linux-kernel - Re: [alsa-devel] [RFC PATCH 0/2] soundwire: add master

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]

Message-ID: <0ec41a5b-6132-6940-f1b3-bac1724b70a4@linux.intel.com>
Date:   Fri, 14 Feb 2020 17:34:40 -0600
From:   Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     alsa-devel@...a-project.org, tiwai@...e.de,
        linux-kernel@...r.kernel.org,
        Ranjani Sridharan <ranjani.sridharan@...ux.intel.com>,
        vkoul@...nel.org, broonie@...nel.org,
        srinivas.kandagatla@...aro.org, jank@...ence.com,
        slawomir.blauciak@...el.com,
        Bard liao <yung-chuan.liao@...ux.intel.com>,
        Rander Wang <rander.wang@...ux.intel.com>
Subject: Re: [alsa-devel] [RFC PATCH 0/2] soundwire: add master_device/driver
 support

On 2/14/20 10:49 AM, Greg KH wrote:
> On Fri, Jan 31, 2020 at 10:20:09PM -0600, Pierre-Louis Bossart wrote:
>> The SoundWire master representation needs to evolve to take into account:
>>
>> a) a May 2019 recommendation from Greg KH to remove platform devices
>>
>> b) the need on Intel platforms to support hardware startup only once
>> the power rail dependencies are settled. The SoundWire master is not
>> an independent IP at all.
>>
>> c) the need to deal with external wakes handled by the PCI subsystem
>> in specific low-power modes
>>
>> In case it wasn't clear, the SoundWire subsystem is currently unusable
>> with v5.5 on devices hitting the shelves very soon (race conditions,
>> power management, suspend/resume, etc). v5.6 will only provide
>> interface changes and no functional improvement. We've circled on the
>> same concepts since September 2019, and I hope this direction is now
>> aligned with the suggestions from Vinod Koul and that we can target
>> v5.7 as the 'SoundWire just works(tm)' version.
>>
>> This series is provided as an RFC since it depends on patches already
>> for review.
> 
> Both of these look sane to me, nice work!
> 
> Reviewed-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>

Thanks Greg, I really appreciate the review and educational guidance on 
the use of platform devices!

If that's alright with Vinod and you, I'd like to defer the integration 
of these patches a bit so that we can improve them a bit e.g.

- I realized earlier this week there's a potential issue with the 
'shutdown' callback, what I did looks duplicated with 'shutdown' called 
from the device core and by the parent.

- it would make sense to add support for this 'master device' in the 
Qualcomm drivers as well, so that we are aligned with a similar device 
hierarchy (PCI -> sdw_master_device and DT/plaforms ->sdw_master_device).

- we probably want to add the sysfs patches that started this discussion 
on the platform devices. This would be very useful to debug new 
platforms, we depend on BIOS/firmware and we can't always know the 
actual settings just by looking at the DSDT static contents (multiple 
tests and overridden values).

My preference in terms of integration in drivers/soundwire would be

1. Intel DAI cleanup (only one kfree missing, will resubmit v3 today)

2. [PATCH 00/10] soundwire: bus: fix race conditions, add suspend-resume
this series solves a lot of issues and should go first.

3. ASoC/SOF integration (still with platform devices)
I narrowed this down to 6 patches, that would help me submit the rest of 
the ASoC/SOF patches in Mark Brown's tree. That would be Intel specific. 
This step would be the first where everything SoundWire-related can be 
enabled in a build, and while we've caught a lot of cross-compilation 
issues it's likely some bots will find corner cases to keep us busy.

4. master_device/driver transition: these updated patches removing 
platform devices + sysfs support + Qualcomm support (the last point 
would depend on the workload/support of Qualcomm/Linaro folks, I don't 
want to commit on their behalf).

5. New SoundWire functionality for Intel platforms (clock-stop/restart 
and synchronized links). The code would be only located in 
drivers/soundwire and be easier to handle. For the synchronized links we 
still have a bit of validation work to do so it should really come last.

Would this work for everyone?

Thanks,
-Pierre