FYI, we noticed the below changes on
git://git.kernel.org/pub/scm/linux/kernel/git/mel/linux-balancenuma mm-numa-protnone-v1r7
commit 5733e3f48a1f2c9ea4eb1eba38c06f65a48891a4 ("mm: Convert p[te|md]_mknonnuma and remaining page table manipulations")
+------------------------------------------+------------+------------+
| | 0f455f9110 | 5733e3f48a |
+------------------------------------------+------------+------------+
| boot_successes | 37 | 10 |
| early-boot-hang | 1 | |
| boot_failures | 0 | 4 |
| kernel_BUG_at_fs/xfs/xfs_aops.c | 0 | 4 |
| invalid_opcode | 0 | 4 |
| RIP:xfs_vm_writepage | 0 | 4 |
| Kernel_panic-not_syncing:Fatal_exception | 0 | 4 |
| backtrace:vfs_fsync_range | 0 | 4 |
| backtrace:SyS_msync | 0 | 4 |
+------------------------------------------+------------+------------+
2014-11-14 12:47:11 truncate /tmp/vm-scalability/sparse-msync-mt -s 67407118336
2014-11-14 12:47:11 ./usemem --runtime 300 -S -t 56 -f /tmp/vm-scalability/sparse-msync-mt -F --prealloc --open-rw 601849270
[ 136.721372] ------------[ cut here ]------------
[ 136.726793] kernel BUG at fs/xfs/xfs_aops.c:1069!
[ 136.733766] invalid opcode: 0000 [#1] SMP
[ 136.738504] Modules linked in: loop ipmi_watchdog snd_pcm snd_timer snd soundcore pcspkr mgag200 syscopyarea sysfillrect sysimgblt ttm drm_kms_helper sb_edac drm edac_core i2c_i801 wmi ipmi_si ipmi_msghandler acpi_power_meter
[ 136.761454] CPU: 5 PID: 7251 Comm: usemem Not tainted 3.18.0-rc4-wl-ath-g788be579 #1
[ 136.770312] Hardware name: Intel Corporation S2600WTT/S2600WTT, BIOS GRNDSDP1.TP2.0025.R02.1403131625 03/13/2014
[ 136.781911] task: ffff880852ab3ac0 ti: ffff88081db58000 task.ti: ffff88081db58000
[ 136.790512] RIP: 0010:[<ffffffff8132b77c>] [<ffffffff8132b77c>] xfs_vm_writepage+0x5fc/0x680
[ 136.800315] RSP: 0018:ffff88081db5bba8 EFLAGS: 00010246
[ 136.806515] RAX: 0020151e0000002d RBX: ffff88103c6202f8 RCX: 000000000000000c
[ 136.814768] RDX: ffff88103c6202f8 RSI: ffff88081db5be40 RDI: ffffea002062e0c0
[ 136.823032] RBP: ffff88081db5bc58 R08: 0000000000000002 R09: ffff88087ffda000
[ 136.831302] R10: 0000000000000035 R11: 0000000000000000 R12: ffff88103c6201a8
[ 136.839580] R13: 0000000000022951 R14: ffff88081db5be40 R15: ffffea002062e0c0
[ 136.847865] FS: 00007f41ec7d4700(0000) GS:ffff88085f4a0000(0000) knlGS:0000000000000000
[ 136.857235] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 136.863994] CR2: 00007f41e8fcda08 CR3: 000000083bc17000 CR4: 00000000001407e0
[ 136.872320] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 136.880650] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 136.888984] Stack:
[ 136.891595] 0000000fb1c70000 ffff88103c6202f8 ffff88081db5be40 00007f4100000001
[ 136.900285] 0000000100000001 ffffea002062e0c0 0000000000001000 0000000022952000
[ 136.908988] ffff88081db5bc38 ffffffff81329e28 ffff88081db5bc58 0000000000000000
[ 136.917701] Call Trace:
[ 136.920846] [<ffffffff81329e28>] ? xfs_vm_set_page_dirty+0x138/0x190
[ 136.928476] [<ffffffff81169e47>] __writepage+0x17/0x50
[ 136.934751] [<ffffffff8116a86c>] write_cache_pages+0x21c/0x4f0
[ 136.941809] [<ffffffff8188c57e>] ? __wait_on_bit+0x7e/0x90
[ 136.948485] [<ffffffff81169e30>] ? global_dirtyable_memory+0x50/0x50
[ 136.956144] [<ffffffff8116ab80>] generic_writepages+0x40/0x60
[ 136.963118] [<ffffffff8132acb3>] xfs_vm_writepages+0x43/0x50
[ 136.970007] [<ffffffff8116c77e>] do_writepages+0x1e/0x30
[ 136.976518] [<ffffffff81160c89>] __filemap_fdatawrite_range+0x59/0x60
[ 136.984297] [<ffffffff81160d8a>] filemap_write_and_wait_range+0x2a/0x70
[ 136.992281] [<ffffffff81335cb7>] xfs_file_fsync+0x57/0x1d0
[ 136.999015] [<ffffffff8120bdfb>] vfs_fsync_range+0x1b/0x30
[ 137.005749] [<ffffffff8119978d>] SyS_msync+0x1ed/0x250
[ 137.012104] [<ffffffff818912a9>] system_call_fastpath+0x12/0x17
[ 137.019341] Code: 8d 78 08 89 ca 48 c7 00 00 00 00 00 48 c7 44 10 f8 00 00 00 00 48 83 e7 f8 48 29 f8 01 c1 31 c0 c1 e9 03 f3 48 ab e9 44 ff ff ff <0f> 0b 31 c0 e9 5b fd ff ff 4d 85 f6 74 0e 4c 89 f7 89 45 98 e8
[ 137.042173] RIP [<ffffffff8132b77c>] xfs_vm_writepage+0x5fc/0x680
[ 137.049667] RSP <ffff88081db5bba8>
[ 137.054165] ---[ end trace f24aad15132e6627 ]---
[ 137.059929] Kernel panic - not syncing: Fatal exception
--yliu
On Wed, Nov 19, 2014 at 10:17:10AM +0800, Yuanhan Liu wrote:
> FYI, we noticed the below changes on
>
> git://git.kernel.org/pub/scm/linux/kernel/git/mel/linux-balancenuma mm-numa-protnone-v1r7
> commit 5733e3f48a1f2c9ea4eb1eba38c06f65a48891a4 ("mm: Convert p[te|md]_mknonnuma and remaining page table manipulations")
>
Thanks. This is very similar to the bug that Sasha reported so I'm
hoping the fix for that will fix this one too. FWIW, I have pushed a
branch called mm-numa-protnone-v2r5 to the same tree and I'm currently
testing it as a release candidate.
--
Mel Gorman
SUSE Labs