rcu(n, index);
if (unlikely(!n))
goto backtrace;
}
@@ -1835,7 +1831,7 @@ static void trie_collect_stats(struct tr
if (n->bits < MAX_STAT_DEPTH)
s->nodesizes[n->bits]++;
- for (i = 0; i < tnode_child_length(n); i++) {
+ for (i = tnode_child_length(n); i--;) {
if (!rcu_access_pointer(n->child[i]))
s->nullpointers++;
}

234
target/linux/generic/pending-3.18/080-18-fib_trie-Add-tracking-value-for-suffix-length.patch
View File

@ -1,234 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Wed, 31 Dec 2014 10:57:08 -0800
Subject: [PATCH] fib_trie: Add tracking value for suffix length
This change adds a tracking value for the maximum suffix length of all
prefixes stored in any given tnode. With this value we can determine if we
need to backtrace or not based on if the suffix is greater than the pos
value.
By doing this we can reduce the CPU overhead for lookups in the local table
as many of the prefixes there are 32b long and have a suffix length of 0
meaning we can immediately backtrace to the root node without needing to
test any of the nodes between it and where we ended up.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -96,6 +96,7 @@ struct tnode {
t_key key;
unsigned char bits; /* 2log(KEYLENGTH) bits needed */
unsigned char pos; /* 2log(KEYLENGTH) bits needed */
+ unsigned char slen;
struct tnode __rcu *parent;
struct rcu_head rcu;
union {
@@ -311,6 +312,7 @@ static struct tnode *leaf_new(t_key key)
* as the nodes are searched
*/
l->key = key;
+ l->slen = 0;
l->pos = 0;
/* set bits to 0 indicating we are not a tnode */
l->bits = 0;
@@ -342,6 +344,7 @@ static struct tnode *tnode_new(t_key key
if (tn) {
tn->parent = NULL;
+ tn->slen = pos;
tn->pos = pos;
tn->bits = bits;
tn->key = (shift < KEYLENGTH) ? (key >> shift) << shift : 0;
@@ -387,6 +390,9 @@ static void put_child(struct tnode *tn,
else if (!wasfull && isfull)
tn->full_children++;
+ if (n && (tn->slen < n->slen))
+ tn->slen = n->slen;
+
rcu_assign_pointer(tn->child[i], n);
}
@@ -635,6 +641,41 @@ static int halve(struct trie *t, struct
return 0;
}
+static unsigned char update_suffix(struct tnode *tn)
+{
+ unsigned char slen = tn->pos;
+ unsigned long stride, i;
+
+ /* search though the list of children looking for nodes that might
+ * have a suffix greater than the one we currently have. This is
+ * why we start with a stride of 2 since a stride of 1 would
+ * represent the nodes with suffix length equal to tn->pos
+ */
+ for (i = 0, stride = 0x2ul ; i < tnode_child_length(tn); i += stride) {
+ struct tnode *n = tnode_get_child(tn, i);
+
+ if (!n || (n->slen <= slen))
+ continue;
+
+ /* update stride and slen based on new value */
+ stride <<= (n->slen - slen);
+ slen = n->slen;
+ i &= ~(stride - 1);
+
+ /* if slen covers all but the last bit we can stop here
+ * there will be nothing longer than that since only node
+ * 0 and 1 << (bits - 1) could have that as their suffix
+ * length.
+ */
+ if ((slen + 1) >= (tn->pos + tn->bits))
+ break;
+ }
+
+ tn->slen = slen;
+
+ return slen;
+}
+
/* From "Implementing a dynamic compressed trie" by Stefan Nilsson of
* the Helsinki University of Technology and Matti Tikkanen of Nokia
* Telecommunications, page 6:
@@ -790,6 +831,19 @@ no_children:
/* drop dead node */
tnode_free_init(tn);
tnode_free(tn);
+ return;
+ }
+
+ /* Return if at least one deflate was run */
+ if (max_work != MAX_WORK)
+ return;
+
+ /* push the suffix length to the parent node */
+ if (tn->slen > tn->pos) {
+ unsigned char slen = update_suffix(tn);
+
+ if (tp && (slen > tp->slen))
+ tp->slen = slen;
}
}
@@ -818,8 +872,58 @@ static inline struct list_head *get_fa_h
return &li->falh;
}
-static void insert_leaf_info(struct hlist_head *head, struct leaf_info *new)
+static void leaf_pull_suffix(struct tnode *l)
+{
+ struct tnode *tp = node_parent(l);
+
+ while (tp && (tp->slen > tp->pos) && (tp->slen > l->slen)) {
+ if (update_suffix(tp) > l->slen)
+ break;
+ tp = node_parent(tp);
+ }
+}
+
+static void leaf_push_suffix(struct tnode *l)
+{
+ struct tnode *tn = node_parent(l);
+
+ /* if this is a new leaf then tn will be NULL and we can sort
+ * out parent suffix lengths as a part of trie_rebalance
+ */
+ while (tn && (tn->slen < l->slen)) {
+ tn->slen = l->slen;
+ tn = node_parent(tn);
+ }
+}
+
+static void remove_leaf_info(struct tnode *l, struct leaf_info *old)
+{
+ struct hlist_node *prev;
+
+ /* record the location of the pointer to this object */
+ prev = rtnl_dereference(hlist_pprev_rcu(&old->hlist));
+
+ /* remove the leaf info from the list */
+ hlist_del_rcu(&old->hlist);
+
+ /* if we emptied the list this leaf will be freed and we can sort
+ * out parent suffix lengths as a part of trie_rebalance
+ */
+ if (hlist_empty(&l->list))
+ return;
+
+ /* if we removed the tail then we need to update slen */
+ if (!rcu_access_pointer(hlist_next_rcu(prev))) {
+ struct leaf_info *li = hlist_entry(prev, typeof(*li), hlist);
+
+ l->slen = KEYLENGTH - li->plen;
+ leaf_pull_suffix(l);
+ }
+}
+
+static void insert_leaf_info(struct tnode *l, struct leaf_info *new)
{
+ struct hlist_head *head = &l->list;
struct leaf_info *li = NULL, *last = NULL;
if (hlist_empty(head)) {
@@ -836,6 +940,12 @@ static void insert_leaf_info(struct hlis
else
hlist_add_before_rcu(&new->hlist, &li->hlist);
}
+
+ /* if we added to the tail node then we need to update slen */
+ if (!rcu_access_pointer(hlist_next_rcu(&new->hlist))) {
+ l->slen = KEYLENGTH - new->plen;
+ leaf_push_suffix(l);
+ }
}
/* rcu_read_lock needs to be hold by caller from readside */
@@ -925,7 +1035,7 @@ static struct list_head *fib_insert_node
/* we have found a leaf. Prefixes have already been compared */
if (IS_LEAF(n)) {
/* Case 1: n is a leaf, and prefixes match*/
- insert_leaf_info(&n->list, li);
+ insert_leaf_info(n, li);
return fa_head;
}
@@ -939,7 +1049,7 @@ static struct list_head *fib_insert_node
return NULL;
}
- insert_leaf_info(&l->list, li);
+ insert_leaf_info(l, li);
/* Case 2: n is a LEAF or a TNODE and the key doesn't match.
*
@@ -1206,7 +1316,7 @@ int fib_table_lookup(struct fib_table *t
/* only record pn and cindex if we are going to be chopping
* bits later. Otherwise we are just wasting cycles.
*/
- if (index) {
+ if (n->slen > n->pos) {
pn = n;
cindex = index;
}
@@ -1225,7 +1335,7 @@ int fib_table_lookup(struct fib_table *t
* between the key and the prefix exist in the region of
* the lsb and higher in the prefix.
*/
- if (unlikely(prefix_mismatch(key, n)))
+ if (unlikely(prefix_mismatch(key, n)) || (n->slen == n->pos))
goto backtrace;
/* exit out and process leaf */
@@ -1425,7 +1535,7 @@ int fib_table_delete(struct fib_table *t
tb->tb_num_default--;
if (list_empty(fa_head)) {
- hlist_del_rcu(&li->hlist);
+ remove_leaf_info(l, li);
free_leaf_info(li);
}

52
target/linux/generic/pending-3.18/080-19-fib_trie-Use-index-0ul-n-bits-instead-of-index-n-bit.patch
View File

@ -1,52 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:08 -0800
Subject: [PATCH] fib_trie: Use index & (~0ul << n->bits) instead of index >>
n->bits
In doing performance testing and analysis of the changes I recently found
that by shifting the index I had created an unnecessary dependency.
I have updated the code so that we instead shift a mask by bits and then
just test against that as that should save us about 2 CPU cycles since we
can generate the mask while the key and pos are being processed.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -961,12 +961,12 @@ static struct tnode *fib_find_node(struc
* prefix plus zeros for the bits in the cindex. The index
* is the difference between the key and this value. From
* this we can actually derive several pieces of data.
- * if !(index >> bits)
- * we know the value is cindex
- * else
+ * if (index & (~0ul << bits))
* we have a mismatch in skip bits and failed
+ * else
+ * we know the value is cindex
*/
- if (index >> n->bits)
+ if (index & (~0ul << n->bits))
return NULL;
/* we have found a leaf. Prefixes have already been compared */
@@ -1301,12 +1301,12 @@ int fib_table_lookup(struct fib_table *t
* prefix plus zeros for the "bits" in the prefix. The index
* is the difference between the key and this value. From
* this we can actually derive several pieces of data.
- * if !(index >> bits)
- * we know the value is child index
- * else
+ * if (index & (~0ul << bits))
* we have a mismatch in skip bits and failed
+ * else
+ * we know the value is cindex
*/
- if (index >> n->bits)
+ if (index & (~0ul << n->bits))
break;
/* we have found a leaf. Prefixes have already been compared */

267
target/linux/generic/pending-3.18/080-20-fib_trie-Fix-RCU-bug-and-merge-similar-bits-of-infla.patch
View File

