linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Jun'ichi Nomura" <j-nomura@ce.jp.nec.com>
To: Tejun Heo <tj@kernel.org>
Cc: Naveen Goswamy <naveen.goswamy@polymtl.ca>,
	Jens Axboe <axboe@kernel.dk>,
	James Bottomley <James.Bottomley@HansenPartnership.com>,
	Stefan Richter <stefanr@s5r6.in-berlin.de>,
	Dave Jones <davej@redhat.com>,
	linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org
Subject: Re: Kernel crashing on eject SD card
Date: Thu, 16 Feb 2012 10:26:38 +0900	[thread overview]
Message-ID: <4F3C5B4E.3080606@ce.jp.nec.com> (raw)
In-Reply-To: <20120215172617.GB24986@google.com>

Hi,

Thank you for review and comments.

On 02/16/12 02:26, Tejun Heo wrote:
> On Wed, Feb 15, 2012 at 11:56:19AM +0900, Jun'ichi Nomura wrote:
>> +int invalidate_partitions(struct gendisk *disk, struct block_device *bdev)
>> +{
>> +	int res;
>> +
>> +	res = drop_partitions(disk, bdev);
>> +	if (res)
>> +		return res;
>> +
> 
> Hmmm... shouldn't we have set_capacity(disk, 0) here?

Added.
I wasn't sure whether I should leave it to drivers.
But it seems capacity 0 for ENOMEDIUM device is reasonable.

>> +	check_disk_size_change(disk, bdev);
>> +	bdev->bd_invalidated = 0;
>> +	/* tell userspace that the media / partition table may have changed */
>> +	kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
> 
> Also, we really shouldn't be generating KOBJ_CHANGE after every
> -ENOMEDIUM open.  This can easily lead to infinite loop.  We should
> generate this iff we actually dropped partitions && modified the size.

invalidate_partitions() is called only when bd_invalidated is set.
So KOBJ_CHANGE is not raised for every ENOMEDIUM open.

I put it explicit in the function to make it safer for
possible misuse.

How about this?

---------------------------------------------------------
Do not call drivers when invalidating partitions for -ENOMEDIUM

When a scsi driver returns -ENOMEDIUM for open(),
__blkdev_get() calls rescan_partitions(), which ends up calling
sd_revalidate_disk() without getting a refcount of scsi_device.

That could lead to oops like this:

  process A                  process B
  ----------------------------------------------
  sys_open
    __blkdev_get
      sd_open
        returns -ENOMEDIUM
                             scsi_remove_device
                               <scsi_device torn down>
      rescan_partitions
        sd_revalidate_disk
          <oops>

Oopses are reported here:
http://marc.info/?l=linux-scsi&m=132388619710052

This patch separates the partition invalidation from rescan_partitions()
and use it for -ENOMEDIUM case. 

Index: linux-3.3/block/partition-generic.c
===================================================================
--- linux-3.3.orig/block/partition-generic.c	2012-02-15 09:00:25.147293790 +0900
+++ linux-3.3/block/partition-generic.c	2012-02-16 10:48:22.257680685 +0900
@@ -389,17 +389,11 @@ static bool disk_unlock_native_capacity(
 	}
 }
 
-int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
+static int drop_partitions(struct gendisk *disk, struct block_device *bdev)
 {
-	struct parsed_partitions *state = NULL;
 	struct disk_part_iter piter;
 	struct hd_struct *part;
-	int p, highest, res;
-rescan:
-	if (state && !IS_ERR(state)) {
-		kfree(state);
-		state = NULL;
-	}
+	int res;
 
 	if (bdev->bd_part_count)
 		return -EBUSY;
@@ -412,6 +406,24 @@ rescan:
 		delete_partition(disk, part->partno);
 	disk_part_iter_exit(&piter);
 
+	return 0;
+}
+
+int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
+{
+	struct parsed_partitions *state = NULL;
+	struct hd_struct *part;
+	int p, highest, res;
+rescan:
+	if (state && !IS_ERR(state)) {
+		kfree(state);
+		state = NULL;
+	}
+
+	res = drop_partitions(disk, bdev);
+	if (res)
+		return res;
+
 	if (disk->fops->revalidate_disk)
 		disk->fops->revalidate_disk(disk);
 	check_disk_size_change(disk, bdev);
@@ -515,6 +527,26 @@ rescan:
 	return 0;
 }
 
