linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jan Kara <jack@suse.cz>
To: Christoph Hellwig <hch@lst.de>
Cc: Qian Cai <quic_qiancai@quicinc.com>, Jens Axboe <axboe@kernel.dk>,
	Tejun Heo <tj@kernel.org>, Jan Kara <jack@suse.cz>,
	linux-block@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>,
	cgroups@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: move the bdi from the request_queue to the gendisk
Date: Wed, 11 Aug 2021 13:25:14 +0200	[thread overview]
Message-ID: <20210811112514.GC14725@quack2.suse.cz> (raw)
In-Reply-To: <20210810200256.GA30809@lst.de>

On Tue 10-08-21 22:02:56, Christoph Hellwig wrote:
> On Tue, Aug 10, 2021 at 03:36:39PM -0400, Qian Cai wrote:
> > 
> > 
> > On 8/9/2021 10:17 AM, Christoph Hellwig wrote:
> > > Hi Jens,
> > > 
> > > this series moves the pointer to the bdi from the request_queue
> > > to the bdi, better matching the life time rules of the different
> > > objects.
> > 
> > Reverting this series fixed an use-after-free in bdev_evict_inode().
> 
> Please try the patch below as a band-aid.  Although the proper fix is
> that non-default bdi_writeback structures grab a reference to the bdi,
> as this was a landmine that might have already caused spurious issues
> before.

Well, non-default bdi_writeback structures do hold bdi reference - see
wb_exit() which drops the reference. I think the problem rather was that a
block device's inode->i_wb was pointing to the default bdi_writeback
structure and that got freed after bdi_put() before block device inode was
shutdown through bdput()... So what I think we need is that if the inode
references the default writeback structure, it actually holds a reference
to the bdi.

								Honza
> 
> diff --git a/block/genhd.c b/block/genhd.c
> index f8def1129501..2e4a9d187196 100644
> --- a/block/genhd.c
> +++ b/block/genhd.c
> @@ -1086,7 +1086,6 @@ static void disk_release(struct device *dev)
>  
>  	might_sleep();
>  
> -	bdi_put(disk->bdi);
>  	if (MAJOR(dev->devt) == BLOCK_EXT_MAJOR)
>  		blk_free_ext_minor(MINOR(dev->devt));
>  	disk_release_events(disk);
> diff --git a/fs/block_dev.c b/fs/block_dev.c
> index 7c969f81327a..c6087dbae6cf 100644
> --- a/fs/block_dev.c
> +++ b/fs/block_dev.c
> @@ -849,11 +849,15 @@ static void init_once(void *data)
>  
>  static void bdev_evict_inode(struct inode *inode)
>  {
> +	struct block_device *bdev = I_BDEV(inode);
> +
>  	truncate_inode_pages_final(&inode->i_data);
>  	invalidate_inode_buffers(inode); /* is it needed here? */
>  	clear_inode(inode);
>  	/* Detach inode from wb early as bdi_put() may free bdi->wb */
>  	inode_detach_wb(inode);
> +	if (!bdev_is_partition(bdev))
> +		bdi_put(bdev->bd_disk->bdi);
>  }
>  
>  static const struct super_operations bdev_sops = {
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

  parent reply	other threads:[~2021-08-11 11:25 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-09 14:17 move the bdi from the request_queue to the gendisk Christoph Hellwig
2021-08-09 14:17 ` [PATCH 1/5] mm: hide laptop_mode_wb_timer entirely behind the BDI API Christoph Hellwig
2021-08-09 14:33   ` Johannes Thumshirn
2021-08-09 15:10   ` Jan Kara
2021-08-10 21:56   ` Guenter Roeck
2021-08-11  5:22     ` Christoph Hellwig
2021-08-09 14:17 ` [PATCH 2/5] block: pass a gendisk to blk_queue_update_readahead Christoph Hellwig
2021-08-09 14:35   ` Johannes Thumshirn
2021-08-09 15:17   ` Jan Kara
2021-08-09 14:17 ` [PATCH 3/5] block: add a queue_has_disk helper Christoph Hellwig
2021-08-09 14:37   ` Johannes Thumshirn
2021-08-09 15:18   ` Jan Kara
2021-08-09 14:17 ` [PATCH 4/5] block: move the bdi from the request_queue to the gendisk Christoph Hellwig
2021-08-09 14:38   ` Johannes Thumshirn
2021-08-09 15:47   ` Jan Kara
2021-08-09 17:57     ` Jens Axboe
2021-08-09 21:29       ` Jan Kara
2021-08-10 16:44     ` Christoph Hellwig
2021-10-14 14:31   ` [sparc64] kernel OOPS (was: [PATCH 4/5] block: move the bdi from the request_queue to the gendisk) Anatoly Pugachev
2021-10-14 14:32     ` Christoph Hellwig
2021-10-14 20:27       ` Anatoly Pugachev
2021-08-09 14:17 ` [PATCH 5/5] block: remove the bd_bdi in struct block_device Christoph Hellwig
2021-08-09 14:55   ` Johannes Thumshirn
2021-08-09 15:49   ` Jan Kara
2021-08-09 21:42 ` move the bdi from the request_queue to the gendisk Jens Axboe
2021-08-10 19:36 ` Qian Cai
2021-08-10 20:02   ` Christoph Hellwig
2021-08-11  2:28     ` Qian Cai
2021-08-11 11:25     ` Jan Kara [this message]
2021-08-11 11:51       ` Christoph Hellwig
2021-08-11 12:47         ` Jan Kara

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=20210811112514.GC14725@quack2.suse.cz \
    --to=jack@suse.cz \
    --cc=akpm@linux-foundation.org \
    --cc=axboe@kernel.dk \
    --cc=cgroups@vger.kernel.org \
    --cc=hch@lst.de \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=quic_qiancai@quicinc.com \
    --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).