[Cc:ed to l-k and plausible maintainers: I'm not subscribed to netdev
so, please, if you prune l-k, leave me in the Cc]
I just upgraded to 2.6.17.8 (built with GCC 4.1.1 on x86 / Athlon IV).
As luck would have it I made a bunch of other changes at the same time,
namely that I LVMed my root filesystem and migrated to an initramfs (see
<http://linux-raid.osdl.org/index.php/RAID_Boot> for a slightly outdated
version of the init script I'm using, I must bring it up to date; among
other things it works on non-RAID boxes now, indeed the machine
experiencing the problems is not using md at all).
While I doubt these changes have caused this problem, they make it
tricky to back down to an earlier kernel, as older kernels I've already
compiled won't boot anymore (I'd have to rebuild one with the initramfs
in it). So I haven't tried to reproduce this on earlier kernels yet.
About half an hour after booting (while bootstrapping OCaml over NFSv3
as it happens), I started to notice long lags (~10s) accessing NFS-
mounted stuff. I was in XEmacs at the time so I wrote it off to a
malfunctioning itimer and decided to fix it later.
Then the build froze. I couldn't very well ignore *that*. Perhaps I
couldn't blame XEmacs after all.
The kernel log showed a heap of BUGs from somewhere inside the skb
management layer, somewhere in UDP fragment processing while
handling NFS requests. It starts like this:
Aug 12 21:31:08 hades warning: kernel: BUG: warning at include/linux/skbuff.h:975/__skb_trim()
Aug 12 21:31:08 hades warning: kernel: <c030ed39> ip_append_data+0x5b3/0x951 <c030fc18> ip_generic_getfrag+0x0/0x96
Aug 12 21:31:08 hades warning: kernel: <c030fc18> ip_generic_getfrag+0x0/0x96 <c0326307> udp_sendmsg+0x42e/0x51d
Aug 12 21:31:08 hades warning: kernel: <c032632f> udp_sendmsg+0x456/0x51d <c032b8f5> inet_sendmsg+0x3b/0x45
Aug 12 21:31:08 hades warning: kernel: <c02f309c> sock_sendmsg+0xbd/0xd5 <c02f309c> sock_sendmsg+0xbd/0xd5
Aug 12 21:31:08 hades warning: kernel: <c0122cac> autoremove_wake_function+0x0/0x35 <c0122cac> autoremove_wake_function+0x0/0x35
Aug 12 21:31:08 hades warning: kernel: <c03264e6> udp_sendpage+0xf0/0xfa <c02f4ad6> kernel_sendmsg+0x26/0x2c
Aug 12 21:31:08 hades warning: kernel: <c033bdff> xs_send_tail+0x4b/0x52 <c033d2b3> xs_udp_send_request+0x1f4/0x295
Aug 12 21:31:08 hades warning: kernel: <c032b8ff> inet_sendpage+0x0/0x83 <c033aed0> xprt_transmit+0xcc/0x1fb
Aug 12 21:31:08 hades warning: kernel: <c033e9b3> rpcauth_wrap_req+0x6c/0x74 <c033ea58> rpcauth_marshcred+0x4b/0x52
Aug 12 21:31:08 hades warning: kernel: <c033b1cb> xprt_prepare_transmit+0x80/0x89 <c01dcd10> nfs3_xdr_writeargs+0x0/0x7f
Aug 12 21:31:08 hades warning: kernel: <c033a20b> call_transmit+0x1ac/0x1e9 <c033dfab> __rpc_execute+0x82/0x1af
Aug 12 21:31:08 hades warning: kernel: <c011b62d> sigprocmask+0x86/0xa3 <c01d8466> nfs_execute_write+0x34/0x48
Aug 12 21:31:08 hades warning: kernel: <c01d9874> nfs_flush_one+0xc1/0x102 <c01d86e3> nfs_flush_list+0x87/0xcf
Aug 12 21:31:08 hades warning: kernel: <c01d97b3> nfs_flush_one+0x0/0x102 <c01d96d8> nfs_sync_inode_wait+0x9a/0x112
