Re: [PATCH v2 2/3] btrfs: use bio::remaining for async checksumming synchronization

[Qu Wenruo <wqu@suse.com>]

在 2026/9/5 08:11, Qu Wenruo 写道:
> 
> 
> 在 2026/9/3 15:53, Daniel Vacek 写道:
>> We can use bio::remaining counter to sync the offloaded checksuming.
>> As a result we can slim down the btrfs_bio structure by 24 bytes
>> and simplify the code a bit.
>>
>> $ pahole | diff
>> -    /* size: 328, cachelines: 6, members: 15 */
>> +    /* size: 304, cachelines: 5, members: 14 */
>>
>> Moreover this will allow us enabling async checksumming with encryption
>> where we need to checksum the bounce bio instead of our regular one
>> embedded in btrfs_bio. And so we need to extend it's lifetime. This is
>> the preffered way to do so.
>>
>> Signed-off-by: Daniel Vacek <neelx@suse.com>
>> Reviewed-by: Qu Wenruo <wqu@suse.com>
>> ---
>>
>> No change since v1
>> ---
>>   fs/btrfs/bio.c       | 4 ----
>>   fs/btrfs/bio.h       | 4 ----
>>   fs/btrfs/file-item.c | 6 ++----
>>   3 files changed, 2 insertions(+), 12 deletions(-)
>>
>> diff --git a/fs/btrfs/bio.c b/fs/btrfs/bio.c
>> index 19b4855969f5..771b7d598aee 100644
>> --- a/fs/btrfs/bio.c
>> +++ b/fs/btrfs/bio.c
>> @@ -103,7 +103,6 @@ static struct btrfs_bio *btrfs_split_bio(struct 
>> btrfs_fs_info *fs_info,
>>       bbio->can_use_append = orig_bbio->can_use_append;
>>       bbio->is_scrub = orig_bbio->is_scrub;
>>       bbio->is_remap = orig_bbio->is_remap;
>> -    bbio->async_csum = orig_bbio->async_csum;
>>       atomic_inc(&orig_bbio->pending_ios);
>>       return bbio;
>> @@ -114,9 +113,6 @@ void btrfs_bio_end_io(struct btrfs_bio *bbio, 
>> blk_status_t status)
>>       /* Make sure we're already in task context. */
>>       ASSERT(in_task());
>> -    if (bbio->async_csum)
>> -        wait_for_completion(&bbio->csum_done);
>> -
>>       bbio->bio.bi_status = status;
>>       if (bbio->bio.bi_pool == &btrfs_clone_bioset) {
>>           struct btrfs_bio *orig_bbio = bbio->private;
>> diff --git a/fs/btrfs/bio.h b/fs/btrfs/bio.h
>> index b7bd377a0162..bbf362b8668b 100644
>> --- a/fs/btrfs/bio.h
>> +++ b/fs/btrfs/bio.h
>> @@ -58,7 +58,6 @@ struct btrfs_bio {
>>               struct btrfs_ordered_extent *ordered;
>>               struct btrfs_ordered_sum *sums;
>>               struct work_struct csum_work;
>> -            struct completion csum_done;
>>               struct bvec_iter csum_saved_iter;
>>               u64 orig_physical;
>>               u64 orig_logical;
>> @@ -93,9 +92,6 @@ struct btrfs_bio {
>>       /* Whether the bio is coming from copy_remapped_data_io(). */
>>       bool is_remap:1;
>> -    /* Whether the csum generation for data write is async. */
>> -    bool async_csum:1;
>> -
>>       /* Whether the bio is written using zone append. */
>>       bool can_use_append:1;
>> diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
>> index 4a7681557ec1..0fed4e0d32d5 100644
>> --- a/fs/btrfs/file-item.c
>> +++ b/fs/btrfs/file-item.c
>> @@ -818,9 +818,8 @@ static void csum_one_bio_work(struct work_struct 
>> *work)
>>       struct btrfs_bio *bbio = container_of(work, struct btrfs_bio, 
>> csum_work);
>>       ASSERT(btrfs_op(&bbio->bio) == BTRFS_MAP_WRITE);
>> -    ASSERT(bbio->async_csum == true);
>>       csum_one_bio(bbio);
>> -    complete(&bbio->csum_done);
>> +    bio_endio(&bbio->bio);
>>   }
>>   /*
>> @@ -854,8 +853,7 @@ int btrfs_csum_one_bio(struct btrfs_bio *bbio, 
>> bool async)
>>           csum_one_bio(bbio);
>>           return 0;
>>       }
>> -    init_completion(&bbio->csum_done);
>> -    bbio->async_csum = true;
>> +    bio_inc_remaining(bio);
> 
> There seems to be a window where the bio can be finished before submission.
> 
> Firstly at btrfs_csum_one_bio() time, bio->bi_endio is not yet 
> initialized, it's only properly assigned at btrfs_submit_bio().
> 
> Then we queue the csum generation work.
> 
> But by some bad timing, the bio submission is delayed, we can have the 
> following sequence:
> 
>          Submission               |       Csum generation
> ---------------------------------+------------------------------------
> btrfs_submit_chunk()             |
> |- btrfs_csum_one_bio()          |
> |  |- bio_inc_remaining()        |
> |  |- schedule_work()            | csum_one_bio_work()
> |                                | |- bio_endio()
> |                                | Now the bio is finished, although
> |                                | bi_end_io is NULL, nothing real
> |                                | happened.

Damn it, I really need some tea before reviewing patches in the morning.

A bio has bi_remaining initialized to 1, so bio_inc_remaining() will 
change it to 2.

After bio_inc_remaining(), the next bio_endio() will not call 
bi_end_io(), but only decrease the bi_remaining back to 1.

So it won't call bi_end_io() in this case.

Please discard the above analysis.
> |- btrfs_submit_bio()
> 
> 
> This can be even worse, if the bio_endio() is called when 
> btrfs_submit_bio() has only partially setup the bio (e.g. bi_end_io() is 
> set, but bi_bdev is not set)
> 
> Not mention now it changed the context where the endio function is called.
> 
> Previously csum_one_bio_work() will never call bi_end_io() function, but 
> now it can.
> 
> The change has a much larger impact than I initially thought.
>>       INIT_WORK(&bbio->csum_work, csum_one_bio_work);
>>       schedule_work(&bbio->csum_work);
>>       return 0;
>