I was playing with btrfs with 2 files of 3.5 GB (using loop), I completely zeroed one of the files. As expected, I had checksum failures, and I run btrfs-vol -b just to see what happened, and I got this (using -rc3): [25765.340492] btrfs csum failed ino 260 off 122880 csum 2566472073 private 3262812575 [25765.340530] btrfs csum failed ino 260 off 118784 csum 2566472073 private 1515197691 [25765.340560] btrfs csum failed ino 260 off 114688 csum 2566472073 private 1467250717 [25765.340589] btrfs csum failed ino 260 off 110592 csum 2566472073 private 551456760 [25765.340618] btrfs csum failed ino 260 off 106496 csum 2566472073 private 2751401677 [25765.340647] btrfs csum failed ino 260 off 102400 csum 2566472073 private 2648834377 [25765.340675] btrfs csum failed ino 260 off 98304 csum 2566472073 private 467965630 [25765.340703] btrfs csum failed ino 260 off 94208 csum 2566472073 private 3164409631 [25765.340732] btrfs csum failed ino 260 off 90112 csum 2566472073 private 2042500030 [25765.340760] btrfs csum failed ino 260 off 86016 csum 2566472073 private 830495785 [25809.292651] btrfs: relocating block group 3541041152 flags 9 [25809.416493] __ratelimit: 158 callbacks suppressed [25809.416498] btrfs csum failed ino 257 off 61440 csum 2566472073 private 1989812792 [25809.416537] btrfs csum failed ino 257 off 57344 csum 2566472073 private 3105762401 [25809.416603] btrfs csum failed ino 257 off 53248 csum 2566472073 private 2215503813 [25809.416719] btrfs csum failed ino 257 off 323584 csum 2566472073 private 883932373 [25809.416731] btrfs csum failed ino 257 off 49152 csum 2566472073 private 1747807236 [25809.416752] btrfs csum failed ino 257 off 319488 csum 2566472073 private 2519068844 [25809.416774] btrfs csum failed ino 257 off 45056 csum 2566472073 private 3487590687 [25809.416783] btrfs csum failed ino 257 off 315392 csum 2566472073 private 2599709027 [25809.416813] btrfs csum failed ino 257 off 311296 csum 2566472073 private 157132003 [25809.416823] btrfs csum failed ino 257 off 40960 csum 2566472073 private 599615997 [25809.657271] ------------[ cut here ]------------ [25809.657294] WARNING: at fs/btrfs/relocation.c:3588 btrfs_relocate_block_group+0x27f/0x390 [btrfs]() [25809.657298] Hardware name: System Product Name [25809.657300] Modules linked in: btrfs loop crc32c libcrc32c ppp_deflate zlib_deflate bsd_comp ppp_async crc_ccitt ppp_generic slhc kvm_intel kvm i915 snd_ca0106 snd_rawmidi drm_kms_helper cfbcopyarea snd_ac97_codec ac97_bus video snd_pcm_oss snd_mixer_oss snd_pcm backlight snd_timer cfbimgblt snd processor button cfbfillrect intel_agp wacom soundcore snd_page_alloc usbhid fan fuse ehci_hcd uhci_hcd thermal thermal_sys usbcore atl1 mii [last unloaded: btrfs] [25809.657350] Pid: 15354, comm: btrfs-vol Not tainted 2.6.32-rc3 #1 [25809.657353] Call Trace: [25809.657361] [<ffffffff8104f9bb>] warn_slowpath_common+0x7b/0xc0 [25809.657366] [<ffffffff8104fa14>] warn_slowpath_null+0x14/0x20 [25809.657377] [<ffffffffa0396b7f>] btrfs_relocate_block_group+0x27f/0x390 [btrfs] [25809.657390] [<ffffffffa037cfab>] btrfs_relocate_chunk+0x8b/0x5e0 [btrfs] [25809.657404] [<ffffffffa0372a0b>] ? map_extent_buffer+0xdb/0xe0 [btrfs] [25809.657417] [<ffffffffa0371b13>] ? unmap_extent_buffer+0x13/0x40 [btrfs] [25809.657431] [<ffffffffa0366c54>] ? btrfs_item_offset+0xe4/0xf0 [btrfs] [25809.657444] [<ffffffffa037db3f>] btrfs_balance+0x1ef/0x270 [btrfs] [25809.657449] [<ffffffff810cc0ef>] ? find_get_page+0x7f/0xf0 [25809.657461] [<ffffffffa0382e10>] btrfs_ioctl+0x510/0xa20 [btrfs] [25809.657465] [<ffffffff810cc3e7>] ? unlock_page+0x27/0x30 [25809.657470] [<ffffffff810e8409>] ? __do_fault+0x449/0x540 [25809.657476] [<ffffffff8111bf62>] vfs_ioctl+0x22/0xa0 [25809.657481] [<ffffffff813cf4e9>] ? sub_preempt_count+0x59/0x60 [25809.657486] [<ffffffff8111c108>] do_vfs_ioctl+0x88/0x570 [25809.657491] [<ffffffff8120ecc9>] ? __up_read+0x99/0xc0 [25809.657497] [<ffffffff8107323e>] ? up_read+0xe/0x10 [25809.657501] [<ffffffff8111c63f>] sys_ioctl+0x4f/0x80 [25809.657507] [<ffffffff8100b5db>] system_call_fastpath+0x16/0x1b [25809.657511] ---[ end trace b3050419e02a5436 ]--- [25809.662186] ------------[ cut here ]------------ [25809.662190] kernel BUG at fs/btrfs/volumes.c:1746! [25809.662193] invalid opcode: 0000 [#1] PREEMPT SMP [25809.662198] last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host3/target3:0:0/3:0:0:0/model [25809.662201] CPU 0 [25809.662203] Modules linked in: btrfs loop crc32c libcrc32c ppp_deflate zlib_deflate bsd_comp ppp_async crc_ccitt ppp_generic slhc kvm_intel kvm i915 snd_ca0106 snd_rawmidi drm_kms_helper cfbcopyarea snd_ac97_codec ac97_bus video snd_pcm_oss snd_mixer_oss snd_pcm backlight snd_timer cfbimgblt snd processor button cfbfillrect intel_agp wacom soundcore snd_page_alloc usbhid fan fuse ehci_hcd uhci_hcd thermal thermal_sys usbcore atl1 mii [last unloaded: btrfs] [25809.662245] Pid: 15354, comm: btrfs-vol Tainted: G W 2.6.32-rc3 #1 System Product Name [25809.662248] RIP: 0010:[<ffffffffa037d4e4>] [<ffffffffa037d4e4>] btrfs_relocate_chunk+0x5c4/0x5e0 [btrfs] [25809.662261] RSP: 0018:ffff880073afdc18 EFLAGS: 00010282 [25809.662263] RAX: 00000000fffffffb RBX: ffff880077210000 RCX: ffffffffa0396ae7 [25809.662266] RDX: 0000000000000007 RSI: ffffffff8110670f RDI: ffffffff811044d5 [25809.662269] RBP: ffff880073afdd08 R08: 0000000000000000 R09: ffff880073afd938 [25809.662271] R10: 0000000000000000 R11: 0000000000000000 R12: ffff880077153000 [25809.662274] R13: 0000000000000000 R14: 00000000d3100000 R15: ffff880073afdd48 [25809.662277] FS: 00007f3e04ec5730(0000) GS:ffff880001800000(0000) knlGS:0000000000000000 [25809.662281] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [25809.662283] CR2: 00007f79213f8000 CR3: 00000000bc95f000 CR4: 00000000000026e0 [25809.662286] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [25809.662289] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [25809.662292] Process btrfs-vol (pid: 15354, threadinfo ffff880073afc000, task ffff880001a84500) [25809.662294] Stack: [25809.662296] ffff880073afdc88 ffffffffa0372a0b ffff880073afdcc0 0000000000000001 [25809.662300] <0> 0000000000000100 0000000000000000 ffff8800b65e1800 ffff880073afdd48 [25809.662306] <0> ffff880077153000 00000000d3100000 00000000000000fb ffff88007b6ee990 [25809.662312] Call Trace: [25809.662324] [<ffffffffa0372a0b>] ? map_extent_buffer+0xdb/0xe0 [btrfs] [25809.662335] [<ffffffffa0371b13>] ? unmap_extent_buffer+0x13/0x40 [btrfs] [25809.662347] [<ffffffffa0366c54>] ? btrfs_item_offset+0xe4/0xf0 [btrfs] [25809.662358] [<ffffffffa037db3f>] btrfs_balance+0x1ef/0x270 [btrfs] [25809.662362] [<ffffffff810cc0ef>] ? find_get_page+0x7f/0xf0 [25809.662373] [<ffffffffa0382e10>] btrfs_ioctl+0x510/0xa20 [btrfs] [25809.662376] [<ffffffff810cc3e7>] ? unlock_page+0x27/0x30 [25809.662380] [<ffffffff810e8409>] ? __do_fault+0x449/0x540 [25809.662385] [<ffffffff8111bf62>] vfs_ioctl+0x22/0xa0 [25809.662388] [<ffffffff813cf4e9>] ? sub_preempt_count+0x59/0x60 [25809.662392] [<ffffffff8111c108>] do_vfs_ioctl+0x88/0x570 [25809.662396] [<ffffffff8120ecc9>] ? __up_read+0x99/0xc0 [25809.662400] [<ffffffff8107323e>] ? up_read+0xe/0x10 [25809.662404] [<ffffffff8111c63f>] sys_ioctl+0x4f/0x80 [25809.662409] [<ffffffff8100b5db>] system_call_fastpath+0x16/0x1b [25809.662411] Code: ff 31 c0 e9 c0 fa ff ff 0f 1f 80 00 00 00 00 0f 0b eb fe 0f 1f 40 00 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe <0f> 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f [25809.662463] RIP [<ffffffffa037d4e4>] btrfs_relocate_chunk+0x5c4/0x5e0 [btrfs] [25809.662475] RSP <ffff880073afdc18> [25809.662478] ---[ end trace b3050419e02a5437 ]--- -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Oct 06, 2009 at 08:48:32PM +0200, Diego Calleja wrote:> I was playing with btrfs with 2 files of 3.5 GB (using loop), I completely > zeroed one of the files. As expected, I had checksum failures, and I run > btrfs-vol -b just to see what happened, and I got this (using -rc3):Thanks, I''ll try to reproduce. Which raid level did you use for data? If not raid1, could you try with raid1? ;) -chris -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday 07 October 2009 05:17:54 Chris Mason escribió:> Thanks, I''ll try to reproduce. Which raid level did you use for data? > If not raid1, could you try with raid1? ;)I''m not sure, since the utils won''t tell. I mkfs''ed and mounted one of the 3.5GB files with no special options, and copied some files on it. I mkfs''ed the second file, put it in a loop device, and added it to the pool with btrfs-vol -a. Then I run btrfs-vol -b and I copied more files and rebalanced several times. Then I zeroed one of the files used as disks, and tried to read the files in the volume (unsucessfully). Then I rebalanced, and I hit the oops. It seems to be very reproducible. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Oct 07, 2009 at 03:51:46PM +0200, Diego Calleja wrote:> On Wednesday 07 October 2009 05:17:54 Chris Mason escribió: > > > Thanks, I''ll try to reproduce. Which raid level did you use for data? > > If not raid1, could you try with raid1? ;) > > I''m not sure, since the utils won''t tell. I mkfs''ed and mounted one of the 3.5GB > files with no special options, and copied some files on it. I mkfs''ed the second > file, put it in a loop device, and added it to the pool with btrfs-vol -a. Then > I run btrfs-vol -b and I copied more files and rebalanced several times. > > Then I zeroed one of the files used as disks, and tried to read the files in > the volume (unsucessfully). Then I rebalanced, and I hit the oops. It seems to > be very reproducible.Ok, in this case you ended up with raid0 on the data. If you: mkfs.btrfs -d raid1 /dev/loop0 /dev/loop1 you''ll get data raid1. -chris -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday 07 October 2009 16:33:09 Chris Mason escribió:> Ok, in this case you ended up with raid0 on the data. If you: > > mkfs.btrfs -d raid1 /dev/loop0 /dev/loop1 you''ll get data raid1.I got a BUG...but it may be a different bug. Unlike the last time, I didn''t even need to zero one of the "disks", the second I rebalanced it oopsed: [ 705.060358] btrfs: relocating block group 4194304 flags 4 [ 1006.106133] btrfs: relocating block group 5704777728 flags 18 [ 1006.336521] btrfs: found 1 extents [ 1006.469967] btrfs: relocating block group 5384896512 flags 20 [ 1006.471043] btrfs allocation failed flags 20, wanted 4096 [ 1006.471048] space_info has 317378560 free, is full [ 1006.471052] space_info total=319881216, pinned=16384, delalloc=0, may_use=0, used=2355200, root=2875392, super=131072, reserved=0 [ 1006.471058] block group 5384896512 has 319881216 bytes, 2355200 used 16384 pinned 0 reserved [ 1006.471063] entry offset 5384896512, bytes 483328, bitmap yes [ 1006.471065] block group has cluster?: yes [ 1006.471067] 1 blocks of free space at or bigger than bytes is [ 1006.471092] ------------[ cut here ]------------ [ 1006.471094] kernel BUG at fs/btrfs/transaction.c:1001! [ 1006.471098] invalid opcode: 0000 [#1] PREEMPT SMP [ 1006.471103] last sysfs file: /sys/devices/virtual/block/loop0/range [ 1006.471106] CPU 0 [ 1006.471109] Modules linked in: ppp_deflate bsd_comp ppp_async crc_ccitt btrfs zlib_deflate crc32c libcrc32c loop ppp_generic slhc kvm_intel kvm i915 snd_ca0106 snd_rawmidi drm_kms_helper snd_ac97_codec ac97_bus cfbcopyarea snd_pcm_oss video snd_mixer_oss backlight snd_pcm snd_timer cfbimgblt processor wacom intel_agp cfbfillrect button snd soundcore snd_page_alloc usbhid fan fuse ehci_hcd uhci_hcd thermal thermal_sys atl1 mii usbcore [ 1006.471159] Pid: 8207, comm: btrfs-vol Not tainted 2.6.32-rc3 #1 System Product Name [ 1006.471162] RIP: 0010:[<ffffffffa0257665>] [<ffffffffa0257665>] btrfs_commit_transaction+0x6c5/0x6f0 [btrfs] [ 1006.471183] RSP: 0018:ffff8800bc285a78 EFLAGS: 00010286 [ 1006.471186] RAX: 00000000ffffffe4 RBX: ffff8800bdef5800 RCX: 0000000000000016 [ 1006.471189] RDX: 0000000000000000 RSI: ffff8800bc2859f8 RDI: ffff8800a86d8060 [ 1006.471192] RBP: ffff8800bc285b18 R08: 0000000000000000 R09: 0000000000000000 [ 1006.471195] R10: 0000000000000000 R11: 0000000000000000 R12: ffff8800a181f5c0 [ 1006.471199] R13: ffff8800a1eb1ae0 R14: ffff8800bc285ab8 R15: ffff8800a1eb1b50 [ 1006.471202] FS: 00007f7c5427f730(0000) GS:ffff880001800000(0000) knlGS:0000000000000000 [ 1006.471206] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1006.471209] CR2: 0000000002812500 CR3: 00000000a2f88000 CR4: 00000000000026f0 [ 1006.471212] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 1006.471215] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 1006.471219] Process btrfs-vol (pid: 8207, threadinfo ffff8800bc284000, task ffff88000b2bae00) [ 1006.471221] Stack: [ 1006.471223] ffff8800bc285ad8 0000000000000206 00000001bf15ad00 0000000001000010 [ 1006.471229] <0> ffff8800a181f5c0 0000000000000004 ffff8800a1eb1b80 ffff8800a1eb1ae0 [ 1006.471235] <0> 0000000000000000 ffff88000b2bae00 ffffffff8106eb30 ffff8800bc285ad0 [ 1006.471242] Call Trace: [ 1006.471250] [<ffffffff8106eb30>] ? autoremove_wake_function+0x0/0x40 [ 1006.471264] [<ffffffffa0256d0e>] ? start_transaction+0x11e/0x170 [btrfs] [ 1006.471285] [<ffffffffa029630c>] relocate_block_group+0xec/0x4b0 [btrfs] [ 1006.471291] [<ffffffff813cf4e9>] ? sub_preempt_count+0x59/0x60 [ 1006.471301] [<ffffffffa025596b>] ? btrfs_clean_old_snapshots+0x8b/0x1c0 [btrfs] [ 1006.471314] [<ffffffffa02714fb>] ? btrfs_wait_ordered_extents+0x1bb/0x1d0 [btrfs] [ 1006.471325] [<ffffffffa0296a5a>] btrfs_relocate_block_group+0x15a/0x390 [btrfs] [ 1006.471336] [<ffffffffa027cfab>] btrfs_relocate_chunk+0x8b/0x5e0 [btrfs] [ 1006.471348] [<ffffffffa0272a0b>] ? map_extent_buffer+0xdb/0xe0 [btrfs] [ 1006.471359] [<ffffffffa0271b13>] ? unmap_extent_buffer+0x13/0x40 [btrfs] [ 1006.471372] [<ffffffffa0266c54>] ? btrfs_item_offset+0xe4/0xf0 [btrfs] [ 1006.471383] [<ffffffffa027db3f>] btrfs_balance+0x1ef/0x270 [btrfs] [ 1006.471387] [<ffffffff810cc0ef>] ? find_get_page+0x7f/0xf0 [ 1006.471398] [<ffffffffa0282e10>] btrfs_ioctl+0x510/0xa20 [btrfs] [ 1006.471401] [<ffffffff810cc3e7>] ? unlock_page+0x27/0x30 [ 1006.471406] [<ffffffff810e8409>] ? __do_fault+0x449/0x540 [ 1006.471412] [<ffffffff8111bf62>] vfs_ioctl+0x22/0xa0 [ 1006.471415] [<ffffffff813cf4e9>] ? sub_preempt_count+0x59/0x60 [ 1006.471419] [<ffffffff8111c108>] do_vfs_ioctl+0x88/0x570 [ 1006.471424] [<ffffffff8120ecc9>] ? __up_read+0x99/0xc0 [ 