linux驱动mutex会阻塞,Linux内核崩溃mutex_lock_slowpath“阻塞超过120秒”.该怎么办？...

我有开箱即用的Debian Lenny非自定义内核2.6.26-2-amd64.

全新的服务器,用于其潜力,cpu和磁盘方面的5％.

意味着它可能不会因为过载而崩溃.

每隔几天它就会在控制台日志中冻结数百条这样的消息：

: [284847.828428] INFO: task apache2:12473 blocked for more than 120 seconds.

: [284847.868468] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

: [284847.912759] apache2 D ffff8101bc6b7ab0 0 12473 14358

: [284847.912763] ffff810160d5bc50 0000000000000082 ffff8101c0002e40 0000000000000000

: [284847.912766] ffff8101a7c42950 ffff810327d92810 ffff8101a7c42bd8 0000000400000044

: [284847.912770] ffff8101c0002e40 00000000000612d0 0000000000000000 00000040000612d0

: [284847.912773] Call Trace:

: [284847.912786] [] __mutex_lock_slowpath+0x64/0x9b

: [284847.912790] [] mutex_lock+0xa/0xb

: [284847.912794] [] do_lookup+0x82/0x1c1

: [284847.912800] [] __link_path_walk+0x87a/0xd19

: [284847.912805] [] kmem_getpages+0x96/0x15f

: [284847.912808] [] ____cache_alloc_node+0x6d/0x106

: [284847.912814] [] path_walk+0x46/0x8b

: [284847.912819] [] do_path_lookup+0x158/0x1cf

: [284847.912822] [] getname+0x140/0x1a7

: [284847.912827] [] __user_walk_fd+0x37/0x4c

: [284847.912831] [] vfs_lstat_fd+0x18/0x47

: [284847.912840] [] sys_newlstat+0x19/0x31

: [284847.912848] [] system_call_after_swapgs+0x8a/0x8f

几乎所有跟踪都将__mutex_lock_slowpath作为顶级.

只有一些有不同的痕迹：

: [284847.737386] INFO: task apache2:12472 blocked for more than 120 seconds.

: [284847.777551] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

: [284847.824881] apache2 D ffff8101bc6b7ab0 0 12472 14358

: [284847.824886] ffff8101b9cc1c50 0000000000000086 ffffffffa0131e0a 0000000000000002

: [284847.824889] ffff8102e7454300 ffff810324c6cad0 ffff8102e7454588 0000000000000000

: [284847.824893] 0000000000000001 0000000000000296 0000000000000003 ffff8101b9cc1c58

: [284847.824896] Call Trace:

: [284847.828403] [] :ext3:__ext3_journal_dirty_Metadata+0x1e/0x46

: [284847.828412] [] __mutex_lock_slowpath+0x64/0x9b

: [284847.828418] [] mutex_lock+0xa/0xb

: [284847.828421] [] do_lookup+0x82/0x1c1

: [284847.828427] [] __link_path_walk+0x87a/0xd19

: [284847.828428] [] find_lock_page+0x1f/0x8a

: [284847.828428] [] filemap_fault+0x1c2/0x33c

: [284847.828428] [] path_walk+0x46/0x8b

: [284847.828428] [] do_path_lookup+0x158/0x1cf

: [284847.828428] [] getname+0x140/0x1a7

: [284847.828428] [] __user_walk_fd+0x37/0x4c

: [284847.828428] [] vfs_lstat_fd+0x18/0x47

: [284847.828428] [] sys_newlstat+0x19/0x31

: [284847.828428] [] system_call_after_swapgs+0x8a/0x8f

kernel: [1912668.466347] INFO: task apache2:17984 blocked for more than 120 seconds.

[1912668.507035] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

: [1912668.555165] apache2 D ffff8101c5637ba0 0 17984 17282

: [1912668.596752] ffff810166a7dd30 0000000000000086 0000000000000000 ffff810166a7dcd8

: [1912668.643341] ffff8101c563c880 ffff81024505f000 0000000000000002 ffff810166a7dd68

: [1912668.699566] 0000000000000086 00000000000cb1a0 0000000000000000 ffff81017f344d60

: [1912668.744773] Call Trace:

: [1912668.761754] [] pick_next_task_fair+0x6e/0x7a

: [1912668.829311] [] bio_alloc_bioset+0x89/0xd9

: [1912668.861930] [] getnstimeofday+0x39/0x98

: [1912668.897005] [] sync_page+0x0/0x41

: [1912668.927868] [] io_schedule+0x5c/0x9e

: [1912668.960286] [] sync_page+0x3c/0x41

: [1912668.991756] [] __wait_on_bit_lock+0x36/0x66

: [1912669.031757] [] __lock_page+0x5e/0x64

: [1912669.064191] [] wake_bit_function+0x0/0x23

: [1912669.100100] [] handle_mm_fault+0x5e4/0x8de

: [1912669.134531] [] autoremove_wake_function+0x0/0x2e

: [1912669.174623] [] fcntl_setlk+0x1cf/0x291

: [1912669.210623] [] autoremove_wake_function+0x0/0x2e

: [1912669.246923] [] sys_fcntl+0x280/0x2f7

在谷歌搜索“mutex_lock_slowpath”之后,我只能找到在某些提交中引入此问题的内核邮件列表讨论.不要参考verison.

讨论最近于2011年1月25日.

我使用的内核是Debian Lenny,一年前.

我该怎么办？这个bug甚至在内核中修复了吗？如果它是如此明显的错误,为什么它很少发生？

>我应该从kernel.org下载最新的内核并进行升级吗？

>我应该使用Debian backports来安装新的“Approved”内核吗？

我错过了什么吗？该怎么办？