linux-kernel - Re: [PATCH 2/2] block: hold queue lock while iterating in diskstats

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]

Message-ID: <Yelb4+r5KuV67tO0@T590>
Date:   Thu, 20 Jan 2022 20:56:03 +0800
From:   Ming Lei <ming.lei@...hat.com>
To:     Daniel Wagner <dwagner@...e.de>
Cc:     linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
        Jens Axboe <axboe@...nel.dk>, Hannes Reinecke <hare@...e.de>
Subject: Re: [PATCH 2/2] block: hold queue lock while iterating in
 diskstats_show

On Thu, Jan 20, 2022 at 11:52:48AM +0100, Daniel Wagner wrote:
> The request queues can be freed while we operate on them. Make sure we

It shouldn't be freed here.

> hold a reference when using blk_mq_queue_tag_busy_iter.
> 
> RIP: blk_mq_queue_tag_busy_iter

Any chance to pose the whole stack trace?

> Call Trace:
>  ? blk_mq_hctx_mark_pending
>  ? diskstats_show
>  ? blk_mq_hctx_mark_pending
>  blk_mq_in_flight
>  diskstats_show
>  ? klist_next
>  seq_read
>  proc_reg_read
>  vfs_read
>  ksys_read
>  do_syscall_64
>  entry_SYSCALL_64_after_hwframe
> 
> Signed-off-by: Daniel Wagner <dwagner@...e.de>
> ---
>  block/genhd.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/block/genhd.c b/block/genhd.c
> index c9d4386dd177..0e163055a4e6 100644
> --- a/block/genhd.c
> +++ b/block/genhd.c
> @@ -1167,10 +1167,13 @@ static int diskstats_show(struct seq_file *seqf, void *v)
>  		if (bdev_is_partition(hd) && !bdev_nr_sectors(hd))
>  			continue;
>  		part_stat_read_all(hd, &stat);
> +		if (blk_queue_enter(gp->queue, BLK_MQ_REQ_NOWAIT))
> +			continue;

The disk device's refcnt is supposed to be grabbed by class device
iterator code, so queue shouldn't be released in diskstats_show().

Not mention blk_queue_enter() won't prevent disk from being released...

thanks,
Ming