Regression 2.6.35-rc6: ALSA Intel HDA/Realtek: missing Beep
in [Kernel]
|
Prev: linux-next: please pull in from lost-spurious-irq branch
Next: x86: Add memory modify constraints to xchg() and cmpxchg()
From: Takashi Iwai on 28 Jul 2010 11:40 At Wed, 28 Jul 2010 16:49:03 +0200, Mario 'BitKoenig' Holbe wrote: > > Hello, > > My PC-Speaker Beep control worked in 2.6.34, but is gone in 2.6.35-rc6. > The sound device is an Asus P5E-V HDMI (Intel G35) onboard Intel HDA w/ > Realtek ALC883 codec. > > The driver does still register a PCBeep input, but the Controls as well > as the sound (:)) are gone. > > HDA Intel 0000:00:1b.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22 > HDA Intel 0000:00:1b.0: irq 47 for MSI/MSI-X > HDA Intel 0000:00:1b.0: setting latency timer to 64 > input: HDA Digital PCBeep as /devices/pci0000:00/0000:00:1b.0/input/input5 > > $ diff card0-codec#0-2.6.34 card0-codec#0-2.6.35-rc6 > 111,114d110 > < Control: name="Beep Playback Volume", index=0, device=0 > < ControlAmp: chs=3, dir=In, idx=5, ofs=0 > < Control: name="Beep Playback Switch", index=0, device=0 > < ControlAmp: chs=3, dir=In, idx=5, ofs=0 > > Attached /proc/asound/card0/codec#0 from 2.6.35-rc6. It's because now the driver checks the SSID your board sets up. Realtek codecs suppose SSID containing some useful bits to inform the h/w setups. The presence of PC beep is one of it. So, it's actually BIOS that clears it. In the earlier version, the driver didn't check this. Actually, the real bug is that it still creates a beep device without mixers. This should be avoided... thanks, Takashi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Takashi Iwai on 29 Jul 2010 01:50 At Wed, 28 Jul 2010 23:27:46 +0200, Mario 'BitKoenig' Holbe wrote: > > [1 <text/plain; us-ascii (quoted-printable)>] > On Wed, Jul 28, 2010 at 06:03:29PM +0200, Takashi Iwai wrote: > > I wrote: > > > Mario 'BitKoenig' Holbe wrote: > > > > My PC-Speaker Beep control worked in 2.6.34, but is gone in 2.6.35-rc6. > > > It's because now the driver checks the SSID your board sets up. > > > So, it's actually BIOS that clears it. > > But the BIOS itself beeps through the sound-card at boot :/ But BIOS tells that the HD-audio codec shouldn't use, so the driver follows it. > > Or, does the following patch fix? It's already in sound git tree, > > Nope, unfortunately it doesn't. Still no Beep controls, no beep through > sound-card. > > May I somehow provide any further data? Please give alsa-info.sh output instead of codec proc file. It's more comprehensive. > Or am I somehow able to tweak it? Is there a module parameter to set > this SSID bit? I mean, it did work before... :) With the patch below, you'll likely have back the system beep sound. But it doesn't go through codec, thus no volume control. thanks, Takashi --- From 8af2591d6342a9e4bb79b4f1236246a79d20ebee Mon Sep 17 00:00:00 2001 From: Takashi Iwai <tiwai(a)suse.de> Date: Wed, 28 Jul 2010 17:37:16 +0200 Subject: [PATCH] ALSA: hda - Don't register beep input device when no beep is available We check now the availability of PC beep and skip the build of beep mixers, but the driver still registers the input device. This should be checked as well. Signed-off-by: Takashi Iwai <tiwai(a)suse.de> --- sound/pci/hda/patch_realtek.c | 32 +++++++++++++++++++------------- 1 files changed, 19 insertions(+), 13 deletions(-) diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c index ff614dd..d7fd846 100644 --- a/sound/pci/hda/patch_realtek.c +++ b/sound/pci/hda/patch_realtek.c @@ -10566,10 +10566,12 @@ static int patch_alc882(struct hda_codec *codec) } } - err = snd_hda_attach_beep_device(codec, 0x1); - if (err < 0) { - alc_free(codec); - return err; + if (spec->cdefine.enable_pcbeep) { + err = snd_hda_attach_beep_device(codec, 0x1); + if (err < 0) { + alc_free(codec); + return err; + } } if (board_config != ALC882_AUTO) @@ -12435,7 +12437,7 @@ static int patch_alc262(struct hda_codec *codec) } } - if (!spec->no_analog) { + if (!spec->no_analog && spec->cdefine.enable_pcbeep) { err = snd_hda_attach_beep_device(codec, 0x1); if (err < 0) { alc_free(codec); @@ -14458,10 +14460,12 @@ static int patch_alc269(struct hda_codec *codec) } } - err = snd_hda_attach_beep_device(codec, 0x1); - if (err < 0) { - alc_free(codec); - return err; + if (spec->cdefine.enable_pcbeep) { + err = snd_hda_attach_beep_device(codec, 0x1); + if (err < 0) { + alc_free(codec); + return err; + } } if (board_config != ALC269_AUTO) @@ -18691,10 +18695,12 @@ static int patch_alc662(struct hda_codec *codec) } } - err = snd_hda_attach_beep_device(codec, 0x1); - if (err < 0) { - alc_free(codec); - return err; + if (spec->cdefine.enable_pcbeep) { + err = snd_hda_attach_beep_device(codec, 0x1); + if (err < 0) { + alc_free(codec); + return err; + } } if (board_config != ALC662_AUTO) -- 1.7.2 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Takashi Iwai on 29 Jul 2010 04:30 At Thu, 29 Jul 2010 10:09:30 +0200, Mario 'BitKoenig' Holbe wrote: > > On Thu, Jul 29, 2010 at 07:44:38AM +0200, Takashi Iwai wrote: > > Mario 'BitKoenig' Holbe wrote: > > > But the BIOS itself beeps through the sound-card at boot :/ > > But BIOS tells that the HD-audio codec shouldn't use, so the driver > > follows it. > > Even grub's beep (play 480 440 1) goes through the sound-card. So it > seems like the BIOS leaves everything set up working as well. Well, my point is that BIOS gives the wrong SSID value. It's bad that the value is compliant with Realtek codecs specification, but it clears the PC-beep bit explicitly. In most cases, SSID value isn't compliant (so BIOS gives a wrong but it's less wrong :), thus the driver takes a fallback to PC-beep enabled. > Please don't get me wrong. I'm not saying the driver does something > wrong, I'm sure it doesn't. I'm just sure that if I could convince it to > behave as if it would have detected the Beep pin everything would work > fine again because it did before... FYI, you can change SSID by writing /sys/class/sound/hwC*D*/subsystem_id file. Then reconfigure the codec via triggering /sys/class/sound/hwC*D*/reconfig. > > Please give alsa-info.sh output instead of codec proc file. It's more > > comprehensive. > > Attached. > > This is from a kernel with both patches applied you sent me. > > > With the patch below, you'll likely have back the system beep sound. > > Nope, no sound. But I guess this wasn't the intention of the patch. Now, > no beep input is registered anymore - which was the intention, I guess. It's the intention. If SSID set by BIOS doesn't indicate the presence of PC-speaker bit *explicitly*, we shouldn't touch beep stuff in codec. > > But it doesn't go through codec, thus no volume control. > > If you mean it should go through the 5V PC Speaker (i.e. pcspkr) - I > don't have such a thing connected. I always appreciated having volume- > and mute control over the beep at night when you can't get away from > work but don't like to wake up anybody just because command completion > beeps, because you pasted something in the wrong window, or whatever. For the driver, the outside connection doesn't matter. The information is given as the spec bit, and just follows it :) So, the fix is likely to override SSID value, or create a special quirk rule to enable PC-beep for known white-list, supposing BIOS won't be fixed in any future... Takashi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Takashi Iwai on 29 Jul 2010 05:50 At Thu, 29 Jul 2010 11:15:17 +0200, Mario 'BitKoenig' Holbe wrote: > > On Thu, Jul 29, 2010 at 10:52:36AM +0200, Takashi Iwai wrote: > > Usually the codec SSID isn't checked in other places, so passing a > > bogus value should be OK. Pass a value like 2: > > Mh, 1 probably :) > > $ cat /etc/modprobe.d/local-alsa.conf > options snd-hda-intel model=asus-p5q > install snd-hda-intel /sbin/modprobe --ignore-install snd-hda-intel $CMDLINE_OPTS && { cd /sys/class/sound/hwC0D0; echo -n 1 > subsystem_id; echo -n 1 > reconfig; : ; } > > Yep, Beep is back :) > > Btw... > > On Thu, Jul 29, 2010 at 10:26:22AM +0200, Takashi Iwai wrote: > > So, the fix is likely to override SSID value, or create a special > > quirk rule to enable PC-beep for known white-list, supposing BIOS > > won't be fixed in any future... > > Well, the Board is from 2007, the last BIOS is from Aug 2009. I don't > think Asus will provide an update just for that :) OK, then try the patch below (over my previous two patches). It enables PC-beep for your device forcibly. Takashi --- diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c index 1744d4d..439d6e7 100644 --- a/sound/pci/hda/patch_realtek.c +++ b/sound/pci/hda/patch_realtek.c @@ -5280,8 +5280,24 @@ static void fillup_priv_adc_nids(struct hda_codec *codec, hda_nid_t *nids, #ifdef CONFIG_SND_HDA_INPUT_BEEP #define set_beep_amp(spec, nid, idx, dir) \ ((spec)->beep_amp = HDA_COMPOSE_AMP_VAL(nid, 3, idx, dir)) + +static struct snd_pci_quirk beep_white_list[] = { + SND_PCI_QUIRK(0x1043, 0x829f, "ASUS", 1), + {} +}; + +static inline int has_cdefine_beep(struct hda_codec *codec) +{ + struct alc_spec *spec = codec->spec; + const struct snd_pci_quirk *q; + q = snd_pci_quirk_lookup(codec->bus->pci, beep_white_list); + if (q) + return q->value; + return spec->cdefine.enable_pcbeep; +} #else #define set_beep_amp(spec, nid, idx, dir) /* NOP */ +#define has_cdefine_beep(codec) 0 #endif /* @@ -10666,7 +10682,7 @@ static int patch_alc882(struct hda_codec *codec) } } - if (spec->cdefine.enable_pcbeep) { + if (has_cdefine_beep(codec)) { err = snd_hda_attach_beep_device(codec, 0x1); if (err < 0) { alc_free(codec); @@ -10721,7 +10737,7 @@ static int patch_alc882(struct hda_codec *codec) set_capture_mixer(codec); - if (spec->cdefine.enable_pcbeep) + if (has_cdefine_beep(codec)) set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT); if (board_config == ALC882_AUTO) @@ -12537,7 +12553,7 @@ static int patch_alc262(struct hda_codec *codec) } } - if (!spec->no_analog && spec->cdefine.enable_pcbeep) { + if (!spec->no_analog && has_cdefine_beep(codec)) { err = snd_hda_attach_beep_device(codec, 0x1); if (err < 0) { alc_free(codec); @@ -12588,7 +12604,7 @@ static int patch_alc262(struct hda_codec *codec) } if (!spec->cap_mixer && !spec->no_analog) set_capture_mixer(codec); - if (!spec->no_analog && spec->cdefine.enable_pcbeep) + if (!spec->no_analog && has_cdefine_beep(codec)) set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT); spec->vmaster_nid = 0x0c; @@ -14593,7 +14609,7 @@ static int patch_alc269(struct hda_codec *codec) } } - if (spec->cdefine.enable_pcbeep) { + if (has_cdefine_beep(codec)) { err = snd_hda_attach_beep_device(codec, 0x1); if (err < 0) { alc_free(codec); @@ -14635,7 +14651,7 @@ static int patch_alc269(struct hda_codec *codec) if (!spec->cap_mixer) set_capture_mixer(codec); - if (spec->cdefine.enable_pcbeep) + if (has_cdefine_beep(codec)) set_beep_amp(spec, 0x0b, 0x04, HDA_INPUT); if (board_config == ALC269_AUTO) @@ -18832,7 +18848,7 @@ static int patch_alc662(struct hda_codec *codec) } } - if (spec->cdefine.enable_pcbeep) { + if (has_cdefine_beep(codec)) { err = snd_hda_attach_beep_device(codec, 0x1); if (err < 0) { alc_free(codec); @@ -18859,7 +18875,7 @@ static int patch_alc662(struct hda_codec *codec) if (!spec->cap_mixer) set_capture_mixer(codec); - if (spec->cdefine.enable_pcbeep) { + if (has_cdefine_beep(codec)) { switch (codec->vendor_id) { case 0x10ec0662: set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT); -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Takashi Iwai on 29 Jul 2010 09:40
At Thu, 29 Jul 2010 15:25:32 +0200, Mario 'BitKoenig' Holbe wrote: > > On Thu, Jul 29, 2010 at 11:42:44AM +0200, Takashi Iwai wrote: > > OK, then try the patch below (over my previous two patches). It > > enables PC-beep for your device forcibly. > > Yes, it does and it works. Thank you very much again. > > I just checked 2 more boards - the P5E-VM HDMI (P5E-V HDMI's little > brother) is already covered by the quirk and P5Q-EM (G45) seems to > specify it correctly (subsystem_id = 0x104382fe - seems I got the logic > wrong before - LSB isn't set here and it works). So, it's likely a bad luck about the SSID of P5E-V. I guess this unexpectedly matches with the value Realtek codec expected, and unluckily the value had side-effects. Anyway, I applied the patch now. thanks, Takashi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |