Skip to contentSkip to conversation

group for tests that are dangerous for verifiers?

DISCUSSION Mark Tinguely 4 participants 7 messages Jun 20, 2013 – Jun 24, 2013 XFS filesystem

lore
Share & tools
lore

Mark Tinguely

author
#0 lore raw

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

mkfs.xfs -f /dev/sda3

meta-data=/dev/sda3              isize=256    agcount=4, agsize=4194496 blks
         =                       sectsz=512   attr=2, projid32bit=0
data     =                       bsize=4096   blocks=16777984, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0
log      =internal log           bsize=4096   blocks=8192, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

db/xfs_db /dev/sda3
xfs_db> sb 0
xfs_db> p
magicnum = 0x58465342
blocksize = 4096
dblocks = 16777984
rblocks = 0
rextents = 0
uuid = fd78c924-79c8-4cc4-af42-c5a8b505ab05 logstart = 16777220
rootino = 128
rbmino = 129
rsumino = 130
rextsize = 1
agblocks = 4194496
agcount = 4
rbmblocks = 0
logblocks = 8192
versionnum = 0xb4a4
sectsize = 512
inodesize = 256
inopblock = 16
fname = "\000\000\000\000\000\000\000\000\000\000\000\000" blocklog = 12
sectlog = 9
inodelog = 8
inopblog = 4
agblklog = 23
rextslog = 0
inprogress = 0
imax_pct = 25
icount = 64
ifree = 61
fdblocks = 16769772
frextents = 0
uquotino = 0
gquotino = 0
qflags = 0
flags = 0
shared_vn = 0
inoalignmt = 2
unit = 0
width = 0
dirblklog = 0
logsectlog = 0
logsectsize = 0
logsunit = 1
features2 = 0xa
bad_features2 = 0xa

72715.142266] XFS (sda3): Mounting Filesystem

[72715.811778] XFS (sda3): Corruption detected. Unmount and run xfs_repair
[72715.818396] XFS (sda3): bad inode magic/vsn daddr 64 #8 (magic=5858)
[72715.824748] XFS: Assertion failed: 0, file: /root/xfs/fs/xfs/xfs_inode.c, line: 417

PID: 271 TASK: ffff88034fb365c0 CPU: 1 COMMAND: "kworker/1:1H"
#0 [ffff88034f407a10] machine_kexec at ffffffff8102a0c0 #1 [ffff88034f407a60] crash_kexec at ffffffff810a4943 #2 [ffff88034f407b30] oops_end at ffffffff8145dc10 #3 [ffff88034f407b60] die at ffffffff81005693 #4 [ffff88034f407b90] do_trap at ffffffff8145d583 #5 [ffff88034f407bf0] do_invalid_op at ffffffff81002de0 #6 [ffff88034f407c90] invalid_op at ffffffff81465ca8

    [exception RIP: assfail+29]
    RIP: ffffffffa04614dd  RSP: ffff88034f407d48  RFLAGS: 00010296
    RAX: 0000000000000047  RBX: ffff88034dedc800  RCX: 0000000000000dda
    RDX: 0000000000000000  RSI: 0000000000000082  RDI: ffffffff81a2dfc0
    RBP: ffff88034f407d48   R8: ffffffff81cd4848   R9: ffffffff81ce7003
    R10: 0000000000000068  R11: 000000000002d89c  R12: 0000000000000008
    R13: ffff88034f57ee40  R14: ffff88034f0ba800  R15: 0000000000000020
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018

#7 [ffff88034f407d50] xfs_inode_buf_verify at ffffffffa04aa400 [xfs] #8 [ffff88034f407db0] xfs_inode_buf_read_verify at ffffffffa04aa4b9 [xfs] #9 [ffff88034f407dc0] xfs_buf_iodone_work at ffffffffa044fd45 [xfs]
#10 [ffff88034f407df0] process_one_work at ffffffff81059c8f #11 [ffff88034f407e50] worker_thread at ffffffff8105ad3b
#12 [ffff88034f407ec0] kthread at ffffffff810612fb
#13 [ffff88034f407f50] ret_from_fork at ffffffff81464b6c