@ -1,267 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:14 -0800
Subject: [PATCH] fib_trie: Fix RCU bug and merge similar bits of inflate/halve
This patch addresses two issues.
The first issue is the fact that I believe I had the RCU freeing sequence
slightly out of order. As a result we could get into an issue if a caller
went into a child of a child of the new node, then backtraced into the to be
freed parent, and then attempted to access a child of a child that may have
been consumed in a resize of one of the new nodes children. To resolve this I
have moved the resize after we have freed the oldtnode. The only side effect
of this is that we will now be calling resize on more nodes in the case of
inflate due to the fact that we don't have a good way to test to see if a
full_tnode on the new node was there before or after the allocation. This
should have minimal impact however since the node should already be
correctly size so it is just the cost of calling should_inflate that we
will be taking on the node which is only a couple of cycles.
The second issue is the fact that inflate and halve were essentially doing
the same thing after the new node was added to the trie replacing the old
one. As such it wasn't really necessary to keep the code in both functions
so I have split it out into two other functions, called replace and
update_children.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -396,8 +396,30 @@ static void put_child(struct tnode *tn,
rcu_assign_pointer(tn->child[i], n);
}
-static void put_child_root(struct tnode *tp, struct trie *t,
- t_key key, struct tnode *n)
+static void update_children(struct tnode *tn)
+{
+ unsigned long i;
+
+ /* update all of the child parent pointers */
+ for (i = tnode_child_length(tn); i;) {
+ struct tnode *inode = tnode_get_child(tn, --i);
+
+ if (!inode)
+ continue;
+
+ /* Either update the children of a tnode that
+ * already belongs to us or update the child
+ * to point to ourselves.
+ */
+ if (node_parent(inode) == tn)
+ update_children(inode);
+ else
+ node_set_parent(inode, tn);
+ }
+}
+
+static inline void put_child_root(struct tnode *tp, struct trie *t,
+ t_key key, struct tnode *n)
{
if (tp)
put_child(tp, get_index(key, tp), n);
@@ -434,10 +456,35 @@ static void tnode_free(struct tnode *tn)
}
}
+static void replace(struct trie *t, struct tnode *oldtnode, struct tnode *tn)
+{
+ struct tnode *tp = node_parent(oldtnode);
+ unsigned long i;
+
+ /* setup the parent pointer out of and back into this node */
+ NODE_INIT_PARENT(tn, tp);
+ put_child_root(tp, t, tn->key, tn);
+
+ /* update all of the child parent pointers */
+ update_children(tn);
+
+ /* all pointers should be clean so we are done */
+ tnode_free(oldtnode);
+
+ /* resize children now that oldtnode is freed */
+ for (i = tnode_child_length(tn); i;) {
+ struct tnode *inode = tnode_get_child(tn, --i);
+
+ /* resize child node */
+ if (tnode_full(tn, inode))
+ resize(t, inode);
+ }
+}
+
static int inflate(struct trie *t, struct tnode *oldtnode)
{
- struct tnode *inode, *node0, *node1, *tn, *tp;
- unsigned long i, j, k;
+ struct tnode *tn;
+ unsigned long i;
t_key m;
pr_debug("In inflate\n");
@@ -446,13 +493,18 @@ static int inflate(struct trie *t, struc
if (!tn)
return -ENOMEM;
+ /* prepare oldtnode to be freed */
+ tnode_free_init(oldtnode);
+
/* Assemble all of the pointers in our cluster, in this case that
* represents all of the pointers out of our allocated nodes that
* point to existing tnodes and the links between our allocated
* nodes.
*/
for (i = tnode_child_length(oldtnode), m = 1u << tn->pos; i;) {
- inode = tnode_get_child(oldtnode, --i);
+ struct tnode *inode = tnode_get_child(oldtnode, --i);
+ struct tnode *node0, *node1;
+ unsigned long j, k;
/* An empty child */
if (inode == NULL)
@@ -464,6 +516,9 @@ static int inflate(struct trie *t, struc
continue;
}
+ /* drop the node in the old tnode free list */
+ tnode_free_append(oldtnode, inode);
+
/* An internal node with two children */
if (inode->bits == 1) {
put_child(tn, 2 * i + 1, tnode_get_child(inode, 1));
@@ -488,9 +543,9 @@ static int inflate(struct trie *t, struc
node1 = tnode_new(inode->key | m, inode->pos, inode->bits - 1);
if (!node1)
goto nomem;
- tnode_free_append(tn, node1);
+ node0 = tnode_new(inode->key, inode->pos, inode->bits - 1);
- node0 = tnode_new(inode->key & ~m, inode->pos, inode->bits - 1);
+ tnode_free_append(tn, node1);
if (!node0)
goto nomem;
tnode_free_append(tn, node0);
@@ -512,53 +567,9 @@ static int inflate(struct trie *t, struc
put_child(tn, 2 * i, node0);
}
- /* setup the parent pointer into and out of this node */
- tp = node_parent(oldtnode);
- NODE_INIT_PARENT(tn, tp);
- put_child_root(tp, t, tn->key, tn);
+ /* setup the parent pointers into and out of this node */
+ replace(t, oldtnode, tn);
- /* prepare oldtnode to be freed */
- tnode_free_init(oldtnode);
-
- /* update all child nodes parent pointers to route to us */
- for (i = tnode_child_length(oldtnode); i;) {
- inode = tnode_get_child(oldtnode, --i);
-
- /* A leaf or an internal node with skipped bits */
- if (!tnode_full(oldtnode, inode)) {
- node_set_parent(inode, tn);
- continue;
- }
-
- /* drop the node in the old tnode free list */
- tnode_free_append(oldtnode, inode);
-
- /* fetch new nodes */
- node1 = tnode_get_child(tn, 2 * i + 1);
- node0 = tnode_get_child(tn, 2 * i);
-
- /* bits == 1 then node0 and node1 represent inode's children */
- if (inode->bits == 1) {
- node_set_parent(node1, tn);
- node_set_parent(node0, tn);
- continue;
- }
-
- /* update parent pointers in child node's children */
- for (k = tnode_child_length(inode), j = k / 2; j;) {
- node_set_parent(tnode_get_child(inode, --k), node1);
- node_set_parent(tnode_get_child(inode, --j), node0);
- node_set_parent(tnode_get_child(inode, --k), node1);
- node_set_parent(tnode_get_child(inode, --j), node0);
- }
-
- /* resize child nodes */
- resize(t, node1);
- resize(t, node0);
- }
-
- /* we completed without error, prepare to free old node */
- tnode_free(oldtnode);
return 0;
nomem:
/* all pointers should be clean so we are done */
@@ -568,7 +579,7 @@ nomem:
static int halve(struct trie *t, struct tnode *oldtnode)
{
- struct tnode *tn, *tp, *inode, *node0, *node1;
+ struct tnode *tn;
unsigned long i;
pr_debug("In halve\n");
@@ -577,14 +588,18 @@ static int halve(struct trie *t, struct
if (!tn)
return -ENOMEM;
+ /* prepare oldtnode to be freed */
+ tnode_free_init(oldtnode);
+
/* Assemble all of the pointers in our cluster, in this case that
* represents all of the pointers out of our allocated nodes that
* point to existing tnodes and the links between our allocated
* nodes.
*/
for (i = tnode_child_length(oldtnode); i;) {
- node1 = tnode_get_child(oldtnode, --i);
- node0 = tnode_get_child(oldtnode, --i);
+ struct tnode *node1 = tnode_get_child(oldtnode, --i);
+ struct tnode *node0 = tnode_get_child(oldtnode, --i);
+ struct tnode *inode;
/* At least one of the children is empty */
if (!node1 || !node0) {
@@ -609,34 +624,8 @@ static int halve(struct trie *t, struct
put_child(tn, i / 2, inode);
}
- /* setup the parent pointer out of and back into this node */
- tp = node_parent(oldtnode);
- NODE_INIT_PARENT(tn, tp);
- put_child_root(tp, t, tn->key, tn);
-
- /* prepare oldtnode to be freed */
- tnode_free_init(oldtnode);
-
- /* update all of the child parent pointers */
- for (i = tnode_child_length(tn); i;) {
- inode = tnode_get_child(tn, --i);
-
- /* only new tnodes will be considered "full" nodes */
- if (!tnode_full(tn, inode)) {
- node_set_parent(inode, tn);
- continue;
- }
-
- /* Two nonempty children */
- node_set_parent(tnode_get_child(inode, 1), inode);
- node_set_parent(tnode_get_child(inode, 0), inode);
-
- /* resize child node */
- resize(t, inode);
- }
-
- /* all pointers should be clean so we are done */
- tnode_free(oldtnode);
+ /* setup the parent pointers into and out of this node */
+ replace(t, oldtnode, tn);
return 0;
}

61
target/linux/generic/pending-3.18/080-21-fib_trie-Fall-back-to-slen-update-on-inflate-halve-f.patch
View File

@ -1,61 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:20 -0800
Subject: [PATCH] fib_trie: Fall back to slen update on inflate/halve failure
This change corrects an issue where if inflate or halve fails we were
exiting the resize function without at least updating the slen for the
node. To correct this I have moved the update of max_size into the while
loop so that it is only decremented on a successful call to either inflate
or halve.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -752,7 +752,7 @@ static void resize(struct trie *t, struc
{
struct tnode *tp = node_parent(tn), *n = NULL;
struct tnode __rcu **cptr;
- int max_work;
+ int max_work = MAX_WORK;
pr_debug("In tnode_resize %p inflate_threshold=%d threshold=%d\n",
tn, inflate_threshold, halve_threshold);
@@ -775,8 +775,7 @@ static void resize(struct trie *t, struc
/* Double as long as the resulting node has a number of
* nonempty nodes that are above the threshold.
*/
- max_work = MAX_WORK;
- while (should_inflate(tp, tn) && max_work--) {
+ while (should_inflate(tp, tn) && max_work) {
if (inflate(t, tn)) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(t->stats->resize_node_skipped);
@@ -784,6 +783,7 @@ static void resize(struct trie *t, struc
break;
}
+ max_work--;
tn = rtnl_dereference(*cptr);
}
@@ -794,8 +794,7 @@ static void resize(struct trie *t, struc
/* Halve as long as the number of empty children in this
* node is above threshold.
*/
- max_work = MAX_WORK;
- while (should_halve(tp, tn) && max_work--) {
+ while (should_halve(tp, tn) && max_work) {
if (halve(t, tn)) {
#ifdef CONFIG_IP_FIB_TRIE_STATS
this_cpu_inc(t->stats->resize_node_skipped);
@@ -803,6 +802,7 @@ static void resize(struct trie *t, struc
break;
}
+ max_work--;
tn = rtnl_dereference(*cptr);
}

206
target/linux/generic/pending-3.18/080-22-fib_trie-Add-collapse-and-should_collapse-to-resize.patch
View File

