]> git.baikalelectronics.ru Git - kernel.git/commit
Revert "jbd2: speedup jbd2_journal_dirty_metadata()"
authorLinus Torvalds <torvalds@linux-foundation.org>
Sat, 27 Jun 2015 16:41:50 +0000 (09:41 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 27 Jun 2015 16:41:50 +0000 (09:41 -0700)
commit7d5e40de6f59fac077cfc23d98aaf778be24ef13
tree47b9548dc8d0199625a8cb0ab7e8a3b567516549
parent3100197bdb246bfdae421ca548fab3798f9db534
Revert "jbd2: speedup jbd2_journal_dirty_metadata()"

This reverts commit 450875ea90dc62828fe51d373cae7db9a7a6d8ce.

This commit seems to be the cause of the following jbd2 assertion
failure:

   ------------[ cut here ]------------
   kernel BUG at fs/jbd2/transaction.c:1325!
   invalid opcode: 0000 [#1] SMP
   Modules linked in: bnep bluetooth fuse ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 ...
   CPU: 7 PID: 5509 Comm: gcc Not tainted 4.1.0-10944-g2a298679b411 #1
   Hardware name:                  /DH87RL, BIOS RLH8710H.86A.0327.2014.0924.1645 09/24/2014
   task: ffff8803bf866040 ti: ffff880308528000 task.ti: ffff880308528000
   RIP: jbd2_journal_dirty_metadata+0x237/0x290
   Call Trace:
     __ext4_handle_dirty_metadata+0x43/0x1f0
     ext4_handle_dirty_dirent_node+0xde/0x160
     ? jbd2_journal_get_write_access+0x36/0x50
     ext4_delete_entry+0x112/0x160
     ? __ext4_journal_start_sb+0x52/0xb0
     ext4_unlink+0xfa/0x260
     vfs_unlink+0xec/0x190
     do_unlinkat+0x24a/0x270
     SyS_unlink+0x11/0x20
     entry_SYSCALL_64_fastpath+0x12/0x6a
   ---[ end trace ae033ebde8d080b4 ]---

which is not easily reproducible (I've seen it just once, and then Ted
was able to reproduce it once).  Revert it while Ted and Jan try to
figure out what is wrong.

Cc: Jan Kara <jack@suse.cz>
Acked-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/jbd2/transaction.c