lkml.org 
[lkml]   [2022]   [Nov]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v2] xfs: extend the freelist before available space check
Hi,

On Thu, Nov 03, 2022 at 09:10:25PM +0800, Gao Xiang wrote:
> There is a long standing issue which could cause fs shutdown due to
> inode extent to btree conversion failure right after an extent
> allocation in the same AG, which is absolutely unexpected due to the
> proper minleft reservation in the previous allocation. Brian once
> addressed one of the root cause [1], however, such symptom can still
> occur after the commit is merged as reported [2], and our cloud
> environment is also suffering from this issue.
>
> From the description of the commit [1], I found that Zirong has an
> in-house stress test reproducer for this issue, therefore I asked him
> to reproduce again and he confirmed that such issue can still be
> reproducable on RHEL 9.
>
> Thanks to him, after dumping the transaction log items, I think
> the root cause is as below:
> 1. Allocate space with the following condition:
> freeblks: 18304 pagf_flcount: 6
> reservation: 18276 need (min_free): 6
> args->minleft: 1
> available = freeblks + agflcount - reservation - need - minleft
> = 18304 + min(6, 6) - 18276 - 6 - 1 = 27
>
> The first allocation check itself is ok;
>
> 2. At that time, the AG state is
> AGF Buffer: (XAGF)
> ver:1 seq#:3 len:2621424
> root BNO:9 CNT:7
> level BNO:2 CNT:2
> 1st:64 last:69 cnt:6 freeblks:18277 longest:6395
>
> agfl (flfirst = 64, fllast = 69, flcount = 6)
> 64:547 65:167 66:1651 67:2040807 68:783 69:604
>
> 3. Then, cntbt needs a new btree block (so take one block
> from agfl), and then the log records a new AGF:
> blkno 62914177, len 1, map_size 1
> 00000000: 58 41 47 46 00 00 00 01 00 00 00 03 00 27 ff f0 XAGF.........'..
> 00000010: 00 00 00 09 00 00 00 07 00 00 00 00 00 00 00 02 ................
> 00000020: 00 00 00 02 00 00 00 00 00 00 00 41 00 00 00 45 ...........A...E
> 00000030: 00 00 00 05 00 00 47 65 00 00 18 fb 00 00 00 09 ......Ge........
> 00000040: 75 dc c1 b5 1a 45 40 2a 80 50 72 f0 59 6e 62 66 u....E@*.Pr.Ynbf
> agf 3 flfirst: 65 (0x41) fllast: 69 (0x45) cnt: 5
> freeblks 18277
>
> 4. agfl 64 (daddr 62918552) was then written as a cntbt block
> log item:
> type#011= 0x123c
> flags#011= 0x8
> blkno 62918552, len 8, map_size 1
> 00000000: 41 42 33 43 00 00 00 fd 00 1f 23 e4 ff ff ff ff AB3C......#.....
> 00000010: 00 00 00 00 03 c0 0f 98 00 00 00 00 00 00 00 00 ................
> 00000020: 75 dc c1 b5 1a 45 40 2a 80 50 72 f0 59 6e 62 66 u....E@*.Pr.Ynbf
>
> 5. Finally, the following inode extent to btree allocation fails:
> Nov 1 07:56:09 dell-per750-08 kernel: ------------[ cut here ]------------
> Nov 1 07:56:09 dell-per750-08 kernel: WARNING: CPU: 15 PID: 49290 at fs/xfs/libxfs/xfs_bmap.c:717 xfs_bmap_extents_to_btree+0xc51/0x1050 [xfs]
> ...
> Nov 1 07:56:10 dell-per750-08 kernel: XFS (sda2): agno 3 agflcount 5 freeblks 18277 reservation 18276 6
>
> since
>
> available = freeblks + agflcount - reservation - need - minleft
> = 18277 + min(5, 6) - 18276 - 6 - 0 = 0 < 1
> kaboom.
>

Perhaps it's still not a correct fix since the second conversion
allocation will fail as:

available = freeblks + agflcount - reservation - need - minleft
= 18276 + min(6, 6) - 18276 - 6 - 0 = 0 < 1

If we don't want to use the last blocks of the AG, we should shorten
args->maxlen to avoid touch these agfl blocks, thoughts?

2300 static bool
2301 xfs_alloc_space_available(
2302 struct xfs_alloc_arg *args,
2303 xfs_extlen_t min_free,
2304 int flags)
2305 {

...

2329 available = (int)(pag->pagf_freeblks + agflcount -
2330 reservation - min_free - args->minleft);

^ here available = 27

2331 if (available < (int)max(args->total, alloc_len))
2332 return false;
2333
2334 /*
2335 * Clamp maxlen to the amount of free space available for the actual
2336 * extent allocation.
2337 */
2338 if (available < (int)args->maxlen && !(flags & XFS_ALLOC_FLAG_CHECK)) {
2339 args->maxlen = available;

^ so args->maxlen = 27 here

and then freeblks from 18304 - 27 = 18277, but with another agfl block
allocated (pagf_flcount from 6 to 5), the inequality will not satisfy:

available = freeblks + agflcount - reservation - need - minleft
= 18277 + min(5, 6) - 18276 - 6 - 0 = 0 < 1

I think one of the correct fix is to fix args->maxlen above though, or
some better preferred idea to fix this?

Thanks,
Gao Xiang

\
 
 \ /
  Last update: 2022-11-03 14:58    [W:0.069 / U:0.440 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site