MIME-version: 1.0
Content-type: text/plain; charset=UTF-8
Content-transfer-encoding: 8BIT
Message-id: <1375868596.22638.35.camel@kjgkr>
Subject: Re: [PATCH] f2fs: fix a deadlock in fsync
From: Jaegeuk Kim <jaegeuk.kim@samsung.com>
Reply-to: jaegeuk.kim@samsung.com
To: Jim Xu <jinuxstyle@gmail.com>
Cc: "linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-f2fs-devel@lists.sourceforge.net" 
	<linux-f2fs-devel@lists.sourceforge.net>
Date: Wed, 07 Aug 2013 18:43:16 +0900
In-reply-to: <7C98DE50-F64F-4FE8-8804-964DDD25561A@gmail.com>
References: <1375704124-22274-1-git-send-email-jinuxstyle@gmail.com>
 <1375793175.22638.18.camel@kjgkr>
 <7C98DE50-F64F-4FE8-8804-964DDD25561A@gmail.com>
Organization: Samsung
DLP-Filter: Pass
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 880
Lines: 22

Hi Jin,

2013-08-07 (수), 12:23 +0800, Jim Xu:
> Hi Kim,
> 
> The patch sounds good to me.
> 
> BTW, as in the deadlock situation I noticed, I did not find any thread was doing checkpoint at that moment. The bdi writeback thread and the fsync user thread was not. Neither was the background f2fs_gc thread because there had enough free segments. Meanwhile, another user thread was blocked in the f2fs_sync_file->sync_node_pages, which, I think, is not a factor contributing to the deadlock.

I indicated one possible scenario.
As you described, many fsync calls can also consume all the fs_lock too.
Thanks,

-- 
Jaegeuk Kim
Samsung

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/