Reran xfstests today on rc4 on nfsv4.1 (to unmodified Fedora 20 server) mount - still get hung testcase and dmesg entries logged repeatedly [ 1081.138806] INFO: task aio-dio-fcntl-r:14913 blocked for more than 120 seconds. [ 1081.138808] Not tainted 3.18.0-031800rc3-generic #201411022335 [ 1081.138809] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 1081.138810] aio-dio-fcntl-r D 0000000000000000 0 14913 14911 0x00000000 [ 1081.138812] ffff88005a5fbac8 0000000000000082 0000000000000000 ffff88015f5cb200 [ 1081.138814] ffff88005a5fbfd8 0000000000014640 ffff88014ee39180 0000000000014640 [ 1081.138817] ffff880174ed8a00 ffff88015f5cb200 ffff88005a5fbae8 ffff88006225ab10 [ 1081.138819] Call Trace: [ 1081.138822] [<ffffffff817a4dd9>] schedule+0x29/0x70 [ 1081.138825] [<ffffffff817a50fe>] schedule_preempt_disabled+0xe/0x10 [ 1081.138829] [<ffffffff817a6db5>] __mutex_lock_slowpath+0x95/0x100 [ 1081.138832] [<ffffffff817a6e43>] mutex_lock+0x23/0x37 [ 1081.138843] [<ffffffffc01a47e6>] nfs_file_direct_write+0xd6/0x360 [nfs] [ 1081.138852] [<ffffffffc01a4a8e>] nfs_direct_IO+0x1e/0x30 [nfs] [ 1081.138855] [<ffffffff811704bb>] generic_file_direct_write+0xbb/0x180 [ 1081.138858] [<ffffffff81170827>] __generic_file_write_iter+0x2a7/0x350 [ 1081.138861] [<ffffffff8117091b>] generic_file_write_iter+0x4b/0xc0 [ 1081.138868] [<ffffffffc019aabb>] nfs_file_write+0xab/0x1c0 [nfs] [ 1081.138877] [<ffffffffc019aa10>] ? nfs_revalidate_file_size.isra.20+0xa0/0xa0 [nfs] [ 1081.138879] [<ffffffff8122f2d6>] aio_run_iocb+0x226/0x3e0 [ 1081.138883] [<ffffffff817a6e36>] ? mutex_lock+0x16/0x37 [ 1081.138885] [<ffffffff8122d971>] ? aio_read_events_ring+0x1e1/0x250 [ 1081.138888] [<ffffffff8122f583>] io_submit_one+0xf3/0x320 [ 1081.138890] [<ffffffff8122fd14>] do_io_submit+0x104/0x1d0 [ 1081.138893] [<ffffffff8122fdf0>] SyS_io_submit+0x10/0x20 [ 1081.138895] [<ffffffff817a8fed>] system_call_fastpath+0x16/0x1b On Thu, Nov 13, 2014 at 10:16 PM, Steve French <smfrench@xxxxxxxxx> wrote: > I was experimenting to see how many new tests the updated xfstests > enabled and when trying it over nfs ran into a strange problem with > 3.18-rc3 on nfs v3 mount > > after the usual generic/035 (nlink count is 1 when should be zero) > test failure with v3 > xfstest generic/036 hung with the following > > [22918.651195] INFO: task aio-dio-fcntl-r:63057 blocked for more than > 120 seconds. > [22918.651197] Not tainted 3.18.0-031800rc3-generic #201411022335 > [22918.651198] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" > disables this message. > [22918.651199] aio-dio-fcntl-r D 0000000000000000 0 63057 63056 0x00000000 > [22918.651202] ffff880111c8bd08 0000000000000046 ffff88009d099648 > ffffea0002742600 > [22918.651205] ffff880111c8bfd8 0000000000014640 ffff88009165bf00 > 0000000000014640 > [22918.651207] ffffffff81c1d4e0 ffff8800ba110a00 ffff880111c8bcf8 > ffff8800808f7c50 > [22918.651209] Call Trace: > [22918.651213] [<ffffffff817a4dd9>] schedule+0x29/0x70 > [22918.651216] [<ffffffff817a50fe>] schedule_preempt_disabled+0xe/0x10 > [22918.651219] [<ffffffff817a6db5>] __mutex_lock_slowpath+0x95/0x100 > [22918.651222] [<ffffffff817a6e43>] mutex_lock+0x23/0x37 > [22918.651239] [<ffffffffc08b086b>] nfs4_file_fsync+0x6b/0xa0 [nfsv4] > [22918.651243] [<ffffffff81216089>] vfs_fsync+0x29/0x40 > [22918.651250] [<ffffffffc07e8cea>] nfs_file_flush+0x8a/0xd0 [nfs] > [22918.651255] [<ffffffff811e378a>] filp_close+0x3a/0x90 > [22918.651261] [<ffffffff8120303a>] put_files_struct.part.12+0x7a/0xd0 > [22918.651264] [<ffffffff812033e5>] put_files_struct+0x15/0x20 > [22918.651266] [<ffffffff812034b2>] exit_files+0x52/0x60 > [22918.651269] [<ffffffff81071811>] do_exit+0x171/0x470 > [22918.651272] [<ffffffff81071ba4>] do_group_exit+0x44/0xa0 > [22918.651274] [<ffffffff81071c17>] SyS_exit_group+0x17/0x20 > [22918.651276] [<ffffffff817a8fed>] system_call_fastpath+0x16/0x1b > > Is this something recently fixed (or regressed)? I don't see much in > the changelog in the last two weeks in fs/nfs that looks plausibly > related. > > Tomorrow I will update to current (rc4) mainline and see if it goes > away - but curious if others had seen this ... (e.g. with test 036) > > -- > Thanks, > > Steve -- Thanks, Steve -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
