Re: Regression: memory corruption on Atmel SAMA5D31
From: Tudor.Ambarus
Date: Mon Jun 20 2022 - 04:44:06 EST
On 6/20/22 10:04, Thorsten Leemhuis wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>
> On 18.05.22 09:51, Peter Rosin wrote:
>> 2022-05-18 at 08:21, Tudor.Ambarus@xxxxxxxxxxxxx wrote:
>>> On 5/17/22 17:50, Peter Rosin wrote:
>>>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>>>>
>>>> 2022-04-11 at 08:21, Tudor.Ambarus@xxxxxxxxxxxxx wrote:
>>>>> There are some concurrency bugs in the at-hdmac (DMA) driver, I'm handling them
>>>>> and will come with a resolution. Disabling the DMA showed the bug is no more
>>>>> reproducible.
>>>>
>>>> Any news?
>>>
>>> I'm now allocated on this, so I started looking around what has to be done.
>>> I'm thinking of using virt-dma to manage the channels and the request queues.
>>> Will get back to you after I'll have something working.
>>
>> Sounds good, thanks!
>
> That was about a month ago. Has any progress been made to get this
> regression fixed?
Hi, Thorsten, Peter,
I was mostly out of office last month, and I'll still be offline this week.
I made some progress, tried to address the bugs in an incremental way. I now
encounter the memory corruption less often, but I still hit it. I made some
drafts at [1] if someone is curios. Anyway, I'm modifying the driver to use
virt-dma and also I'm trying to move the election of a new transfer in the
irq handler instead of in tasklet. I couldn't find a quick non-invasive fix,
so still in progress.
Cheers,
ta
git@xxxxxxxxxx:ambarus/linux-0day.git, branch dma-regression-hdmac-v5.18-rc7-4th-attempt
>
> Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
>
> P.S.: As the Linux kernel's regression tracker I deal with a lot of
> reports and sometimes miss something important when writing mails like
> this. If that's the case here, don't hesitate to tell me in a public
> reply, it's in everyone's interest to set the public record straight.
>
> #regzbot poke
>
>