PID: 20866 TASK: ffff880351f24180 CPU: 1 COMMAND: "mount"
#0 [ffff88035175f7c8] __schedule at ffffffff8145b6fe
#1 [ffff88035175f850] schedule at ffffffff8145bb34
#2 [ffff88035175f860] schedule_timeout at ffffffff81459bd5 #3 [ffff88035175f910] wait_for_completion at ffffffff8145ac6d #4 [ffff88035175f970] xfs_buf_iowait at ffffffffa0450a09 [xfs] #5 [ffff88035175f9b0] _xfs_buf_read at ffffffffa0450bf9 [xfs] #6 [ffff88035175f9d0] xfs_buf_read_map at ffffffffa0450cf3 [xfs] #7 [ffff88035175fa20] xfs_trans_read_buf_map at ffffffffa04ca0e9 [xfs] #8 [ffff88035175fa90] xfs_imap_to_bp at ffffffffa04aa52b [xfs] #9 [ffff88035175fb10] xfs_iread at ffffffffa04b0e5d [xfs]
#10 [ffff88035175fb80] xfs_iget_cache_miss at ffffffffa04586a0 [xfs] #11 [ffff88035175fbf0] xfs_iget at ffffffffa0459589 [xfs] #12 [ffff88035175fc80] xfs_mountfs at ffffffffa04bad49 [xfs] #13 [ffff88035175fcf0] xfs_fs_fill_super at ffffffffa0464ad7 [xfs] #14 [ffff88035175fd30] mount_bdev at ffffffff81155f85
#15 [ffff88035175fdb0] xfs_fs_mount at ffffffffa04626d0 [xfs] #16 [ffff88035175fdc0] mount_fs at ffffffff81156bee
#17 [ffff88035175fe10] vfs_kern_mount at ffffffff8116fe21 #18 [ffff88035175fe60] do_new_mount at ffffffff8117136f
#19 [ffff88035175fec0] do_mount at ffffffff811729b6
#20 [ffff88035175ff20] sys_mount at ffffffff81172a8b
#21 [ffff88035175ff80] system_call_fastpath at ffffffff81464c12

    RIP: 00007fbcdfd09daa  RSP: 00007fffcb9f6f38  RFLAGS: 00010202
    RAX: 00000000000000a5  RBX: ffffffff81464c12  RCX: ffffffffffffffc8
    RDX: 00007fbce0a6ce80  RSI: 00007fbce0a6ce60  RDI: 00007fbce0a6ce40
    RBP: 00000000c0ed0400   R8: 0000000000000000   R9: 0101010101010101
    R10: ffffffffc0ed0400  R11: 0000000000000202  R12: 00007fbce0a6ce60
    R13: 00007fbce0a6cde0  R14: 0000000000000400  R15: 0000000000000000
    ORIG_RAX: 00000000000000a5  CS: 0033  SS: 002b

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#1 lore raw
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

But is the assert needed?

I have to remove the assert to avoid a recursive-fault oops on 32-bit x86. A test run is included at the end of this letter, and you might see if the behavior is OK.

There are two patches here. One is the assert removal, made after all of Dave's recent changes for 3.11. The second is a quick-and-dirty patch to fix xfs/111 for the make install version of xfstests. The latter one is not used for the sample output here.

Thanks for considering this case. A review on something other than 32-bit x86 would be quite helpful.

If nothing else, might Dave take a look at the xfs_repair output from my results here? The one-line difference is "fatal error -- couldn't iget realtime bitmap inode -- error - 117". On one hand, I was working with rtdevs recently, and the rtdevs probably weren't zeroed out. OTOH, I was able to fix it using the latest released xfsprogs from an alternate partition (kernel 3.9.5).

Thanks, and have a good weekend!

Michael

Quoted 1 line

From e51ad981dd93966a2a24f62678ab8d6163ac2111 Mon Sep 17 00:00:00 2001

From: "Michael L. Semon" <[email protected]>
Date: Fri, 21 Jun 2013 12:03:01 -0400
Subject: [PATCH] Remove assert to avoid recursive-fault oops on 32-bit x86

This patch removes an assert exclusively for the purpose of getting past xfstests xfs/111 without it causing a recursive-fault oops. The fall-through case for this removal has not been fatal but still should be reviewed by an expert.

Signed-off-by Michael L. Semon <[email protected]>
---
 fs/xfs/xfs_inode_buf.c | 1 -
 1 file changed, 1 deletion(-)
fs/xfs/xfs_inode_buf.c +0 -1
index 99c6acb..426f1a7 100644
@@ -92,7 +92,6 @@ xfs_inode_buf_verify(
 				"bad inode magic/vsn daddr %lld #%d (magic=%x)",
 				(unsigned long long)bp->b_bn, i,
 				be16_to_cpu(dip->di_magic));
-			ASSERT(0);
 #endif
 		}
 	}
Signature · 1 line
1.8.3
Quoted 1 line

From e44e5e46ae9f4f29a69485088b873a2d5d55726b Mon Sep 17 00:00:00 2001

From: "Michael L. Semon" <[email protected]>
Date: Fri, 21 Jun 2013 12:44:57 -0400
Subject: [PATCH] xfstests: give xfs/111 a file to copy and trash later

xfs/111 copies src/itrash.c to $SCRATCH_DEV so it can be trashed later. However, src/itrash.c is not installed by make install, so use instead a file that is known to be installed.

