[alsa-devel] [PATCH] ALSA: hda - Fix missing ELD info when using jackpoll_ms parameter
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com --- sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c index 4098196..f0a8ea8 100644 --- a/sound/pci/hda/patch_hdmi.c +++ b/sound/pci/hda/patch_hdmi.c @@ -1142,32 +1142,34 @@ static void hdmi_setup_audio_infoframe(struct hda_codec *codec,
static bool hdmi_present_sense(struct hdmi_spec_per_pin *per_pin, int repoll);
-static void hdmi_intrinsic_event(struct hda_codec *codec, unsigned int res) +static void jack_callback(struct hda_codec *codec, struct hda_jack_tbl *jack) { struct hdmi_spec *spec = codec->spec; + int pin_idx = pin_nid_to_pin_index(spec, jack->nid); + if (pin_idx < 0) + return; + + if (hdmi_present_sense(get_pin(spec, pin_idx), 1)) + snd_hda_jack_report_sync(codec); +} + +static void hdmi_intrinsic_event(struct hda_codec *codec, unsigned int res) +{ int tag = res >> AC_UNSOL_RES_TAG_SHIFT; - int pin_nid; - int pin_idx; struct hda_jack_tbl *jack; int dev_entry = (res & AC_UNSOL_RES_DE) >> AC_UNSOL_RES_DE_SHIFT;
jack = snd_hda_jack_tbl_get_from_tag(codec, tag); if (!jack) return; - pin_nid = jack->nid; jack->jack_dirty = 1;
_snd_printd(SND_PR_VERBOSE, "HDMI hot plug event: Codec=%d Pin=%d Device=%d Inactive=%d Presence_Detect=%d ELD_Valid=%d\n", - codec->addr, pin_nid, dev_entry, !!(res & AC_UNSOL_RES_IA), + codec->addr, jack->nid, dev_entry, !!(res & AC_UNSOL_RES_IA), !!(res & AC_UNSOL_RES_PD), !!(res & AC_UNSOL_RES_ELDV));
- pin_idx = pin_nid_to_pin_index(spec, pin_nid); - if (pin_idx < 0) - return; - - if (hdmi_present_sense(get_pin(spec, pin_idx), 1)) - snd_hda_jack_report_sync(codec); + jack_callback(codec, jack); }
static void hdmi_non_intrinsic_event(struct hda_codec *codec, unsigned int res) @@ -2080,7 +2082,8 @@ static int generic_hdmi_init(struct hda_codec *codec) hda_nid_t pin_nid = per_pin->pin_nid;
hdmi_init_pin(codec, pin_nid); - snd_hda_jack_detect_enable(codec, pin_nid, pin_nid); + snd_hda_jack_detect_enable_callback(codec, pin_nid, pin_nid, + codec->jackpoll_interval > 0 ? jack_callback : NULL); } return 0; }
At Wed, 4 Dec 2013 10:19:41 +0800, David Henningsson wrote:
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com
sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
Hrm, I vaguely remember that I gave a very similar patch for testing, but it failed by some reason...
Takashi
diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c index 4098196..f0a8ea8 100644 --- a/sound/pci/hda/patch_hdmi.c +++ b/sound/pci/hda/patch_hdmi.c @@ -1142,32 +1142,34 @@ static void hdmi_setup_audio_infoframe(struct hda_codec *codec,
static bool hdmi_present_sense(struct hdmi_spec_per_pin *per_pin, int repoll);
-static void hdmi_intrinsic_event(struct hda_codec *codec, unsigned int res) +static void jack_callback(struct hda_codec *codec, struct hda_jack_tbl *jack) { struct hdmi_spec *spec = codec->spec;
- int pin_idx = pin_nid_to_pin_index(spec, jack->nid);
- if (pin_idx < 0)
return;- if (hdmi_present_sense(get_pin(spec, pin_idx), 1))
snd_hda_jack_report_sync(codec);+}
+static void hdmi_intrinsic_event(struct hda_codec *codec, unsigned int res) +{ int tag = res >> AC_UNSOL_RES_TAG_SHIFT;
int pin_nid;
int pin_idx; struct hda_jack_tbl *jack; int dev_entry = (res & AC_UNSOL_RES_DE) >> AC_UNSOL_RES_DE_SHIFT;
jack = snd_hda_jack_tbl_get_from_tag(codec, tag); if (!jack) return;
pin_nid = jack->nid; jack->jack_dirty = 1;
_snd_printd(SND_PR_VERBOSE, "HDMI hot plug event: Codec=%d Pin=%d Device=%d Inactive=%d Presence_Detect=%d ELD_Valid=%d\n",
codec->addr, pin_nid, dev_entry, !!(res & AC_UNSOL_RES_IA),
codec->addr, jack->nid, dev_entry, !!(res & AC_UNSOL_RES_IA),
- pin_idx = pin_nid_to_pin_index(spec, pin_nid);
- if (pin_idx < 0)
return;- if (hdmi_present_sense(get_pin(spec, pin_idx), 1))
snd_hda_jack_report_sync(codec);
- jack_callback(codec, jack);
}
static void hdmi_non_intrinsic_event(struct hda_codec *codec, unsigned int res) @@ -2080,7 +2082,8 @@ static int generic_hdmi_init(struct hda_codec *codec) hda_nid_t pin_nid = per_pin->pin_nid;
hdmi_init_pin(codec, pin_nid);
snd_hda_jack_detect_enable(codec, pin_nid, pin_nid);
snd_hda_jack_detect_enable_callback(codec, pin_nid, pin_nid,codec->jackpoll_interval > 0 ? jack_callback : NULL);}
1.7.9.5
2013-12-04 15:17, Takashi Iwai skrev:
At Wed, 4 Dec 2013 10:19:41 +0800, David Henningsson wrote:
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com
sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
Hrm, I vaguely remember that I gave a very similar patch for testing, but it failed by some reason...