Aug 12 21:31:08 hades warning: kernel: <c01d1d17> nfs_file_flush+0x7b/0xb4 <c0144955> filp_close+0x33/0x59
Aug 12 21:31:08 hades warning: kernel: <c01026af> sysenter_past_esp+0x54/0x75
... and then goes on like this, as the system tries to flush out the
data that Just Won't Go:
Aug 12 21:31:08 hades warning: kernel: BUG: warning at include/linux/skbuff.h:975/__skb_trim()
Aug 12 21:31:08 hades warning: kernel: <c030ed39> ip_append_data+0x5b3/0x951 <c030fc18> ip_generic_getfrag+0x0/0x96
Aug 12 21:31:08 hades warning: kernel: <c030fc18> ip_generic_getfrag+0x0/0x96 <c0326307> udp_sendmsg+0x42e/0x51d
Aug 12 21:31:08 hades warning: kernel: <c032632f> udp_sendmsg+0x456/0x51d <c032b8f5> inet_sendmsg+0x3b/0x45
Aug 12 21:31:08 hades warning: kernel: <c02f309c> sock_sendmsg+0xbd/0xd5 <c02f309c> sock_sendmsg+0xbd/0xd5
Aug 12 21:31:08 hades warning: kernel: <c0122cac> autoremove_wake_function+0x0/0x35 <c0122cac> autoremove_wake_function+0x0/0x35
Aug 12 21:31:08 hades warning: kernel: <c03264e6> udp_sendpage+0xf0/0xfa <c02f4ad6> kernel_sendmsg+0x26/0x2c
Aug 12 21:31:08 hades warning: kernel: <c033bdff> xs_send_tail+0x4b/0x52 <c033d2b3> xs_udp_send_request+0x1f4/0x295
Aug 12 21:31:08 hades warning: kernel: <c032b8ff> inet_sendpage+0x0/0x83 <c01dcd10> nfs3_xdr_writeargs+0x0/0x7f
Aug 12 21:31:08 hades warning: kernel: <c033aed0> xprt_transmit+0xcc/0x1fb <c033e9b3> rpcauth_wrap_req+0x6c/0x74
Aug 12 21:31:08 hades warning: kernel: <c033ea58> rpcauth_marshcred+0x4b/0x52 <c033b1cb> xprt_prepare_transmit+0x80/0x89
Aug 12 21:31:08 hades warning: kernel: <c01dcd10> nfs3_xdr_writeargs+0x0/0x7f <c033a20b> call_transmit+0x1ac/0x1e9
Aug 12 21:31:08 hades warning: kernel: <c033dfab> __rpc_execute+0x82/0x1af <c0347c70> schedule+0x47a/0x519
Aug 12 21:31:08 hades warning: kernel: <c01203bd> run_workqueue+0x9d/0xfc <c033e0d8> rpc_async_schedule+0x0/0x5
Aug 12 21:31:08 hades warning: kernel: <c0120758> worker_thread+0x0/0x10c <c0120833> worker_thread+0xdb/0x10c
Aug 12 21:31:08 hades warning: kernel: <c0110d1f> default_wake_function+0x0/0xc <c0122bea> kthread+0x91/0xbd
Aug 12 21:31:08 hades warning: kernel: <c0122b59> kthread+0x0/0xbd <c0100ba5> kernel_thread_helper+0x5/0xb
This latter BUG then repeated every second or two: eventually NFS
stalled (perhaps all UDP traffic was stalled: I didn't check.)
Possibly-interesting network parameters:
1: gordianet: <BROADCAST,MULTICAST,UP,10000> mtu 1458 qdisc pfifo_fast qlen 100
link/ether 00:50:04:32:9e:2f brd ff:ff:ff:ff:ff:ff
inet 192.168.14.18/24 brd 192.168.14.255 scope global gordianet
The NFS mounts I was accessing at the time, obviously hitting the UDP
fragment processing layer rather hard given an MTU of 1458:
/home/.loki.wkstn.nix from loki:/home/.loki.wkstn.nix
Flags: rw,vers=3,rsize=32768,wsize=32768,hard,intr,proto=udp,timeo=7,retrans=3,addr=loki
/usr/packages from package:/usr/packages
Flags: rw,vers=3,rsize=32768,wsize=32768,hard,intr,proto=udp,timeo=7,retrans=3,addr=package
(`package' is a CNAME for `loki', so these are the same machine.)