1006.471428] [<ffffffff8107323e>] ? up_read+0xe/0x10 [ 1006.471432] [<ffffffff8111c63f>] sys_ioctl+0x4f/0x80 [ 1006.471438] [<ffffffff8100b5db>] system_call_fastpath+0x16/0x1b [ 1006.471440] Code: 00 00 00 00 e9 c4 fe ff ff be ab 03 00 00 48 c7 c7 1a 84 29 a0 e8 ac 83 df e0 e9 e8 fa ff ff 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe <0f> 0b eb fe 0f 0b eb fe 0f 0b eb fe 0f 0b eb fe be d7 03 00 00 [ 1006.471492] RIP [<ffffffffa0257665>] btrfs_commit_transaction+0x6c5/0x6f0 [btrfs] [ 1006.471504] RSP <ffff8800bc285a78> [ 1006.471507] ---[ end trace 14afea9f89710f14 ]--- -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Oct 07, 2009 at 09:10:28PM +0200, Diego Calleja wrote:> On Wednesday 07 October 2009 16:33:09 Chris Mason escribió: > > Ok, in this case you ended up with raid0 on the data. If you: > > > > mkfs.btrfs -d raid1 /dev/loop0 /dev/loop1 you''ll get data raid1. > > I got a BUG...but it may be a different bug. Unlike the last time, > I didn''t even need to zero one of the "disks", the second I > rebalanced it oopsed:I''m afraid this is good old enospc. Balancing still needs some work to be completely safe. -chris -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday 07 October 2009 21:45:29 Chris Mason escribió:> I''m afraid this is good old enospc. Balancing still needs some work to > be completely safe.I''ve tried using less data and raid1, but I can''t reproduce it. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
By the way, i think it''d be useful if debug-tree would tell which
policy
the fs is applying to each chunk. Something like this:
item 4 key (FIRST_CHUNK_TREE CHUNK_ITEM 8379826176) itemoff 3495
itemsize 112
chunk length 319881216 owner 2 type 17 (data on RAID1)
num_stripes 2
stripe 0 devid 2 offset 2265382912
stripe 1 devid 1 offset 989003776
Index: btrfs-progs-unstable/print-tree.c
==================================================================---
btrfs-progs-unstable.orig/print-tree.c 2009-10-07 16:44:53.013210615 +0200
+++ btrfs-progs-unstable/print-tree.c 2009-10-07 21:37:46.162546261 +0200
@@ -79,15 +79,37 @@
return 0;
}
+static void print_chunk_type(u64 chunk_flags)
+{
+ printf(" type %llu (", chunk_flags);
+
+ if (chunk_flags & BTRFS_BLOCK_GROUP_DATA)
+ printf("data");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_METADATA)
+ printf("metadata");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_SYSTEM)
+ printf("system??");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_DUP)
+ printf("dup??");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_RAID0)
+ printf(" on RAID0");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_RAID1)
+ printf(" on RAID1");
+ if (chunk_flags & BTRFS_BLOCK_GROUP_RAID10)
+ printf(" on RAID10");
+ printf(") ");
+}
+
static void print_chunk(struct extent_buffer *eb, struct btrfs_chunk *chunk)
{
int num_stripes = btrfs_chunk_num_stripes(eb, chunk);
int i;
- printf("\t\tchunk length %llu owner %llu type %llu num_stripes
%d\n",
+ printf("\t\tchunk length %llu owner %llu",
(unsigned long long)btrfs_chunk_length(eb, chunk),
- (unsigned long long)btrfs_chunk_owner(eb, chunk),
- (unsigned long long)btrfs_chunk_type(eb, chunk),
- num_stripes);
+ (unsigned long long)btrfs_chunk_owner(eb, chunk));
+ print_chunk_type((unsigned long long)btrfs_chunk_type(eb, chunk));
+ printf("num_stripes %d\n", num_stripes);
+
for (i = 0 ; i < num_stripes ; i++) {
printf("\t\t\tstripe %d devid %llu offset %llu\n", i,
(unsigned long long)btrfs_stripe_devid_nr(eb, chunk, i),
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs"
in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html