On 03/03/2015 11:35 AM, David Sterba wrote:
> On Tue, Mar 03, 2015 at 07:02:58PM +0800, Liu Bo wrote:
>> On Mon, Mar 02, 2015 at 12:46:55PM -0500, Josef Bacik wrote:
>>> Dave could hit this assert consistently running btrfs/078.  This is because
>>> when we update the block groups we could truncate the free space, which would
>>> try to delete the csums for that range and dirty the csum root.  For this to
>>> happen we have to have already written out the csum root so it's kind of hard to
>>> hit this case.  This patch fixes this by changing the logic to only write the
>>> dirty block groups if the dirty_cowonly_roots list is empty.  This will get us
>>> the same effect as before since we add the extent root last, and will cover the
>>> case that we dirty some other root again but not the extent root.  Thanks,
>>
>> Free space inode is NODATASUM, so its searching csum tree in
>> __btrfs_free_extent() is really unnecessary, by skipping that, csum tree
>> won't be inserted into dirty_cowonly_roots list again, so it also passed btrfs/078,
>> at least on my box.
>
> I can hit the bug [1] even with Josef's patch, I'm can test your fix if you
> want.
>
> MKFS_OPTIONS  -- -O skinny-metadata,extref -m single -d single --mixed /dev/sdc1
> MOUNT_OPTIONS -- -o enospc_debug,space_cache,noatime /dev/sdc1 /mnt/a2
>
> on top of 3.19-rc5 with Chris' for-linus.
>

Can you try this on top of this patch and send me the logs?  Thanks,

Josef