Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Sender: cang=codeaurora.org@mg.codeaurora.org
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date:   Wed, 29 Sep 2021 11:31:16 +0800
From:   Can Guo <cang@codeaurora.org>
To:     Bart Van Assche <bvanassche@acm.org>
Cc:     asutoshd@codeaurora.org, nguyenb@codeaurora.org,
        hongwus@codeaurora.org, ziqichen@codeaurora.org,
        linux-scsi@vger.kernel.org, kernel-team@android.com,
        Alim Akhtar <alim.akhtar@samsung.com>,
        Avri Altman <avri.altman@wdc.com>,
        "James E.J. Bottomley" <jejb@linux.ibm.com>,
        "Martin K. Petersen" <martin.petersen@oracle.com>,
        Bean Huo <beanhuo@micron.com>,
        Stanley Chu <stanley.chu@mediatek.com>,
        Jaegeuk Kim <jaegeuk@kernel.org>,
        Adrian Hunter <adrian.hunter@intel.com>,
        open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] scsi: ufs: Fix a possible dead lock in clock scaling
In-Reply-To: <cc9cb9e7-68bd-3bfa-9310-5fbf99a86544@acm.org>
References: <1631843521-2863-1-git-send-email-cang@codeaurora.org>
 <cc9cb9e7-68bd-3bfa-9310-5fbf99a86544@acm.org>
Message-ID: <fbc4d03a07f03fe4fbe697813111471f@codeaurora.org>
User-Agent: Roundcube Webmail/1.3.9
Precedence: bulk

Hi Bart,

On 2021-09-18 01:27, Bart Van Assche wrote:
> On 9/16/21 6:51 PM, Can Guo wrote:
>> Assume a scenario where task A and B call ufshcd_devfreq_scale()
>> simultaneously. After task B calls downgrade_write() [1], but before 
>> it
>> calls down_read() [3], if task A calls down_write() [2], when task B 
>> calls
>> down_read() [3], it will lead to dead lock.
> 
> Something is wrong with the above description. The downgrade_write() 
> call is
> not followed by down_read() but by up_read(). Additionally, I don't see 
> how
> concurrent calls of ufshcd_devfreq_scale() could lead to a deadlock.

As mentioned in the commit msg, the down_read() [3] is from 
ufshcd_wb_ctrl().

Task A -
down_write [2]
ufshcd_clock_scaling_prepare
ufshcd_devfreq_scale
ufshcd_clkscale_enable_store

Task B -
down_read [3]
ufshcd_exec_dev_cmd
ufshcd_query_flag
ufshcd_wb_ctrl
downgrade_write [1]
ufshcd_devfreq_scale
ufshcd_devfreq_target
devfreq_set_target
update_devfreq
devfreq_performance_handler
governor_store


> If one thread calls downgrade_write() and another thread calls 
> down_write()
> immediately, that down_write() call will block until the other thread 
> has called up_read()
> without triggering a deadlock.

Since the down_write() caller is blocked, the down_read() caller, which 
comes after
down_write(), is blocked too, no? downgrade_write() keeps lock owner as 
it is, but
it does not change the fact that readers and writers can be blocked by 
each other.

> 
> Thanks,
> 
> Bart.

Thanks,

Can.