That indeed sounds vague. :-) At this point, is there anything required on my part to help getting this bug fixed? We run into it when we're backporting kernel features, i e, using graphics driver from one kernel version and hda driver from another kernel version... (and in this weird kernel mixup we haven't figured out why we don't get an interrupt, so as a workaround I'm using jackpoll_ms until we've figured that out.)
At Wed, 04 Dec 2013 15:24:28 +0800, David Henningsson wrote:
2013-12-04 15:17, Takashi Iwai skrev:
At Wed, 4 Dec 2013 10:19:41 +0800, David Henningsson wrote:
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com
sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
Hrm, I vaguely remember that I gave a very similar patch for testing, but it failed by some reason...
That indeed sounds vague. :-) At this point, is there anything required on my part to help getting this bug fixed? We run into it when we're backporting kernel features, i e, using graphics driver from one kernel version and hda driver from another kernel version... (and in this weird kernel mixup we haven't figured out why we don't get an interrupt, so as a workaround I'm using jackpoll_ms until we've figured that out.)
It's in a thread in June, about [PATCH 1/2] ALSA: hdmi - poll eld at resume time
Takashi
2013-12-04 15:42, Takashi Iwai skrev:
At Wed, 04 Dec 2013 15:24:28 +0800, David Henningsson wrote:
2013-12-04 15:17, Takashi Iwai skrev:
At Wed, 4 Dec 2013 10:19:41 +0800, David Henningsson wrote:
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com
sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
Hrm, I vaguely remember that I gave a very similar patch for testing, but it failed by some reason...
That indeed sounds vague. :-) At this point, is there anything required on my part to help getting this bug fixed? We run into it when we're backporting kernel features, i e, using graphics driver from one kernel version and hda driver from another kernel version... (and in this weird kernel mixup we haven't figured out why we don't get an interrupt, so as a workaround I'm using jackpoll_ms until we've figured that out.)
It's in a thread in June, about [PATCH 1/2] ALSA: hdmi - poll eld at resume time
Takashi
If you mean this patch:
http://mailman.alsa-project.org/pipermail/alsa-devel/2013-June/063336.html
...it would probably resolve this issue, but instead cause problems when jackpoll is not used, because snd_hda_jack_report_sync does not call the jack callback. Essentially, you need to call hdmi_jack_detect_cb from hdmi_intrinsic_event and your patch should work too.
I think - this analysis is all done by quickly looking at your patch, not by actual testing.
At Wed, 04 Dec 2013 15:56:05 +0800, David Henningsson wrote:
2013-12-04 15:42, Takashi Iwai skrev:
At Wed, 04 Dec 2013 15:24:28 +0800, David Henningsson wrote:
2013-12-04 15:17, Takashi Iwai skrev:
At Wed, 4 Dec 2013 10:19:41 +0800, David Henningsson wrote:
In the case of using jackpoll_ms instead of unsol events, the jack was correctly detected, but ELD info was not refreshed on plug-in.
And without ELD info, no proper restriction of pcm, which can in turn break sound output on some devices.
Signed-off-by: David Henningsson david.henningsson@canonical.com
sound/pci/hda/patch_hdmi.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-)
@Takashi; feel free to add stable if you think it's worth it. (After all using the jackpoll_ms parameter is not something we usually do.) It's still a bug fix so sending to Linus tree should be okay, I believe.
Hrm, I vaguely remember that I gave a very similar patch for testing, but it failed by some reason...
That indeed sounds vague. :-) At this point, is there anything required on my part to help getting this bug fixed? We run into it when we're backporting kernel features, i e, using graphics driver from one kernel version and hda driver from another kernel version... (and in this weird kernel mixup we haven't figured out why we don't get an interrupt, so as a workaround I'm using jackpoll_ms until we've figured that out.)
It's in a thread in June, about [PATCH 1/2] ALSA: hdmi - poll eld at resume time
Takashi
If you mean this patch:
http://mailman.alsa-project.org/pipermail/alsa-devel/2013-June/063336.html
...it would probably resolve this issue, but instead cause problems when jackpoll is not used, because snd_hda_jack_report_sync does not call the jack callback. Essentially, you need to call hdmi_jack_detect_cb from hdmi_intrinsic_event and your patch should work too.
I think - this analysis is all done by quickly looking at your patch, not by actual testing.
OK, then I took your patch. It shouldn't break the things, at least, through a quick glance.
thanks,
Takashi
participants (2)
-
David Henningsson -
Takashi Iwai