ext4 BUG in dom0 Kernel 2.6.32.36

From: [email protected]
To: [email protected]; [email protected]
CC: [email protected]; [email protected]
Subject: ext4 BUG in dom0 Kernel 2.6.32.36
Date: Tue, 6 Sep 2011 15:24:14 +0800

Hi:

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack below)
32.36 kernel commit: http://git.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=commit;h=ae333e97552c81ab10395ad1ffc6d6daaadb144a

The bug only show up in our cluster environments which includes 300 physical machines, one server will run into this bug per day. Running ontop of every server, there are about 30 VMS, each of which has heavy IO workload inside.(we are doing some kinds of stress tests)

We have our own distribute file system as the storage of cluster, every VM'image file will be spilt into several files with equal size in physical disk, and every creation of file use ext4 fallocation(fallocation size 1MB). So I believe there will be quite a lot of uninitialized extent to be initialized during the test.

After go through the src code. Call routinue is ext4_da_sritepages->mpage_da_map_blocks->ext4_get_blocks->ext4_ext_get_blocks-> ext4_ext_handle_uninitialized_extents->ext4_ext_convert_to_initialized->ext4_ext_insert_extent

if ext4_ext_handle_uninitialized_extents is called, then line 3306 must be satisfied. that is we have in_range(iblock, ee_block, ee_len) = true.
so iblock >= ee_block

fs/ext4/extents.c
3306 <+++<+++if (in_range(iblock, ee_block, ee_len)) { 3307 <+++<+++<+++newblock = iblock - ee_block + ee_start; 3308 <+++<+++<+++/* number of remaining blocks in the extent */ 3309 <+++<+++<+++allocated = ee_len - (iblock - ee_block); 3310 <+++<+++<+++ext_debug("%u fit into %u:%d -> %llu\n", iblock, 3311 <+++<+++<+++<+++<+++ee_block, ee_len, newblock); 3312
3313 <+++<+++<+++/* Do not put uninitialized extent in the cache */ 3314 <+++<+++<+++if (!ext4_ext_is_uninitialized(ex)) { 3315 <+++<+++<+++<+++ext4_ext_put_in_cache(inode, ee_block, 3316 <+++<+++<+++<+++<+++<+++<+++ee_len, ee_start,
3317 <+++<+++<+++<+++<+++<+++<+++EXT4_EXT_CACHE_EXTENT); 3318 <+++<+++<+++<+++goto out;
3319 <+++<+++<+++}
3320 <+++<+++<+++ret = ext4_ext_handle_uninitialized_extents(handle, 3321 <+++<+++<+++<+++<+++inode, iblock, max_blocks, path, 3322 <+++<+++<+++<+++<+++flags, allocated, bh_result, newblock); 3323 <+++<+++<+++return ret;
3324 <+++<+++}

the newext is from line 2678, its ee_block is iblock + max_blocks the nearex is path[depth].p_ext(line 1683)

BUG_ON 1716 means iblock + max_blocks = ee_block.
So maybe that means we have iblock = ee_block and max_blocks = 0.

1716 <+++<+++BUG_ON(newext->ee_block == nearex->ee_block);
1717 <+++<+++len = (EXT_MAX_EXTENT(eh) - nearex) * sizeof(struct ext4_extent); 1718 <+++<+++len = len < 0 ? 0 : len;
1719 <+++<+++ext_debug("insert %d:%llu:[%d]%d before: nearest 0x%p, " 1720 <+++<+++<+++<+++"move %d from 0x%p to 0x%p\n",
1721 <+++<+++<+++<+++le32_to_cpu(newext->ee_block),
1722 <+++<+++<+++<+++ext_pblock(newext),
1723 <+++<+++<+++<+++ext4_ext_is_uninitialized(newext),
1724 <+++<+++<+++<+++ext4_ext_get_actual_len(newext),
1725 <+++<+++<+++<+++nearex, len, nearex + 1, nearex + 2);
1726 <+++<+++memmove(nearex + 1, nearex, len);
1727 <+++<+++path[depth].p_ext = nearex;
1728 <+++}

