linux-kernel - Re: [PATCH 1/2] dmaengine: hdmac: factorise memset descriptor allocation

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:	Wed, 7 Oct 2015 18:33:17 +0100
From:	Maxime Ripard <maxime.ripard@...e-electrons.com>
To:	Vinod Koul <vinod.koul@...el.com>
Cc:	Nicolas Ferre <nicolas.ferre@...el.com>,
	Alexandre Belloni <alexandre.belloni@...e-electrons.com>,
	dmaengine@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-arm-kernel@...ts.infradead.org,
	Ludovic Desroches <ludovic.desroches@...el.com>,
	Cyrille Pitchen <cyrille.pitchen@...el.com>
Subject: Re: [PATCH 1/2] dmaengine: hdmac: factorise memset descriptor
 allocation

Hi Vinod,

On Wed, Oct 07, 2015 at 02:54:13PM +0100, Vinod Koul wrote:
> On Thu, Oct 01, 2015 at 04:52:37PM +0200, Maxime Ripard wrote:
> 
> > +static struct at_desc *atc_create_memset_desc(struct dma_chan *chan,
> > +					      dma_addr_t psrc,
> > +					      dma_addr_t pdst,
> > +					      size_t len)
> > +{
> > +	struct at_dma_chan *atchan = to_at_dma_chan(chan);
> > +	struct at_desc *desc;
> > +	size_t xfer_count;
> > +
> > +	u32 ctrla = ATC_SRC_WIDTH(2) |
> > +		ATC_DST_WIDTH(2);
> 
> why is this over two lines :)
> 
> > +	u32 ctrlb = ATC_DEFAULT_CTRLB | ATC_IEN |
> > +		ATC_SRC_ADDR_MODE_FIXED |
> > +		ATC_DST_ADDR_MODE_INCR |
> > +		ATC_FC_MEM2MEM;
> > +
> > +	xfer_count = len >> 2;
> > +	if (xfer_count > ATC_BTSIZE_MAX) {
> > +		dev_err(chan2dev(chan), "%s: buffer is too big\n",
> > +			__func__);
> > +		return NULL;
> > +	}
> 
> This is fine, but this is driver limitation. We should really split the
> txn to multiple descriptors here..

Both these things are actually just copy from the previous code.

I can probably fix the first style issue in this patch, but splitting
the memset descriptors should go in a separate patch.

> 
> > @@ -914,46 +953,26 @@ atc_prep_dma_memset(struct dma_chan *chan, dma_addr_t dest, int value,
> >  		return NULL;
> >  	}
> >  
> > -	xfer_count = len >> 2;
> > -	if (xfer_count > ATC_BTSIZE_MAX) {
> > -		dev_err(chan2dev(chan), "%s: buffer is too big\n",
> > +	vaddr = dma_pool_alloc(atdma->memset_pool, GFP_ATOMIC, &paddr);
> 
> Mostly people use _zalloc variant, any reason why you don't want that

Not necessarily, but I don't need zeroed memory either, since I
overwrite the content right after the allocation.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

Download attachment "signature.asc" of type "application/pgp-signature" (820 bytes)