From: Christoph Hellwig <hch@lst.de>
To: linux-scsi@vger.kernel.org
Cc: Hannes Reinecke <hare@suse.de>, Mike Snitzer <snitzer@redhat.com>,
Mike Christie <michaelc@cs.wisc.edu>,
"Martin K. Petersen" <martin.petersen@oracle.com>
Subject: [PATCH 9/9] scsi_dh: don't allow to detach device handlers at runtime
Date: Thu, 30 Apr 2015 19:32:31 +0200 [thread overview]
Message-ID: <1430415151-30948-10-git-send-email-hch@lst.de> (raw)
In-Reply-To: <1430415151-30948-1-git-send-email-hch@lst.de>
The I/O submission and completion pathes call into the device handler
without any synchronization agains detachment. So disallow detaching
device handlers at runtime.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/scsi/device_handler/scsi_dh.c | 31 ++++++++++++++-----------------
1 file changed, 14 insertions(+), 17 deletions(-)
diff --git a/drivers/scsi/device_handler/scsi_dh.c b/drivers/scsi/device_handler/scsi_dh.c
index 0d6ab33..1e945aa 100644
--- a/drivers/scsi/device_handler/scsi_dh.c
+++ b/drivers/scsi/device_handler/scsi_dh.c
@@ -142,17 +142,6 @@ static int scsi_dh_handler_attach(struct scsi_device *sdev,
}
/*
- * scsi_dh_handler_detach - Detach a device handler from a device
- * @sdev - SCSI device the device handler should be detached from
- */
-static void scsi_dh_handler_detach(struct scsi_device *sdev)
-{
- sdev->handler->detach(sdev);
- sdev_printk(KERN_NOTICE, sdev, "%s: Detached\n", sdev->handler->name);
- module_put(sdev->handler->module);
-}
-
-/*
* Functions for sysfs attribute 'dh_state'
*/
static ssize_t
@@ -179,8 +168,9 @@ store_dh_state(struct device *dev, struct device_attribute *attr,
/*
* Detach from a device handler
*/
- scsi_dh_handler_detach(sdev);
- err = 0;
+ sdev_printk(KERN_WARNING, sdev,
+ "can't detach handler %s!\n", buf);
+ err = -EINVAL;
} else if (!strncmp(buf, "activate", 8)) {
/*
* Activate a device handler
@@ -230,8 +220,11 @@ int scsi_dh_add_device(struct scsi_device *sdev)
void scsi_dh_remove_device(struct scsi_device *sdev)
{
- if (sdev->handler)
- scsi_dh_handler_detach(sdev);
+ if (sdev->handler) {
+ sdev->handler->detach(sdev);
+ module_put(sdev->handler->module);
+ }
+
device_remove_file(&sdev->sdev_gendev, &scsi_dh_state_attr);
}
@@ -393,15 +386,19 @@ int scsi_dh_attach(struct request_queue *q, const char *name)
goto out_put_device;
}
+ if (err)
+ return err;
+
if (sdev->handler) {
if (sdev->handler == scsi_dh)
goto out_put_device;
sdev_printk(KERN_WARNING, sdev,
- "replacing device handler %s with %s!, "
+ "can't replace device handler %s with %s!, "
"please fix the device handler tables.\n",
sdev->handler->name, name);
- scsi_dh_handler_detach(sdev);
+ err = -EINVAL;
+ goto out_put_device;
}
err = scsi_dh_handler_attach(sdev, scsi_dh);
--
1.9.1
next prev parent reply other threads:[~2015-04-30 17:34 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-04-30 17:32 integrate scsi_dh better into the scsi core Christoph Hellwig
2015-04-30 17:32 ` [PATCH 1/9] dm-mpath: check kstrdup return value in parse_hw_handler Christoph Hellwig
2015-04-30 18:35 ` Hannes Reinecke
2015-05-01 13:57 ` Mike Snitzer
2015-05-01 15:07 ` Martin K. Petersen
2015-05-01 15:29 ` Martin K. Petersen
2015-04-30 17:32 ` [PATCH 2/9] dm-mpath, scsi_dh: don't let dm detach device handlers Christoph Hellwig
2015-04-30 18:21 ` Mike Snitzer
2015-05-01 9:20 ` Christoph Hellwig
2015-05-01 7:10 ` Hannes Reinecke
2015-05-01 13:58 ` Mike Snitzer
2015-05-01 15:30 ` Martin K. Petersen
2015-04-30 17:32 ` [PATCH 3/9] dm-mpath: don't call scsi_dh_attach when we want to retain the attached handler Christoph Hellwig
2015-04-30 18:25 ` Mike Snitzer
2015-05-01 7:10 ` Hannes Reinecke
2015-05-01 15:34 ` Martin K. Petersen
2015-05-01 16:46 ` Christoph Hellwig
2015-04-30 17:32 ` [PATCH 4/9] dm-mpath, scsi_dh: request scsi_dh modules in scsi_dh, not dm-mpath Christoph Hellwig
2015-04-30 18:28 ` Mike Snitzer
2015-05-01 7:11 ` Hannes Reinecke
2015-05-01 13:59 ` Mike Snitzer
2015-05-01 15:35 ` Martin K. Petersen
2015-04-30 17:32 ` [PATCH 5/9] scsi_dh: integrate into the core SCSI code Christoph Hellwig
2015-05-01 7:13 ` Hannes Reinecke
2015-05-01 15:36 ` Martin K. Petersen
2015-05-04 7:50 ` Hannes Reinecke
2015-05-05 15:25 ` Christoph Hellwig
2015-04-30 17:32 ` [PATCH 6/9] scsi_dh: move device matching to the core code Christoph Hellwig
2015-04-30 18:32 ` Mike Snitzer
2015-05-01 7:15 ` Hannes Reinecke
2015-05-01 15:39 ` Martin K. Petersen
2015-04-30 17:32 ` [PATCH 7/9] scsi_dh: kill struct scsi_dh_data Christoph Hellwig
2015-05-01 7:17 ` Hannes Reinecke
2015-05-01 15:41 ` Martin K. Petersen
2015-04-30 17:32 ` [PATCH 8/9] scsi_dh: add a common helper to get a scsi_device from a request_queue Christoph Hellwig
2015-05-01 7:18 ` Hannes Reinecke
2015-05-01 15:45 ` Martin K. Petersen
2015-04-30 17:32 ` Christoph Hellwig [this message]
2015-05-01 7:19 ` [PATCH 9/9] scsi_dh: don't allow to detach device handlers at runtime Hannes Reinecke
2015-05-01 15:46 ` Martin K. Petersen
2015-04-30 18:31 ` integrate scsi_dh better into the scsi core Mike Snitzer
2015-05-01 15:47 ` Martin K. Petersen
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=1430415151-30948-10-git-send-email-hch@lst.de \
--to=hch@lst.de \
--cc=hare@suse.de \
--cc=linux-scsi@vger.kernel.org \
--cc=martin.petersen@oracle.com \
--cc=michaelc@cs.wisc.edu \
--cc=snitzer@redhat.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;
as well as URLs for NNTP newsgroup(s).