Summary: | Please mark =sys-kernel/gentoo-sources-2.6.34-r1 and =sys-kernel/vanilla-sources-2.6.34 stable | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Mike Pagano <mpagano> |
Component: | New packages | Assignee: | Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | axiator, denilsonsa, lpmichalik, nikoli |
Priority: | High | Keywords: | STABLEREQ |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | 327299 | ||
Bug Blocks: | 327777 |
Description
Mike Pagano
2010-06-24 13:05:50 UTC
I've seen a major issue with NFS and 2.6.34 which has caused me to revert back to 2.6.33. After normal usage with NFS-mounted home directories, processes hang as do any attempts to run commands like lsof or sync. It becomes impossible to shutdown cleanly. Jun 23 22:57:43 rivendell kernel: INFO: task flush-0:22:5393 blocked for more than 120 seconds. Jun 23 22:57:43 rivendell kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. Jun 23 22:57:43 rivendell kernel: flush-0:22 D 00000001137d54b4 0 5393 2 0x00000080 Jun 23 22:57:43 rivendell kernel: ffff88026f6e5c60 0000000000000046 ffff880264672370 ffff88026f6e5c10 Jun 23 22:57:43 rivendell kernel: ffff880200000000 0000000000004000 0000000000013a80 ffff88026f6e5fd8 Jun 23 22:57:43 rivendell kernel: ffff88026f6e5fd8 ffff880270dbf700 0000000000013a80 ffff88026f6e5fd8 Jun 23 22:57:43 rivendell kernel: Call Trace: Jun 23 22:57:43 rivendell kernel: [<ffffffff81217875>] ? nfs_write_inode+0x55/0x260 Jun 23 22:57:43 rivendell kernel: [<ffffffff81113b50>] ? inode_wait+0x0/0x10 Jun 23 22:57:43 rivendell kernel: [<ffffffff81113b59>] inode_wait+0x9/0x10 Jun 23 22:57:43 rivendell kernel: [<ffffffff81597057>] __wait_on_bit+0x57/0x80 Jun 23 22:57:43 rivendell kernel: [<ffffffff8111e876>] inode_wait_for_writeback+0x86/0xb0 Jun 23 22:57:43 rivendell kernel: [<ffffffff8105ea90>] ? wake_bit_function+0x0/0x40 Jun 23 22:57:43 rivendell kernel: [<ffffffff8111f8ab>] wb_writeback+0x17b/0x1e0 Jun 23 22:57:43 rivendell kernel: [<ffffffff8111fcbc>] wb_do_writeback+0x1bc/0x1c0 Jun 23 22:57:43 rivendell kernel: [<ffffffff8111fd0b>] bdi_writeback_task+0x4b/0xd0 Jun 23 22:57:43 rivendell kernel: [<ffffffff810d9800>] ? bdi_start_fn+0x0/0xe0 Jun 23 22:57:43 rivendell kernel: [<ffffffff810d986c>] bdi_start_fn+0x6c/0xe0 Jun 23 22:57:43 rivendell kernel: [<ffffffff810d9800>] ? bdi_start_fn+0x0/0xe0 Jun 23 22:57:43 rivendell kernel: [<ffffffff8105e54e>] kthread+0x8e/0xa0 Jun 23 22:57:43 rivendell kernel: [<ffffffff81003b54>] kernel_thread_helper+0x4/0x10 Jun 23 22:57:43 rivendell kernel: [<ffffffff8105e4c0>] ? kthread+0x0/0xa0 Jun 23 22:57:43 rivendell kernel: [<ffffffff81003b50>] ? kernel_thread_helper+0x0/0x10 Jun 23 22:57:43 rivendell kernel: INFO: task bash:5526 blocked for more than 120 seconds. Jun 23 22:57:43 rivendell kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. Jun 23 22:57:43 rivendell kernel: bash D 00000001137d3a35 0 5526 1 0x00000080 Jun 23 22:57:43 rivendell kernel: ffff88026ebc9b88 0000000000000082 ffff88026ebc9ae8 ffff88026ebc9b38 Jun 23 22:57:43 rivendell kernel: ffff880200000000 0000000000004000 0000000000013a80 ffff88026ebc9fd8 Jun 23 22:57:43 rivendell kernel: ffff88026ebc9fd8 ffff8802648c2200 0000000000013a80 