Loaded modules:
Module Size Used by
radeon 101152 2
drm 58836 3 radeon
The .config:
CONFIG_X86_32=y
CONFIG_SEMAPHORE_SLEEPERS=y
CONFIG_X86=y
CONFIG_MMU=y
CONFIG_GENERIC_ISA_DMA=y
CONFIG_GENERIC_IOMAP=y
CONFIG_GENERIC_HWEIGHT=y
CONFIG_ARCH_MAY_HAVE_PC_FDC=y
CONFIG_DMI=y
CONFIG_EXPERIMENTAL=y
CONFIG_BROKEN_ON_SMP=y
CONFIG_LOCK_KERNEL=y
CONFIG_INIT_ENV_ARG_LIMIT=32
CONFIG_LOCALVERSION=""
CONFIG_LOCALVERSION_AUTO=y
CONFIG_SWAP=y
CONFIG_SYSVIPC=y
CONFIG_POSIX_MQUEUE=y
CONFIG_BSD_PROCESS_ACCT=y
CONFIG_SYSCTL=y
CONFIG_RELAY=y
CONFIG_INITRAMFS_SOURCE="usr/initramfs.hades"
CONFIG_INITRAMFS_ROOT_UID=99
CONFIG_INITRAMFS_ROOT_GID=101
CONFIG_UID16=y
CONFIG_VM86=y
CONFIG_CC_OPTIMIZE_FOR_SIZE=y
CONFIG_KALLSYMS=y
CONFIG_HOTPLUG=y
CONFIG_PRINTK=y
CONFIG_BUG=y
CONFIG_ELF_CORE=y
CONFIG_BASE_FULL=y
CONFIG_FUTEX=y
CONFIG_EPOLL=y
CONFIG_SHMEM=y
CONFIG_SLAB=y
CONFIG_BASE_SMALL=0
CONFIG_MODULES=y
CONFIG_MODULE_UNLOAD=y
CONFIG_KMOD=y
CONFIG_BLK_DEV_IO_TRACE=y
CONFIG_IOSCHED_NOOP=y
CONFIG_IOSCHED_AS=m
CONFIG_IOSCHED_DEADLINE=m
CONFIG_IOSCHED_CFQ=y
CONFIG_DEFAULT_CFQ=y
CONFIG_DEFAULT_IOSCHED="cfq"
CONFIG_X86_PC=y
CONFIG_MK7=y
CONFIG_X86_CMPXCHG=y
CONFIG_X86_XADD=y
CONFIG_X86_L1_CACHE_SHIFT=6
CONFIG_RWSEM_XCHGADD_ALGORITHM=y
CONFIG_GENERIC_CALIBRATE_DELAY=y
CONFIG_X86_WP_WORKS_OK=y
CONFIG_X86_INVLPG=y
CONFIG_X86_BSWAP=y
CONFIG_X86_POPAD_OK=y
CONFIG_X86_CMPXCHG64=y
CONFIG_X86_GOOD_APIC=y
CONFIG_X86_INTEL_USERCOPY=y
CONFIG_X86_USE_PPRO_CHECKSUM=y
CONFIG_X86_USE_3DNOW=y
CONFIG_X86_TSC=y
CONFIG_HPET_TIMER=y
CONFIG_HPET_EMULATE_RTC=y
CONFIG_PREEMPT=y
CONFIG_PREEMPT_BKL=y
CONFIG_X86_UP_APIC=y
CONFIG_X86_UP_IOAPIC=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y
CONFIG_X86_MCE=y
CONFIG_X86_MCE_NONFATAL=y
CONFIG_NOHIGHMEM=y
CONFIG_PAGE_OFFSET=0xC0000000
CONFIG_PROC_MM=y
CONFIG_ARCH_FLATMEM_ENABLE=y
CONFIG_ARCH_SPARSEMEM_ENABLE=y
CONFIG_ARCH_SELECT_MEMORY_MODEL=y
CONFIG_SELECT_MEMORY_MODEL=y
CONFIG_FLATMEM_MANUAL=y
CONFIG_FLATMEM=y
CONFIG_FLAT_NODE_MEM_MAP=y
CONFIG_SPARSEMEM_STATIC=y
CONFIG_SPLIT_PTLOCK_CPUS=4
CONFIG_MTRR=y
CONFIG_REGPARM=y
CONFIG_HZ_1000=y
CONFIG_HZ=1000
CONFIG_PHYSICAL_START=0x100000
CONFIG_PM=y
CONFIG_PM_LEGACY=y
CONFIG_PCI=y
CONFIG_PCI_GOANY=y
CONFIG_PCI_BIOS=y
CONFIG_PCI_DIRECT=y
CONFIG_ISA_DMA_API=y
CONFIG_BINFMT_ELF=y
CONFIG_BINFMT_MISC=y
CONFIG_NET=y
CONFIG_PACKET=y
CONFIG_PACKET_MMAP=y
CONFIG_UNIX=y
CONFIG_INET=y
CONFIG_IP_MULTICAST=y
CONFIG_IP_FIB_HASH=y
CONFIG_INET_DIAG=y
CONFIG_INET_TCP_DIAG=y
CONFIG_TCP_CONG_BIC=y
CONFIG_STANDALONE=y
CONFIG_PREVENT_FIRMWARE_BUILD=y
CONFIG_FW_LOADER=m
CONFIG_PARPORT=y
CONFIG_PARPORT_PC=y
CONFIG_PARPORT_1284=y
CONFIG_BLK_DEV_FD=y
CONFIG_BLK_DEV_LOOP=m
CONFIG_BLK_DEV_CRYPTOLOOP=m
CONFIG_BLK_DEV_NBD=m
CONFIG_CDROM_PKTCDVD=y
CONFIG_CDROM_PKTCDVD_BUFFERS=16
CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDEDISK=y
CONFIG_IDEDISK_MULTI_MODE=y
CONFIG_BLK_DEV_IDECD=y
CONFIG_IDE_GENERIC=y
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_OFFBOARD=y
CONFIG_BLK_DEV_GENERIC=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_PDC202XX_NEW=y
CONFIG_BLK_DEV_VIA82CXXX=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_IDEDMA_AUTO=y
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_MD=y
CONFIG_BLK_DEV_DM=y
CONFIG_DM_CRYPT=y
CONFIG_DM_SNAPSHOT=y
CONFIG_DM_MIRROR=y
CONFIG_DM_ZERO=y
CONFIG_NETDEVICES=y
CONFIG_DUMMY=m
CONFIG_NET_ETHERNET=y
CONFIG_MII=y
CONFIG_NET_VENDOR_3COM=y
CONFIG_VORTEX=y
CONFIG_INPUT=y
CONFIG_INPUT_MOUSEDEV=y
CONFIG_INPUT_MOUSEDEV_PSAUX=y
CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024
CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768
CONFIG_INPUT_JOYDEV=y
CONFIG_INPUT_KEYBOARD=y
CONFIG_KEYBOARD_ATKBD=y
CONFIG_INPUT_MOUSE=y
CONFIG_MOUSE_PS2=y
CONFIG_INPUT_JOYSTICK=y
CONFIG_JOYSTICK_ANALOG=y
CONFIG_INPUT_MISC=y
CONFIG_SERIO=y
CONFIG_SERIO_I8042=y
CONFIG_SERIO_LIBPS2=y
CONFIG_GAMEPORT=y
CONFIG_GAMEPORT_EMU10K1=y
CONFIG_VT=y
CONFIG_VT_CONSOLE=y
CONFIG_HW_CONSOLE=y
CONFIG_SERIAL_8250=y
CONFIG_SERIAL_8250_PCI=y
CONFIG_SERIAL_8250_NR_UARTS=4
CONFIG_SERIAL_8250_RUNTIME_UARTS=4
CONFIG_SERIAL_CORE=y
CONFIG_UNIX98_PTYS=y
CONFIG_PRINTER=y
CONFIG_RTC=y
CONFIG_AGP=y
CONFIG_AGP_VIA=y
CONFIG_DRM=m
CONFIG_DRM_RADEON=m
CONFIG_I2C=y
CONFIG_I2C_CHARDEV=y
CONFIG_I2C_ISA=y
CONFIG_I2C_VIAPRO=y
CONFIG_HWMON=y
CONFIG_HWMON_VID=y
CONFIG_SENSORS_W83627HF=y
CONFIG_VIDEO_V4L2=y
CONFIG_VIDEO_SELECT=y
CONFIG_VGA_CONSOLE=y
