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:   Tue, 03 Nov 2020 13:28:10 +0800
From:   Can Guo <cang@codeaurora.org>
To:     Jaegeuk Kim <jaegeuk@kernel.org>
Cc:     linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org,
        kernel-team@android.com, alim.akhtar@samsung.com,
        avri.altman@wdc.com, bvanassche@acm.org
Subject: Re: [PATCH v4 1/5] scsi: ufs: atomic update for clkgating_enable
In-Reply-To: <20201028194352.GA3060274@google.com>
References: <20201026195124.363096-1-jaegeuk@kernel.org>
 <20201026195124.363096-2-jaegeuk@kernel.org>
 <20d1c2ca06e95beb207fd4ba1b61dc80@codeaurora.org>
 <20201027033311.GA1745317@google.com>
 <76df977d164683c7404d2dc702f2e5ad@codeaurora.org>
 <20201028194352.GA3060274@google.com>
Message-ID: <2c73eb84991227b7e42ba4c1d268d390@codeaurora.org>
User-Agent: Roundcube Webmail/1.3.9
Precedence: bulk

On 2020-10-29 03:43, Jaegeuk Kim wrote:
> On 10/27, Can Guo wrote:
>> On 2020-10-27 11:33, Jaegeuk Kim wrote:
>> > On 10/27, Can Guo wrote:
>> > > On 2020-10-27 03:51, Jaegeuk Kim wrote:
>> > > > From: Jaegeuk Kim <jaegeuk@google.com>
>> > > >
>> > > > When giving a stress test which enables/disables clkgating, we hit
>> > > > device
>> > > > timeout sometimes. This patch avoids subtle racy condition to address
>> > > > it.
>> > > >
>> > > > Note that, this requires a patch to address the device stuck by
>> > > > REQ_CLKS_OFF in
>> > > > __ufshcd_release().
>> > > >
>> > > > The fix is "scsi: ufs: avoid to call REQ_CLKS_OFF to CLKS_OFF".
>> > >
>> > > Why don't you just squash the fix into this one?
>> >
>> > I'm seeing this patch just revealed that problem.
>> 
>> That scenario (back to back calling of ufshcd_release()) only happens
>> when you stress the clkgate_enable sysfs node, so let's keep the fix
>> as one to make things simple. What do you think?
> 
> If we don't have this patch, do you think this issue won't happen at 
> all?
> 

At least I've never seen this scenario these years, otherwise I would 
have
put up a fix.

Thanks,

Can Guo.

>> 
>> Thanks,
>> 
>> Can Guo.
>> 
>> >
>> > >
>> > > Thanks,
>> > >
>> > > Can Guo.
>> > >
>> > > >
>> > > > Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
>> > > > ---
>> > > >  drivers/scsi/ufs/ufshcd.c | 12 ++++++------
>> > > >  1 file changed, 6 insertions(+), 6 deletions(-)
>> > > >
>> > > > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
>> > > > index cc8d5f0c3fdc..6c9269bffcbd 100644
>> > > > --- a/drivers/scsi/ufs/ufshcd.c
>> > > > +++ b/drivers/scsi/ufs/ufshcd.c
>> > > > @@ -1808,19 +1808,19 @@ static ssize_t
>> > > > ufshcd_clkgate_enable_store(struct device *dev,
>> > > >  		return -EINVAL;
>> > > >
>> > > >  	value = !!value;
>> > > > +
>> > > > +	spin_lock_irqsave(hba->host->host_lock, flags);
>> > > >  	if (value == hba->clk_gating.is_enabled)
>> > > >  		goto out;
>> > > >
>> > > > -	if (value) {
>> > > > -		ufshcd_release(hba);
>> > > > -	} else {
>> > > > -		spin_lock_irqsave(hba->host->host_lock, flags);
>> > > > +	if (value)
>> > > > +		__ufshcd_release(hba);
>> > > > +	else
>> > > >  		hba->clk_gating.active_reqs++;
>> > > > -		spin_unlock_irqrestore(hba->host->host_lock, flags);
>> > > > -	}
>> > > >
>> > > >  	hba->clk_gating.is_enabled = value;
>> > > >  out:
>> > > > +	spin_unlock_irqrestore(hba->host->host_lock, flags);
>> > > >  	return count;
>> > > >  }