From Bugzilla Helper: User-Agent: Mozilla/5.0 Galeon/1.2.11 (X11; Linux i686; U;) Gecko/20030708 Description of problem: [root@lvr1 root]# Assertion failure in do_get_write_access() at transaction.c:612: "!(((jh2bh(jh))->b_state & (1UL << BH_Lock)) != 0)" ------------[ cut here ]------------ kernel BUG at transaction.c:612! invalid operand: 0000 Kernel 2.4.9-e.25enterprise CPU: 3 EIP: 0010:[<f885da8e>] Not tainted EFLAGS: 00010282 EIP is at do_get_write_access [jbd] 0x20e eax: 00000024 ebx: eb10c3e0 ecx: c02f7844 edx: 00006b3c esi: eb10c3e0 edi: cbbac490 ebp: eeb27400 esp: f013f9f8 ds: 0018 es: 0018 ss: 0018 Process sendmail (pid: 25873, stackpage=f013f000) Stack: f8866f10 00000264 00000000 00000000 00000000 f57508e0 eeb27494 eeb27400 f6b6f1e0 cbbac490 f885df47 f6b6f1e0 cbbac490 00000000 00000000 f6b6f1e0 f013fabc ea03eac0 f88727a1 f6b6f1e0 eb10c3e0 f885e5f7 e4cff370 cae24cc0 Call Trace: [<f8866f10>] .LC7 [jbd] 0x0 [<f885df47>] journal_get_write_access_Rsmp_a7d05437 [jbd] 0x37 [<f88727a1>] ext3_reserve_inode_write [ext3] 0x31 [<f885e5f7>] journal_dirty_metadata_Rsmp_08cf9292 [jbd] 0x1a7 [<c01debbe>] dev_queue_xmit [kernel] 0x14e [<f8872838>] ext3_mark_inode_dirty [ext3] 0x18 [<f88728f8>] ext3_dirty_inode [ext3] 0x98 [<f885f270>] __journal_file_buffer [jbd] 0xe0 [<c015c1ba>] __mark_inode_dirty [kernel] 0x2a [<f886c977>] ext3_free_blocks [ext3] 0x657 [<c01e6d74>] qdisc_restart [kernel] 0x14 [<f886fd0b>] ext3_get_block_handle [ext3] 0xdb [<c02021c5>] tcp_transmit_skb [kernel] 0x565 [<c01fdd97>] tcp_clean_rtx_queue [kernel] 0x247 [<c01f2e00>] ip_queue_xmit [kernel] 0x2b0 [<f887006a>] ext3_getblk [ext3] 0xda [<f8807e99>] __scsi_end_request [scsi_mod] 0x1b9 [<f886fd0b>] ext3_get_block_handle [ext3] 0xdb [<f885deed>] do_get_write_access [jbd] 0x66d [<c0113b5f>] smp_call_function_interrupt [kernel] 0x2f [<c024763a>] call_call_function_interrupt [kernel] 0x5 [<f887148c>] ext3_clear_blocks [ext3] 0x13c [<f885deed>] do_get_write_access [jbd] 0x66d [<f885e5f7>] journal_dirty_metadata_Rsmp_08cf9292 [jbd] 0x1a7 [<c01199d2>] __wake_up [kernel] 0x42 [<f88715b8>] ext3_free_data [ext3] 0x118 [<c021df75>] unix_dgram_sendmsg [kernel] 0x3a5 [<f88727a1>] ext3_reserve_inode_write [ext3] 0x31 [<c01492e7>] refile_buffer [kernel] 0x17 [<c0113b5f>] smp_call_function_interrupt [kernel] 0x2f [<c024763a>] call_call_function_interrupt [kernel] 0x5 [<f88719aa>] ext3_truncate [ext3] 0x13a [<f885d1c6>] start_this_handle [jbd] 0x66 [<f885d285>] start_this_handle [jbd] 0x125 [<f885d38f>] journal_start_Rsmp_ec53be73 [jbd] 0xbf [<f886f1e8>] start_transaction [ext3] 0x58 [<f886f387>] ext3_delete_inode [ext3] 0xe7 [<f887dd20>] ext3_sops [ext3] 0x0 [<f887dd20>] ext3_sops [ext3] 0x0 [<c015dccc>] iput_free [kernel] 0x14c [<f8862a8a>] log_wait_commit_Rsmp_7dd6c3ad [jbd] 0x5a [<c015ad8b>] dentry_iput [kernel] 0x4b [<c015ae76>] dput [kernel] 0xc6 [<c015a641>] locks_remove_posix [kernel] 0x141 [<f886dc64>] ext3_release_file [ext3] 0x14 [<c0147b48>] __fput [kernel] 0x68 [<c014671e>] filp_close [kernel] 0x9e [<c014678b>] sys_close [kernel] 0x5b [<c01073c3>] system_call [kernel] 0x33 Code: 0f 0b 58 5a c7 44 24 08 e2 ff ff ff 8b 54 24 24 b8 01 00 00 <0>Kernel panic: not continuing Version-Release number of selected component (if applicable): e25 How reproducible: Didn't try Steps to Reproduce: 1.e25 2.e25 3. Additional info:
There is a patch which will probably fix this, but until recently we didn't know _why_ it would help --- the bug that the patch was designed to fix didn't exist until 2.4.10+ kernels. But we have recently identified a real problem in 2.4.9 which the patch will fix, and we have started testing a back-port of that patch to the AS-2.1 tree.
An unsupported engineering kernel containing this fix is now available for testing and evaluation at http://people.redhat.com/~jbaron/.private/testing/2.4.9-e.27.18.test/
Created attachment 95317 [details] part 1/2: fix ext3 locking in do_get_write_access
Created attachment 95318 [details] part 2/2: fix buffer refile locking problem exposed by part1 fix
Any plans on getting the transaction.c bugs fix in QU3?
Yes, it is in the development tree for QU3.
This issue has been addressed in QU3. Please reopen this ticket if you find evidence that the issue persists in the QU3 release. If there are other outstanding issues in this ticket, please open a new BZ ticket to address them.
The same error in e27
The fix was not in e27. e27.17 was the first side-branch build with it included; e28 was the first mainline kernel with it.