CONFIG_DUMMY_CONSOLE=y
CONFIG_SOUND=y
CONFIG_SND=y
CONFIG_SND_TIMER=y
CONFIG_SND_PCM=y
CONFIG_SND_HWDEP=y
CONFIG_SND_RAWMIDI=y
CONFIG_SND_SEQUENCER=y
CONFIG_SND_SEQ_DUMMY=m
CONFIG_SND_OSSEMUL=y
CONFIG_SND_MIXER_OSS=y
CONFIG_SND_PCM_OSS=y
CONFIG_SND_PCM_OSS_PLUGINS=y
CONFIG_SND_SEQUENCER_OSS=y
CONFIG_SND_RTCTIMER=y
CONFIG_SND_SEQ_RTCTIMER_DEFAULT=y
CONFIG_SND_SUPPORT_OLD_API=y
CONFIG_SND_VERBOSE_PROCFS=y
CONFIG_SND_MPU401_UART=m
CONFIG_SND_AC97_CODEC=y
CONFIG_SND_AC97_BUS=y
CONFIG_SND_MPU401=m
CONFIG_SND_EMU10K1=y
CONFIG_SND_INTEL8X0=y
CONFIG_USB_ARCH_HAS_HCD=y
CONFIG_USB_ARCH_HAS_OHCI=y
CONFIG_USB_ARCH_HAS_EHCI=y
CONFIG_USB=y
CONFIG_USB_DEVICEFS=y
CONFIG_USB_BANDWIDTH=y
CONFIG_USB_DYNAMIC_MINORS=y
CONFIG_USB_EHCI_HCD=y
CONFIG_USB_UHCI_HCD=y
CONFIG_USB_PRINTER=y
CONFIG_USB_STORAGE=y
CONFIG_EXT2_FS=y
CONFIG_EXT2_FS_XATTR=y
CONFIG_EXT2_FS_POSIX_ACL=y
CONFIG_EXT3_FS=y
CONFIG_EXT3_FS_XATTR=y
CONFIG_EXT3_FS_POSIX_ACL=y
CONFIG_JBD=y
CONFIG_FS_MBCACHE=y
CONFIG_REISERFS_FS=y
CONFIG_REISERFS_FS_XATTR=y
CONFIG_REISERFS_FS_POSIX_ACL=y
CONFIG_FS_POSIX_ACL=y
CONFIG_INOTIFY=y
CONFIG_QUOTA=y
CONFIG_QFMT_V2=y
CONFIG_QUOTACTL=y
CONFIG_DNOTIFY=y
CONFIG_FUSE_FS=y
CONFIG_ISO9660_FS=y
CONFIG_JOLIET=y
CONFIG_ZISOFS=y
CONFIG_ZISOFS_FS=y
CONFIG_UDF_FS=y
CONFIG_UDF_NLS=y
CONFIG_FAT_FS=y
CONFIG_MSDOS_FS=y
CONFIG_VFAT_FS=m
CONFIG_FAT_DEFAULT_CODEPAGE=437
CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1"
CONFIG_PROC_FS=y
CONFIG_SYSFS=y
CONFIG_TMPFS=y
CONFIG_HUGETLBFS=y
CONFIG_HUGETLB_PAGE=y
CONFIG_RAMFS=y
CONFIG_CONFIGFS_FS=y
CONFIG_NFS_FS=y
CONFIG_NFS_V3=y
CONFIG_NFS_V3_ACL=y
CONFIG_NFSD=y
CONFIG_NFSD_V2_ACL=y
CONFIG_NFSD_V3=y
CONFIG_NFSD_V3_ACL=y
CONFIG_NFSD_TCP=y
CONFIG_LOCKD=y
CONFIG_LOCKD_V4=y
CONFIG_EXPORTFS=y
CONFIG_NFS_ACL_SUPPORT=y
CONFIG_NFS_COMMON=y
CONFIG_SUNRPC=y
CONFIG_PARTITION_ADVANCED=y
CONFIG_MSDOS_PARTITION=y
CONFIG_NLS=y
CONFIG_NLS_DEFAULT="iso8859-1"
CONFIG_NLS_CODEPAGE_437=y
CONFIG_NLS_ASCII=m
CONFIG_NLS_ISO8859_1=y
CONFIG_NLS_ISO8859_15=m
CONFIG_PROFILING=y
CONFIG_OPROFILE=y
CONFIG_LOG_BUF_SHIFT=14
CONFIG_DEBUG_BUGVERBOSE=y
CONFIG_DEBUG_FS=y
CONFIG_UNWIND_INFO=y
CONFIG_EARLY_PRINTK=y
CONFIG_STACK_BACKTRACE_COLS=2