@ -1,206 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:26 -0800
Subject: [PATCH] fib_trie: Add collapse() and should_collapse() to resize
This patch really does two things.
First it pulls the logic for determining if we should collapse one node out
of the tree and the actual code doing the collapse into a separate pair of
functions. This helps to make the changes to these areas more readable.
Second it encodes the upper 32b of the empty_children value onto the
full_children value in the case of bits == KEYLENGTH. By doing this we are
able to handle the case of a 32b node where empty_children would appear to
be 0 when it was actually 1ul << 32.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -83,7 +83,8 @@
#define MAX_STAT_DEPTH 32
-#define KEYLENGTH (8*sizeof(t_key))
+#define KEYLENGTH (8*sizeof(t_key))
+#define KEY_MAX ((t_key)~0)
typedef unsigned int t_key;
@@ -102,8 +103,8 @@ struct tnode {
union {
/* The fields in this struct are valid if bits > 0 (TNODE) */
struct {
- unsigned int full_children; /* KEYLENGTH bits needed */
- unsigned int empty_children; /* KEYLENGTH bits needed */
+ t_key empty_children; /* KEYLENGTH bits needed */
+ t_key full_children; /* KEYLENGTH bits needed */
struct tnode __rcu *child[0];
};
/* This list pointer if valid if bits == 0 (LEAF) */
@@ -302,6 +303,16 @@ static struct tnode *tnode_alloc(size_t
return vzalloc(size);
}
+static inline void empty_child_inc(struct tnode *n)
+{
+ ++n->empty_children ? : ++n->full_children;
+}
+
+static inline void empty_child_dec(struct tnode *n)
+{
+ n->empty_children-- ? : n->full_children--;
+}
+
static struct tnode *leaf_new(t_key key)
{
struct tnode *l = kmem_cache_alloc(trie_leaf_kmem, GFP_KERNEL);
@@ -335,7 +346,7 @@ static struct leaf_info *leaf_info_new(i
static struct tnode *tnode_new(t_key key, int pos, int bits)
{
- size_t sz = offsetof(struct tnode, child[1 << bits]);
+ size_t sz = offsetof(struct tnode, child[1ul << bits]);
struct tnode *tn = tnode_alloc(sz);
unsigned int shift = pos + bits;
@@ -348,8 +359,10 @@ static struct tnode *tnode_new(t_key key
tn->pos = pos;
tn->bits = bits;
tn->key = (shift < KEYLENGTH) ? (key >> shift) << shift : 0;
- tn->full_children = 0;
- tn->empty_children = 1<<bits;
+ if (bits == KEYLENGTH)
+ tn->full_children = 1;
+ else
+ tn->empty_children = 1ul << bits;
}
pr_debug("AT %p s=%zu %zu\n", tn, sizeof(struct tnode),
@@ -375,11 +388,11 @@ static void put_child(struct tnode *tn,
BUG_ON(i >= tnode_child_length(tn));
- /* update emptyChildren */
+ /* update emptyChildren, overflow into fullChildren */
if (n == NULL && chi != NULL)
- tn->empty_children++;
- else if (n != NULL && chi == NULL)
- tn->empty_children--;
+ empty_child_inc(tn);
+ if (n != NULL && chi == NULL)
+ empty_child_dec(tn);
/* update fullChildren */
wasfull = tnode_full(tn, chi);
@@ -630,6 +643,24 @@ static int halve(struct trie *t, struct
return 0;
}
+static void collapse(struct trie *t, struct tnode *oldtnode)
+{
+ struct tnode *n, *tp;
+ unsigned long i;
+
+ /* scan the tnode looking for that one child that might still exist */
+ for (n = NULL, i = tnode_child_length(oldtnode); !n && i;)
+ n = tnode_get_child(oldtnode, --i);
+
+ /* compress one level */
+ tp = node_parent(oldtnode);
+ put_child_root(tp, t, oldtnode->key, n);
+ node_set_parent(n, tp);
+
+ /* drop dead node */
+ node_free(oldtnode);
+}
+
static unsigned char update_suffix(struct tnode *tn)
{
unsigned char slen = tn->pos;
@@ -729,10 +760,12 @@ static bool should_inflate(const struct
/* Keep root node larger */
threshold *= tp ? inflate_threshold : inflate_threshold_root;
- used += tn->full_children;
used -= tn->empty_children;
+ used += tn->full_children;
- return tn->pos && ((50 * used) >= threshold);
+ /* if bits == KEYLENGTH then pos = 0, and will fail below */
+
+ return (used > 1) && tn->pos && ((50 * used) >= threshold);
}
static bool should_halve(const struct tnode *tp, const struct tnode *tn)
@@ -744,13 +777,29 @@ static bool should_halve(const struct tn
threshold *= tp ? halve_threshold : halve_threshold_root;
used -= tn->empty_children;
- return (tn->bits > 1) && ((100 * used) < threshold);
+ /* if bits == KEYLENGTH then used = 100% on wrap, and will fail below */
+
+ return (used > 1) && (tn->bits > 1) && ((100 * used) < threshold);
+}
+
+static bool should_collapse(const struct tnode *tn)
+{
+ unsigned long used = tnode_child_length(tn);
+
+ used -= tn->empty_children;
+
+ /* account for bits == KEYLENGTH case */
+ if ((tn->bits == KEYLENGTH) && tn->full_children)
+ used -= KEY_MAX;
+
+ /* One child or none, time to drop us from the trie */
+ return used < 2;
}
#define MAX_WORK 10
static void resize(struct trie *t, struct tnode *tn)
{
- struct tnode *tp = node_parent(tn), *n = NULL;
+ struct tnode *tp = node_parent(tn);
struct tnode __rcu **cptr;
int max_work = MAX_WORK;
@@ -764,14 +813,6 @@ static void resize(struct trie *t, struc
cptr = tp ? &tp->child[get_index(tn->key, tp)] : &t->trie;
BUG_ON(tn != rtnl_dereference(*cptr));
- /* No children */
- if (tn->empty_children > (tnode_child_length(tn) - 1))
- goto no_children;
-
- /* One child */
- if (tn->empty_children == (tnode_child_length(tn) - 1))
- goto one_child;
-
/* Double as long as the resulting node has a number of
* nonempty nodes that are above the threshold.
*/
@@ -807,19 +848,8 @@ static void resize(struct trie *t, struc
}
/* Only one child remains */
- if (tn->empty_children == (tnode_child_length(tn) - 1)) {
- unsigned long i;
-one_child:
- for (i = tnode_child_length(tn); !n && i;)
- n = tnode_get_child(tn, --i);
-no_children:
- /* compress one level */
- put_child_root(tp, t, tn->key, n);
- node_set_parent(n, tp);
-
- /* drop dead node */
- tnode_free_init(tn);
- tnode_free(tn);
+ if (should_collapse(tn)) {
+ collapse(t, tn);
return;
}

34
target/linux/generic/pending-3.18/080-23-fib_trie-Use-empty_children-instead-of-counting-empt.patch
View File

@ -1,34 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:33 -0800
Subject: [PATCH] fib_trie: Use empty_children instead of counting empty nodes
in stats collection
It doesn't make much sense to count the pointers ourselves when
empty_children already has a count for the number of NULL pointers stored
in the tnode. As such save ourselves the cycles and just use
empty_children.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1954,16 +1954,10 @@ static void trie_collect_stats(struct tr
hlist_for_each_entry_rcu(li, &n->list, hlist)
++s->prefixes;
} else {
- unsigned long i;
-
s->tnodes++;
if (n->bits < MAX_STAT_DEPTH)
s->nodesizes[n->bits]++;
-
- for (i = tnode_child_length(n); i--;) {
- if (!rcu_access_pointer(n->child[i]))
- s->nullpointers++;
- }
+ s->nullpointers += n->empty_children;
}
}
rcu_read_unlock();

79
target/linux/generic/pending-3.18/080-24-fib_trie-Move-fib_find_alias-to-file-where-it-is-use.patch
View File

@ -1,79 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:39 -0800
Subject: [PATCH] fib_trie: Move fib_find_alias to file where it is used
The function fib_find_alias is only accessed by functions in fib_trie.c as
such it makes sense to relocate it and cast it as static so that the
compiler can take advantage of optimizations it can do to it as a local
function.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -32,7 +32,6 @@ int fib_dump_info(struct sk_buff *skb, u
unsigned int);
void rtmsg_fib(int event, __be32 key, struct fib_alias *fa, int dst_len,
u32 tb_id, const struct nl_info *info, unsigned int nlm_flags);
-struct fib_alias *fib_find_alias(struct list_head *fah, u8 tos, u32 prio);
static inline void fib_result_assign(struct fib_result *res,
struct fib_info *fi)
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -410,24 +410,6 @@ errout:
rtnl_set_sk_err(info->nl_net, RTNLGRP_IPV4_ROUTE, err);
}
-/* Return the first fib alias matching TOS with
- * priority less than or equal to PRIO.
- */
-struct fib_alias *fib_find_alias(struct list_head *fah, u8 tos, u32 prio)
-{
- if (fah) {
- struct fib_alias *fa;
- list_for_each_entry(fa, fah, fa_list) {
- if (fa->fa_tos > tos)
- continue;
- if (fa->fa_info->fib_priority >= prio ||
- fa->fa_tos < tos)
- return fa;
- }
- }
- return NULL;
-}
-
static int fib_detect_death(struct fib_info *fi, int order,
struct fib_info **last_resort, int *last_idx,
int dflt)
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -998,6 +998,26 @@ static struct tnode *fib_find_node(struc
return n;
}
+/* Return the first fib alias matching TOS with
+ * priority less than or equal to PRIO.
+ */
+static struct fib_alias *fib_find_alias(struct list_head *fah, u8 tos, u32 prio)
+{
+ struct fib_alias *fa;
+
+ if (!fah)
+ return NULL;
+
+ list_for_each_entry(fa, fah, fa_list) {
+ if (fa->fa_tos > tos)
+ continue;
+ if (fa->fa_info->fib_priority >= prio || fa->fa_tos < tos)
+ return fa;
+ }
+
+ return NULL;
+}
+
static void trie_rebalance(struct trie *t, struct tnode *tn)
{
struct tnode *tp;

116
target/linux/generic/pending-3.18/080-25-fib_trie-Various-clean-ups-for-handling-slen.patch
View File

