linux-kernel - Re: Tearing down DMA transfer setup after DMA client has finished

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 for Android: free password hash cracker in your pocket

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

Message-ID: <yw1xvauv1yew.fsf@unicorn.mansr.com>
Date:   Wed, 07 Dec 2016 16:44:55 +0000
From:   Måns Rullgård <mans@...sr.com>
To:     Vinod Koul <vinod.koul@...el.com>
Cc:     Mason <slash.tmp@...e.fr>, Russell King <linux@....linux.org.uk>,
        dmaengine@...r.kernel.org,
        Linus Walleij <linus.walleij@...aro.org>,
        Dan Williams <dan.j.williams@...el.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Jon Mason <jdmason@...zu.us>, Mark Brown <broonie@...nel.org>,
        Lars-Peter Clausen <lars@...afoo.de>,
        Lee Jones <lee.jones@...aro.org>,
        Laurent Pinchart <laurent.pinchart@...asonboard.com>,
        Arnd Bergmann <arnd@...db.de>,
        Maxime Ripard <maxime.ripard@...e-electrons.com>,
        Dave Jiang <dave.jiang@...el.com>,
        Peter Ujfalusi <peter.ujfalusi@...com>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@...sung.com>,
        Sebastian Frias <sf84@...oste.net>,
        Thibaud Cornic <thibaud_cornic@...madesigns.com>
Subject: Re: Tearing down DMA transfer setup after DMA client has finished

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

> On Tue, Dec 06, 2016 at 01:42:31PM +0100, Mason wrote:
>> On 06/12/2016 06:12, Vinod Koul wrote:
>> 
>> > On Tue, Nov 29, 2016 at 07:25:02PM +0100, Mason wrote:
>> > 
>> >> Is there a way to write a driver within the existing framework?
>> > 
>> > I think so, looking back at comments from Russell, I do tend to agree with
>> > that. Is there a specific reason why sbox can't be tied to alloc and free
>> > channels?
>> 
>> Here's a recap of the situation.
>> 
>> The "SBOX+MBUS" HW is used in several iterations of the tango SoC:
>
> btw is SBOX setup dependent upon the peripheral connected to?

The sbox is basically a crossbar that connects each of a number of input
ports to any of a number of output ports.  A few of the inputs and
outputs are dma channels reading or writing to memory while the rest are
peripheral devices.  To perform a mem-to-device transfer, you pick a dma
read channel, program the sbox to connect it to the chosen device, and
finally program the dma channel with address and size to transfer.

>> tango3
>>   2 memory channels available
>>   6 devices ("clients"?) may request an MBUS channel
>
> But only 2 can get a channel at any time..
>
>> tango4 (one more channel)
>>   3 memory channels available
>>   7 devices may request an MBUS channel :
>>     NFC0, NFC1, SATA0, SATA1, memcpy, (IDE0, IDE1)
>
> Same here
>
> Only thing is users shouldn't hold on to channel and freeup when not in use.
>
>> Notes:
>> The current NFC driver supports only one controller.
>> IDE is mostly obsolete at this point.
>> 
>> tango5 (SATA gets own dedicated MBUS channel pair)
>>   3 memory channels available
>>   5 devices may request an MBUS channel :
>>     NFC0, NFC1, memcpy, (IDE0, IDE1)
>> 
>> 
>> If I understand the current DMA driver (written by Mans), client
>> drivers are instructed to use a specific channel in the DT, and
>> the DMA driver muxes access to that channel. The DMA driver
>> manages a per-channel queue of outstanding DMA transfer requests,
>> and a new transfer is started friom within the DMA ISR
>> (modulo the fact that the interrupt does not signal completion
>> of the transfer, as explained else-thread).
>> 
>> What you're proposing, Vinod, is to make a channel exclusive
>> to a driver, as long as the driver has not explicitly released
>> the channel, via dma_release_channel(), right?
>
> Precisely, but yes the downside of that is concurrent access are
> limited, but am not sure if driver implements virtual channels and
> allows that..

My driver implements virtual channels.  The problem is that the physical
dma channels signal completion slightly too soon, at least with
mem-to-device transfers.  Apparently we need to keep the sbox routing
until the peripheral indicates that it has actually received all the
data.

-- 
Måns Rullgård