2678 <+++<+++ex3 = &newex;
2679 <+++<+++ex3->ee_block = cpu_to_le32(iblock + max_blocks); 2680 <+++<+++ext4_ext_store_pblock(ex3, newblock + max_blocks); 2681 <+++<+++ex3->ee_len = cpu_to_le16(allocated - max_blocks); 2682 <+++<+++ext4_ext_mark_uninitialized(ex3);
2683 <+++<+++err = ext4_ext_insert_extent(handle, inode, path, ex3, 0); 2684 <+++<+++if (err == -ENOSPC && may_zeroout) {
2685 <+++<+++<+++err = ext4_ext_zeroout(inode, &orig_ex);

if max_blocks = 0; it means 2225, mpd->b_size >> mpd->inode->i_blkbits is 0.

fs/ext4/inode.c
2220 static int mpage_da_map_blocks(struct mpage_da_data *mpd) 2221 {
2222 <+++int err, blks, get_blocks_flags;
2223 <+++struct buffer_head new;
2224 <+++sector_t next = mpd->b_blocknr;
2225 <+++unsigned max_blocks = mpd->b_size >> mpd->inode->i_blkbits; 2226 <+++loff_t disksize = EXT4_I(mpd->inode)->i_disksize; 2227 <+++handle_t *handle = NULL;
2228

Could it be possilbe, right now I am tring to reproduce this problem in a much easiler way, any suggestion?

Many thanks.

------------[ cut here ]------------
kernel BUG at fs/ext4/extents.c:1716!
invalid opcode: 0000 [#1] SMP
last sysfs file: /sys/block/tapdevk/stat
CPU 3
Modules linked in: xt_iprange xt_mac arptable_filter arp_tables xt_physdev nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack

iptable_filter ip_tables bridge autofs4 ipmi_devintf ipmi_si ipmi_msghandler lockd sunrpc bonding ipv6 8021q garp stp llc xenfs dm_multipath fuse xen_netback xen_blkback blktap blkback_pagemap loop nbd video output sbs sbshc parport_pc lp parport joydev ses enclosure snd_seq_dummy snd_seq_oss bnx2 snd_seq_midi_event snd_seq snd_seq_device dcdbas snd_pcm_oss snd_mixer_oss serio_raw snd_pcm snd_timer snd soundcore snd_page_alloc iTCO_wdt iTCO_vendor_support pcspkr shpchp [last unloaded: freq_table]
Pid: 9073, comm: flush-8:16 Not tainted 2.6.32.36xen #1 PowerEdge R710