@ -1,116 +0,0 @@
From: Alexander Duyck <alexander.h.duyck@redhat.com>
Date: Thu, 22 Jan 2015 15:51:45 -0800
Subject: [PATCH] fib_trie: Various clean-ups for handling slen
While doing further work on the fib_trie I noted a few items.
First I was using calls that were far more complicated than they needed to
be for determining when to push/pull the suffix length. I have updated the
code to reflect the simplier logic.
The second issue is that I realised we weren't necessarily handling the
case of a leaf_info struct surviving a flush. I have updated the logic so
that now we will call pull_suffix in the event of having a leaf info value
left in the leaf after flushing it.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -917,27 +917,20 @@ static void leaf_push_suffix(struct tnod
static void remove_leaf_info(struct tnode *l, struct leaf_info *old)
{
- struct hlist_node *prev;
-
- /* record the location of the pointer to this object */
- prev = rtnl_dereference(hlist_pprev_rcu(&old->hlist));
+ /* record the location of the previous list_info entry */
+ struct hlist_node **pprev = old->hlist.pprev;
+ struct leaf_info *li = hlist_entry(pprev, typeof(*li), hlist.next);
/* remove the leaf info from the list */
hlist_del_rcu(&old->hlist);
- /* if we emptied the list this leaf will be freed and we can sort
- * out parent suffix lengths as a part of trie_rebalance
- */
- if (hlist_empty(&l->list))
+ /* only access li if it is pointing at the last valid hlist_node */
+ if (hlist_empty(&l->list) || (*pprev))
return;
- /* if we removed the tail then we need to update slen */
- if (!rcu_access_pointer(hlist_next_rcu(prev))) {
- struct leaf_info *li = hlist_entry(prev, typeof(*li), hlist);
-
- l->slen = KEYLENGTH - li->plen;
- leaf_pull_suffix(l);
- }
+ /* update the trie with the latest suffix length */
+ l->slen = KEYLENGTH - li->plen;
+ leaf_pull_suffix(l);
}
static void insert_leaf_info(struct tnode *l, struct leaf_info *new)
@@ -961,7 +954,7 @@ static void insert_leaf_info(struct tnod
}
/* if we added to the tail node then we need to update slen */
- if (!rcu_access_pointer(hlist_next_rcu(&new->hlist))) {
+ if (l->slen < (KEYLENGTH - new->plen)) {
l->slen = KEYLENGTH - new->plen;
leaf_push_suffix(l);
}
@@ -1613,6 +1606,7 @@ static int trie_flush_leaf(struct tnode
struct hlist_head *lih = &l->list;
struct hlist_node *tmp;
struct leaf_info *li = NULL;
+ unsigned char plen = KEYLENGTH;
hlist_for_each_entry_safe(li, tmp, lih, hlist) {
found += trie_flush_list(&li->falh);
@@ -1620,8 +1614,14 @@ static int trie_flush_leaf(struct tnode
if (list_empty(&li->falh)) {
hlist_del_rcu(&li->hlist);
free_leaf_info(li);
+ continue;
}
+
+ plen = li->plen;
}
+
+ l->slen = KEYLENGTH - plen;
+
return found;
}
@@ -1700,13 +1700,22 @@ int fib_table_flush(struct fib_table *tb
for (l = trie_firstleaf(t); l; l = trie_nextleaf(l)) {
found += trie_flush_leaf(l);
- if (ll && hlist_empty(&ll->list))
- trie_leaf_remove(t, ll);
+ if (ll) {
+ if (hlist_empty(&ll->list))
+ trie_leaf_remove(t, ll);
+ else
+ leaf_pull_suffix(ll);
+ }
+
ll = l;
}
- if (ll && hlist_empty(&ll->list))
- trie_leaf_remove(t, ll);
+ if (ll) {
+ if (hlist_empty(&ll->list))
+ trie_leaf_remove(t, ll);
+ else
+ leaf_pull_suffix(ll);
+ }
pr_debug("trie_flush found=%d\n", found);
return found;

89
target/linux/generic/pending-3.18/081-01-pppoe-Use-workqueue-to-die-properly-when-a-PADT-is-r.patch
View File

@ -1,89 +0,0 @@
From: Simon Farnsworth <simon@farnz.org.uk>
Date: Sun, 1 Mar 2015 10:54:39 +0000
Subject: [PATCH] pppoe: Use workqueue to die properly when a PADT is received
When a PADT frame is received, the socket may not be in a good state to
close down the PPP interface. The current implementation handles this by
simply blocking all further PPP traffic, and hoping that the lack of traffic
will trigger the user to investigate.
Use schedule_work to get to a process context from which we clear down the
PPP interface, in a fashion analogous to hangup on a TTY-based PPP
interface. This causes pppd to disconnect immediately, and allows tools to
take immediate corrective action.
Note that pppd's rp_pppoe.so plugin has code in it to disable the session
when it disconnects; however, as a consequence of this patch, the session is
already disabled before rp_pppoe.so is asked to disable the session. The
result is a harmless error message:
Failed to disconnect PPPoE socket: 114 Operation already in progress
This message is safe to ignore, as long as the error is 114 Operation
already in progress; in that specific case, it means that the PPPoE session
has already been disabled before pppd tried to disable it.
Signed-off-by: Simon Farnsworth <simon@farnz.org.uk>
Tested-by: Dan Williams <dcbw@redhat.com>
Tested-by: Christoph Schulz <develop@kristov.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -454,6 +454,18 @@ out:
return NET_RX_DROP;
}
+static void pppoe_unbind_sock_work(struct work_struct *work)
+{
+ struct pppox_sock *po = container_of(work, struct pppox_sock,
+ proto.pppoe.padt_work);
+ struct sock *sk = sk_pppox(po);
+
+ lock_sock(sk);
+ pppox_unbind_sock(sk);
+ release_sock(sk);
+ sock_put(sk);
+}
+
/************************************************************************
*
* Receive a PPPoE Discovery frame.
@@ -499,7 +511,8 @@ static int pppoe_disc_rcv(struct sk_buff
}
bh_unlock_sock(sk);
- sock_put(sk);
+ if (!schedule_work(&po->proto.pppoe.padt_work))
+ sock_put(sk);
}
abort:
@@ -612,6 +625,8 @@ static int pppoe_connect(struct socket *
lock_sock(sk);
+ INIT_WORK(&po->proto.pppoe.padt_work, pppoe_unbind_sock_work);
+
error = -EINVAL;
if (sp->sa_protocol != PX_PROTO_OE)
goto end;
--- a/include/linux/if_pppox.h
+++ b/include/linux/if_pppox.h
@@ -19,6 +19,7 @@
#include <linux/netdevice.h>
#include <linux/ppp_channel.h>
#include <linux/skbuff.h>
+#include <linux/workqueue.h>
#include <uapi/linux/if_pppox.h>
static inline struct pppoe_hdr *pppoe_hdr(const struct sk_buff *skb)
@@ -32,6 +33,7 @@ struct pppoe_opt {
struct pppoe_addr pa; /* what this socket is bound to*/
struct sockaddr_pppox relay; /* what socket data will be
relayed to (PPPoE relaying) */
+ struct work_struct padt_work;/* Work item for handling PADT */
};
struct pptp_opt {

25
target/linux/generic/pending-3.18/081-02-pppoe-Lacks-DST-MAC-address-check.patch
View File

@ -1,25 +0,0 @@
From: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Date: Mon, 20 Apr 2015 21:07:48 +0200
Subject: [PATCH] pppoe: Lacks DST MAC address check
A pppoe session is identified by its session ID and MAC address.
Currently pppoe does not check if the received pkg has the correct
MAC address. This is a problem when the eth I/F is in promisc mode
as then any DST MAC address is accepted.
Signed-off-by: Joakim Tjernlund <joakim.tjernlund@transmode.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -379,6 +379,9 @@ static int pppoe_rcv_core(struct sock *s
* can't change.
*/
+ if (skb->pkt_type == PACKET_OTHERHOST)
+ goto abort_kfree;
+
if (sk->sk_state & PPPOX_BOUND) {
ppp_input(&po->chan, skb);
} else if (sk->sk_state & PPPOX_RELAY) {

28
target/linux/generic/pending-3.18/081-03-pppoe-drop-pppoe-device-in-pppoe_unbind_sock_work.patch
View File

@ -1,28 +0,0 @@
From: Felix Fietkau <nbd@nbd.name>
Date: Sat, 9 May 2015 23:08:38 +0200
Subject: [PATCH] pppoe: drop pppoe device in pppoe_unbind_sock_work
After receiving a PADT and the socket is closed, user space will no
longer drop the reference to the pppoe device.
This leads to errors like this:
[ 488.570000] unregister_netdevice: waiting for eth0.2 to become free. Usage count = 2
Fixes: 287f3a943fe ("pppoe: Use workqueue to die properly when a PADT is received")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -464,6 +464,10 @@ static void pppoe_unbind_sock_work(struc
struct sock *sk = sk_pppox(po);
lock_sock(sk);
+ if (po->pppoe_dev) {
+ dev_put(po->pppoe_dev);
+ po->pppoe_dev = NULL;
+ }
pppox_unbind_sock(sk);
release_sock(sk);
sock_put(sk);

45
target/linux/generic/pending-3.18/081-06-ppp-don-t-set-sk_state-to-PPPOX_ZOMBIE-in-pppoe_disc.patch
View File

@ -1,45 +0,0 @@
From: Guillaume Nault <g.nault@alphalink.fr>
Date: Thu, 19 Nov 2015 12:52:56 +0100
Subject: [PATCH] ppp: don't set sk_state to PPPOX_ZOMBIE in pppoe_disc_rcv()
Since 287f3a943fef ("pppoe: Use workqueue to die properly when a PADT
is received"), pppoe_disc_rcv() disconnects the socket by scheduling
pppoe_unbind_sock_work(). This is enough to stop socket transmission
and makes the PPPOX_ZOMBIE state uncessary.
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -500,27 +500,9 @@ static int pppoe_disc_rcv(struct sk_buff
pn = pppoe_pernet(dev_net(dev));
po = get_item(pn, ph->sid, eth_hdr(skb)->h_source, dev->ifindex);
- if (po) {
- struct sock *sk = sk_pppox(po);
-
- bh_lock_sock(sk);
-
- /* If the user has locked the socket, just ignore
- * the packet. With the way two rcv protocols hook into
- * one socket family type, we cannot (easily) distinguish
- * what kind of SKB it is during backlog rcv.
- */
- if (sock_owned_by_user(sk) == 0) {
- /* We're no longer connect at the PPPOE layer,
- * and must wait for ppp channel to disconnect us.
- */
- sk->sk_state = PPPOX_ZOMBIE;
- }
-
- bh_unlock_sock(sk);
+ if (po)
if (!schedule_work(&po->proto.pppoe.padt_work))
- sock_put(sk);
- }
+ sock_put(sk_pppox(po));
abort:
kfree_skb(skb);

51
target/linux/generic/pending-3.18/081-07-ppp-remove-PPPOX_ZOMBIE-socket-state.patch
View File

@ -1,51 +0,0 @@
From: Guillaume Nault <g.nault@alphalink.fr>
Date: Thu, 19 Nov 2015 12:53:21 +0100
Subject: [PATCH] ppp: remove PPPOX_ZOMBIE socket state
PPPOX_ZOMBIE is never set anymore.
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -311,7 +311,7 @@ static void pppoe_flush_dev(struct net_d
lock_sock(sk);
if (po->pppoe_dev == dev &&
- sk->sk_state & (PPPOX_CONNECTED | PPPOX_BOUND | PPPOX_ZOMBIE)) {
+ sk->sk_state & (PPPOX_CONNECTED | PPPOX_BOUND)) {
pppox_unbind_sock(sk);
sk->sk_state_change(sk);
po->pppoe_dev = NULL;
@@ -775,7 +775,7 @@ static int pppoe_ioctl(struct socket *so
struct pppox_sock *relay_po;
err = -EBUSY;
- if (sk->sk_state & (PPPOX_BOUND | PPPOX_ZOMBIE | PPPOX_DEAD))
+ if (sk->sk_state & (PPPOX_BOUND | PPPOX_DEAD))
break;
err = -ENOTCONN;
--- a/drivers/net/ppp/pppox.c
+++ b/drivers/net/ppp/pppox.c
@@ -58,7 +58,7 @@ void pppox_unbind_sock(struct sock *sk)
{
/* Clear connection to ppp device, if attached. */
- if (sk->sk_state & (PPPOX_BOUND | PPPOX_CONNECTED | PPPOX_ZOMBIE)) {
+ if (sk->sk_state & (PPPOX_BOUND | PPPOX_CONNECTED)) {
ppp_unregister_channel(&pppox_sk(sk)->chan);
sk->sk_state = PPPOX_DEAD;
}
--- a/include/linux/if_pppox.h
+++ b/include/linux/if_pppox.h
@@ -91,7 +91,6 @@ enum {
PPPOX_CONNECTED = 1, /* connection established ==TCP_ESTABLISHED */
PPPOX_BOUND = 2, /* bound to ppp device */
PPPOX_RELAY = 4, /* forwarding is enabled */
- PPPOX_ZOMBIE = 8, /* dead, but still bound to ppp device */
PPPOX_DEAD = 16 /* dead, useless, please clean me up!*/
};

