Testing "discards" atop ext4 on a 2.6.28-rc8 kernel:
Dec 17 18:03:51 kernel: Bad page state in process 'kjournald2'
Dec 17 18:03:51 kernel: page:ffffe200004695a0
flags:0x0100000000000080 mapping:0000000000000000 mapcount:0 count:0
Dec 17 18:03:51 kernel: Trying to fix it up, but a reboot is needed
Dec 17 18:03:51 kernel: Backtrace:
Dec 17 18:03:51 kernel: Pid: 19919, comm: kjournald2 Tainted: P
W 2.6.28-rc8 #2
Dec 17 18:03:51 kernel: Call Trace:
Dec 17 18:03:51 kernel: [<ffffffff80274349>] bad_page+0x6f/0x9d
Dec 17 18:03:51 kernel: [<ffffffff80274f19>] free_hot_cold_page+0xc8/0x1a0
Dec 17 18:03:51 kernel: [<ffffffff80275046>] free_hot_page+0xb/0xd
Dec 17 18:03:51 kernel: [<ffffffff8027532e>] __free_pages+0x18/0x21
Dec 17 18:03:51 kernel: [<ffffffff80275369>] free_pages+0x32/0x37
Dec 17 18:03:51 kernel: [<ffffffffa065577d>]
jbd2_journal_commit_transaction+0xb63/0xf06 [jbd2]
Dec 17 18:03:51 kernel: [<ffffffff804ae925>] ? thread_return+0x3d/0xa2
Was iteratively running:
iozone -s 1g -i 0 -+u -I -t 100 -x -r 1024
... the error occurred during the initial write phase (writing 100
files of 1GB each), where the files were all at ~215MB.
Chris