RIP: e030:[<ffffffff811a6184>] [<ffffffff811a6184>] ext4_ext_insert_extent+0xac1/0xbe0
RSP: e02b:ffff8801499cd580 EFLAGS: 00010246
RAX: 0000000000002948 RBX: 0000000000000000 RCX: ffff8801499cd780
RDX: ffff8801499cd360 RSI: ffff88007dedb310 RDI: 0000000000000017
RBP: ffff8801499cd650 R08: ffff8801499cd340 R09: ffff880063488930
R10: 000000018100f8bf R11: dead000000200200 R12: ffff88005a29700c
R13: ffff88005a297000 R14: ffff8801158198c0 R15: ffff88003e9ea1b0
FS: 00007fd3cc4bf6e0(0000) GS:ffff88002808f000(0000) knlGS:0000000000000000
CS: e033 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 000000000042a09e CR3: 00000000bf3bd000 CR4: 0000000000002660
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process flush-8:16 (pid: 9073, threadinfo ffff8801499cc000, task ffff880149ad5b40)
Stack:
ffff8801499cd780 ffff88003e9ea180 ffff8801c5b47300 01ffffff81103c0c
<0> ffff88003e9ea180 000000017dedb2a0 ffff880115819800 ffff88007dedb2a0
<0> ffff8801499cd5d0 ffffffff811c12ea ffff8801499cd5f0 ffffffff811c16ea
Call Trace:
[<ffffffff811c12ea>] ? jbd_unlock_bh_journal_head+0x16/0x18
[<ffffffff811c16ea>] ? jbd2_journal_put_journal_head+0x4d/0x52
[<ffffffff811bb7d6>] ? jbd2_journal_get_write_access+0x31/0x38
[<ffffffff811a88e9>] ? __ext4_journal_get_write_access+0x4c/0x5f
[<ffffffff811a6ce3>] ext4_ext_handle_uninitialized_extents+0xa40/0xef5
[<ffffffff8100f175>] ? xen_force_evtchn_callback+0xd/0xf
[<ffffffff8100f8d2>] ? check_events+0x12/0x20
[<ffffffff81042fcf>] ? need_resched+0x23/0x2d
[<ffffffff811a74e1>] ext4_ext_get_blocks+0x265/0x6eb
[<ffffffff81042fcf>] ? need_resched+0x23/0x2d
[<ffffffff81188b55>] ext4_get_blocks+0x140/0x204
[<ffffffff81188d2f>] mpage_da_map_blocks+0xb7/0x681
[<ffffffff810d3b29>] ? find_get_pages_tag+0x48/0xcc
[<ffffffff8100f8d2>] ? check_events+0x12/0x20
[<ffffffff810da8df>] ? pagevec_lookup_tag+0x27/0x30
[<ffffffff810d87cc>] ? write_cache_pages+0x175/0x35e
[<ffffffff811893f0>] ? __mpage_da_writepage+0x0/0x164
[<ffffffff81103c0c>] ? kmem_cache_alloc+0x94/0xf6
[<ffffffff811bbc40>] ? jbd2_journal_start+0xa1/0xcd
[<ffffffff8119957f>] ? ext4_journal_start_sb+0xdc/0x111
[<ffffffff81186852>] ? ext4_meta_trans_blocks+0x74/0xce
[<ffffffff8118bc42>] ext4_da_writepages+0x47a/0x6a7
[<ffffffff810d8a00>] do_writepages+0x21/0x2a
[<ffffffff8112cdb8>] writeback_single_inode+0xc8/0x1e3
[<ffffffff8112d5e4>] writeback_inodes_wb+0x30b/0x37e
[<ffffffff8102f82d>] ? paravirt_end_context_switch+0x17/0x31
[<ffffffff8100b459>] ? xen_end_context_switch+0x1e/0x22
[<ffffffff8112d788>] wb_writeback+0x131/0x1bb
[<ffffffff81064029>] ? try_to_del_timer_sync+0x73/0x81
[<ffffffff8112d9ef>] wb_do_writeback+0x13c/0x153
[<ffffffff8106425b>] ? process_timeout+0x0/0x10
[<ffffffff810e78d1>] ? bdi_start_fn+0x0/0xd0
[<ffffffff8112da32>] bdi_writeback_task+0x2c/0xb3
[<ffffffff810e793b>] bdi_start_fn+0x6a/0xd0
[<ffffffff810754b7>] kthread+0x6e/0x76
[<ffffffff81013daa>] child_rip+0xa/0x20
[<ffffffff81012f91>] ? int_ret_from_sys_call+0x7/0x1b
[<ffffffff8101371d>] ? retint_restore_args+0x5/0x6
[<ffffffff81013da0>] ? child_rip+0x0/0x20
Code: 8d 04 85 f4 ff ff ff 85 c0 0f 49 d8 48 63 d3 e8 47 c7 07 00 49 8d 44 24 0c 49 89 47 10 eb 3a bb f4 ff ff ff e9 c2 00 00 00 75 04
<0f> 0b eb fe 41 0f b7 45 04 49 8d 7c 24 0c 48 6b c0 0c 4c 89 e6
RIP [<ffffffff811a6184>] ext4_ext_insert_extent+0xac1/0xbe0
RSP <ffff8801499cd580>
---[ end trace 035c7d09ed95fb32 ]---

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Hi:

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack below)

Date: Tue, 6 Sep 2011 10:53:47 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]
Subject: Re: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 5 lines · MaoXiaoyun

Hi:

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack below)

Did you try the 3.0 kernel?