82
target/linux/generic/pending-3.18/081-08-pppoe-fix-memory-corruption-in-padt-work-structure.patch
View File

@ -1,82 +0,0 @@
From: Guillaume Nault <g.nault@alphalink.fr>
Date: Thu, 3 Dec 2015 16:49:32 +0100
Subject: [PATCH] pppoe: fix memory corruption in padt work structure
pppoe_connect() mustn't touch the padt_work field of pppoe sockets
because that work could be already pending.
[ 21.473147] BUG: unable to handle kernel NULL pointer dereference at 00000004
[ 21.474523] IP: [<c1043177>] process_one_work+0x29/0x31c
[ 21.475164] *pde = 00000000
[ 21.475513] Oops: 0000 [#1] SMP
[ 21.475910] Modules linked in: pppoe pppox ppp_generic slhc crc32c_intel aesni_intel virtio_net xts aes_i586 lrw gf128mul ablk_helper cryptd evdev acpi_cpufreq processor serio_raw button ext4 crc16 mbcache jbd2 virtio_blk virtio_pci virtio_ring virtio
[ 21.476168] CPU: 2 PID: 164 Comm: kworker/2:2 Not tainted 4.4.0-rc1 #1
[ 21.476168] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Debian-1.8.2-1 04/01/2014
[ 21.476168] task: f5f83c00 ti: f5e28000 task.ti: f5e28000
[ 21.476168] EIP: 0060:[<c1043177>] EFLAGS: 00010046 CPU: 2
[ 21.476168] EIP is at process_one_work+0x29/0x31c
[ 21.484082] EAX: 00000000 EBX: f678b2a0 ECX: 00000004 EDX: 00000000
[ 21.484082] ESI: f6c69940 EDI: f5e29ef0 EBP: f5e29f0c ESP: f5e29edc
[ 21.484082] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[ 21.484082] CR0: 80050033 CR2: 000000a4 CR3: 317ad000 CR4: 00040690
[ 21.484082] Stack:
[ 21.484082] 00000000 f6c69950 00000000 f6c69940 c0042338 f5e29f0c c1327945 00000000
[ 21.484082] 00000008 f678b2a0 f6c69940 f678b2b8 f5e29f30 c1043984 f5f83c00 f6c69970
[ 21.484082] f678b2a0 c10437d3 f6775e80 f678b2a0 c10437d3 f5e29fac c1047059 f5e29f74
[ 21.484082] Call Trace:
[ 21.484082] [<c1327945>] ? _raw_spin_lock_irq+0x28/0x30
[ 21.484082] [<c1043984>] worker_thread+0x1b1/0x244
[ 21.484082] [<c10437d3>] ? rescuer_thread+0x229/0x229
[ 21.484082] [<c10437d3>] ? rescuer_thread+0x229/0x229
[ 21.484082] [<c1047059>] kthread+0x8f/0x94
[ 21.484082] [<c1327a32>] ? _raw_spin_unlock_irq+0x22/0x26
[ 21.484082] [<c1327ee9>] ret_from_kernel_thread+0x21/0x38
[ 21.484082] [<c1046fca>] ? kthread_parkme+0x19/0x19
[ 21.496082] Code: 5d c3 55 89 e5 57 56 53 89 c3 83 ec 24 89 d0 89 55 e0 8d 7d e4 e8 6c d8 ff ff b9 04 00 00 00 89 45 d8 8b 43 24 89 45 dc 8b 45 d8 <8b> 40 04 8b 80 e0 00 00 00 c1 e8 05 24 01 88 45 d7 8b 45 e0 8d
[ 21.496082] EIP: [<c1043177>] process_one_work+0x29/0x31c SS:ESP 0068:f5e29edc
[ 21.496082] CR2: 0000000000000004
[ 21.496082] ---[ end trace e362cc9cf10dae89 ]---
Reported-by: Andrew <nitr0@seti.kr.ua>
Fixes: 287f3a943fef ("pppoe: Use workqueue to die properly when a PADT is received")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
--- a/drivers/net/ppp/pppoe.c
+++ b/drivers/net/ppp/pppoe.c
@@ -550,6 +550,9 @@ static int pppoe_create(struct net *net,
sk->sk_family = PF_PPPOX;
sk->sk_protocol = PX_PROTO_OE;
+ INIT_WORK(&pppox_sk(sk)->proto.pppoe.padt_work,
+ pppoe_unbind_sock_work);
+
return 0;
}
@@ -614,8 +617,6 @@ static int pppoe_connect(struct socket *
lock_sock(sk);
- INIT_WORK(&po->proto.pppoe.padt_work, pppoe_unbind_sock_work);
-
error = -EINVAL;
if (sp->sa_protocol != PX_PROTO_OE)
goto end;
@@ -645,8 +646,13 @@ static int pppoe_connect(struct socket *
po->pppoe_dev = NULL;
}
- memset(sk_pppox(po) + 1, 0,
- sizeof(struct pppox_sock) - sizeof(struct sock));
+ po->pppoe_ifindex = 0;
+ memset(&po->pppoe_pa, 0, sizeof(po->pppoe_pa));
+ memset(&po->pppoe_relay, 0, sizeof(po->pppoe_relay));
+ memset(&po->chan, 0, sizeof(po->chan));
+ po->next = NULL;
+ po->num = 0;
+
sk->sk_state = PPPOX_NONE;
}

101
target/linux/generic/pending-3.18/082-ipv6-ip6_fragment-fix-headroom-tests-and-skb-leak.patch
View File

@ -1,101 +0,0 @@
From: Florian Westphal <fw@strlen.de>
Date: Thu, 17 Sep 2015 11:24:48 +0100
Subject: [PATCH] ipv6: ip6_fragment: fix headroom tests and skb leak
David Woodhouse reports skb_under_panic when we try to push ethernet
header to fragmented ipv6 skbs:
skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000
data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan
[..]
ip6_finish_output2+0x196/0x4da
David further debugged this:
[..] offending fragments were arriving here with skb_headroom(skb)==10.
Which is reasonable, being the Solos ADSL card's header of 8 bytes
followed by 2 bytes of PPP frame type.
The problem is that if netfilter ipv6 defragmentation is used, skb_cow()
in ip6_forward will only see reassembled skb.
Therefore, headroom is overestimated by 8 bytes (we pulled fragment
header) and we don't check the skbs in the frag_list either.
We can't do these checks in netfilter defrag since outdev isn't known yet.
Furthermore, existing tests in ip6_fragment did not consider the fragment
or ipv6 header size when checking headroom of the fraglist skbs.
While at it, also fix a skb leak on memory allocation -- ip6_fragment
must consume the skb.
I tested this e1000 driver hacked to not allocate additional headroom
(we end up in slowpath, since LL_RESERVED_SPACE is 16).
If 2 bytes of headroom are allocated, fastpath is taken (14 byte
ethernet header was pulled, so 16 byte headroom available in all
fragments).
Reported-by: David Woodhouse <dwmw2@infradead.org>
Diagnosed-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Closes 20532
---
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -600,20 +600,22 @@ int ip6_fragment(struct sk_buff *skb, in
}
mtu -= hlen + sizeof(struct frag_hdr);
+ hroom = LL_RESERVED_SPACE(rt->dst.dev);
if (skb_has_frag_list(skb)) {
int first_len = skb_pagelen(skb);
struct sk_buff *frag2;
if (first_len - hlen > mtu ||
((first_len - hlen) & 7) ||
- skb_cloned(skb))
+ skb_cloned(skb) ||
+ skb_headroom(skb) < (hroom + sizeof(struct frag_hdr)))
goto slow_path;
skb_walk_frags(skb, frag) {
/* Correct geometry. */
if (frag->len > mtu ||
((frag->len & 7) && frag->next) ||
- skb_headroom(frag) < hlen)
+ skb_headroom(frag) < (hlen + hroom + sizeof(struct frag_hdr)))
goto slow_path_clean;
/* Partially cloned skb? */
@@ -630,8 +632,6 @@ int ip6_fragment(struct sk_buff *skb, in
err = 0;
offset = 0;
- frag = skb_shinfo(skb)->frag_list;
- skb_frag_list_init(skb);
/* BUILD HEADER */
*prevhdr = NEXTHDR_FRAGMENT;
@@ -639,8 +639,11 @@ int ip6_fragment(struct sk_buff *skb, in
if (!tmp_hdr) {
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_FRAGFAILS);
- return -ENOMEM;
+ err = -ENOMEM;
+ goto fail;
}
+ frag = skb_shinfo(skb)->frag_list;
+ skb_frag_list_init(skb);
__skb_pull(skb, hlen);
fh = (struct frag_hdr *)__skb_push(skb, sizeof(struct frag_hdr));
@@ -738,7 +741,6 @@ slow_path:
*/
*prevhdr = NEXTHDR_FRAGMENT;
- hroom = LL_RESERVED_SPACE(rt->dst.dev);
troom = rt->dst.dev->needed_tailroom;
/*

54
target/linux/generic/pending-3.18/083-solos-pci-Increase-headroom-on-received-packets.patch
View File

