Re: Broadcom WiFi SDIO performance regression after commit "mmc: sdhci: Remove finish_tasklet"

From: Dmitry Osipenko
Date: Mon Aug 31 2020 - 14:53:42 EST


31.08.2020 18:08, Adrian Hunter пишет:
> On 27/08/20 12:36 pm, Dmitry Osipenko wrote:
>> 27.08.2020 09:45, Adrian Hunter пишет:
>>> On 27/08/20 9:07 am, Dmitry Osipenko wrote:
>>>> Hello!
>>>>
>>>> I was debugging WiFi performance problems on Acer A500 tablet device
>>>> that has BCM4329 WiFi chip which is connected to NVIDIA Terga20 SoC via
>>>> SDIO and found that the following commit causes a solid 5-10 Mbit/s of
>>>> WiFi throughput regression after 5.2 kernel:
>>>
>>> What is that in percentage terms?
>>
>> That is about 20%.
>>
>>>> commit c07a48c2651965e84d35cf193dfc0e5f7892d612
>>>> Author: Adrian Hunter <adrian.hunter@xxxxxxxxx>
>>>> Date: Fri Apr 5 15:40:20 2019 +0300
>>>>
>>>> mmc: sdhci: Remove finish_tasklet
>>>>
>>>> Remove finish_tasklet. Requests that require DMA-unmapping or
>>>> sdhci_reset
>>>> are completed either in the IRQ thread or a workqueue if the
>>>> completion is
>>>> not initiated by the IRQ.
>>>>
>>>> Reverting the offending commit on top of recent linux-next resolves the
>>>> problem.
>>>>
>>>> Ulf / Adrian, do you have any ideas what could be done in regards to
>>>> restoring the SDIO performance? Should we just revert the offending commit?
>>>>
>>>
>>> Unfortunately I think we are past the point of returning to the tasklet.
>>>
>>> sdhci can complete requests in the irq handler but only if ->pre_req() and
>>> ->post_req() are used, which is not supported by SDIO at present. pre_req
>>> and post_req were introduced to reduce latency for the block driver, so it
>>> seems reasonable perhaps to look at using them in SDIO as well.
>>>
>>
>> I'll try to take a look at pre/post_req(), but I'm not very familiar
>> with the MMC code, so it may take quite some time. Will be great if you
>> could help with making a patch that I could test!
>>
>
> You could start by seeing if using pre/post_req helps at all, as below.
> If that doesn't help, then it might need more analysis.

Hello, Adrian! I tested yours patch and 100% fixes the problem! Thank
you very much! Please make a proper patch and feel free to add my t-b!

Tested-by: Dmitry Osipenko <digetx@xxxxxxxxx>