linux-kernel - Re: [PATCH] [RFC] dmaengine: add fifo

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, 7 Jun 2019 10:24:00 +0100
From:   Jon Hunter <jonathanh@...dia.com>
To:     Dmitry Osipenko <digetx@...il.com>,
        Peter Ujfalusi <peter.ujfalusi@...com>,
        Sameer Pujar <spujar@...dia.com>, Vinod Koul <vkoul@...nel.org>
CC:     <dan.j.williams@...el.com>, <tiwai@...e.com>,
        <dmaengine@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <sharadg@...dia.com>, <rlokhande@...dia.com>, <dramesh@...dia.com>,
        <mkumard@...dia.com>, linux-tegra <linux-tegra@...r.kernel.org>
Subject: Re: [PATCH] [RFC] dmaengine: add fifo_size member


On 06/06/2019 18:25, Dmitry Osipenko wrote:
> 06.06.2019 19:53, Jon Hunter пишет:
>>
>> On 06/06/2019 17:44, Dmitry Osipenko wrote:
>>> 06.06.2019 19:32, Jon Hunter пишет:
>>>>
>>>> On 06/06/2019 16:18, Dmitry Osipenko wrote:
>>>>
>>>> ...
>>>>
>>>>>>> If I understood everything correctly, the FIFO buffer is shared among
>>>>>>> all of the ADMA clients and hence it should be up to the ADMA driver to
>>>>>>> manage the quotas of the clients. So if there is only one client that
>>>>>>> uses ADMA at a time, then this client will get a whole FIFO buffer, but
>>>>>>> once another client starts to use ADMA, then the ADMA driver will have
>>>>>>> to reconfigure hardware to split the quotas.
>>>>>>
>>>>>> The FIFO quotas are managed by the ADMAIF driver (does not exist in
>>>>>> mainline currently but we are working to upstream this) because it is
>>>>>> this device that owns and needs to configure the FIFOs. So it is really
>>>>>> a means to pass the information from the ADMAIF to the ADMA.
>>>>>
>>>>> So you'd want to reserve a larger FIFO for an audio channel that has a
>>>>> higher audio rate since it will perform reads more often. You could also
>>>>> prioritize one channel over the others, like in a case of audio call for
>>>>> example.
>>>>>
>>>>> Is the shared buffer smaller than may be needed by clients in a worst
>>>>> case scenario? If you could split the quotas statically such that each
>>>>> client won't ever starve, then seems there is no much need in the
>>>>> dynamic configuration.
>>>>
>>>> Actually, this is still very much relevant for the static case. Even if
>>>> we defined a static configuration of the FIFO mapping in the ADMAIF
>>>> driver we still need to pass this information to the ADMA. I don't
>>>> really like the idea of having it statically defined in two different
>>>> drivers.
>>>
>>> Ah, so you need to apply the same configuration in two places. Correct?
>>>
>>> Are ADMAIF and ADMA really two different hardware blocks? Or you
>>> artificially decoupled the ADMA driver?
>>
>> These are two different hardware modules with their own register sets.
>> Yes otherwise, it would be a lot simpler!
> 
> The register sets are indeed separated, but it looks like that ADMAIF is
> really a part of ADMA that is facing to Audio Crossbar. No? What is the
> purpose of ADMAIF? Maybe you could amend the ADMA hardware description
> with the ADMAIF addition until it's too late.

The ADMA can perform the following transfers (per the CH_CONFIG
register) ...

MEMORY_TO_MEMORY
AHUB_TO_MEMORY
MEMORY_TO_AHUB
AHUB_TO_AHUB

Hence it is possible to use the ADMA to do memory-to-memory transfers
that do not involve the ADMAIF.

So no the ADMAIF is not part of the ADMA. It is purely the interface to
the crossbar (AHUB/APE), but from a hardware standpoint they are
separate. And so no we will not amend the hardware description.

Jon

-- 
nvpublic