@ -1,54 +0,0 @@
From: David Woodhouse <dwmw2@infradead.org>
Date: Thu, 17 Sep 2015 11:19:53 +0100
Subject: [PATCH] solos-pci: Increase headroom on received packets
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
A comment in include/linux/skbuff.h says that:
* Various parts of the networking layer expect at least 32 bytes of
* headroom, you should not reduce this.
This was demonstrated by a panic when handling fragmented IPv6 packets:
http://marc.info/?l=linux-netdev&m=144236093519172&w=2
It's not entirely clear if that comment is still valid — and if it is,
perhaps netif_rx() ought to be enforcing it with a warning.
But either way, it is rather stupid from a performance point of view
for us to be receiving packets into a buffer which doesn't have enough
room to prepend an Ethernet header — it means that *every* incoming
packet is going to be need to be reallocated. So let's fix that.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
---
--- a/drivers/atm/solos-pci.c
+++ b/drivers/atm/solos-pci.c
@@ -805,7 +805,12 @@ static void solos_bh(unsigned long card_
continue;
}
- skb = alloc_skb(size + 1, GFP_ATOMIC);
+ /* Use netdev_alloc_skb() because it adds NET_SKB_PAD of
+ * headroom, and ensures we can route packets back out an
+ * Ethernet interface (for example) without having to
+ * reallocate. Adding NET_IP_ALIGN also ensures that both
+ * PPPoATM and PPPoEoBR2684 packets end up aligned. */
+ skb = netdev_alloc_skb_ip_align(NULL, size + 1);
if (!skb) {
if (net_ratelimit())
dev_warn(&card->dev->dev, "Failed to allocate sk_buff for RX\n");
@@ -869,7 +874,10 @@ static void solos_bh(unsigned long card_
/* Allocate RX skbs for any ports which need them */
if (card->using_dma && card->atmdev[port] &&
!card->rx_skb[port]) {
- struct sk_buff *skb = alloc_skb(RX_DMA_SIZE, GFP_ATOMIC);
+ /* Unlike the MMIO case (qv) we can't add NET_IP_ALIGN
+ * here; the FPGA can only DMA to addresses which are
+ * aligned to 4 bytes. */
+ struct sk_buff *skb = dev_alloc_skb(RX_DMA_SIZE);
if (skb) {
SKB_CB(skb)->dma_addr =
pci_map_single(card->dev, skb->data,

67
target/linux/generic/pending-3.18/087-regmap-make-LZO-cache-optional.patch
View File

@ -1,67 +0,0 @@
From de88e9b0354c2e3ff8eae3f97afe43a34f5ed239 Mon Sep 17 00:00:00 2001
From: Jonas Gorski <jonas.gorski@gmail.com>
Date: Sat, 13 May 2017 13:03:21 +0200
Subject: [PATCH] regmap: make LZO cache optional
Commit 2cbbb579bcbe3 ("regmap: Add the LZO cache support") added support
for LZO compression in regcache, but there were never any users added
afterwards. Since LZO support itself has its own size, it currently is
rather a deoptimization.
So make it optional by introducing a symbol that can be selected by
drivers wanting to make use of it.
Saves e.g. ~46 kB on MIPS (size of LZO support + regcache LZO code).
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
---
I tried using google to find any users (even out-of-tree ones), but at
best I found a single driver submission that was switched to RBTREE in
subsequent resubmissions (MFD_SMSC).
One could maybe also just drop the code because of no users for 5 years,
but that would be up to the maintainer(s) to decide.
drivers/base/regmap/Kconfig | 5 ++++-
drivers/base/regmap/Makefile | 3 ++-
drivers/base/regmap/regcache.c | 2 ++
3 files changed, 8 insertions(+), 2 deletions(-)
--- a/drivers/base/regmap/Kconfig
+++ b/drivers/base/regmap/Kconfig
@@ -4,9 +4,12 @@
config REGMAP
default y if (REGMAP_I2C || REGMAP_SPI || REGMAP_SPMI || REGMAP_MMIO || REGMAP_IRQ)
+ select IRQ_DOMAIN if REGMAP_IRQ
+ bool
+
+config REGCACHE_COMPRESSED
select LZO_COMPRESS
select LZO_DECOMPRESS
- select IRQ_DOMAIN if REGMAP_IRQ
bool
config REGMAP_I2C
--- a/drivers/base/regmap/Makefile
+++ b/drivers/base/regmap/Makefile
@@ -1,5 +1,6 @@
obj-$(CONFIG_REGMAP) += regmap.o regcache.o
-obj-$(CONFIG_REGMAP) += regcache-rbtree.o regcache-lzo.o regcache-flat.o
+obj-$(CONFIG_REGMAP) += regcache-rbtree.o regcache-flat.o
+obj-$(CONFIG_REGCACHE_COMPRESSED) += regcache-lzo.o
obj-$(CONFIG_DEBUG_FS) += regmap-debugfs.o
obj-$(CONFIG_REGMAP_I2C) += regmap-i2c.o
obj-$(CONFIG_REGMAP_SPI) += regmap-spi.o
--- a/drivers/base/regmap/regcache.c
+++ b/drivers/base/regmap/regcache.c
@@ -21,7 +21,9 @@
static const struct regcache_ops *cache_types[] = {
&regcache_rbtree_ops,
+#if IS_ENABLED(CONFIG_REGCACHE_COMPRESSED)
&regcache_lzo_ops,
+#endif
&regcache_flat_ops,
};

109
target/linux/generic/pending-3.18/090-overlayfs-fallback-to-readonly-when-full.patch
View File

@ -1,109 +0,0 @@
[linux-unionfs added to Cc]
On Tue, May 19, 2015 at 09:51:20AM +0200, Bastian Bittorf wrote:
> Hi Miklos,
>
> sorry for writing directly to you, feel free to forward
> this to the appropriate mailinglist.
>
> we have a problem with mainline overlay filesystem on kernel 3.18:
> https://dev.openwrt.org/ticket/19564
>
> 2 things are odd:
> when the working filesystem is full, overlays fails with:
>
> overlayfs: failed to create directory /overlay/work/work
>
> what is strange, that we call it with:
>
> mount(overlay, "/mnt", "overlay", MS_NOATIME, lowerdir)
>
> see here:
> http://nbd.name/gitweb.cgi?p=fstools.git;a=blob;f=libfstools/mount.c;h=81176ce399b4cd8e2d347c0008c13dec92407f55;hb=e6004000ff15d7bd32cf5663e8690fc94d7ec747#l125
>
> do you have an idea whats wrong?
> 1) is it really needed, that we need space for creating dir "/overlay/work"?
> 2) why does overlay need "/overlay/work/work"?
The work directory is needed for atomic copy-up and similar. It is not actually
necessary to mount a read-only overlay. Post 4.0 it is possible to mount the
overlay without workdir (but even then it won't happen automatically in case the
upper fs is full, so this should be fixed in the latest kernel too).
Could you please try the following patch? If the workdir can't be created it
will fall back to mounting the overlay read-only.
Thanks,
Miklos
---
fs/overlayfs/copy_up.c | 3 +++
fs/overlayfs/dir.c | 9 +++++++++
fs/overlayfs/super.c | 12 +++++++++---
3 files changed, 21 insertions(+), 3 deletions(-)
--- a/fs/overlayfs/copy_up.c
+++ b/fs/overlayfs/copy_up.c
@@ -315,6 +315,9 @@ int ovl_copy_up_one(struct dentry *paren
struct cred *override_cred;
char *link = NULL;
+ if (WARN_ON(!workdir))
+ return -EROFS;
+
ovl_path_upper(parent, &parentpath);
upperdir = parentpath.dentry;
--- a/fs/overlayfs/dir.c
+++ b/fs/overlayfs/dir.c
@@ -222,6 +222,9 @@ static struct dentry *ovl_clear_empty(st
struct kstat stat;
int err;
+ if (WARN_ON(!workdir))
+ return ERR_PTR(-EROFS);
+
err = ovl_lock_rename_workdir(workdir, upperdir);
if (err)
goto out;
@@ -322,6 +325,9 @@ static int ovl_create_over_whiteout(stru
struct dentry *newdentry;
int err;
+ if (WARN_ON(!workdir))
+ return -EROFS;
+
err = ovl_lock_rename_workdir(workdir, upperdir);
if (err)
goto out;
@@ -507,6 +513,9 @@ static int ovl_remove_and_whiteout(struc
int err;
int flags = 0;
+ if (WARN_ON(!workdir))
+ return -EROFS;
+
if (is_dir) {
opaquedir = ovl_check_empty_and_clear(dentry);
err = PTR_ERR(opaquedir);
--- a/fs/overlayfs/super.c
+++ b/fs/overlayfs/super.c
@@ -760,9 +760,15 @@ static int ovl_fill_super(struct super_b
ufs->workdir = ovl_workdir_create(ufs->upper_mnt, workpath.dentry);
err = PTR_ERR(ufs->workdir);
if (IS_ERR(ufs->workdir)) {
- pr_err("overlayfs: failed to create directory %s/%s\n",
- ufs->config.workdir, OVL_WORKDIR_NAME);
- goto out_put_lower_mnt;
+ if (err == -ENOSPC || err == -EROFS) {
+ pr_warning("overlayfs: failed to create work directory (%s), mounting read-only\n", err == ENOSPC ? "ENOSPC" : "EROFS");
+ sb->s_flags |= MS_RDONLY;
+ ufs->workdir = NULL;
+ } else {
+ pr_err("overlayfs: failed to create directory %s/%s\n",
+ ufs->config.workdir, OVL_WORKDIR_NAME);
+ goto out_put_lower_mnt;
+ }
}
/*

10
target/linux/generic/pending-3.18/091-mtd-spi-nor-add-support-Spansion_S25FL164K.patch
View File

@ -1,10 +0,0 @@
--- a/drivers/mtd/spi-nor/spi-nor.c
+++ b/drivers/mtd/spi-nor/spi-nor.c
@@ -567,6 +567,7 @@ static const struct spi_device_id spi_no
{ "s25fl008k", INFO(0xef4014, 0, 64 * 1024, 16, SECT_4K) },
{ "s25fl016k", INFO(0xef4015, 0, 64 * 1024, 32, SECT_4K) },
{ "s25fl064k", INFO(0xef4017, 0, 64 * 1024, 128, SECT_4K) },
+ { "s25fl164k", INFO(0x014017, 0, 64 * 1024, 128, SECT_4K) },
/* SST -- large erase sizes are "overlays", "sectors" are 4K */
{ "sst25vf040b", INFO(0xbf258d, 0, 64 * 1024, 8, SECT_4K | SST_WRITE) },

47
target/linux/generic/pending-3.18/092-01-spi-Check-to-see-if-the-device-is-processing-a-messa.patch
View File

@ -1,47 +0,0 @@
From: Mark Brown <broonie@kernel.org>
Date: Tue, 9 Dec 2014 19:46:56 +0000
Subject: [PATCH] spi: Check to see if the device is processing a message
before we idle
cur_msg is updated under the queue lock and holds the message we are
currently processing. Since currently we only ever do removals in the
pump kthread it doesn't matter in what order we do things but we want
to be able to push things out from the submitting thread so pull the
check to see if we're currently handling a message before we check to
see if the queue is idle.
Signed-off-by: Mark Brown <broonie@kernel.org>
---
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -891,8 +891,16 @@ static void spi_pump_messages(struct kth
bool was_busy = false;
int ret;
- /* Lock queue and check for queue work */
+ /* Lock queue */
spin_lock_irqsave(&master->queue_lock, flags);
+
+ /* Make sure we are not already running a message */
+ if (master->cur_msg) {
+ spin_unlock_irqrestore(&master->queue_lock, flags);
+ return;
+ }
+
+ /* Check if the queue is idle */
if (list_empty(&master->queue) || !master->running) {
if (!master->busy) {
spin_unlock_irqrestore(&master->queue_lock, flags);
@@ -916,11 +924,6 @@ static void spi_pump_messages(struct kth
return;
}
- /* Make sure we are not already running a message */
- if (master->cur_msg) {
- spin_unlock_irqrestore(&master->queue_lock, flags);
- return;
- }
/* Extract head of queue */
master->cur_msg =
list_first_entry(&master->queue, struct spi_message, queue);

