linux-kernel - RE: [PATCH V3 1/2] i2c: qup: Fix broken dma when CONFIG_DEBUG

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

Message-ID: <002201d1b79a$22862bc0$67928340$@codeaurora.org>
Date:	Fri, 27 May 2016 03:31:13 +0530
From:	"Sricharan" <sricharan@...eaurora.org>
To:	"'Wolfram Sang'" <wsa@...-dreams.de>
Cc:	<linux-arm-msm@...r.kernel.org>, <ntelkar@...eaurora.org>,
	<agross@...eaurora.org>, <linux-kernel@...r.kernel.org>,
	<nkaje@...eaurora.org>, <absahu@...eaurora.org>,
	<linux-i2c@...r.kernel.org>, <galak@...eaurora.org>,
	<andy.gross@...aro.org>, <linux-arm-kernel@...ts.infradead.org>
Subject: RE: [PATCH V3 1/2] i2c: qup: Fix broken dma when CONFIG_DEBUG_SG is enabled

Hi,

>> sg_set_buf expects that the buf parameter passed in should be from
>> lowmem and a valid pageframe. This is not true for pages from
>> dma_alloc_coherent which can be carveouts, hence the check fails.
>
>OK, given you mean dma_pool_alloc here, the check fails for the
>pageframe because of the pool? Is my understanding correct?
>
Yes right. Since those are carveouts, there is no valid pageframe,
 so the check fails.

>> Change allocation of sg buffers from dma_coherent memory to kzalloc
>> to fix the issue.
>
>But why can you drop the coherency?
>
 The coherency is not dropped here. dma_map/unmap used
 makes the buffer coherent before passing it to
 dmaengine. Previously it was not required.
 I can add this in description if its not clear.

 >> @@ -1268,6 +1260,8 @@ static int qup_i2c_xfer_v2(struct i2c_adapter *adap,
>>  		}
>>  	}
>>
>> +	idx = 0;
>> +
>
>This looks like an unrelated change.
    Ha, wrong. This should have been in a separate patch. This was to fix a
    initialization issue in dma mode. Sorry, should not have been here, will
    move it to out.

Regards,
 Sricharan