Re: [PATCH] clk: tegra210: Fix default rates for HDA clocks

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2440 bytes --]

On Wed, Jun 05, 2019 at 12:30:31PM +0100, Jon Hunter wrote:
> 
> On 31/05/2019 15:58, Jon Hunter wrote:
> > 
> > On 29/05/2019 14:46, Thierry Reding wrote:
> >> On Wed, May 29, 2019 at 10:18:21AM +0100, Jon Hunter wrote:
> >>> Currently the default clock rates for the HDA and HDA2CODEC_2X clocks
> >>> are both 19.2MHz. However, the default rates for these clocks should
> >>> actually be 51MHz and 48MHz, respectively. Correct the default clock
> >>> rates for these clocks by specifying them in the clock init table for
> >>> Tegra210.
> >>>
> >>> Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
> >>> ---
> >>>  drivers/clk/tegra/clk-tegra210.c | 2 ++
> >>>  1 file changed, 2 insertions(+)
> >>
> >> Does this fix anything? Should this be backported to stable releases?
> > 
> > Good point. We are aligning the clock configuration with what we ship.
> > So I thought for completeness it would be good to test HDA playback
> > across the various sample-rates we support (32kHz to 192kHz) but with or
> > without this patch I am not hearing anything. Let me check on this with
> > Sameer as I would like to see if we need to mark this for stable or not.
> > 
> >> Acked-by: Thierry Reding <treding@nvidia.com>
> 
> I have confirmed that this does fix HDA playback on Tegra210. Without
> this fix, I am seeing the following messages during playback and
> playback is distorted ...
> 
> Write error: -32,Broken pipe
> [   15.069335] tegra-mc 70019000.memory-controller: hdar: read
> @0x0000000000000000: EMEM address decode error (EMEM decode error)
> Write error: -32,Broken pipe
> [   15.465362] tegra-mc 70019000.memory-controller: hdar: read
> @0x0000000000000000: EMEM address decode error (EMEM decode error)
> Write error: -32,Broken pipe
> [   15.858615] tegra-mc 70019000.memory-controller: hdar: read
> @0x0000000000000000: EMEM address decode error (EMEM decode error)
> W
> 
> Do you want me to update the change and resend?

Honestly I'm not sure if it's worth it. I haven't seen any bug reports
for this and we haven't had audio over HDMI support for very long, so a
backport may not be necessary.

I guess there'd be some use to backport this so that our stable kernel
testing passes these. So it's really up to you. I have a slight tendency
towards backporting, because it's really tiny and then we just have it
out of the way and it's not going to haunt us.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]