ffff88026ebc9fd8 Jun 23 22:57:43 rivendell kernel: Call Trace: Jun 23 22:57:43 rivendell kernel: [<ffffffff810c2ac0>] ? sync_page+0x0/0x50 Jun 23 22:57:43 rivendell kernel: [<ffffffff81596802>] io_schedule+0x42/0x60 Jun 23 22:57:43 rivendell kernel: [<ffffffff810c2af8>] sync_page+0x38/0x50 Jun 23 22:57:43 rivendell kernel: [<ffffffff81596f12>] __wait_on_bit_lock+0x52/0xb0 Jun 23 22:57:43 rivendell kernel: [<ffffffff810c2aa2>] __lock_page+0x62/0x70 Jun 23 22:57:43 rivendell kernel: [<ffffffff8105ea90>] ? wake_bit_function+0x0/0x40 Jun 23 22:57:43 rivendell kernel: [<ffffffff810c2bd2>] ? unlock_page+0x22/0x30 Jun 23 22:57:43 rivendell kernel: [<ffffffff810cc8c3>] invalidate_inode_pages2_range+0x333/0x360 Jun 23 22:57:43 rivendell kernel: [<ffffffff810dbf09>] ? do_wp_page+0x4b9/0x830 Jun 23 22:57:43 rivendell kernel: [<ffffffff810cc902>] invalidate_inode_pages2+0x12/0x20 Jun 23 22:57:43 rivendell kernel: [<ffffffff8120a6b8>] nfs_revalidate_mapping+0xc8/0x120 Jun 23 22:57:43 rivendell kernel: [<ffffffff8120844e>] nfs_file_read+0x6e/0x100 Jun 23 22:57:43 rivendell kernel: [<ffffffff810fe762>] do_sync_read+0xd2/0x110 Jun 23 22:57:43 rivendell kernel: [<ffffffff8159b824>] ? do_page_fault+0x184/0x420 Jun 23 22:57:43 rivendell kernel: [<ffffffff8126cf11>] ? security_file_permission+0x11/0x20 Jun 23 22:57:43 rivendell kernel: [<ffffffff810feeb3>] vfs_read+0xb3/0x170 Jun 23 22:57:43 rivendell kernel: [<ffffffff810963b1>] ? audit_syscall_entry+0x241/0x270 Jun 23 22:57:43 rivendell kernel: [<ffffffff810fefbc>] sys_read+0x4c/0x80 Jun 23 22:57:43 rivendell kernel: [<ffffffff81002dab>] system_call_fastpath+0x16/0x1b I think the fix may be http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=c5efa5fc91f1f6d1d47e65f39e7ec6d1157c777d based on http://kerneltrap.com/mailarchive/linux-kernel/2010/5/22/4573766/thread but have yet to confirm this. I see this on both x86_64 and ppc with 2.6.34, but not 2.6.33. (In reply to comment #1) > I've seen a major issue with NFS and 2.6.34 which has caused me to revert back > to 2.6.33. Thanks, Andrew. Let's open up a new bug for this and please confirm if that patch fixes the issue. I will include this patch in gentoo-sources upon your confirmation. Please hold up, arch teams. feel free to re-CC us as soon as a new version is good to go then Will re-add archs when ready. Sorry for the bugspam I don't see this on HPPA, but 2.6.34 happens to be a kernel that actually works for HPPA, whereas 2.6.33 does not. Stable for HPPA. ok, it's been nine days without a confirmation on the patch, so let's continue to stabilize. I will address any new bugs with a subsequent release. Arch's, please continue to stabilize. x86 stable amd64 done Sorry for not getting back sooner but this issue isn't the easiest to replicate. However, I have seen the issue with 2.6.34 even with the patch and have had to revert back to 2.6.33. 2.6.34 ends up in a state where you can't shutdown and safely umount filesystems. I don't think that's safe. If you set up a machine/VM with an NFS home directory, you should see this issue through normal use. alpha/arm/ia64/sh/sparc stable Marked ppc/ppc64 stable. Closing since we were the last arches. |