CONFIG_X86_FIND_SMP_CONFIG=y
CONFIG_X86_MPPARSE=y
CONFIG_DOUBLEFAULT=y
CONFIG_CRYPTO=y
CONFIG_CRC32=y
CONFIG_ZLIB_INFLATE=y
CONFIG_GENERIC_HARDIRQS=y
CONFIG_GENERIC_IRQ_PROBE=y
CONFIG_X86_BIOS_REBOOT=y
CONFIG_KTIME_SCALAR=y
On 12 Aug 2006, [email protected] mused:
> Then the build froze. I couldn't very well ignore *that*. Perhaps I
> couldn't blame XEmacs after all.
It just happened again. It's reproducibly triggered, at least on this
system, by the ocaml-3.09.02 configure script running over NFS (probably
NFS over UDP is necessary as well).
In theory I could now use this to minimise the problem, but the sheer
number of reboots this is likely to need is dissuading me: I'll put it
off until someone actually wants such a minimal crash case :)
--
`We're sysadmins. We deal with the inconceivable so often I can clearly
see the need to define levels of inconceivability.' --- Rik Steenwinkel
On 12 Aug 2006, [email protected] wrote:
> On 12 Aug 2006, [email protected] mused:
>> Then the build froze. I couldn't very well ignore *that*. Perhaps I
>> couldn't blame XEmacs after all.
>
> It just happened again. It's reproducibly triggered, at least on this
> system, by the ocaml-3.09.02 configure script running over NFS (probably
> NFS over UDP is necessary as well).
... also triggered by running a CVS update on a large tree, and pretty
much by doing anything heavy-duty with NFS on UDP at all.
NFS on TCP is, as expected, unaffected by this bug.
--
`We're sysadmins. We deal with the inconceivable so often I can clearly
see the need to define levels of inconceivability.' --- Rik Steenwinkel
On Sat, Aug 12, 2006 at 09:19:19PM +0000, Nix wrote:
>
> The kernel log showed a heap of BUGs from somewhere inside the skb
> management layer, somewhere in UDP fragment processing while
> handling NFS requests. It starts like this:
>
> Aug 12 21:31:08 hades warning: kernel: BUG: warning at include/linux/skbuff.h:975/__skb_trim()
> Aug 12 21:31:08 hades warning: kernel: <c030ed39> ip_append_data+0x5b3/0x951 <c030fc18> ip_generic_getfrag+0x0/0x96
Oops, I missed this code path when I disallowed skb_trim from operating
on a paged skb. This patch should fix the problem.
