lkml.org 
[lkml]   [2019]   [Aug]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    SubjectRe: [linux-sunxi] Re: [PATCH v4 6/9] ASoC: sun4i-i2s: Add multi-lane functionality
    Date
    Dne sreda, 31. julij 2019 ob 14:29:53 CEST je Maxime Ripard napisal(a):
    > On Tue, Jul 30, 2019 at 07:57:10PM +0200, Jernej Škrabec wrote:
    > > Dne torek, 04. junij 2019 ob 11:38:44 CEST je Code Kipper napisal(a):
    > > > On Tue, 4 Jun 2019 at 11:02, Christopher Obbard <chris@64studio.com>
    wrote:
    > > > > On Tue, 4 Jun 2019 at 09:43, Code Kipper <codekipper@gmail.com> wrote:
    > > > > > On Tue, 4 Jun 2019 at 09:58, Maxime Ripard
    > > > > > <maxime.ripard@bootlin.com>
    > >
    > > wrote:
    > > > > > > On Mon, Jun 03, 2019 at 07:47:32PM +0200, codekipper@gmail.com
    wrote:
    > > > > > > > From: Marcus Cooper <codekipper@gmail.com>
    > > > > > > >
    > > > > > > > The i2s block supports multi-lane i2s output however this
    > > > > > > > functionality
    > > > > > > > is only possible in earlier SoCs where the pins are exposed and
    > > > > > > > for
    > > > > > > > the i2s block used for HDMI audio on the later SoCs.
    > > > > > > >
    > > > > > > > To enable this functionality, an optional property has been
    > > > > > > > added to
    > > > > > > > the bindings.
    > > > > > > >
    > > > > > > > Signed-off-by: Marcus Cooper <codekipper@gmail.com>
    > > > > > >
    > > > > > > I'd like to have Mark's input on this, but I'm really worried
    > > > > > > about
    > > > > > > the interaction with the proper TDM support.
    > > > > > >
    > > > > > > Our fundamental issue is that the controller can have up to 8
    > > > > > > channels, but either on 4 lines (instead of 1), or 8 channels on 1
    > > > > > > (like proper TDM) (or any combination between the two, but that
    > > > > > > should
    > > > > > > be pretty rare).
    > > > > >
    > > > > > I understand...maybe the TDM needs to be extended to support this to
    > > > > > consider channel mapping and multiple transfer lines. I was thinking
    > > > > > about the later when someone was requesting support on IIRC a while
    > > > > > ago, I thought masking might of been a solution. These can wait as
    > > > > > the
    > > > > > only consumer at the moment is LibreELEC and we can patch it there.
    > > > >
    > > > > Hi Marcus,
    > > > >
    > > > > FWIW, the TI McASP driver has support for TDM & (i think?) multiple
    > > > > transfer lines which are called serializers.
    > > > > Maybe this can help with inspiration?
    > > > > see
    > > > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tre
    > > > > e/s
    > > > > ound/soc/ti/davinci-mcasp.c sample DTS:
    > > > >
    > > > > &mcasp0 {
    > > > >
    > > > > #sound-dai-cells = <0>;
    > > > > status = "okay";
    > > > > pinctrl-names = "default";
    > > > > pinctrl-0 = <&mcasp0_pins>;
    > > > >
    > > > > op-mode = <0>;
    > > > > tdm-slots = <8>;
    > > > > serial-dir = <
    > > > >
    > > > > 2 0 1 0
    > > > > 0 0 0 0
    > > > > 0 0 0 0
    > > > > 0 0 0 0
    > > > > >
    > > > > >;
    > > > >
    > > > > tx-num-evt = <1>;
    > > > > rx-num-evt = <1>;
    > > > >
    > > > > };
    > > > >
    > > > > Cheers!
    > > >
    > > > Thanks, this looks good.
    > >
    > > I would really like to see this issue resolved, because HDMI audio support
    > > in mainline Linux for those SoCs is long overdue.
    > >
    > > However, there is a possibility to still add HDMI audio suport (stereo
    > > only) even if this issue is not completely solved. If we agree that
    > > configuration of channels would be solved with additional property as
    > > Christopher suggested, support for >2 channels can be left for a later
    > > time when support for that property would be implemented. Currently,
    > > stereo HDMI audio support can be added with a few patches.
    > >
    > > I think all I2S cores are really the same, no matter if internally
    > > connected to HDMI controller or routed to pins, so it would make sense to
    > > use same compatible for all of them. It's just that those I2S cores which
    > > are routed to pins can use only one lane and >2 channels can be used only
    > > in TDM mode of operation, if I understand this correctly.
    > >
    > > New property would have to be optional, so it's omission would result in
    > > same channel configuration as it is already present, to preserve
    > > compatibility with old device trees. And this mode is already sufficient
    > > for stereo HDMI audio support.
    >
    > Yeah, it looks like a good plan.
    >
    > > Side note: HDMI audio with current sun4i-i2s driver has a delay (about a
    > > second), supposedly because DW HDMI controller automatically generates CTS
    > > value based on I2S clock (auto CTS value generation is enabled per
    > > DesignWare recomendation for DW HDMI I2S interface).
    >
    > Is that a constant delay during the audio playback, or only at startup?

    I think it's just at startup, e.g. if you're watching movie, audio is in sync,
    it's just that first second or so is silent.

    >
    > > I2S driver from BSP Linux solves that by having I2S clock output
    > > enabled all the time. Should this be flagged with some additional
    > > property in DT?
    >
    > I'd say that if the codec has that requirement, then it should be
    > between the codec and the DAI, the DT doesn't really have anything to
    > do with this.

    Ok, but how to communicate that fact to I2S driver then? BSP driver solves
    that by using different compatible, but as I said before, I2S cores are not
    really different, so this seems wrong.

    Best regards,
    Jernej


    \
     
     \ /
      Last update: 2019-08-01 07:32    [W:2.657 / U:0.352 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site