Quoted 5 lines

Date: Tue, 6 Sep 2011 10:53:47 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected];

[email protected]

Quoted 7 lines

Subject: Re: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 4 lines · MaoXiaoyun

Hi:

Quoted 1 line

Quoted 1 line

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack

below)

Quoted 2 lines

Did you try the 3.0 kernel?

No, I am afried the change would be to much for our current env. May result in other stable issue.
So, I want to dig out what really happen. Hopes.

Date: Tue, 6 Sep 2011 11:55:02 -0700
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]
Subject: Re: [Xen-devel] RE: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 16 lines · MaoXiaoyun

Quoted 5 lines

Date: Tue, 6 Sep 2011 10:53:47 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected];

[email protected]

Quoted 8 lines

Subject: Re: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 5 lines · MaoXiaoyun

Hi:

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack

below)

Quoted 1 line

Quoted 4 lines

Quoted 1 line

Did you try the 3.0 kernel?

No, I am afried the change would be to much for our current env. May result in other stable issue.
So, I want to dig out what really happen. Hopes.

Another question is whether this is a regression compared to earlier versions of 2.6.32? Do you know if this problem exists in a non-Xen environment?

Thanks,
J
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]
Subject: RE: [Xen-devel] RE: ext4 BUG in dom0 Kernel 2.6.32.36
Date: Wed, 7 Sep 2011 10:35:21 +0800

----------------------------------------

Quoted 23 lines

Date: Tue, 6 Sep 2011 11:55:02 -0700
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]
Subject: Re: [Xen-devel] RE: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 16 lines · MaoXiaoyun

Quoted 5 lines

Date: Tue, 6 Sep 2011 10:53:47 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected];

[email protected]

Quoted 8 lines

Subject: Re: ext4 BUG in dom0 Kernel 2.6.32.36

Quoted 5 lines · MaoXiaoyun

Hi:

I've met an ext4 Bug in dom0 kernel 2.6.32.36. (See kernel stack

below)

Quoted 1 line

Quoted 9 lines

Quoted 4 lines

Quoted 1 line

Did you try the 3.0 kernel?

No, I am afried the change would be to much for our current env. May result in other stable issue.
So, I want to dig out what really happen. Hopes.

Another question is whether this is a regression compared to earlier versions of 2.6.32? Do you know if this problem exists in a non-Xen environment?

There are some others reports this issue in non-xen env.
http://markmail.org/message/ywr4nfgiuvgdcr7y
http://www.spinics.net/lists/linux-ext4/msg21066.html

The difficulty is I haven't find a efficient way to reproduce it. (Currently it only show in our cluster, redeploy our cluster may cost 3days more. )

Quoted 6 lines

Thanks,
J
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Hi:

We met an ext4 BUG_ON in extents.c:1716 which crash kernel flush thread, and result in disk unvailiable.

BUG details refer to: http://www.gossamer-threads.com/lists/xen/devel/217091?do=post_view_threaded

Attached is the fix, verified in our env.

Without this patch, more than 3 servers hit BUG_ON in our hundreds of servers every day.

many thanks.

Date: Mon, 26 Sep 2011 10:28:08 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [patch 1/1] ext4-fix-dirty-extent-when-origin-leaf-extent-reac.patch

Quoted 4 lines · MaoXiaoyun

Hi:

Quoted 5 lines

We met an ext4 BUG_ON in extents.c:1716 which crash kernel flush thread, and result in disk unvailiable.