Greg, we need this for 2.6.17 stable as well if Dave is OK with it.
[INET]: Use pskb_trim_unique when trimming paged unique skbs
The IPv4/IPv6 datagram output path was using skb_trim to trim paged
packets because they know that the packet has not been cloned yet
(since the packet hasn't been given to anything else in the system).
This broke because skb_trim no longer allows paged packets to be
trimmed. Paged packets must be given to one of the pskb_trim functions
instead.
This patch adds a new pskb_trim_unique function to cover the IPv4/IPv6
datagram output path scenario and replaces the corresponding skb_trim
calls with it.
Signed-off-by: Herbert Xu <[email protected]>
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 19c96d4..7f99995 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1040,6 +1040,21 @@ static inline int pskb_trim(struct sk_bu
}
/**
+ * pskb_trim_unique - remove end from a paged unique (not cloned) buffer
+ * @skb: buffer to alter
+ * @len: new length
+ *
+ * This is identical to pskb_trim except that the caller knows that
+ * the skb is not cloned so we should never get an error due to out-
+ * of-memory.
+ */
+static inline void pskb_trim_unique(struct sk_buff *skb, unsigned int len)
+{
+ int err = pskb_trim(skb, len);
+ BUG_ON(err);
+}
+
+/**
* skb_orphan - orphan a buffer
* @skb: buffer to orphan
*
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 9bf307a..4c20f55 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -947,7 +947,7 @@ alloc_new_skb:
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
data += fraggap;
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
copy = datalen - transhdrlen - fraggap;
@@ -1142,7 +1142,7 @@ ssize_t ip_append_page(struct sock *sk,
data, fraggap, 0);
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
/*
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 69451af..4fb47a2 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -1095,7 +1095,7 @@ alloc_new_skb:
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
data += fraggap;
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
copy = datalen - transhdrlen - fraggap;
if (copy < 0) {
On Sun, 13 Aug 2006, Herbert Xu suggested tentatively:
> Oops, I missed this code path when I disallowed skb_trim from operating
> on a paged skb. This patch should fix the problem.
>
> Greg, we need this for 2.6.17 stable as well if Dave is OK with it.
>
> [INET]: Use pskb_trim_unique when trimming paged unique skbs
[...]
Yes, that seems to have fixed it. Thank you!
--
`We're sysadmins. We deal with the inconceivable so often I can clearly
see the need to define levels of inconceivability.' --- Rik Steenwinkel
From: Herbert Xu <[email protected]>
Date: Sun, 13 Aug 2006 22:59:11 +1000
> On Sat, Aug 12, 2006 at 09:19:19PM +0000, Nix wrote:
> >
> > The kernel log showed a heap of BUGs from somewhere inside the skb
> > management layer, somewhere in UDP fragment processing while
> > handling NFS requests. It starts like this:
> >
> > Aug 12 21:31:08 hades warning: kernel: BUG: warning at include/linux/skbuff.h:975/__skb_trim()
> > Aug 12 21:31:08 hades warning: kernel: <c030ed39> ip_append_data+0x5b3/0x951 <c030fc18> ip_generic_getfrag+0x0/0x96
>
> Oops, I missed this code path when I disallowed skb_trim from operating
> on a paged skb. This patch should fix the problem.
>
> Greg, we need this for 2.6.17 stable as well if Dave is OK with it.
>
> [INET]: Use pskb_trim_unique when trimming paged unique skbs
...
> Signed-off-by: Herbert Xu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Looks good, Greg please push to -stable. I'll push this into
2.6.18 under seperate cover.
