Linux SCSI subsystem development
 help / color / mirror / Atom feed
From: James Bottomley <jejb@linux.vnet.ibm.com>
To: Wei Fang <fangwei1@huawei.com>, martin.petersen@oracle.com
Cc: linux-scsi@vger.kernel.org, bart.vanassche@sandisk.com,
	emilne@redhat.com
Subject: Re: [PATCH v2] scsi: avoid a permanent stop of the scsi device's request queue
Date: Fri, 09 Dec 2016 08:02:42 -0800	[thread overview]
Message-ID: <1481299362.2403.19.camel@linux.vnet.ibm.com> (raw)
In-Reply-To: <1481276138-570-1-git-send-email-fangwei1@huawei.com>

On Fri, 2016-12-09 at 17:35 +0800, Wei Fang wrote:
> A scan work can run simultaneously with fc_remote_port_delete().
> If a scsi device is added to the ->__devices list in the scan work,
> it can be touched and will be blocked in scsi_target_block(), which
> will be called in fc_remote_port_delete(), and QUEUE_FLAG_STOPPED
> flag will be setted to the scsi device's request queue.
>  
> The scsi device is being setted to the SDEV_RUNNING state in
> scsi_sysfs_add_sdev() at the end of the scan work. When the remote
> port reappears, scsi_target_unblock() will be called, but the
> QUEUE_FLAG_STOPPED flag will not be cleared, since
> scsi_internal_device_unblock() ignores SCSI devices in SDEV_RUNNING
> state. It results in a permanent stop of the scsi device's request
> queue. Every requests sended to it will be blocked.

This is a bit unclear as a description of the problem

> Since the scsi device shouldn't be unblocked in this case, fix
> it by removing scsi_device_set_state() in scsi_sysfs_add_sdev().

So is this as a description of the solution, because the reader doesn't
know there's a prior place where SDEV_RUNNING was previously set.

How about

---
A race between scanning and fc_remote_port_delete() may result in a
permanent stop if the device gets blocked before scsi_sysfs_add_lun()
and unblocked after.  The reason is that blocking a device sets both
the SDEV_BLOCKED state and the QUEUE_FLAG_STOPPED.  However,
scsi_sysfs_add_sdev() unconditionally sets SDEV_RUNNING which causes
the device to be ignored by scsi_target_unblock() and thus never have
its QUEUE_FLAG_STOPPED cleared leading to a device which is apparently
running but has a stopped queue.

We actually have two places where SDEV_RUNNING is set: once in
scsi_add_lun() which respects the blocked flag and once in
scsi_sysfs_add_sdev() which doesn't.  Since the second set is entirely
spurious, simply remove it to fix the problem.
---

James


  parent reply	other threads:[~2016-12-09 16:02 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-12-09  9:35 [PATCH v2] scsi: avoid a permanent stop of the scsi device's request queue Wei Fang
2016-12-09 15:24 ` James Bottomley
2016-12-09 16:02 ` James Bottomley [this message]
2016-12-12  1:54   ` Wei Fang

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=1481299362.2403.19.camel@linux.vnet.ibm.com \
    --to=jejb@linux.vnet.ibm.com \
    --cc=bart.vanassche@sandisk.com \
    --cc=emilne@redhat.com \
    --cc=fangwei1@huawei.com \
    --cc=linux-scsi@vger.kernel.org \
    --cc=martin.petersen@oracle.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