BUG details refer to: http://www.gossamer-threads.com/lists/xen/devel/217091?do=post_view_threaded

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

Quoted 5 lines

Without this patch, more than 3 servers hit BUG_ON in our hundreds of servers every day.

many thanks.

Quoted 1 line

Quoted 1 line · MaoXiaoyun

----------------------------------------

Quoted 4 lines

Date: Mon, 26 Sep 2011 10:28:08 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected];

[email protected]

Quoted 8 lines

Subject: Re: [patch 1/1] ext4-fix-dirty-extent-when-origin-leaf-extent-reac.patch

Quoted 5 lines · MaoXiaoyun

Hi:

We met an ext4 BUG_ON in extents.c:1716 which crash kernel flush thread,

and result in disk unvailiable.

Quoted 2 lines

Quoted 2 lines

BUG details refer to:

http://www.gossamer-threads.com/lists/xen/devel/217091?do=post_view_threaded

Quoted 1 line

Quoted 1 line

Quoted 3 lines

Quoted 1 line

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32,

right?

Quoted 1 line

The patch is for 2.6.39. It can be patched on 2.6.32 too. Thanks.

Quoted 2 lines

Quoted 2 lines

Without this patch, more than 3 servers hit BUG_ON in our hundreds of

servers every day.

Quoted 5 lines

Quoted 3 lines

many thanks.

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

Quoted 29 lines

Quoted 1 line

Quoted 1 line · MaoXiaoyun

----------------------------------------

Quoted 4 lines

Date: Mon, 26 Sep 2011 10:28:08 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected];

[email protected]

Quoted 8 lines

Subject: Re: [patch 1/1] ext4-fix-dirty-extent-when-origin-leaf-extent-reac.patch

Quoted 5 lines · MaoXiaoyun

Hi:

We met an ext4 BUG_ON in extents.c:1716 which crash kernel flush thread,

and result in disk unvailiable.

Quoted 2 lines

Quoted 2 lines

BUG details refer to:

http://www.gossamer-threads.com/lists/xen/devel/217091?do=post_view_threaded

Quoted 4 lines

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32,

right?

Quoted 1 line

Quoted 2 lines

The patch is for 2.6.39. It can be patched on 2.6.32 too. Thanks.

So why don't you suggest applying this to the stable tree maintainers instead? xen-devel really isn't the right forum for this sort of bug fixes, particularly when the underlying kernel.org tree is still being maintained.

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

Quoted 5 lines · Konrad Rzeszutek Wilk

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

I'm curious --- is there a good reason why Xen users are using an upstream 2.6.32 kernel? If they are using a distro kernel, fine, but then the distro kernel should be providing the support. But at this point, 2.6.32 is so positively *ancient* that, I'm personally not interesting in providing free, unpaid distro support for users who aren't willing to either (a) pay $$$ and get a supported distro kernel, or (b) use a much more modern kernel. At this point, Guest and Host Xen support is available in 3.0 kernels, so there's really no excuse, right?

- Ted

Quoted 17 lines · Ted Ts'o

Quoted 5 lines · Konrad Rzeszutek Wilk

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

I'm curious --- is there a good reason why Xen users are using an upstream 2.6.32 kernel? If they are using a distro kernel, fine, but then the distro kernel should be providing the support. But at this point, 2.6.32 is so positively *ancient* that, I'm personally not interesting in providing free, unpaid distro support for users who aren't willing to either (a) pay $$$ and get a supported distro kernel, or (b) use a much more modern kernel. At this point, Guest and Host Xen support is available in 3.0 kernels, so there's really no excuse, right?

Quoted 2 lines

- Ted

In my case, for Dom0 I use the 2.6.32 kernel from my distrib, because it's stable.
I have stability problems with the kernel 3.0 on Dom0, and as I haven't physical access neither kvm or serial port, I don't know what to report... It just hang randomly, without any line in logs.

Does the Xen support on 3.0 kernels should be considered stable ?

Date: Tue, 27 Sep 2011 15:35:23 -0400
From: [email protected]
To: [email protected]
CC: [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [patch 1/1] ext4-fix-dirty-extent-when-origin-leaf-extent-reac.patch

Quoted 5 lines · Konrad Rzeszutek Wilk

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

I'm curious --- is there a good reason why Xen users are using an upstream 2.6.32 kernel? If they are using a distro kernel, fine, but then the distro kernel should be providing the support. But at this point, 2.6.32 is so positively *ancient* that, I'm personally not interesting in providing free, unpaid distro support for users who aren't willing to either (a) pay $$$ and get a supported distro kernel, or (b) use a much more modern kernel. At this point, Guest and Host Xen support is available in 3.0 kernels, so there's really no excuse, right?

- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 5 lines · Konrad Rzeszutek Wilk

Quoted 5 lines

Quoted 2 lines

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

I'm curious --- is there a good reason why Xen users are using an upstream 2.6.32 kernel? If they are using a distro kernel, fine, but then the distro kernel should be providing the support. But at this point, 2.6.32 is so positively *ancient* that, I'm personally not interesting in providing free, unpaid distro support for users who aren't willing to either (a) pay $$$ and get a supported distro kernel, or (b) use a much more modern kernel. At this point, Guest and Host Xen support is available in 3.0 kernels, so there's really no excuse, right?

From 600d493b14ebd776cf8ea0e9dcdccc0d54200403 Mon Sep 17 00:00:00 2001

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

From 600d493b14ebd776cf8ea0e9dcdccc0d54200403 Mon Sep 17 00:00:00 2001
From: Zheng Liu <[email protected]>
Date: Wed, 28 Sep 2011 16:26:05 +0800
Subject: [PATCH] ext4: fix BUG_ON() in ext4_ext_insert_extent()

We will meet with a BUG_ON() if following script is run.

mkfs.ext4 -b 4096 /dev/sdb1 1000000
mount -t ext4 /dev/sdb1 /mnt/sdb1
fallocate -l 100M /mnt/sdb1/test
sync
for((i=0;i<170;i++))
do

        dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1 seek=`expr $i \* 2`

done
umount /mnt/sdb1
mount -t ext4 /dev/sdb1 /mnt/sdb1
dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1 seek=341 umount /mnt/sdb1
mount /dev/sdb1 /mnt/sdb1
dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1 seek=340 sync

The reason is that it forgot to mark dirty when splitting two extents in ext4_ext_convert_to_initialized(). Althrough ex has been updated in memory, it is not dirtied both in ext4_ext_convert_to_initialized() and ext4_ext_insert_extent(). The disk layout is corrupted. Then it will meet with a BUG_ON() when writting at the start of that extent again.

Cc [email protected] #for 2.6.39

Cc Greg Kroah-Hartman <[email protected]>

Cc "Theodore Ts'o" <[email protected]>

Cc Xiaoyun Mao <[email protected]>

Cc Yingbin Wang <[email protected]>

Cc Jia Wan <[email protected]>

Signed-off-by Zheng Liu <[email protected]>

---
 fs/ext4/extents.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

fs/ext4/extents.c +1 -0

index 4890d6f..cd20425 100644
@@ -2607,6 +2607,7 @@ static int ext4_ext_convert_to_initialized(handle_t *handle,
 		ex1 = ex;
 		ex1->ee_len = cpu_to_le16(map->m_lblk - ee_block);
 		ext4_ext_mark_uninitialized(ex1);
+		ext4_ext_dirty(handle, inode, path + depth);
 		ex2 = &newex;
 	}
 	/*

Signature · 1 line

1.7.4.1

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

Quoted 6 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Quoted 4 lines

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 12 lines · Greg KH

Quoted 10 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

Quoted 1 line

No, .39 has not been maintained for quite some time now, sorry.

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Quoted 14 lines · Zheng Liu

Quoted 13 lines · Greg KH

Quoted 10 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

Quoted 5 lines

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Ah, ok, that makes sense, can you provide me a patch that will apply to the .32 and .33-longterm kernels?

From ff98f00676657b797c426c80804a3fde3f86ea83 Mon Sep 17 00:00:00 2001

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 21 lines · Greg KH

Quoted 19 lines · Zheng Liu

Quoted 7 lines · Greg KH

Quoted 6 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Quoted 6 lines

Quoted 4 lines

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Quoted 2 lines

Ah, ok, that makes sense, can you provide me a patch that will apply to the .32 and .33-longterm kernels?

Hi Greg,

I couldn't download the .32 and .33-longterm kernels from kernel.org because the full sources were not found on that server. Thus this patch is generated from the .32-mainline kernel and can be applied to the .32 and .33 kernels.

Quoted 22 lines · Zheng Liu

Quoted 21 lines · Greg KH

Quoted 19 lines · Zheng Liu

Quoted 7 lines · Greg KH

Quoted 6 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Quoted 6 lines

Quoted 4 lines

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Quoted 8 lines

Quoted 2 lines

Ah, ok, that makes sense, can you provide me a patch that will apply to the .32 and .33-longterm kernels?

Hi Greg,

I couldn't download the .32 and .33-longterm kernels from kernel.org because the full sources were not found on that server. Thus this patch is generated from the .32-mainline kernel and can be applied to the .32 and .33 kernels.

The .32 and .33 longterm kernels are part of the linux-stable tree on git.kernel.org, they are in their own branch. Please redo this against those trees, as I'm pretty sure that there will be conflicts, due to all of the different changes since the .0 releases.

From bc522003378af679afd227ff87497dfb4fd4d652 Mon Sep 17 00:00:00 2001

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 32 lines · Greg KH

Quoted 22 lines · Zheng Liu

Quoted 21 lines · Greg KH

Quoted 19 lines · Zheng Liu

Quoted 7 lines · Greg KH

Quoted 6 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Quoted 6 lines

Quoted 4 lines

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Quoted 8 lines

Quoted 2 lines

Ah, ok, that makes sense, can you provide me a patch that will apply to the .32 and .33-longterm kernels?

Hi Greg,

I couldn't download the .32 and .33-longterm kernels from kernel.org because the full sources were not found on that server. Thus this patch is generated from the .32-mainline kernel and can be applied to the .32 and .33 kernels.

Quoted 4 lines

The .32 and .33 longterm kernels are part of the linux-stable tree on git.kernel.org, they are in their own branch. Please redo this against those trees, as I'm pretty sure that there will be conflicts, due to all of the different changes since the .0 releases.

Hi Greg

This patch is for .32 longterm kernel. Please try it again.

regards,
Zheng

Quoted 1 line

From bc522003378af679afd227ff87497dfb4fd4d652 Mon Sep 17 00:00:00 2001

From: Zheng Liu <[email protected]>
Date: Fri, 28 Oct 2011 19:41:19 +0800
Subject: [PATCH] ext4: fix BUG_ON() in ext4_ext_insert_extent()

We will meet with a BUG_ON() if following script is run.

mkfs.ext4 -b 4096 /dev/sdb1 1000000
mount -t ext4 /dev/sdb1 /mnt/sdb1
fallocate -l 100M /mnt/sdb1/test
sync
for((i=0;i<170;i++))
do
dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1
seek=expr $i \* 2
done
umount /mnt/sdb1
mount -t ext4 /dev/sdb1 /mnt/sdb1
dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1 seek=341 umount /mnt/sdb1
mount /dev/sdb1 /mnt/sdb1
dd if=/dev/zero of=/mnt/sdb1/test conv=notrunc bs=256k count=1 seek=340 sync

The reason is that it forgot to mark dirty when splitting two extents in ext4_ext_convert_to_initialized(). Althrough ex has been updated in memory,
it is not dirtied both in ext4_ext_convert_to_initialized() and ext4_ext_insert_extent(). The disk layout is corrupted. Then it will meet with
a BUG_ON() when writting at the start of that extent again.