Signed-off-by Michael L. Semon <[email protected]>
---
 tests/xfs/111 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/xfs/111 +1 -1
index bedead8..c7ca489 100755
@@ -53,7 +53,7 @@ echo Create some files
 I=0
 while [ $I -lt 1000 ]
 do
-    cp src/itrash.c $SCRATCH_MNT/${I}
+    cp src/itrash $SCRATCH_MNT/${I}
     let I=$I+1
 done
 umount $SCRATCH_DEV
Signature · 293 lines
1.8.3

Script started on Fri 21 Jun 2013 12:23:26 PM EDT
root@plbearer:/var/lib/xfstests# ./check -xfs xfs/111
FSTYP         -- xfs (debug)
PLATFORM      -- Linux/i686 plbearer 3.10.0-rc6+
MKFS_OPTIONS  -- -f -llogdev=/dev/sda6 -bsize=4096 /dev/sdb6
MOUNT_OPTIONS -- -ologdev=/dev/sda6 /dev/sdb6 /mnt/xfstests-scratch

xfs/111	 - output mismatch (see /var/lib/xfstests/results/xfs/111.out.bad)
    --- tests/xfs/111.out	2013-06-21 12:19:39.085344653 -0400
    +++ /var/lib/xfstests/results/xfs/111.out.bad	2013-06-21 12:23:53.762849921 -0400
    @@ -6,6 +6,1006 @@
     log      =LDEV bsize=XXX blocks=XXX
     realtime =RDEV extsz=XXX blocks=XXX, rtextents=XXX
     Create some files
    +cp: cannot stat 'src/itrash.c': No such file or directory
    +cp: cannot stat 'src/itrash.c': No such file or directory
    +cp: cannot stat 'src/itrash.c': No such file or directory
    +cp: cannot stat 'src/itrash.c': No such file or directory
     ...
     (Run 'diff -u tests/xfs/111.out /var/lib/xfstests/results/xfs/111.out.bad' to see the entire diff)
Ran: xfs/111
Failures: xfs/111
Failed 1 of 1 tests
root@plbearer:/var/lib/xfstests# dmesg
[ 1339.824439] XFS (sdb5): Mounting Filesystem
[ 1339.903058] XFS (sdb5): Ending clean mount
[ 1341.637015] XFS (sdb6): Mounting Filesystem
[ 1341.708437] XFS (sdb6): Ending clean mount
[ 1342.211400] XFS (sdb5): Mounting Filesystem
[ 1342.276526] XFS (sdb5): Ending clean mount
[ 1344.500213] XFS (sdb6): Mounting Filesystem
[ 1344.585894] XFS (sdb6): Ending clean mount
[ 1348.470883] XFS (sdb6): Mounting Filesystem
[ 1348.539515] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.539527] XFS (sdb6): bad inode magic/vsn daddr 64 #8 (magic=5858)
[ 1348.540856] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.540863] XFS (sdb6): bad inode magic/vsn daddr 64 #9 (magic=5858)
[ 1348.542140] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.542147] XFS (sdb6): bad inode magic/vsn daddr 64 #10 (magic=5858)
[ 1348.543394] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.543400] XFS (sdb6): bad inode magic/vsn daddr 64 #11 (magic=5858)
[ 1348.544627] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.544633] XFS (sdb6): bad inode magic/vsn daddr 64 #12 (magic=5858)
[ 1348.545854] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.545861] XFS (sdb6): bad inode magic/vsn daddr 64 #13 (magic=5858)
[ 1348.547017] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.547023] XFS (sdb6): bad inode magic/vsn daddr 64 #14 (magic=5858)
[ 1348.548176] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.548183] XFS (sdb6): bad inode magic/vsn daddr 64 #15 (magic=5858)
[ 1348.549328] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.549334] XFS (sdb6): bad inode magic/vsn daddr 64 #16 (magic=5858)
[ 1348.550484] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.550491] XFS (sdb6): bad inode magic/vsn daddr 64 #17 (magic=5858)
[ 1348.551582] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.551588] XFS (sdb6): bad inode magic/vsn daddr 64 #18 (magic=5858)
[ 1348.552694] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.552700] XFS (sdb6): bad inode magic/vsn daddr 64 #19 (magic=5858)
[ 1348.553780] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.553786] XFS (sdb6): bad inode magic/vsn daddr 64 #20 (magic=5858)
[ 1348.554855] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.554861] XFS (sdb6): bad inode magic/vsn daddr 64 #21 (magic=5858)
[ 1348.555927] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.555933] XFS (sdb6): bad inode magic/vsn daddr 64 #22 (magic=5858)
[ 1348.556989] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.556995] XFS (sdb6): bad inode magic/vsn daddr 64 #23 (magic=5858)