184
target/linux/generic/pending-3.18/092-02-spi-Pump-transfers-inside-calling-context-for-spi_sy.patch
View File

@ -1,184 +0,0 @@
From: Mark Brown <broonie@kernel.org>
Date: Tue, 9 Dec 2014 21:38:05 +0000
Subject: [PATCH] spi: Pump transfers inside calling context for spi_sync()
If we are using the standard SPI message pump (which all drivers should be
transitioning over to) then special case the message enqueue and instead of
starting the worker thread to push messages to the hardware do so in the
context of the caller if the controller is idle. This avoids a context
switch in the common case where the controller has a single user in a
single thread, for short PIO transfers there may be no need to context
switch away from the calling context to complete the transfer.
The code is a bit more complex than is desirable in part due to the need
to handle drivers not using the standard queue and in part due to handling
the various combinations of bus locking and asynchronous submission in
interrupt context.
It is still suboptimal since it will still wake the message pump for each
transfer in order to schedule idling of the hardware and if multiple
contexts are using the controller simultaneously a caller may end up
pumping a message for some random other thread rather than for itself,
and if the thread ends up deferring due to another context idling the
hardware then it will just busy wait. It can, however, have the benefit
of aggregating power up and down of the hardware when a caller performs
a series of transfers back to back without any need for the use of
spi_async().
Signed-off-by: Mark Brown <broonie@kernel.org>
---
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -882,6 +882,9 @@ EXPORT_SYMBOL_GPL(spi_finalize_current_t
* needs processing and if so call out to the driver to initialize hardware
* and transfer each message.
*
+ * Note that it is called both from the kthread itself and also from
+ * inside spi_sync(); the queue extraction handling at the top of the
+ * function should deal with this safely.
*/
static void spi_pump_messages(struct kthread_work *work)
{
@@ -900,6 +903,13 @@ static void spi_pump_messages(struct kth
return;
}
+ /* If another context is idling the device then defer */
+ if (master->idling) {
+ queue_kthread_work(&master->kworker, &master->pump_messages);
+ spin_unlock_irqrestore(&master->queue_lock, flags);
+ return;
+ }
+
/* Check if the queue is idle */
if (list_empty(&master->queue) || !master->running) {
if (!master->busy) {
@@ -907,7 +917,9 @@ static void spi_pump_messages(struct kth
return;
}
master->busy = false;
+ master->idling = true;
spin_unlock_irqrestore(&master->queue_lock, flags);
+
kfree(master->dummy_rx);
master->dummy_rx = NULL;
kfree(master->dummy_tx);
@@ -921,6 +933,10 @@ static void spi_pump_messages(struct kth
pm_runtime_put_autosuspend(master->dev.parent);
}
trace_spi_master_idle(master);
+
+ spin_lock_irqsave(&master->queue_lock, flags);
+ master->idling = false;
+ spin_unlock_irqrestore(&master->queue_lock, flags);
return;
}
@@ -1166,12 +1182,9 @@ static int spi_destroy_queue(struct spi_
return 0;
}
-/**
- * spi_queued_transfer - transfer function for queued transfers
- * @spi: spi device which is requesting transfer
- * @msg: spi message which is to handled is queued to driver queue
- */
-static int spi_queued_transfer(struct spi_device *spi, struct spi_message *msg)
+static int __spi_queued_transfer(struct spi_device *spi,
+ struct spi_message *msg,
+ bool need_pump)
{
struct spi_master *master = spi->master;
unsigned long flags;
@@ -1186,13 +1199,23 @@ static int spi_queued_transfer(struct sp
msg->status = -EINPROGRESS;
list_add_tail(&msg->queue, &master->queue);
- if (!master->busy)
+ if (!master->busy && need_pump)
queue_kthread_work(&master->kworker, &master->pump_messages);
spin_unlock_irqrestore(&master->queue_lock, flags);
return 0;
}
+/**
+ * spi_queued_transfer - transfer function for queued transfers
+ * @spi: spi device which is requesting transfer
+ * @msg: spi message which is to handled is queued to driver queue
+ */
+static int spi_queued_transfer(struct spi_device *spi, struct spi_message *msg)
+{
+ return __spi_queued_transfer(spi, msg, true);
+}
+
static int spi_master_initialize_queue(struct spi_master *master)
{
int ret;
@@ -2104,19 +2127,46 @@ static int __spi_sync(struct spi_device
DECLARE_COMPLETION_ONSTACK(done);
int status;
struct spi_master *master = spi->master;
+ unsigned long flags;
+
+ status = __spi_validate(spi, message);
+ if (status != 0)
+ return status;
message->complete = spi_complete;
message->context = &done;
+ message->spi = spi;
if (!bus_locked)
mutex_lock(&master->bus_lock_mutex);
- status = spi_async_locked(spi, message);
+ /* If we're not using the legacy transfer method then we will
+ * try to transfer in the calling context so special case.
+ * This code would be less tricky if we could remove the
+ * support for driver implemented message queues.
+ */
+ if (master->transfer == spi_queued_transfer) {
+ spin_lock_irqsave(&master->bus_lock_spinlock, flags);
+
+ trace_spi_message_submit(message);
+
+ status = __spi_queued_transfer(spi, message, false);
+
+ spin_unlock_irqrestore(&master->bus_lock_spinlock, flags);
+ } else {
+ status = spi_async_locked(spi, message);
+ }
if (!bus_locked)
mutex_unlock(&master->bus_lock_mutex);
if (status == 0) {
+ /* Push out the messages in the calling context if we
+ * can.
+ */
+ if (master->transfer == spi_queued_transfer)
+ spi_pump_messages(&master->pump_messages);
+
wait_for_completion(&done);
status = message->status;
}
--- a/include/linux/spi/spi.h
+++ b/include/linux/spi/spi.h
@@ -260,6 +260,7 @@ static inline void spi_unregister_driver
* @pump_messages: work struct for scheduling work to the message pump
* @queue_lock: spinlock to syncronise access to message queue
* @queue: message queue
+ * @idling: the device is entering idle state
* @cur_msg: the currently in-flight message
* @cur_msg_prepared: spi_prepare_message was called for the currently
* in-flight message
@@ -425,6 +426,7 @@ struct spi_master {
spinlock_t queue_lock;
struct list_head queue;
struct spi_message *cur_msg;
+ bool idling;
bool busy;
bool running;
bool rt;

83
target/linux/generic/pending-3.18/092-03-spi-Only-idle-the-message-pump-in-the-worker-kthread.patch
View File

@ -1,83 +0,0 @@
From: Mark Brown <broonie@kernel.org>
Date: Wed, 10 Dec 2014 13:46:33 +0000
Subject: [PATCH] spi: Only idle the message pump in the worker kthread
In order to avoid the situation where the kthread is waiting for another
context to make the hardware idle let the message pump know if it's being
called from the worker thread context and if it isn't then defer to the
worker thread instead of idling the hardware immediately. This will ensure
that if this situation happens we block rather than busy waiting.
Signed-off-by: Mark Brown <broonie@kernel.org>
---
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -875,8 +875,9 @@ void spi_finalize_current_transfer(struc
EXPORT_SYMBOL_GPL(spi_finalize_current_transfer);
/**
- * spi_pump_messages - kthread work function which processes spi message queue
- * @work: pointer to kthread work struct contained in the master struct
+ * __spi_pump_messages - function which processes spi message queue
+ * @master: master to process queue for
+ * @in_kthread: true if we are in the context of the message pump thread
*
* This function checks if there is any spi message in the queue that
* needs processing and if so call out to the driver to initialize hardware
@@ -886,10 +887,8 @@ EXPORT_SYMBOL_GPL(spi_finalize_current_t
* inside spi_sync(); the queue extraction handling at the top of the
* function should deal with this safely.
*/
-static void spi_pump_messages(struct kthread_work *work)
+static void __spi_pump_messages(struct spi_master *master, bool in_kthread)
{
- struct spi_master *master =
- container_of(work, struct spi_master, pump_messages);
unsigned long flags;
bool was_busy = false;
int ret;
@@ -916,6 +915,15 @@ static void spi_pump_messages(struct kth
spin_unlock_irqrestore(&master->queue_lock, flags);
return;
}
+
+ /* Only do teardown in the thread */
+ if (!in_kthread) {
+ queue_kthread_work(&master->kworker,
+ &master->pump_messages);
+ spin_unlock_irqrestore(&master->queue_lock, flags);
+ return;
+ }
+
master->busy = false;
master->idling = true;
spin_unlock_irqrestore(&master->queue_lock, flags);
@@ -1004,6 +1012,18 @@ static void spi_pump_messages(struct kth
}
}
+/**
+ * spi_pump_messages - kthread work function which processes spi message queue
+ * @work: pointer to kthread work struct contained in the master struct
+ */
+static void spi_pump_messages(struct kthread_work *work)
+{
+ struct spi_master *master =
+ container_of(work, struct spi_master, pump_messages);
+
+ __spi_pump_messages(master, true);
+}
+
static int spi_init_queue(struct spi_master *master)
{
struct sched_param param = { .sched_priority = MAX_RT_PRIO - 1 };
@@ -2165,7 +2185,7 @@ static int __spi_sync(struct spi_device
* can.
*/
if (master->transfer == spi_queued_transfer)
- spi_pump_messages(&master->pump_messages);
+ __spi_pump_messages(master, false);
wait_for_completion(&done);
status = message->status;

146
target/linux/generic/pending-3.18/095-api-fix-compatibility-of-linux-in.h-with-netinet-in..patch
View File

