[<prev] [next>] [day] [month] [year] [list]
Message-ID: <282da79a-eda7-f534-a6ed-8ac38fcc2c8b@web.de>
Date: Sat, 25 Apr 2020 21:30:22 +0200
From: Markus Elfring <Markus.Elfring@....de>
To: Xiyu Yang <xiyuyang19@...an.edu.cn>,
Xin Tan <tanxin.ctf@...il.com>, linux-mmc@...r.kernel.org,
linux-arm-kernel@...ts.infradead.org
Cc: linux-kernel@...r.kernel.org, kernel-janitors@...r.kernel.org,
Andreas Färber <afaerber@...e.de>,
Kangjie Lu <kjlu@....edu>,
Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>,
Ulf Hansson <ulf.hansson@...aro.org>,
Yuan Zhang <yuanxzhang@...an.edu.cn>
Subject: Re: [PATCH] mmc: owl-mmc: Fix dma_chan refcnt leak in owl_mmc_probe()
> Fix this issue by jumping to "err_put_dma" label when those error
> scenarios occur.
I suggest to reconsider your jump target selection.
…
> +++ b/drivers/mmc/host/owl-mmc.c
…
> @@ -643,19 +643,22 @@ static int owl_mmc_probe(struct platform_device *pdev)
> return 0;
>
> +err_put_dma:
> + if (owl_host->dma)
> + dma_release_channel(owl_host->dma);
I interpret the source code in the way that you would like to call
this function for the desired exception handling only after a call
of the function “dma_request_chan” succeeded.
Thus I would expect that the passed pointer will usually be still valid.
(Can the proposed null pointer check be omitted then?)
How do you think about the following change possibility?
+err_release_channel:
+ dma_release_channel(owl_host->dma);
Would you like to add the tag “Fixes” to the change description?
Regards,
Markus
Powered by blists - more mailing lists