[ 1348.558037] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.558045] XFS (sdb6): bad inode magic/vsn daddr 64 #24 (magic=5858)
[ 1348.559075] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.559081] XFS (sdb6): bad inode magic/vsn daddr 64 #25 (magic=5858)
[ 1348.560139] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.560146] XFS (sdb6): bad inode magic/vsn daddr 64 #26 (magic=5858)
[ 1348.561164] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.561170] XFS (sdb6): bad inode magic/vsn daddr 64 #27 (magic=5858)
[ 1348.562169] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.562176] XFS (sdb6): bad inode magic/vsn daddr 64 #28 (magic=5858)
[ 1348.563152] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.563168] XFS (sdb6): bad inode magic/vsn daddr 64 #29 (magic=5858)
[ 1348.564125] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.564131] XFS (sdb6): bad inode magic/vsn daddr 64 #30 (magic=5858)
[ 1348.565065] XFS (sdb6): Corruption detected. Unmount and run xfs_repair
[ 1348.565072] XFS (sdb6): bad inode magic/vsn daddr 64 #31 (magic=5858)
[ 1348.566455] XFS (sdb6): metadata I/O error: block 0x40 ("xfs_trans_read_buf_map") error 117 numblks 16
[ 1348.566472] XFS (sdb6): xfs_imap_to_bp: xfs_trans_read_buf() returned error 117.
[ 1348.566484] XFS (sdb6): failed to read root inode
[ 1349.479741] XFS (sdb5): Mounting Filesystem
[ 1349.550037] XFS (sdb5): Ending clean mount
root@plbearer:/var/lib/xfstests# xfs_repair -l $SCRATCH_LOGDEV $SCRATCH_DEV
Phase 1 - find and verify superblock...
Phase 2 - using external log on /dev/sda6
        - zero log...
        - scan filesystem freespace and inode maps...
        - found root inode chunk
Phase 3 - for each AG...
        - scan and clear agi unlinked lists...
        - process known inodes and perform inode discovery...
        - agno = 0
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
xfs_inode_buf_verify: XFS_CORRUPTION_ERROR
bad magic number 0x5858 on inode 136
bad magic number 0x5858 on inode 137
bad magic number 0x5858 on inode 138
bad magic number 0x5858 on inode 139
bad magic number 0x5858 on inode 140
bad magic number 0x5858 on inode 141
bad magic number 0x5858 on inode 142
bad magic number 0x5858 on inode 143
bad magic number 0x5858 on inode 144
bad magic number 0x5858 on inode 145
bad magic number 0x5858 on inode 146
bad magic number 0x5858 on inode 147
bad magic number 0x5858 on inode 148
bad magic number 0x5858 on inode 149
bad magic number 0x5858 on inode 150
bad magic number 0x5858 on inode 151
bad magic number 0x5858 on inode 152
bad magic number 0x5858 on inode 153
bad magic number 0x5858 on inode 154
bad magic number 0x5858 on inode 155
bad magic number 0x5858 on inode 156
bad magic number 0x5858 on inode 157
bad magic number 0x5858 on inode 158
bad magic number 0x5858 on inode 159
bad magic number 0x5858 on inode 160
bad magic number 0x5858 on inode 161
bad magic number 0x5858 on inode 162
bad magic number 0x5858 on inode 163
bad magic number 0x5858 on inode 164
bad magic number 0x5858 on inode 165
bad magic number 0x5858 on inode 166
bad magic number 0x5858 on inode 167
bad magic number 0x5858 on inode 168
bad magic number 0x5858 on inode 169
bad magic number 0x5858 on inode 170
bad magic number 0x5858 on inode 171
bad magic number 0x5858 on inode 172
bad magic number 0x5858 on inode 173
bad magic number 0x5858 on inode 174
bad magic number 0x5858 on inode 175
bad magic number 0x5858 on inode 176
bad magic number 0x5858 on inode 177
bad magic number 0x5858 on inode 178
bad magic number 0x5858 on inode 179
bad magic number 0x5858 on inode 180
bad magic number 0x5858 on inode 181
bad magic number 0x5858 on inode 182
bad magic number 0x5858 on inode 183
bad magic number 0x5858 on inode 184
bad magic number 0x5858 on inode 185
bad magic number 0x5858 on inode 186
bad magic number 0x5858 on inode 187
bad magic number 0x5858 on inode 188
bad magic number 0x5858 on inode 189
bad magic number 0x5858 on inode 190
bad magic number 0x5858 on inode 191