Cc [email protected] #for 2.6.32

Cc Greg Kroah-Hartman <[email protected]>

Cc "Theodore Ts'o" <[email protected]>

Cc Xiaoyun Mao <[email protected]>

Cc Yingbin Wang <[email protected]>

Cc Jia Wan <[email protected]>

Signed-off-by Zheng Liu <[email protected]>

Sorry for dragging this out, but what commit id does this correspond to in Linus's tree? I can't seem to figure it out.

Or does it not correspond to anything? If so, I need a sentence that says why it doesn't for the patch changelog.

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

On Wed, Nov 02, 2011 at 02:00:24PM -0700, Greg KH wrote:
[snip]

Quoted 4 lines

Sorry for dragging this out, but what commit id does this correspond to in Linus's tree? I can't seem to figure it out.

Quoted 2 lines

Or does it not correspond to anything? If so, I need a sentence that says why it doesn't for the patch changelog.

Hi Greg,

This patch doesn't apply for upstream because the code has been refactored since 3.0.

Quoted 22 lines · Zheng Liu

Quoted 21 lines · Greg KH

Quoted 19 lines · Zheng Liu

Quoted 7 lines · Greg KH

Quoted 6 lines · Ted Ts'o

Quoted 4 lines · Tao Ma

actually this bug does show up in 2.6.39 and I think stable tree still needs this fix. After some careful test, my colleague has generated the patch. Please considering ack it so that Greg can add it into the stable tree.

Quoted 6 lines

Quoted 4 lines

Sorry for the delay, but yes. This patch would be good for the stable tree for 2.6.39 (if Greg is still accepting patches for 2.6.39-stable). It doesn't apply for upstream ext4 since the code has been changed/refactored since then, but it's a good fix.

No, .39 has not been maintained for quite some time now, sorry.

Hi Greg,

Thank you for your attention. Actually this bug is between from .32 to .39. Please considering to apply this patch to other stable or longterm trees.

Quoted 8 lines

Quoted 2 lines

Ah, ok, that makes sense, can you provide me a patch that will apply to the .32 and .33-longterm kernels?

Hi Greg,

I couldn't download the .32 and .33-longterm kernels from kernel.org because the full sources were not found on that server. Thus this patch is generated from the .32-mainline kernel and can be applied to the .32 and .33 kernels.

The .32 and .33 longterm kernels are part of the linux-stable tree on git.kernel.org, they are in their own branch. Please redo this against those trees, as I'm pretty sure that there will be conflicts, due to all of the different changes since the .0 releases.

From 86c78ef9cd861146a48e8a643601c165b0e80849 Mon Sep 17 00:00:00 2001

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 4 lines · Konrad Rzeszutek Wilk

Quoted 2 lines

Attached is the fix, verified in our env.

So.. you are asking for this upstream git commit to be back-ported to 2.6.32, right?

I'm curious --- is there a good reason why Xen users are using an upstream 2.6.32 kernel? If they are using a distro kernel, fine, but then the distro kernel should be providing the support. But at this point, 2.6.32 is so positively *ancient* that, I'm personally not interesting in providing free, unpaid distro support for users who aren't willing to either (a) pay $$$ and get a supported distro kernel, or (b) use a much more modern kernel. At this point, Guest and Host Xen support is available in 3.0 kernels, so there's really no excuse, right?

Since kernel.org 2.6.32 is still considered to be a maintained long-term-stable kernel, I keep the xen.git version up-to-date with stable-2.6.32 bugfixes and occasional separate Xen-specific fixes. But I'd really prefer to avoid having any non-Xen private changes in that tree, in favour of getting everything from upstream stable.

Do you not consider it worth continuing support of the 2.6.32 stable tree with respect to ext4?