linux-kernel - Re: [PATCH] dmaengine: edma: Reject slave configs using a buswidth of 8 and larger

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]

Date:   Fri, 22 Sep 2017 09:16:13 +0300
From:   Peter Ujfalusi <peter.ujfalusi@...com>
To:     Vinod Koul <vinod.koul@...el.com>,
        Stefan Brüns <stefan.bruens@...h-aachen.de>
CC:     <dmaengine@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        Dan Williams <dan.j.williams@...el.com>
Subject: Re: [PATCH] dmaengine: edma: Reject slave configs using a buswidth of
 8 and larger

Stefan, Vinod,


Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki. Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki

On 2017-09-21 21:29, Peter Ujfalusi wrote:
> 
> 
> On 2017-09-21 20:17, Vinod Koul wrote:
>> On Sun, Sep 17, 2017 at 04:25:10PM +0200, Stefan Brüns wrote:
>>> The driver would happily accept buswidth of 16/32/64 bytes and program
>>> garbage to its registers.
>>
>> Peter?
> 
> Well, eDMA as such has no limitation on the buswidth, the addr_width
> become ACNT, counter for the inner loop - which we use to represent the
> word length. In theory we could support 32K-1 long word with eDMA, which
> we actually do in case of memcpy.
> 
> But the driver does state that it supports bus width of 1, 2, 3 and 4 bytes.
> 
> These are form the past and I never felt the need to check them.
> 
> I think it is fine to reject salve config when asking for longer dev
> width. I might revisit later if the need comes..
> 
> Acked-by: Peter Ujfalusi <peter.ujfalusi@...com>

If it is not too late, I have second thoughts about this..
It might be better to remove the src_addr_width/dst_addr_width check
from edma_slave_config() and add all DMA_SLAVE_BUSWIDTH_* to
EDMA_DMA_BUSWIDTHS.

> 
>>
>>>
>>> Signed-off-by: Stefan Brüns <stefan.bruens@...h-aachen.de>
>>> ---
>>>  drivers/dma/edma.c | 4 ++--
>>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/dma/edma.c b/drivers/dma/edma.c
>>> index 3879f80a4815..c944ea339425 100644
>>> --- a/drivers/dma/edma.c
>>> +++ b/drivers/dma/edma.c
>>> @@ -887,8 +887,8 @@ static int edma_slave_config(struct dma_chan *chan,
>>>  {
>>>  	struct edma_chan *echan = to_edma_chan(chan);
>>>  
>>> -	if (cfg->src_addr_width == DMA_SLAVE_BUSWIDTH_8_BYTES ||
>>> -	    cfg->dst_addr_width == DMA_SLAVE_BUSWIDTH_8_BYTES)
>>> +	if (cfg->src_addr_width >= DMA_SLAVE_BUSWIDTH_8_BYTES ||
>>> +	    cfg->dst_addr_width >= DMA_SLAVE_BUSWIDTH_8_BYTES)
>>>  		return -EINVAL;
>>>  
>>>  	memcpy(&echan->cfg, cfg, sizeof(echan->cfg));
>>> -- 
>>> 2.14.1
>>>
>>
> 
> - Péter
> 

- Péter