+int invalidate_partitions(struct gendisk *disk, struct block_device *bdev)
+{
+	int res;
+
+	if (!bdev->bd_invalidated)
+		return 0;
+
+	res = drop_partitions(disk, bdev);
+	if (res)
+		return res;
+
+	set_capacity(disk, 0);
+	check_disk_size_change(disk, bdev);
+	bdev->bd_invalidated = 0;
+	/* tell userspace that the media / partition table may have changed */
+	kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
+
+	return 0;
+}
+
 unsigned char *read_dev_sector(struct block_device *bdev, sector_t n, Sector *p)
 {
 	struct address_space *mapping = bdev->bd_inode->i_mapping;
Index: linux-3.3/include/linux/genhd.h
===================================================================
--- linux-3.3.orig/include/linux/genhd.h	2012-02-09 12:21:53.000000000 +0900
+++ linux-3.3/include/linux/genhd.h	2012-02-16 10:47:43.783681813 +0900
@@ -596,6 +596,7 @@ extern char *disk_name (struct gendisk *
 
 extern int disk_expand_part_tbl(struct gendisk *disk, int target);
 extern int rescan_partitions(struct gendisk *disk, struct block_device *bdev);
+extern int invalidate_partitions(struct gendisk *disk, struct block_device *bdev);
 extern struct hd_struct * __must_check add_partition(struct gendisk *disk,
 						     int partno, sector_t start,
 						     sector_t len, int flags,
Index: linux-3.3/fs/block_dev.c
===================================================================
--- linux-3.3.orig/fs/block_dev.c	2012-02-09 12:21:53.000000000 +0900
+++ linux-3.3/fs/block_dev.c	2012-02-16 10:47:52.602681441 +0900
@@ -1183,8 +1183,12 @@ static int __blkdev_get(struct block_dev
 			 * The latter is necessary to prevent ghost
 			 * partitions on a removed medium.
 			 */
-			if (bdev->bd_invalidated && (!ret || ret == -ENOMEDIUM))
-				rescan_partitions(disk, bdev);
+			if (bdev->bd_invalidated) {
+				if (!ret)
+					rescan_partitions(disk, bdev);
+				else if (ret == -ENOMEDIUM)
+					invalidate_partitions(disk, bdev);
+			}
 			if (ret)
 				goto out_clear;
 		} else {
@@ -1214,8 +1218,12 @@ static int __blkdev_get(struct block_dev
 			if (bdev->bd_disk->fops->open)
 				ret = bdev->bd_disk->fops->open(bdev, mode);
 			/* the same as first opener case, read comment there */
-			if (bdev->bd_invalidated && (!ret || ret == -ENOMEDIUM))
-				rescan_partitions(bdev->bd_disk, bdev);
+			if (bdev->bd_invalidated) {
+				if (!ret)
+					rescan_partitions(bdev->bd_disk, bdev);
+				else if (ret == -ENOMEDIUM)
+					invalidate_partitions(bdev->bd_disk, bdev);
+			}
 			if (ret)
 				goto out_unlock_bdev;
 		}

  reply	other threads:[~2012-02-16  1:28 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-02-08  0:19 Kernel crashing on eject SD card Naveen Goswamy
2012-02-12 21:08 ` Stefan Richter
2012-02-12 21:20   ` Stefan Richter
2012-02-13  1:46     ` Naveen Goswamy
2012-02-13  2:18     ` Dave Jones
2012-02-13 17:40       ` Naveen Goswamy
2012-02-14 11:14         ` Jun'ichi Nomura
2012-02-14 13:31           ` Stefan Richter
2012-02-14 16:28           ` Tejun Heo
2012-02-15  2:56             ` Jun'ichi Nomura
2012-02-15 17:26               ` Tejun Heo
2012-02-16  1:26                 ` Jun'ichi Nomura [this message]
2012-02-16 16:36                   ` Tejun Heo
2012-03-01 18:58                   ` Luis Henriques
2012-03-02  0:12                     ` Jun'ichi Nomura
2012-03-02  9:35                       ` Luis Henriques
2012-03-02  9:41                       ` Jens Axboe

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=4F3C5B4E.3080606@ce.jp.nec.com \
    --to=j-nomura@ce.jp.nec.com \
    --cc=James.Bottomley@HansenPartnership.com \
    --cc=axboe@kernel.dk \
    --cc=davej@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=naveen.goswamy@polymtl.ca \
    --cc=stefanr@s5r6.in-berlin.de \
    --cc=tj@kernel.org \
    /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).