Re: [PATCH v5 03/14] ASoC: SOF: Add driver debug support.

From: Mark Brown <broonie@kernel.org>
To: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Cc: alsa-devel@alsa-project.org, Alan Cox <alan@linux.intel.com>,
	Takashi Iwai <tiwai@suse.de>,
	Pan Xiuli <xiuli.pan@linux.intel.com>,
	Daniel Baluta <daniel.baluta@gmail.com>,
	Ranjani Sridharan <ranjani.sridharan@linux.intel.com>,
	liam.r.girdwood@linux.intel.com, vkoul@kernel.org,
	andriy.shevchenko@linux.intel.com,
	sound-open-firmware@alsa-project.org
Subject: Re: [PATCH v5 03/14] ASoC: SOF: Add driver debug support.
Date: Tue, 2 Apr 2019 12:44:13 +0700	[thread overview]
Message-ID: <20190402054413.GA2059@sirena.org.uk> (raw)
In-Reply-To: <631c6cba-2f81-6f5f-3d5f-f9f5bbb171c7@linux.intel.com>

[-- Attachment #1.1: Type: text/plain, Size: 790 bytes --]

On Mon, Apr 01, 2019 at 01:15:46PM -0400, Pierre-Louis Bossart wrote:
> On 3/29/19 12:04 PM, Takashi Iwai wrote:

> > Doesn't pos need to be aligned to 32bit as well (at least at actually
> > reading from iomem)?

> Good point,  we can either return an error if the count in both a multiple
> of 32 bits, or do some caching to allow for arbitrary size access. The
> former is very simple but not sure if this is accepted behavior for debugfs.

You really want something that will work with arbatrary userspace tools
if you can rather than requiring special magic read sizes - things like
cat and cp should probably work.  A common thing is to do reads/writes
of the correct sizes into a buffer and then pick the relevant bits out
of the buffer to satisfy what userspace actually asked for.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]