[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAEnQRZBpQPoi5OH--c=ubKYc6B3rspmVnb846UTFW7P5SE62ww@mail.gmail.com>
Date: Tue, 6 Aug 2019 14:15:03 +0300
From: Daniel Baluta <daniel.baluta@...il.com>
To: Nicolin Chen <nicoleotsuka@...il.com>
Cc: Mark Brown <broonie@...nel.org>,
Daniel Baluta <daniel.baluta@....com>,
Devicetree List <devicetree@...r.kernel.org>,
Linux-ALSA <alsa-devel@...a-project.org>,
Pengutronix Kernel Team <kernel@...gutronix.de>,
Timur Tabi <timur@...nel.org>, Rob Herring <robh@...nel.org>,
"S.j. Wang" <shengjiu.wang@....com>,
"Angus Ainslie (Purism)" <angus@...ea.ca>,
Takashi Iwai <tiwai@...e.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
dl-linux-imx <linux-imx@....com>,
Viorel Suman <viorel.suman@....com>,
Fabio Estevam <festevam@...il.com>,
Mihai Serban <mihai.serban@...il.com>,
Lucas Stach <l.stach@...gutronix.de>
Subject: Re: [alsa-devel] [PATCH v2 1/7] ASoC: fsl_sai: Add registers
definition for multiple datalines
On Tue, Jul 30, 2019 at 10:59 AM Nicolin Chen <nicoleotsuka@...il.com> wrote:
>
> On Mon, Jul 29, 2019 at 09:20:01PM +0100, Mark Brown wrote:
> > On Mon, Jul 29, 2019 at 10:57:43PM +0300, Daniel Baluta wrote:
> > > On Mon, Jul 29, 2019 at 10:42 PM Nicolin Chen <nicoleotsuka@...il.com> wrote:
> > > > On Sun, Jul 28, 2019 at 10:24:23PM +0300, Daniel Baluta wrote:
> >
> > > > > @@ -704,7 +711,14 @@ static bool fsl_sai_readable_reg(struct device *dev, unsigned int reg)
> > > > > case FSL_SAI_TCR3:
> > > > > case FSL_SAI_TCR4:
> > > > > case FSL_SAI_TCR5:
> > > > > - case FSL_SAI_TFR:
> > > > > + case FSL_SAI_TFR0:
> >
> > > > A tricky thing here is that those SAI instances on older SoC don't
> > > > support multi data lines physically, while seemly having registers
> > > > pre-defined. So your change doesn't sound doing anything wrong to
> > > > them at all, I am still wondering if it is necessary to apply them
> > > > to newer compatible only though, as for older compatibles of SAI,
> > > > these registers would be useless and confusing if being exposed.
> >
> > > > What do you think?
> >
> > > Yes, I thought about this too. But, I tried to keep the code as short
> > > as possible and technically it is not wrong. When 1 data line is supported
> > > for example application will only care about TDR0, TFR0, etc.
> >
> > So long as it's safe to read the registers (you don't get a bus error or
> > anything) I'd say it's more trouble than it's worth to have separate
> > regmap configuations just for this. The main reasons for restricting
> > readability are where there's physical problems with doing the reads or
> > to keep the size of the debugfs files under control for usability and
> > performance reasons.
>
> Thanks for the input, Mark.
>
> Daniel, did you get a chance to test it on older SoCs? At least
> nothing breaks like bus errors?
Tested on imx6sx-sdb, everything looks good. No bus errors.
Powered by blists - more mailing lists