bad magic number 0x5858 on inode 136, resetting magic number
bad magic number 0x5858 on inode 137, resetting magic number
bad magic number 0x5858 on inode 138, resetting magic number
bad magic number 0x5858 on inode 139, resetting magic number
bad magic number 0x5858 on inode 140, resetting magic number
bad magic number 0x5858 on inode 141, resetting magic number
bad magic number 0x5858 on inode 142, resetting magic number
bad magic number 0x5858 on inode 143, resetting magic number
bad magic number 0x5858 on inode 144, resetting magic number
bad magic number 0x5858 on inode 145, resetting magic number
bad magic number 0x5858 on inode 146, resetting magic number
bad magic number 0x5858 on inode 147, resetting magic number
bad magic number 0x5858 on inode 148, resetting magic number
bad magic number 0x5858 on inode 149, resetting magic number
bad magic number 0x5858 on inode 150, resetting magic number
bad magic number 0x5858 on inode 151, resetting magic number
bad magic number 0x5858 on inode 152, resetting magic number
bad magic number 0x5858 on inode 153, resetting magic number
bad magic number 0x5858 on inode 154, resetting magic number
bad magic number 0x5858 on inode 155, resetting magic number
bad magic number 0x5858 on inode 156, resetting magic number
bad magic number 0x5858 on inode 157, resetting magic number
bad magic number 0x5858 on inode 158, resetting magic number
bad magic number 0x5858 on inode 159, resetting magic number
bad magic number 0x5858 on inode 160, resetting magic number
bad magic number 0x5858 on inode 161, resetting magic number
bad magic number 0x5858 on inode 162, resetting magic number
bad magic number 0x5858 on inode 163, resetting magic number
bad magic number 0x5858 on inode 164, resetting magic number
bad magic number 0x5858 on inode 165, resetting magic number
bad magic number 0x5858 on inode 166, resetting magic number
bad magic number 0x5858 on inode 167, resetting magic number
bad magic number 0x5858 on inode 168, resetting magic number
bad magic number 0x5858 on inode 169, resetting magic number
bad magic number 0x5858 on inode 170, resetting magic number
bad magic number 0x5858 on inode 171, resetting magic number
bad magic number 0x5858 on inode 172, resetting magic number
bad magic number 0x5858 on inode 173, resetting magic number
bad magic number 0x5858 on inode 174, resetting magic number
bad magic number 0x5858 on inode 175, resetting magic number
bad magic number 0x5858 on inode 176, resetting magic number
bad magic number 0x5858 on inode 177, resetting magic number
bad magic number 0x5858 on inode 178, resetting magic number
bad magic number 0x5858 on inode 179, resetting magic number
bad magic number 0x5858 on inode 180, resetting magic number
bad magic number 0x5858 on inode 181, resetting magic number
bad magic number 0x5858 on inode 182, resetting magic number
bad magic number 0x5858 on inode 183, resetting magic number
bad magic number 0x5858 on inode 184, resetting magic number
bad magic number 0x5858 on inode 185, resetting magic number
bad magic number 0x5858 on inode 186, resetting magic number
bad magic number 0x5858 on inode 187, resetting magic number
bad magic number 0x5858 on inode 188, resetting magic number
bad magic number 0x5858 on inode 189, resetting magic number
bad magic number 0x5858 on inode 190, resetting magic number
bad magic number 0x5858 on inode 191, resetting magic number
        - agno = 1
        - agno = 2
        - agno = 3
        - process newly discovered inodes...
Phase 4 - check for duplicate blocks...
        - setting up duplicate extent list...
        - check for inodes claiming duplicate blocks...
        - agno = 0
        - agno = 1
        - agno = 2
        - agno = 3
Phase 5 - rebuild AG headers and trees...
        - reset superblock...
Phase 6 - check inode connectivity...
        - resetting contents of realtime bitmap and summary inodes
xfs_imap_to_bp: xfs_trans_read_buf() returned error 117.
cache_node_purge: refcount was 1, not zero (node=0x937b600)

fatal error -- couldn't iget realtime bitmap inode -- error - 117
root@plbearer:/var/lib/xfstests# exit

Script done on Fri 21 Jun 2013 12:25:26 PM EDT

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#2 lore raw
In reply to Mark Tinguely
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

But it only asserts on a debug kernel...

This isn't the only place where corruption could ASSERT on debug; see xlog_recover_add_to_trans() for example.

But if the test intentionally corrupts it and that leads to an ASSERT that does seem problematic for anyone testing w/ debug enabled.

I guess I'd vote for removing the ASSERT unless there's some reason it should be there - Dave?

-Eric

Quoted 121 lines

mkfs.xfs -f /dev/sda3

meta-data=/dev/sda3              isize=256    agcount=4, agsize=4194496 blks
         =                       sectsz=512   attr=2, projid32bit=0
data     =                       bsize=4096   blocks=16777984, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0
log      =internal log           bsize=4096   blocks=8192, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

db/xfs_db /dev/sda3
xfs_db> sb 0
xfs_db> p
magicnum = 0x58465342
blocksize = 4096
dblocks = 16777984
rblocks = 0
rextents = 0
uuid = fd78c924-79c8-4cc4-af42-c5a8b505ab05 logstart = 16777220
rootino = 128
rbmino = 129
rsumino = 130
rextsize = 1
agblocks = 4194496
agcount = 4
rbmblocks = 0
logblocks = 8192
versionnum = 0xb4a4
sectsize = 512
inodesize = 256
inopblock = 16
fname = "\000\000\000\000\000\000\000\000\000\000\000\000" blocklog = 12
sectlog = 9
inodelog = 8
inopblog = 4
agblklog = 23
rextslog = 0
inprogress = 0
imax_pct = 25
icount = 64
ifree = 61
fdblocks = 16769772
frextents = 0
uquotino = 0
gquotino = 0
qflags = 0
flags = 0
shared_vn = 0
inoalignmt = 2
unit = 0
width = 0
dirblklog = 0
logsectlog = 0
logsectsize = 0
logsunit = 1
features2 = 0xa
bad_features2 = 0xa

72715.142266] XFS (sda3): Mounting Filesystem

[72715.811778] XFS (sda3): Corruption detected. Unmount and run xfs_repair
[72715.818396] XFS (sda3): bad inode magic/vsn daddr 64 #8 (magic=5858)
[72715.824748] XFS: Assertion failed: 0, file: /root/xfs/fs/xfs/xfs_inode.c, line: 417

PID: 271 TASK: ffff88034fb365c0 CPU: 1 COMMAND: "kworker/1:1H"
#0 [ffff88034f407a10] machine_kexec at ffffffff8102a0c0 #1 [ffff88034f407a60] crash_kexec at ffffffff810a4943 #2 [ffff88034f407b30] oops_end at ffffffff8145dc10 #3 [ffff88034f407b60] die at ffffffff81005693 #4 [ffff88034f407b90] do_trap at ffffffff8145d583 #5 [ffff88034f407bf0] do_invalid_op at ffffffff81002de0 #6 [ffff88034f407c90] invalid_op at ffffffff81465ca8

    [exception RIP: assfail+29]
    RIP: ffffffffa04614dd  RSP: ffff88034f407d48  RFLAGS: 00010296
    RAX: 0000000000000047  RBX: ffff88034dedc800  RCX: 0000000000000dda
    RDX: 0000000000000000  RSI: 0000000000000082  RDI: ffffffff81a2dfc0
    RBP: ffff88034f407d48   R8: ffffffff81cd4848   R9: ffffffff81ce7003
    R10: 0000000000000068  R11: 000000000002d89c  R12: 0000000000000008
    R13: ffff88034f57ee40  R14: ffff88034f0ba800  R15: 0000000000000020
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018

#7 [ffff88034f407d50] xfs_inode_buf_verify at ffffffffa04aa400 [xfs] #8 [ffff88034f407db0] xfs_inode_buf_read_verify at ffffffffa04aa4b9 [xfs] #9 [ffff88034f407dc0] xfs_buf_iodone_work at ffffffffa044fd45 [xfs]
#10 [ffff88034f407df0] process_one_work at ffffffff81059c8f #11 [ffff88034f407e50] worker_thread at ffffffff8105ad3b
#12 [ffff88034f407ec0] kthread at ffffffff810612fb
#13 [ffff88034f407f50] ret_from_fork at ffffffff81464b6c

