linux-kernel - Re: [PATCH] dmaengine: ti: Fix runtime PM imbalance on error

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Sat, 22 Jan 2022 16:47:39 +0800 (GMT+08:00)
From:   刘永志 <lyz_cs@....edu.cn>
To:     péter ujfalusi <peter.ujfalusi@...il.com>
Cc:     vkoul@...nel.org, dmaengine@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dmaengine: ti: Fix runtime PM imbalance on error

> -----原始邮件-----
> 发件人: "Péter Ujfalusi" <peter.ujfalusi@...il.com>
> 发送时间: 2022-01-22 16:09:53 (星期六)
> 收件人: "Yongzhi Liu" <lyz_cs@....edu.cn>, vkoul@...nel.org
> 抄送: dmaengine@...r.kernel.org, linux-kernel@...r.kernel.org
> 主题: Re: [PATCH] dmaengine: ti: Fix runtime PM imbalance on error
> 
> 
> 
> On 1/16/22 13:31, Yongzhi Liu wrote:
> > pm_runtime_get_sync() increments the runtime PM usage counter even
> > when it returns an error code, thus a matching decrement is needed on
> > the error handling path to keep the counter balanced.
> 
> The patch is correct, however the commit message is a bit incorrect.
> 
> We are not adding any visible matching decrement, we are switching to
> pm_runtime_resume_and_get() which only increments the usage counter if
> pm_runtime_resume() is successful.
> Granted internally it does a pm_runtime_put_noidle() if resume call fails.
> 
> Switch to pm_runtime_resume_and_get() to keep the device's use caunt
> balanced?
> 

Thanks for your reply. I agree with your change and rewrite the commit message as follows.

[why]
pm_runtime_get_sync() increments the runtime PM usage counter even
when it returns an error code, however pm_runtime_resume_and_get() 
only increments the usage counter if pm_runtime_resume() is successful,
which granted internally does a pm_runtime_put_noidle() if resume call fails.

[how]
Fix this by switching to pm_runtime_resume_and_get() to keep the device's use count
balanced.


> > 
> > Signed-off-by: Yongzhi Liu <lyz_cs@....edu.cn>
> > ---
> >  drivers/dma/ti/edma.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/dma/ti/edma.c b/drivers/dma/ti/edma.c
> > index 08e47f4..a73f779 100644
> > --- a/drivers/dma/ti/edma.c
> > +++ b/drivers/dma/ti/edma.c
> > @@ -2398,9 +2398,9 @@ static int edma_probe(struct platform_device *pdev)
> >  	platform_set_drvdata(pdev, ecc);
> >  
> >  	pm_runtime_enable(dev);
> > -	ret = pm_runtime_get_sync(dev);
> > +	ret = pm_runtime_resume_and_get(dev);
> >  	if (ret < 0) {
> > -		dev_err(dev, "pm_runtime_get_sync() failed\n");
> > +		dev_err(dev, "pm_runtime_resume_and_get() failed\n");
> >  		pm_runtime_disable(dev);
> >  		return ret;
> >  	}
> 
> -- 
> Péter