On Sun, Aug 13, 2006 at 10:59:11PM +1000, Herbert Xu wrote:
> On Sat, Aug 12, 2006 at 09:19:19PM +0000, Nix wrote:
> >
> > The kernel log showed a heap of BUGs from somewhere inside the skb
> > management layer, somewhere in UDP fragment processing while
> > handling NFS requests. It starts like this:
> >
> > Aug 12 21:31:08 hades warning: kernel: BUG: warning at include/linux/skbuff.h:975/__skb_trim()
> > Aug 12 21:31:08 hades warning: kernel: <c030ed39> ip_append_data+0x5b3/0x951 <c030fc18> ip_generic_getfrag+0x0/0x96
>
> Oops, I missed this code path when I disallowed skb_trim from operating
> on a paged skb. This patch should fix the problem.
>
> Greg, we need this for 2.6.17 stable as well if Dave is OK with it.
This patch doesn't apply at all to the latest 2.6.17-stable kernel tree.
Care to rediff it?
thanks,
greg k-h
On Fri, Aug 25, 2006 at 04:03:16PM -0700, Greg KH wrote:
>
> This patch doesn't apply at all to the latest 2.6.17-stable kernel tree.
> Care to rediff it?
Hmm, I just rebased and it actually applied as is to 2.6.17.11 :)
Anyway, here is the result:
[INET]: Use pskb_trim_unique when trimming paged unique skbs
The IPv4/IPv6 datagram output path was using skb_trim to trim paged
packets because they know that the packet has not been cloned yet
(since the packet hasn't been given to anything else in the system).
This broke because skb_trim no longer allows paged packets to be
trimmed. Paged packets must be given to one of the pskb_trim functions
instead.
This patch adds a new pskb_trim_unique function to cover the IPv4/IPv6
datagram output path scenario and replaces the corresponding skb_trim
calls with it.
Signed-off-by: Herbert Xu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 2c31bb0..a1ce843 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1009,6 +1009,21 @@ static inline int pskb_trim(struct sk_bu
}
/**
+ * pskb_trim_unique - remove end from a paged unique (not cloned) buffer
+ * @skb: buffer to alter
+ * @len: new length
+ *
+ * This is identical to pskb_trim except that the caller knows that
+ * the skb is not cloned so we should never get an error due to out-
+ * of-memory.
+ */
+static inline void pskb_trim_unique(struct sk_buff *skb, unsigned int len)
+{
+ int err = pskb_trim(skb, len);
+ BUG_ON(err);
+}
+
+/**
* skb_orphan - orphan a buffer
* @skb: buffer to orphan
*
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index cff9c3a..d987a27 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -946,7 +946,7 @@ alloc_new_skb:
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
data += fraggap;
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
copy = datalen - transhdrlen - fraggap;
@@ -1139,7 +1139,7 @@ ssize_t ip_append_page(struct sock *sk,
data, fraggap, 0);
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
/*
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index e460489..56eddb3 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -1047,7 +1047,7 @@ alloc_new_skb:
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
data += fraggap;
- skb_trim(skb_prev, maxfraglen);
+ pskb_trim_unique(skb_prev, maxfraglen);
}
copy = datalen - transhdrlen - fraggap;
if (copy < 0) {
On Sat, Aug 26, 2006 at 11:18:39AM +1000, Herbert Xu wrote:
> On Fri, Aug 25, 2006 at 04:03:16PM -0700, Greg KH wrote:
> >
> > This patch doesn't apply at all to the latest 2.6.17-stable kernel tree.
> > Care to rediff it?
>
> Hmm, I just rebased and it actually applied as is to 2.6.17.11 :)
> Anyway, here is the result:
No, are you sure you are using the right tree?
> diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
> index cff9c3a..d987a27 100644
> --- a/net/ipv4/ip_output.c
> +++ b/net/ipv4/ip_output.c
> @@ -946,7 +946,7 @@ alloc_new_skb:
> skb_prev->csum = csum_sub(skb_prev->csum,
> skb->csum);
> data += fraggap;
> - skb_trim(skb_prev, maxfraglen);
> + pskb_trim_unique(skb_prev, maxfraglen);
> }
Oh wait, this is already in the queue, no wonder it's failing... Sorry
about this, my fault...
greg k-h