@ -1,146 +0,0 @@
From 279c6c7fa64f5763e6b9f05e7ab3840092e702e7 Mon Sep 17 00:00:00 2001
From: Stephen Hemminger <stephen@networkplumber.org>
Date: Mon, 29 Jun 2015 14:57:48 -1000
Subject: [PATCH] api: fix compatibility of linux/in.h with netinet/in.h
u
This fixes breakage to iproute2 build with recent kernel headers
caused by:
commit a263653ed798216c0069922d7b5237ca49436007
Author: Pablo Neira Ayuso <pablo@netfilter.org>
Date: Wed Jun 17 10:28:27 2015 -0500
netfilter: don't pull include/linux/netfilter.h from netns headers
The issue is that definitions in linux/in.h overlap with those
in netinet/in.h. This patch solves this by introducing the same
mechanism as was used to solve the same problem with linux/in6.h
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/uapi/linux/in.h | 16 +++++++++++++---
include/uapi/linux/libc-compat.h | 22 ++++++++++++++++++++++
2 files changed, 35 insertions(+), 3 deletions(-)
--- a/include/uapi/linux/in.h
+++ b/include/uapi/linux/in.h
@@ -19,8 +19,10 @@
#define _UAPI_LINUX_IN_H
#include <linux/types.h>
+#include <linux/libc-compat.h>
#include <linux/socket.h>
+#if __UAPI_DEF_IN_IPPROTO
/* Standard well-defined IP protocols. */
enum {
IPPROTO_IP = 0, /* Dummy protocol for TCP */
@@ -73,12 +75,14 @@ enum {
#define IPPROTO_RAW IPPROTO_RAW
IPPROTO_MAX
};
+#endif
-
+#if __UAPI_DEF_IN_ADDR
/* Internet address. */
struct in_addr {
__be32 s_addr;
};
+#endif
#define IP_TOS 1
#define IP_TTL 2
@@ -154,6 +158,7 @@ struct in_addr {
/* Request struct for multicast socket ops */
+#if __UAPI_DEF_IP_MREQ
struct ip_mreq {
struct in_addr imr_multiaddr; /* IP multicast address of group */
struct in_addr imr_interface; /* local IP address of interface */
@@ -205,14 +210,18 @@ struct group_filter {
#define GROUP_FILTER_SIZE(numsrc) \
(sizeof(struct group_filter) - sizeof(struct __kernel_sockaddr_storage) \
+ (numsrc) * sizeof(struct __kernel_sockaddr_storage))
+#endif
+#if __UAPI_DEF_IN_PKTINFO
struct in_pktinfo {
int ipi_ifindex;
struct in_addr ipi_spec_dst;
struct in_addr ipi_addr;
};
+#endif
/* Structure describing an Internet (IP) socket address. */
+#if __UAPI_DEF_SOCKADDR_IN
#define __SOCK_SIZE__ 16 /* sizeof(struct sockaddr) */
struct sockaddr_in {
__kernel_sa_family_t sin_family; /* Address family */
@@ -224,8 +233,9 @@ struct sockaddr_in {
sizeof(unsigned short int) - sizeof(struct in_addr)];
};
#define sin_zero __pad /* for BSD UNIX comp. -FvK */
+#endif
-
+#if __UAPI_DEF_IN_CLASS
/*
* Definitions of the bits in an Internet address integer.
* On subnets, host and network parts are found according
@@ -276,7 +286,7 @@ struct sockaddr_in {
#define INADDR_ALLHOSTS_GROUP 0xe0000001U /* 224.0.0.1 */
#define INADDR_ALLRTRS_GROUP 0xe0000002U /* 224.0.0.2 */
#define INADDR_MAX_LOCAL_GROUP 0xe00000ffU /* 224.0.0.255 */
-
+#endif
/* <asm/byteorder.h> contains the htonl type stuff.. */
#include <asm/byteorder.h>
--- a/include/uapi/linux/libc-compat.h
+++ b/include/uapi/linux/libc-compat.h
@@ -56,6 +56,13 @@
/* GLIBC headers included first so don't define anything
* that would already be defined. */
+#define __UAPI_DEF_IN_ADDR 0
+#define __UAPI_DEF_IN_IPPROTO 0
+#define __UAPI_DEF_IN_PKTINFO 0
+#define __UAPI_DEF_IP_MREQ 0
+#define __UAPI_DEF_SOCKADDR_IN 0
+#define __UAPI_DEF_IN_CLASS 0
+
#define __UAPI_DEF_IN6_ADDR 0
/* The exception is the in6_addr macros which must be defined
* if the glibc code didn't define them. This guard matches
@@ -76,6 +83,13 @@
/* Linux headers included first, and we must define everything
* we need. The expectation is that glibc will check the
* __UAPI_DEF_* defines and adjust appropriately. */
+#define __UAPI_DEF_IN_ADDR 1
+#define __UAPI_DEF_IN_IPPROTO 1
+#define __UAPI_DEF_IN_PKTINFO 1
+#define __UAPI_DEF_IP_MREQ 1
+#define __UAPI_DEF_SOCKADDR_IN 1
+#define __UAPI_DEF_IN_CLASS 1
+
#define __UAPI_DEF_IN6_ADDR 1
/* We unconditionally define the in6_addr macros and glibc must
* coordinate. */
@@ -99,6 +113,14 @@
* that we need. */
#else /* !defined(__GLIBC__) */
+/* Definitions for in.h */
+#define __UAPI_DEF_IN_ADDR 1
+#define __UAPI_DEF_IN_IPPROTO 1
+#define __UAPI_DEF_IN_PKTINFO 1
+#define __UAPI_DEF_IP_MREQ 1
+#define __UAPI_DEF_SOCKADDR_IN 1
+#define __UAPI_DEF_IN_CLASS 1
+
/* Definitions for in6.h */
#define __UAPI_DEF_IN6_ADDR 1
#define __UAPI_DEF_IN6_ADDR_ALT 1

182
target/linux/generic/pending-3.18/099-module_arch_freeing_init-new-hook-for-archs-before-m.patch
View File

@ -1,182 +0,0 @@
From: Rusty Russell <rusty@rustcorp.com.au>
Date: Tue, 20 Jan 2015 09:07:04 +1030
Subject: [PATCH] module_arch_freeing_init(): new hook for archs before module->module_init freed.
Archs have been abusing module_free() to clean up their arch-specific
allocations. Since module_free() is also (ab)used by BPF and trace code,
let's keep it to simple allocations, and provide a hook called before
that.
This means that avr32, ia64, parisc and s390 no longer need to implement
their own module_free() at all. avr32 doesn't need module_finalize()
either.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Haavard Skinnemoen <hskinnemoen@gmail.com>
Cc: Hans-Christian Egtvedt <egtvedt@samfundet.no>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: "James E.J. Bottomley" <jejb@parisc-linux.org>
Cc: Helge Deller <deller@gmx.de>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-parisc@vger.kernel.org
Cc: linux-s390@vger.kernel.org
Origin: backport, https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=d453cded05ee219b77815ea194dc36efa5398bca
---
arch/avr32/kernel/module.c | 13 +------------
arch/ia64/kernel/module.c | 6 ++----
arch/parisc/kernel/module.c | 6 +-----
arch/s390/kernel/module.c | 10 +++-------
arch/tile/kernel/module.c | 2 +-
include/linux/moduleloader.h | 2 ++
kernel/module.c | 7 +++++++
7 files changed, 17 insertions(+), 29 deletions(-)
--- a/arch/avr32/kernel/module.c
+++ b/arch/avr32/kernel/module.c
@@ -19,12 +19,10 @@
#include <linux/moduleloader.h>
#include <linux/vmalloc.h>
-void module_free(struct module *mod, void *module_region)
+void module_arch_freeing_init(struct module *mod)
{
vfree(mod->arch.syminfo);
mod->arch.syminfo = NULL;
-
- vfree(module_region);
}
static inline int check_rela(Elf32_Rela *rela, struct module *module,
@@ -291,12 +289,3 @@ int apply_relocate_add(Elf32_Shdr *sechd
return ret;
}
-
-int module_finalize(const Elf_Ehdr *hdr, const Elf_Shdr *sechdrs,
- struct module *module)
-{
- vfree(module->arch.syminfo);
- module->arch.syminfo = NULL;
-
- return 0;
-}
--- a/arch/ia64/kernel/module.c
+++ b/arch/ia64/kernel/module.c
@@ -305,14 +305,12 @@ plt_target (struct plt_entry *plt)
#endif /* !USE_BRL */
void
-module_free (struct module *mod, void *module_region)
+module_arch_freeing_init (struct module *mod)
{
- if (mod && mod->arch.init_unw_table &&
- module_region == mod->module_init) {
+ if (mod->arch.init_unw_table) {
unw_remove_unwind_table(mod->arch.init_unw_table);
mod->arch.init_unw_table = NULL;
}
- vfree(module_region);
}
/* Have we already seen one of these relocations? */
--- a/arch/parisc/kernel/module.c
+++ b/arch/parisc/kernel/module.c
@@ -298,14 +298,10 @@ static inline unsigned long count_stubs(
}
#endif
-
-/* Free memory returned from module_alloc */
-void module_free(struct module *mod, void *module_region)
+void module_arch_freeing_init(struct module *mod)
{
kfree(mod->arch.section);
mod->arch.section = NULL;
-
- vfree(module_region);
}
/* Additional bytes needed in front of individual sections */
--- a/arch/s390/kernel/module.c
+++ b/arch/s390/kernel/module.c
@@ -55,14 +55,10 @@ void *module_alloc(unsigned long size)
}
#endif
-/* Free memory returned from module_alloc */
-void module_free(struct module *mod, void *module_region)
+void module_arch_freeing_init(struct module *mod)
{
- if (mod) {
- vfree(mod->arch.syminfo);
- mod->arch.syminfo = NULL;
- }
- vfree(module_region);
+ vfree(mod->arch.syminfo);
+ mod->arch.syminfo = NULL;
}
static void check_rela(Elf_Rela *rela, struct module *me)
--- a/arch/tile/kernel/module.c
+++ b/arch/tile/kernel/module.c
@@ -83,7 +83,7 @@ void module_free(struct module *mod, voi
0, 0, 0, NULL, NULL, 0);
/*
- * FIXME: If module_region == mod->module_init, trim exception
+ * FIXME: Add module_arch_freeing_init to trim exception
* table entries.
*/
}
--- a/include/linux/moduleloader.h
+++ b/include/linux/moduleloader.h
@@ -82,4 +82,6 @@ int module_finalize(const Elf_Ehdr *hdr,
/* Any cleanup needed when module leaves. */
void module_arch_cleanup(struct module *mod);
+/* Any cleanup before freeing mod->module_init */
+void module_arch_freeing_init(struct module *mod);
#endif
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -1840,6 +1840,10 @@ void __weak module_arch_cleanup(struct m
{
}
+void __weak module_arch_freeing_init(struct module *mod)
+{
+}
+
/* Free a module, remove from lists, etc. */
static void free_module(struct module *mod)
{
@@ -1872,6 +1876,7 @@ static void free_module(struct module *m
/* This may be NULL, but that's OK */
unset_module_init_ro_nx(mod);
+ module_arch_freeing_init(mod);
module_free(mod, mod->module_init);
kfree(mod->args);
percpu_modfree(mod);
@@ -2983,6 +2988,7 @@ static struct module *layout_and_allocat
static void module_deallocate(struct module *mod, struct load_info *info)
{
percpu_modfree(mod);
+ module_arch_freeing_init(mod);
module_free(mod, mod->module_init);
module_free(mod, mod->module_core);
}
@@ -3105,6 +3111,7 @@ static int do_init_module(struct module
rcu_assign_pointer(mod->kallsyms, &mod->core_kallsyms);
#endif
unset_module_init_ro_nx(mod);
+ module_arch_freeing_init(mod);
module_free(mod, mod->module_init);
mod->module_init = NULL;
mod->init_size = 0;

Some files were not shown because too many files have changed in this diff Show More