LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Minchan Kim <minchan@kernel.org>
To: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	linux-kernel@vger.kernel.org, linux-mm@kvack.org,
	Dan Streetman <ddstreet@ieee.org>,
	Seth Jennings <sjennings@variantweb.net>,
	Nitin Gupta <ngupta@vflare.org>, Juneho Choi <juno.choi@lge.com>,
	Gunho Lee <gunho.lee@lge.com>,
	Luigi Semenzato <semenzato@google.com>,
	Jerome Marchand <jmarchan@redhat.com>
Subject: Re: [PATCH v1 01/10] zram: avoid calling of zram_meta_free under init_lock
Date: Fri, 23 Jan 2015 10:15:55 +0900	[thread overview]
Message-ID: <20150123011555.GB18918@blaptop> (raw)
In-Reply-To: <20150123010336.GA18918@blaptop>

On Fri, Jan 23, 2015 at 10:03:36AM +0900, Minchan Kim wrote:
> Hello,
> 
> On Wed, Jan 21, 2015 at 11:21:53PM +0900, Sergey Senozhatsky wrote:
> > On (01/21/15 15:14), Minchan Kim wrote:
> > > We don't need to call zram_meta_free under init_lock.
> > > What we need to prevent race is setting NULL into zram->meta
> > > (ie, init_done). This patch does it.
> > > 
> > > Signed-off-by: Minchan Kim <minchan@kernel.org>
> > > ---
> > >  drivers/block/zram/zram_drv.c | 5 +++--
> > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
> > > index 9250b3f..7e03d86 100644
> > > --- a/drivers/block/zram/zram_drv.c
> > > +++ b/drivers/block/zram/zram_drv.c
> > > @@ -719,6 +719,8 @@ static void zram_reset_device(struct zram *zram, bool reset_capacity)
> > >  	}
> > >  
> > >  	meta = zram->meta;
> > > +	zram->meta = NULL;
> > > +
> > >  	/* Free all pages that are still in this zram device */
> > >  	for (index = 0; index < zram->disksize >> PAGE_SHIFT; index++) {
> > >  		unsigned long handle = meta->table[index].handle;
> > > @@ -731,8 +733,6 @@ static void zram_reset_device(struct zram *zram, bool reset_capacity)
> > >  	zcomp_destroy(zram->comp);
> > >  	zram->max_comp_streams = 1;
> > >  
> > > -	zram_meta_free(zram->meta);
> > > -	zram->meta = NULL;
> > >  	/* Reset stats */
> > >  	memset(&zram->stats, 0, sizeof(zram->stats));
> > >  
> > > @@ -741,6 +741,7 @@ static void zram_reset_device(struct zram *zram, bool reset_capacity)
> > >  		set_capacity(zram->disk, 0);
> > >  
> > >  	up_write(&zram->init_lock);
> > > +	zram_meta_free(meta);
> > 
> > Hello,
> > 
> > since we detached ->meta from zram, this one doesn't really need
> > ->init_lock protection:
> > 
> > 	/* Free all pages that are still in this zram device */
> > 	for (index = 0; index < zram->disksize >> PAGE_SHIFT; index++) {
> > 		unsigned long handle = meta->table[index].handle;
> > 		if (!handle)
> > 			continue;
> > 
> > 		zs_free(meta->mem_pool, handle);
> > 	}
> > 
> > 
> > 	-ss
> 
> Good catch.
> 
> As well, we could move zcomp_destroy and memset(&zram->stats)
> out of the lock but zram_rw_page, ZRAM_ATTR_RO, disksize_show
> and orig_data_size_show have a race bug which access stats
> out of the lock so that it could show the stale vaule.
> Although it's not a significant, there is no reason to hesitate the fix. :)

Argh, sent wrong version.
zram->stats should be protected by the lock but other stat show
functions don't hold a lock so it's racy.

> 
> I will fix it. Thanks!
> 
> 
> > 
> > >  	/*
> > >  	 * Revalidate disk out of the init_lock to avoid lockdep splat.
> > > -- 
> > > 1.9.3
> > > 

  reply	other threads:[~2015-01-23  1:16 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-01-21  6:14 [PATCH v1 00/10] zsmalloc compaction support Minchan Kim
2015-01-21  6:14 ` [PATCH v1 01/10] zram: avoid calling of zram_meta_free under init_lock Minchan Kim
2015-01-21 14:21   ` Sergey Senozhatsky
2015-01-23  1:03     ` Minchan Kim
2015-01-23  1:15       ` Minchan Kim [this message]
2015-01-21  6:14 ` [PATCH v1 02/10] zsmalloc: decouple handle and object Minchan Kim
2015-01-26  2:53   ` Ganesh Mahendran
2015-01-27  3:27     ` Minchan Kim
2015-01-21  6:14 ` [PATCH v1 03/10] zsmalloc: implement reverse mapping Minchan Kim
2015-01-21  6:14 ` [PATCH v1 04/10] zsmalloc: factor out obj_[malloc|free] Minchan Kim
2015-01-21  6:14 ` [PATCH v1 05/10] zsmalloc: add status bit Minchan Kim
2015-01-21  6:14 ` [PATCH v1 06/10] zsmalloc: support compaction Minchan Kim
2015-01-21  6:14 ` [PATCH v1 07/10] zsmalloc: adjust ZS_ALMOST_FULL Minchan Kim
2015-01-21  6:14 ` [PATCH v1 08/10] zram: support compaction Minchan Kim
2015-01-21  6:14 ` [PATCH v1 09/10] zsmalloc: add fullness into stat Minchan Kim
2015-01-21  6:14 ` [PATCH v1 10/10] zsmalloc: record handle in page->private for huge object Minchan Kim

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=20150123011555.GB18918@blaptop \
    --to=minchan@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=ddstreet@ieee.org \
    --cc=gunho.lee@lge.com \
    --cc=jmarchan@redhat.com \
    --cc=juno.choi@lge.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=ngupta@vflare.org \
    --cc=semenzato@google.com \
    --cc=sergey.senozhatsky@gmail.com \
    --cc=sjennings@variantweb.net \
    --subject='Re: [PATCH v1 01/10] zram: avoid calling of zram_meta_free under init_lock' \
    /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

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).