linux-kernel - Re: [PATCH] dmaengine: pxa_dma: fix the maximum requestor line

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:	Tue, 09 Feb 2016 21:22:08 +0100
From:	Robert Jarzmik <robert.jarzmik@...e.fr>
To:	Vinod Koul <vinod.koul@...el.com>
Cc:	Daniel Mack <daniel@...que.org>,
	Haojian Zhuang <haojian.zhuang@...il.com>,
	linux-arm-kernel@...ts.infradead.org, dmaengine@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dmaengine: pxa_dma: fix the maximum requestor line

Vinod Koul <vinod.koul@...el.com> writes:

> On Tue, Feb 09, 2016 at 08:20:05AM +0100, Robert Jarzmik wrote:
>> Vinod Koul <vinod.koul@...el.com> writes:
>> 
>> > On Mon, Feb 08, 2016 at 03:18:51PM +0100, Robert Jarzmik wrote:
>> >> The current number of requestor lines is limited to 31. This was an
>> >> error of a previous commit, as this number is platform dependent, and is
>> >> actually :
>> >>  - for pxa25x: 40 requestor lines
>> >>  - for pxa27x: 74 requestor lines
>> >>  - for pxa3xx: 100 requestor lines
>> >> 
>> >> As the driver doesn't need to know the exact number, but only an
>> >
>> > and why would that be a good assumption?
>> Well, the driver doesn't use the exact value. Would that be 128, 1000 or 10000,
>> it wouldn't change its behavior. The clients either pass -1UL (ie. no flow
>> control) or a requestor line number.
>
> Somehow that does not sound right to me. You seem to have no way of knowing
> what is the actual max request line for a platforms. What if on pxa25x user
> requests more than 40 lines?
>
>> > Btw shouldn't this data come from DT?
>> It can if you wish, but in this case I must amend the platform data case too, ie
>> mmp_dma_platdata. And this has an impact on MMP architecture. I'd rather have
>> the simpler approach unless you really want the exact number of requestor lines
>> to be passed in platform_data+DT.
>
> The information that you have X lines on a platform needs to be described
> and queried. I see current approach error prone

Actually I have thought over your point, and you're quite right (this is your
case of more than 40 lines).

The exception I have found is :
 - a device driver is wrongly written
 - it requests the requestor line 0x1000 (below 0x8000)
 - pxa_dma computes the register in pxad_drcmr(0x1000)
   => this gives : base + 0x1000 + 4 * 0x1000 = base + 0x5000
   => this is a not existing iomapped register

So yes, I will respin the patchset, with :
 - one patch adding the number of requestors in mmp_dma.h
 - one patch adding the number of requestors in devicetree description
 - one patch adding the number of requestors in arch/arm/mach-pxa (my tree)
 - one patch adding the use of number of requestors in pxa_dma.c

Cheers.

-- 
Robert