diff mbox

[v3,RESEND] f2fs: add bug_on when f2fs_gc even fails to get one victim

Message ID 1510880948-154538-1-git-send-email-yunlong.song@huawei.com (mailing list archive)
State New, archived
Headers show

Commit Message

Yunlong Song Nov. 17, 2017, 1:09 a.m. UTC
This can help to find potential bugs on some corner case.

Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
---
 fs/f2fs/gc.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Yunlong Song Nov. 22, 2017, 3:50 a.m. UTC | #1
ping again...

On 2017/11/17 9:09, Yunlong Song wrote:
> This can help to find potential bugs on some corner case.
>
> Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
> ---
>   fs/f2fs/gc.c | 1 +
>   1 file changed, 1 insertion(+)
>
> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
> index 5d5bba4..c89128b 100644
> --- a/fs/f2fs/gc.c
> +++ b/fs/f2fs/gc.c
> @@ -1035,6 +1035,7 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
>   		goto stop;
>   	}
>   	if (!__get_victim(sbi, &segno, gc_type)) {
> +		f2fs_bug_on(sbi, !total_freed && has_not_enough_free_secs(sbi, 0, 0));
>   		ret = -ENODATA;
>   		goto stop;
>   	}
Chao Yu Nov. 23, 2017, 1:05 p.m. UTC | #2
On 2017/11/22 11:50, Yunlong Song wrote:
> ping again...
> 
> On 2017/11/17 9:09, Yunlong Song wrote:
>> This can help to find potential bugs on some corner case.

Could you test this patch with fstest suit? if there are any testcases
can trigger this bug_on, it will be better to fix them all together.

Thanks,

>>
>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
>> ---
>>   fs/f2fs/gc.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
>> index 5d5bba4..c89128b 100644
>> --- a/fs/f2fs/gc.c
>> +++ b/fs/f2fs/gc.c
>> @@ -1035,6 +1035,7 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
>>           goto stop;
>>       }
>>       if (!__get_victim(sbi, &segno, gc_type)) {
>> +        f2fs_bug_on(sbi, !total_freed && has_not_enough_free_secs(sbi, 0, 0));
>>           ret = -ENODATA;
>>           goto stop;
>>       }
>
Yunlong Song Nov. 25, 2017, 9:12 a.m. UTC | #3
Ok, I have found a panic with this bug_on for generic/027 today:

