linux-kernel - Re: [PATCH v2] block: fix trace completion for chained bio

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 23 Mar 2021 08:22:59 +1100
From:   NeilBrown <neilb@...e.de>
To:     edwardh <edwardh@...ology.com>, axboe@...nel.dk, neilb@...e.com
Cc:     linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
        s3t@...ology.com, bingjingc@...ology.com, cccheng@...ology.com,
        Edward Hsieh <edwardh@...ology.com>
Subject: Re: [PATCH v2] block: fix trace completion for chained bio

On Wed, Mar 03 2021, edwardh wrote:

> From: Edward Hsieh <edwardh@...ology.com>
>
> For chained bio, trace_block_bio_complete in bio_endio is currently called
> only by the parent bio once upon all chained bio completed.
> However, the sector and size for the parent bio are modified in bio_split.
> Therefore, the size and sector of the complete events might not match the
> queue events in blktrace.
>
> The original fix of bio completion trace <fbbaf700e7b1> ("block: trace
> completion of all bios.") wants multiple complete events to correspond
> to one queue event but missed this.
>
> md/raid5 read with bio cross chunks can reproduce this issue.
>
> To fix, move trace completion into the loop for every chained bio to call.

Thanks.  I think this is correct as far as tracing goes.
However the code still looks a bit odd.

The comment for the handling of bio_chain_endio suggests that the *only*
purpose for that is to avoid deep recursion.  That suggests it should be
at the end of the function.
As it is blk_throtl_bio_endio() and bio_unint() are only called on the
last bio in a chain.
That seems wrong.

I'd be more comfortable if the patch moved the bio_chain_endio()
handling to the end, after all of that.
So the function would end.

if (bio->bi_end_io == bio_chain_endio) {
   bio = __bio_chain_endio(bio);
   goto again;
} else if (bio->bi_end_io)
   bio->bi_end_io(bio);

Jens:  can you see any reason why that functions must only be called on
the last bio in the chain?

Thanks,
NeilBrown


>
> Fixes: fbbaf700e7b1 ("block: trace completion of all bios.")
> Reviewed-by: Wade Liang <wadel@...ology.com>
> Reviewed-by: BingJing Chang <bingjingc@...ology.com>
> Signed-off-by: Edward Hsieh <edwardh@...ology.com>
> ---
>  block/bio.c | 13 ++++++-------
>  1 file changed, 6 insertions(+), 7 deletions(-)
>
> diff --git a/block/bio.c b/block/bio.c
> index a1c4d29..2ff72cb 100644
> --- a/block/bio.c
> +++ b/block/bio.c
> @@ -1397,8 +1397,7 @@ static inline bool bio_remaining_done(struct bio *bio)
>   *
>   *   bio_endio() can be called several times on a bio that has been chained
>   *   using bio_chain().  The ->bi_end_io() function will only be called the
> - *   last time.  At this point the BLK_TA_COMPLETE tracing event will be
> - *   generated if BIO_TRACE_COMPLETION is set.
> + *   last time.
>   **/
>  void bio_endio(struct bio *bio)
>  {
> @@ -1411,6 +1410,11 @@ void bio_endio(struct bio *bio)
>  	if (bio->bi_bdev)
>  		rq_qos_done_bio(bio->bi_bdev->bd_disk->queue, bio);
>  
> +	if (bio->bi_bdev && bio_flagged(bio, BIO_TRACE_COMPLETION)) {
> +		trace_block_bio_complete(bio->bi_bdev->bd_disk->queue, bio);
> +		bio_clear_flag(bio, BIO_TRACE_COMPLETION);
> +	}
> +
>  	/*
>  	 * Need to have a real endio function for chained bios, otherwise
>  	 * various corner cases will break (like stacking block devices that
> @@ -1424,11 +1428,6 @@ void bio_endio(struct bio *bio)
>  		goto again;
>  	}
>  
> -	if (bio->bi_bdev && bio_flagged(bio, BIO_TRACE_COMPLETION)) {
> -		trace_block_bio_complete(bio->bi_bdev->bd_disk->queue, bio);
> -		bio_clear_flag(bio, BIO_TRACE_COMPLETION);
> -	}
> -
>  	blk_throtl_bio_endio(bio);
>  	/* release cgroup info */
>  	bio_uninit(bio);
> -- 
> 2.7.4

Download attachment "signature.asc" of type "application/pgp-signature" (854 bytes)