PID: 20866 TASK: ffff880351f24180 CPU: 1 COMMAND: "mount"
#0 [ffff88035175f7c8] __schedule at ffffffff8145b6fe
#1 [ffff88035175f850] schedule at ffffffff8145bb34
#2 [ffff88035175f860] schedule_timeout at ffffffff81459bd5 #3 [ffff88035175f910] wait_for_completion at ffffffff8145ac6d #4 [ffff88035175f970] xfs_buf_iowait at ffffffffa0450a09 [xfs] #5 [ffff88035175f9b0] _xfs_buf_read at ffffffffa0450bf9 [xfs] #6 [ffff88035175f9d0] xfs_buf_read_map at ffffffffa0450cf3 [xfs] #7 [ffff88035175fa20] xfs_trans_read_buf_map at ffffffffa04ca0e9 [xfs] #8 [ffff88035175fa90] xfs_imap_to_bp at ffffffffa04aa52b [xfs] #9 [ffff88035175fb10] xfs_iread at ffffffffa04b0e5d [xfs]
#10 [ffff88035175fb80] xfs_iget_cache_miss at ffffffffa04586a0 [xfs] #11 [ffff88035175fbf0] xfs_iget at ffffffffa0459589 [xfs] #12 [ffff88035175fc80] xfs_mountfs at ffffffffa04bad49 [xfs] #13 [ffff88035175fcf0] xfs_fs_fill_super at ffffffffa0464ad7 [xfs] #14 [ffff88035175fd30] mount_bdev at ffffffff81155f85
#15 [ffff88035175fdb0] xfs_fs_mount at ffffffffa04626d0 [xfs] #16 [ffff88035175fdc0] mount_fs at ffffffff81156bee
#17 [ffff88035175fe10] vfs_kern_mount at ffffffff8116fe21 #18 [ffff88035175fe60] do_new_mount at ffffffff8117136f
#19 [ffff88035175fec0] do_mount at ffffffff811729b6
#20 [ffff88035175ff20] sys_mount at ffffffff81172a8b
#21 [ffff88035175ff80] system_call_fastpath at ffffffff81464c12

    RIP: 00007fbcdfd09daa  RSP: 00007fffcb9f6f38  RFLAGS: 00010202
    RAX: 00000000000000a5  RBX: ffffffff81464c12  RCX: ffffffffffffffc8
    RDX: 00007fbce0a6ce80  RSI: 00007fbce0a6ce60  RDI: 00007fbce0a6ce40
    RBP: 00000000c0ed0400   R8: 0000000000000000   R9: 0101010101010101
    R10: ffffffffc0ed0400  R11: 0000000000000202  R12: 00007fbce0a6ce60
    R13: 00007fbce0a6cde0  R14: 0000000000000400  R15: 0000000000000000
    ORIG_RAX: 00000000000000a5  CS: 0033  SS: 002b
Quoted 5 lines

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#3 lore raw
Quoted 7 lines · Eric Sandeen
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

But it only asserts on a debug kernel...

Right, and it has done so for years - blaming verifiers for triggering the assert failure is simply shooting the messenger.

Quoted 6 lines

This isn't the only place where corruption could ASSERT on debug; see xlog_recover_add_to_trans() for example.

But if the test intentionally corrupts it and that leads to an ASSERT that does seem problematic for anyone testing w/ debug enabled.

Yup, it runs src/itrash.c which corrupts every inode it can find.

That's the reason this test is not part of the auto group - it's a test that will cause the system to stop. We've got other tests that are not part of the auto group for exactly the same reason - they cause some kind of terminal failure and so aren't candidates for regression testing.

Quoted 2 lines

I guess I'd vote for removing the ASSERT unless there's some reason it should be there - Dave?

I'm fine with it being removed - we catch the failure just fine. If that then makes 111 work as a regression test (i.e. doesn't trigger the bad-inode bulkstat loop it was designed to test) then perhaps we can consider making that part of the auto group, too...

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#4 lore raw
Quoted 11 lines · Dave Chinner
Quoted 7 lines · Eric Sandeen
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

But it only asserts on a debug kernel...

Right, and it has done so for years - blaming verifiers for triggering the assert failure is simply shooting the messenger.

But this test *intentionally* corrupts, right? So it's prudent to not run a test which you *know* will explode if it runs as designed.

Quoted 7 lines
Quoted 6 lines

This isn't the only place where corruption could ASSERT on debug; see xlog_recover_add_to_trans() for example.

But if the test intentionally corrupts it and that leads to an ASSERT that does seem problematic for anyone testing w/ debug enabled.

Quoted 7 lines

Yup, it runs src/itrash.c which corrupts every inode it can find.

That's the reason this test is not part of the auto group - it's a test that will cause the system to stop. We've got other tests that are not part of the auto group for exactly the same reason - they cause some kind of terminal failure and so aren't candidates for regression testing.

Then maybe just part of the normal dangerous group would be enough. Except this isn't transient (today) - it's not a case where old kernels may oops, it's where it's *designed* to oops on this test, with a debug kernel.

So I guess I could see a debug-dangerous group ;)

Quoted 7 lines
Quoted 2 lines

I guess I'd vote for removing the ASSERT unless there's some reason it should be there - Dave?

