Comment 4 for bug 1095242

Revision history for this message
Andy Whitcroft (apw) wrote : [Acked] [PATCH (Quantal SRU)] ALSA: hda - bug fix for invalid connection list of Haswell HDMI codec pins

On Wed, Jan 02, 2013 at 10:17:35AM +0100, David Henningsson wrote:
> From: Mengdong Lin <email address hidden>
>
> Haswell HDMI codec pins may report invalid connection list entries, which
> will cause failure to play audio via HDMI or Display Port.
>
> So this patch adds fixup for Haswell to workaround this hardware issue:
> enable DP1.2 mode and override the pins' connection list entries with proper
> value.
>
> Signed-off-by: Mengdong Lin <email address hidden>
> Signed-off-by: Xingchao Wang <email address hidden>
> Signed-off-by: Takashi Iwai <email address hidden>
> (cherry picked from commit 6ffe168f822cf7f777987cddc00ade542fd73bf0)
>
> BugLink: https://bugs.launchpad.net/bugs/1095242
> Signed-off-by: David Henningsson <email address hidden>
> ---
>
> We're being advised by Intel to include the following patch for the Quantal kernel, to fix a Haswell/Sharkbay HDMI/DP audio issue. As there are no sharkbay machines out there yet, and the patch contains a "if codec->vendor_id == 80862807" I assume this will cause no regressions.
>
> The commit is already in the 3.8 kernel, so this is just for the 3.5 SRU.
>
> sound/pci/hda/patch_hdmi.c | 28 ++++++++++++++++++++++++++++
> 1 file changed, 28 insertions(+)
>
> diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
> index dc249b2..a7e4da4 100644
> --- a/sound/pci/hda/patch_hdmi.c
> +++ b/sound/pci/hda/patch_hdmi.c
> @@ -1323,6 +1323,30 @@ static const struct hda_codec_ops generic_hdmi_patch_ops = {
> .unsol_event = hdmi_unsol_event,
> };
>
> +static void intel_haswell_fixup_connect_list(struct hda_codec *codec)
> +{
> + unsigned int vendor_param;
> + hda_nid_t list[3] = {0x2, 0x3, 0x4};
> +
> + vendor_param = snd_hda_codec_read(codec, 0x08, 0, 0xf81, 0);
> + if (vendor_param == -1 || vendor_param & 0x02)
> + return;
> +
> + /* enable DP1.2 mode */
> + vendor_param |= 0x02;
> + snd_hda_codec_read(codec, 0x08, 0, 0x781, vendor_param);
> +
> + vendor_param = snd_hda_codec_read(codec, 0x08, 0, 0xf81, 0);
> + if (vendor_param == -1 || !(vendor_param & 0x02))
> + return;
> +
> + /* override 3 pins connection list */
> + snd_hda_override_conn_list(codec, 0x05, 3, list);
> + snd_hda_override_conn_list(codec, 0x06, 3, list);
> + snd_hda_override_conn_list(codec, 0x07, 3, list);
> +}
> +
> +
> static int patch_generic_hdmi(struct hda_codec *codec)
> {
> struct hdmi_spec *spec;
> @@ -1332,6 +1356,10 @@ static int patch_generic_hdmi(struct hda_codec *codec)
> return -ENOMEM;
>
> codec->spec = spec;
> +
> + if (codec->vendor_id == 0x80862807)
> + intel_haswell_fixup_connect_list(codec);
> +
> if (hdmi_parse_codec(codec) < 0) {
> codec->spec = NULL;
> kfree(spec);

Seems to be already upstream, highly machine specific, and recommended by the manufacturer,
therefore:

Acked-by: Andy Whitcroft <email address hidden>

-apw