[ 5157.753224] F2FS-fs (loop2): Mounted with checkpoint version = 2e2
generic/027    [ 5168.741251] run fstests generic/027 at 2017-11-25 04:46:40
[ 5189.445989] F2FS-fs (loop3): Found nat_bits in checkpoint
[ 5189.510872] F2FS-fs (loop3): Mounted with checkpoint version = 165da00b
[ 5250.613849] ------------[ cut here ]------------
[ 5250.616840] kernel BUG at 
/opt/s00293685/src/kernel/jaegeuk/f2fs/fs/f2fs/gc.c:1038!
[ 5250.628467] invalid opcode: 0000 [#1] SMP
[ 5250.628467] Modules linked in:
[ 5250.628467] CPU: 7 PID: 3173 Comm: xfs_io Not tainted 4.14.0-rc4+ #128
[ 5250.628467] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), 
BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014
[ 5250.628467] task: ffff880130f2be80 task.stack: ffffc9000acd0000
[ 5250.628467] RIP: 0010:f2fs_gc+0x9da/0xa80
[ 5250.628467] RSP: 0018:ffffc9000acd3b48 EFLAGS: 00000246
[ 5250.628467] RAX: 000000000000001b RBX: ffff880134fa2648 RCX: 
ffff880134fa2f00
[ 5250.628467] RDX: 0000000000000006 RSI: 0000000000000200 RDI: 
0000000000000001
[ 5250.628467] RBP: ffffc9000acd3c38 R08: 000000000000001b R09: 
0000000000000001
[ 5250.628467] R10: 0000000000000000 R11: 0000000000000001 R12: 
0000000000000000
[ 5250.628467] R13: 0000000000000001 R14: ffff880138472000 R15: 
0000000000000002
[ 5250.628467] FS:  0000000001666880(0000) GS:ffff88013fdc0000(0000) 
knlGS:0000000000000000
[ 5250.628467] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 5250.628467] CR2: 00000000006ef120 CR3: 0000000130f48000 CR4: 
00000000000006e0
[ 5250.628467] Call Trace:
[ 5250.628467]  f2fs_balance_fs+0x13c/0x1f0
[ 5250.628467]  f2fs_create+0x146/0x260
[ 5250.628467]  path_openat+0xe31/0x12c0
[ 5250.628467]  do_filp_open+0x7e/0xd0
[ 5250.628467]  ? kmem_cache_alloc+0x92/0x160
[ 5250.628467]  ? getname_flags+0x4f/0x1f0
[ 5250.628467]  do_sys_open+0x115/0x1f0
[ 5250.628467]  SyS_open+0x1e/0x20
[ 5250.628467]  entry_SYSCALL_64_fastpath+0x13/0x94
[ 5250.628467] RIP: 0033:0x4171d0
[ 5250.628467] RSP: 002b:00007fff9a45b678 EFLAGS: 00000246 ORIG_RAX: 
0000000000000002
[ 5250.628467] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 
00000000004171d0
[ 5250.628467] RDX: 0000000000000180 RSI: 0000000000000042 RDI: 
00007fff9a45c1cb
[ 5250.628467] RBP: 00007fff9a45c1bf R08: 00007fff9a45b7f0 R09: 
0000000000000001
[ 5250.628467] R10: 00000000004bd8d3 R11: 0000000000000246 R12: 
0000000000000006
[ 5250.628467] R13: 00007fff9a45b830 R14: 0000000000000180 R15: 
0000000000000000
[ 5250.628467] Code: 00 bb c3 ff ff ff e9 2c fa ff ff 4d 8b 27 bb fb ff 
ff ff c7 44 24 7c 00 00 00 00 c7 84 24 80 00 00 00 00 00 00 00 e9 0c fa 
ff ff <0f> 0b 41 8b 96 fc 03 00 00 41 8b be f4 03 00 00 4c 8b 21 45 8b
[ 5250.628467] RIP: f2fs_gc+0x9da/0xa80 RSP: ffffc9000acd3b48
[ 5250.685538] ---[ end trace 00b8c84c59632b32 ]---

Let me fix it one by one.

On 2017/11/23 21:05, Chao Yu wrote:
> On 2017/11/22 11:50, Yunlong Song wrote:
>> ping again...
>>
>> On 2017/11/17 9:09, Yunlong Song wrote:
>>> This can help to find potential bugs on some corner case.
> Could you test this patch with fstest suit? if there are any testcases
> can trigger this bug_on, it will be better to fix them all together.
>
> Thanks,
>
>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
>>> ---
>>>    fs/f2fs/gc.c | 1 +
>>>    1 file changed, 1 insertion(+)
>>>
>>> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
>>> index 5d5bba4..c89128b 100644
>>> --- a/fs/f2fs/gc.c
>>> +++ b/fs/f2fs/gc.c
>>> @@ -1035,6 +1035,7 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
>>>            goto stop;
>>>        }
>>>        if (!__get_victim(sbi, &segno, gc_type)) {
>>> +        f2fs_bug_on(sbi, !total_freed && has_not_enough_free_secs(sbi, 0, 0));
>>>            ret = -ENODATA;
>>>            goto stop;
>>>        }
> .
>
diff mbox

Patch

diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index 5d5bba4..c89128b 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -1035,6 +1035,7 @@  int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
 		goto stop;
 	}
 	if (!__get_victim(sbi, &segno, gc_type)) {
+		f2fs_bug_on(sbi, !total_freed && has_not_enough_free_secs(sbi, 0, 0));
 		ret = -ENODATA;
 		goto stop;
 	}