I'm fine with it being removed - we catch the failure just fine. If that then makes 111 work as a regression test (i.e. doesn't trigger the bad-inode bulkstat loop it was designed to test) then perhaps we can consider making that part of the auto group, too...

Removing it sounds like the best option then.

Thanks,
-Eric

Quoted 4 lines

Cheers,

Dave.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#5 lore raw
Quoted 8 lines · Eric Sandeen
Quoted 7 lines · Dave Chinner
Quoted 6 lines · Eric Sandeen
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

But it only asserts on a debug kernel...

Right, and it has done so for years - blaming verifiers for triggering the assert failure is simply shooting the messenger.

But this test *intentionally* corrupts, right? So it's prudent to not run a test which you *know* will explode if it runs as designed.

Common sense, really.

Quoted 9 lines
Quoted 9 lines
Quoted 6 lines

This isn't the only place where corruption could ASSERT on debug; see xlog_recover_add_to_trans() for example.

But if the test intentionally corrupts it and that leads to an ASSERT that does seem problematic for anyone testing w/ debug enabled.

Yup, it runs src/itrash.c which corrupts every inode it can find.

Quoted 7 lines
Quoted 5 lines

That's the reason this test is not part of the auto group - it's a test that will cause the system to stop. We've got other tests that are not part of the auto group for exactly the same reason - they cause some kind of terminal failure and so aren't candidates for regression testing.

Then maybe just part of the normal dangerous group would be enough.

It will only run from the ioctl group today (bulkstat, I guess), so I'd say that adding it to the dangerous group doesn't add any real value except documentation. And it's just as easy to remove the ASSERT() as it is really unnecessary....

Quoted 9 lines

Except this isn't transient (today) - it's not a case where old kernels may oops, it's where it's *designed* to oops on this test, with a debug kernel.

So I guess I could see a debug-dangerous group ;)

Quoted 3 lines
Quoted 2 lines

I guess I'd vote for removing the ASSERT unless there's some reason it should be there - Dave?

Quoted 6 lines
Quoted 4 lines

I'm fine with it being removed - we catch the failure just fine. If that then makes 111 work as a regression test (i.e. doesn't trigger the bad-inode bulkstat loop it was designed to test) then perhaps we can consider making that part of the auto group, too...

Removing it sounds like the best option then.

*nod*

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
author
#6 lore raw
Quoted 11 lines · Dave Chinner
Quoted 10 lines · Eric Sandeen
Quoted 9 lines · Dave Chinner
Quoted 7 lines · Eric Sandeen
Quoted 4 lines · Mark Tinguely

Do we need a xfstest verifier dangerous group?

xfstest 111 purposely damages inodes. In hindsight it make sense that it asserts when running with verifiers.

But it only asserts on a debug kernel...

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Right, and it has done so for years - blaming verifiers for triggering the assert failure is simply shooting the messenger.

But this test *intentionally* corrupts, right? So it's prudent to not run a test which you *know* will explode if it runs as designed.

Common sense, really.

... and the reason for the open question. Should there be a group notation that says the verifiers will *correctly* find a real and known problem on this test.

Quoted 32 lines
Quoted 10 lines
Quoted 10 lines
Quoted 7 lines

This isn't the only place where corruption could ASSERT on debug; see xlog_recover_add_to_trans() for example.

But if the test intentionally corrupts it and that leads to an ASSERT that does seem problematic for anyone testing w/ debug enabled.

Yup, it runs src/itrash.c which corrupts every inode it can find.

Quoted 7 lines
Quoted 5 lines

That's the reason this test is not part of the auto group - it's a test that will cause the system to stop. We've got other tests that are not part of the auto group for exactly the same reason - they cause some kind of terminal failure and so aren't candidates for regression testing.

Then maybe just part of the normal dangerous group would be enough.

It will only run from the ioctl group today (bulkstat, I guess), so I'd say that adding it to the dangerous group doesn't add any real value except documentation. And it's just as easy to remove the ASSERT() as it is really unnecessary....

Quoted 9 lines

Except this isn't transient (today) - it's not a case where old kernels may oops, it's where it's *designed* to oops on this test, with a debug kernel.

So I guess I could see a debug-dangerous group ;)

Quoted 3 lines
Quoted 2 lines

I guess I'd vote for removing the ASSERT unless there's some reason it should be there - Dave?

Quoted 9 lines
Quoted 6 lines
Quoted 4 lines

I'm fine with it being removed - we catch the failure just fine. If that then makes 111 work as a regression test (i.e. doesn't trigger the bad-inode bulkstat loop it was designed to test) then perhaps we can consider making that part of the auto group, too...

Removing it sounds like the best option then.

*nod*

That works too.

Thanks,

--Mark.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs