public inbox for linux-block@vger.kernel.org
 help / color / mirror / Atom feed
From: Ming Lei <ming.lei@redhat.com>
To: Caleb Sander Mateos <csander@purestorage.com>
Cc: Jens Axboe <axboe@kernel.dk>,
	linux-block@vger.kernel.org,
	Uday Shankar <ushankar@purestorage.com>
Subject: Re: [PATCH V2 02/16] ublk: look up ublk task via its pid in timeout handler
Date: Sun, 13 Jul 2025 10:31:08 +0800	[thread overview]
Message-ID: <aHMabA97fb9dCR6y@fedora> (raw)
In-Reply-To: <CADUfDZpvqsQUNrcGefQo+dakaC-aijnXfSYSAPV2sEtoUFfWKA@mail.gmail.com>

On Fri, Jul 11, 2025 at 09:20:31AM -0400, Caleb Sander Mateos wrote:
> On Mon, Jul 7, 2025 at 9:18 PM Ming Lei <ming.lei@redhat.com> wrote:
> >
> > Look up ublk process via its pid in timeout handler, so we can avoid to
> > touch io->task, because it is fragile to touch task structure.
> >
> > It is fine to kill ublk server process and this way is simpler.
> >
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> >  drivers/block/ublk_drv.c | 21 +++++++++++++--------
> >  1 file changed, 13 insertions(+), 8 deletions(-)
> >
> > diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
> > index 65daa6ed3a8e..d7b5ee96978a 100644
> > --- a/drivers/block/ublk_drv.c
> > +++ b/drivers/block/ublk_drv.c
> > @@ -1367,14 +1367,19 @@ static void ublk_queue_cmd_list(struct ublk_io *io, struct rq_list *l)
> >  static enum blk_eh_timer_return ublk_timeout(struct request *rq)
> >  {
> >         struct ublk_queue *ubq = rq->mq_hctx->driver_data;
> > -       struct ublk_io *io = &ubq->ios[rq->tag];
> > -
> > -       if (ubq->flags & UBLK_F_UNPRIVILEGED_DEV) {
> > -               send_sig(SIGKILL, io->task, 0);
> > -               return BLK_EH_DONE;
> > -       }
> > -
> > -       return BLK_EH_RESET_TIMER;
> > +       struct task_struct *p;
> > +       struct pid *pid;
> > +
> > +       if (!(ubq->flags & UBLK_F_UNPRIVILEGED_DEV))
> > +               return BLK_EH_RESET_TIMER;
> > +
> > +       rcu_read_lock();
> > +       pid = find_vpid(ubq->dev->dev_info.ublksrv_pid);
> 
> It looks like ublksrv_pid is set based on whatever the ublk server
> provides in the UBLK_U_CMD_START_DEV/UBLK_U_CMD_END_USER_RECOVERY
> command. I don't see any validation that this is actually the ublk
> server's PID. So couldn't a buggy/malicious ublk server that doesn't
> provide its own PID cause ublk_timeout() to kill some other process
> and leave the ublk I/O pending forever?

Good catch!

Looks we can store the real ublk server pid in ublk_ch_open(), then validate
the passed `ublksrv_pid` in ublk_ctrl_start_dev(), or even always use
the real ublk server pid stored from  ublk_ch_open().

Will do it in next version.


Thanks,
Ming


  reply	other threads:[~2025-07-13  2:31 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-08  1:17 [PATCH V2 00/16] ublk: cleanup for supporting batch IO command Ming Lei
2025-07-08  1:17 ` [PATCH V2 01/16] ublk: move fake timeout logic into __ublk_complete_rq() Ming Lei
2025-07-08  1:17 ` [PATCH V2 02/16] ublk: look up ublk task via its pid in timeout handler Ming Lei
2025-07-11 13:20   ` Caleb Sander Mateos
2025-07-13  2:31     ` Ming Lei [this message]
2025-07-08  1:17 ` [PATCH V2 03/16] ublk: let ublk_fill_io_cmd() cover more things Ming Lei
2025-07-11 13:25   ` Caleb Sander Mateos
2025-07-13  2:34     ` Ming Lei
2025-07-08  1:17 ` [PATCH V2 04/16] ublk: avoid to pass `struct ublksrv_io_cmd *` to ublk_commit_and_fetch() Ming Lei
2025-07-08  1:17 ` [PATCH V2 05/16] ublk: move auto buffer register handling into one dedicated helper Ming Lei
2025-07-11 13:47   ` Caleb Sander Mateos
2025-07-13  2:38     ` Ming Lei
2025-07-08  1:17 ` [PATCH V2 06/16] ublk: store auto buffer register data into `struct ublk_io` Ming Lei
2025-07-08  1:17 ` [PATCH V2 07/16] ublk: add helper ublk_check_fetch_buf() Ming Lei
2025-07-08  1:17 ` [PATCH V2 08/16] ublk: remove ublk_commit_and_fetch() Ming Lei
2025-07-08  1:17 ` [PATCH V2 09/16] ublk: pass 'const struct ublk_io *' to ublk_[un]map_io() Ming Lei
2025-07-08  1:17 ` [PATCH V2 10/16] selftests: ublk: remove `tag` parameter of ->tgt_io_done() Ming Lei
2025-07-08  1:17 ` [PATCH V2 11/16] selftests: ublk: pass 'ublk_thread *' to ->queue_io() and ->tgt_io_done() Ming Lei
2025-07-08  1:17 ` [PATCH V2 12/16] selftests: ublk: pass 'ublk_thread *' to more common helpers Ming Lei
2025-07-08  1:17 ` [PATCH V2 13/16] selftests: ublk: remove ublk queue self-defined flags Ming Lei
2025-07-08  1:17 ` [PATCH V2 14/16] selftests: ublk: improve flags naming Ming Lei
2025-07-08  1:17 ` [PATCH V2 15/16] selftests: ublk: add helper ublk_handle_uring_cmd() for handle ublk command Ming Lei
2025-07-08  1:17 ` [PATCH V2 16/16] selftests: ublk: add utils.h Ming Lei

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aHMabA97fb9dCR6y@fedora \
    --to=ming.lei@redhat.com \
    --cc=axboe@kernel.dk \
    --cc=csander@purestorage.com \
    --cc=linux-block@vger.kernel.org \
    --cc=ushankar@purestorage.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox