February 2011 - Alsa-devel - mailman.alsa-project.org

[alsa-devel] question about ctl.pulse ?
by Raymond Yau 02 Feb '11

02 Feb '11

Refer to http://git.alsa-project.org/?p=alsa-plugins.git;a=blob;f=doc/README-pulse;h… The mixer control plugin also accepts the parameters "source" and "sink" for when you need to specify a sink/source combination with different names. If you need to do this with PCM:s then specify two PCM:s with different "device". Does it mean that the application only get playback controls from sink0, sink1 and capture controls from source0, source1, source2 ctl.sink0 { type pulse sink "0" } ctl.source0 { type pulse source "0" } ctl.sink1 { type pulse sink "1" } ctl.source1 { type pulse source "1" } ctl.sink2 { type pulse sink "2" } ctl.source2 { type pulse source "2" }

2 1

[alsa-devel] [PATCH] asoc: davinci: da830/omap-l138: correct cpu_dai_name
by Vaibhav Bedia 02 Feb '11

02 Feb '11

McASP1 is used on the DA830/OMAP-L137 platform for the codec. This is different from the DA850/OMAP-L138 platform which uses McASP0. This is fixed by adding a new snd_soc_dai_link struct. Signed-off-by: Vaibhav Bedia <vaibhav.bedia(a)ti.com> --- sound/soc/davinci/davinci-evm.c | 18 +++++++++++++++--- 1 files changed, 15 insertions(+), 3 deletions(-) diff --git a/sound/soc/davinci/davinci-evm.c b/sound/soc/davinci/davinci-evm.c index b36f0b3..fe79842 100644 --- a/sound/soc/davinci/davinci-evm.c +++ b/sound/soc/davinci/davinci-evm.c @@ -218,7 +218,19 @@ static struct snd_soc_dai_link dm6467_evm_dai[] = { .ops = &evm_spdif_ops, }, }; -static struct snd_soc_dai_link da8xx_evm_dai = { + +static struct snd_soc_dai_link da830_evm_dai = { + .name = "TLV320AIC3X", + .stream_name = "AIC3X", + .cpu_dai_name = "davinci-mcasp.1", + .codec_dai_name = "tlv320aic3x-hifi", + .codec_name = "tlv320aic3x-codec.1-0018", + .platform_name = "davinci-pcm-audio", + .init = evm_aic3x_init, + .ops = &evm_ops, +}; + +static struct snd_soc_dai_link da850_evm_dai = { .name = "TLV320AIC3X", .stream_name = "AIC3X", .cpu_dai_name= "davinci-mcasp.0", @@ -259,13 +271,13 @@ static struct snd_soc_card dm6467_snd_soc_card_evm = { static struct snd_soc_card da830_snd_soc_card = { .name = "DA830/OMAP-L137 EVM", - .dai_link = &da8xx_evm_dai, + .dai_link = &da830_evm_dai, .num_links = 1, }; static struct snd_soc_card da850_snd_soc_card = { .name = "DA850/OMAP-L138 EVM", - .dai_link = &da8xx_evm_dai, + .dai_link = &da850_evm_dai, .num_links = 1, }; -- 1.6.2.4

3 2

[alsa-devel] alsa-lib TLVs and valgrind
by Sebastian H. 02 Feb '11

02 Feb '11

Hello list I'm about to release a new version of alsamixer-qt4 (now named qasmixer) and ran the program through valgrind. valgrind --track-origins=yes src/qasmixer &> valgrind.log The log file contains tons of warnings like this ==3432== Conditional jump or move depends on uninitialised value(s) ... ... (qasmixer function calls) ... ==3432== Uninitialised value was created by a heap allocation ... malloc (vg_replace_malloc.c:236) ... init_db_range (simple_none.c:1111) ... get_dB_range (simple_none.c:1156) ... get_dB_range_ops (simple_none.c:1173) ... snd_mixer_selem_get_playback_dB_range (simple.c:298) ... ... (qasmixer function calls) ... The fresh alsa-lib-10-24.1 was used for this, compiled with debug symbols enabled. Drivers are from Debian/squeeze. This issue doesn't seem critical for the application but it totally floods the valgrind.log file. Best, Sebastian Holtermann

2 4

Re: [alsa-devel] need input on USB Audio class device.
by Clemens Ladisch 02 Feb '11

02 Feb '11

Sakthi Subramanian wrote: > i have opened a capture device, and the captured data i have written to a > file. > That file is of RAW format. > I have opened the raw file in audacity, with 8 channel it shows only 2 > channels having the captured data. Raw-format files do not have any information about the format, so Audacity has to guess. Just write a .wav file with the proper headers. Regards, Clemens

1 0

[alsa-devel] [PATCH] ASoC: soc-core: Remove display_register() callback
by Dimitris Papastamos 02 Feb '11

02 Feb '11

Remove the display_register() callback as there are no users of it in the upstream kernel and makes it difficult to unify the format of the codec_reg file. Signed-off-by: Dimitris Papastamos <dp(a)opensource.wolfsonmicro.com> --- include/sound/soc.h | 2 -- sound/soc/soc-core.c | 31 +++++++++++++------------------ 2 files changed, 13 insertions(+), 20 deletions(-) diff --git a/include/sound/soc.h b/include/sound/soc.h index 4b6c0a8..2e6463f 100644 --- a/include/sound/soc.h +++ b/include/sound/soc.h @@ -536,8 +536,6 @@ struct snd_soc_codec_driver { /* codec IO */ unsigned int (*read)(struct snd_soc_codec *, unsigned int); int (*write)(struct snd_soc_codec *, unsigned int, unsigned int); - int (*display_register)(struct snd_soc_codec *, char *, - size_t, unsigned int); int (*volatile_register)(struct snd_soc_codec *, unsigned int); int (*readable_register)(struct snd_soc_codec *, unsigned int); short reg_cache_size; diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index 205cbd7..4c061c3 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -110,24 +110,19 @@ static ssize_t soc_codec_reg_show(struct snd_soc_codec *codec, char *buf) if (count >= PAGE_SIZE - 1) break; - if (codec->driver->display_register) { - count += codec->driver->display_register(codec, buf + count, - PAGE_SIZE - count, i); - } else { - /* If the read fails it's almost certainly due to - * the register being volatile and the device being - * powered off. - */ - ret = snd_soc_read(codec, i); - if (ret >= 0) - count += snprintf(buf + count, - PAGE_SIZE - count, - "%.*x", wordsize, ret); - else - count += snprintf(buf + count, - PAGE_SIZE - count, - "<no data: %d>", ret); - } + /* If the read fails it's almost certainly due to + * the register being volatile and the device being + * powered off. + */ + ret = snd_soc_read(codec, i); + if (ret >= 0) + count += snprintf(buf + count, + PAGE_SIZE - count, + "%.*x", wordsize, ret); + else + count += snprintf(buf + count, + PAGE_SIZE - count, + "<no data: %d>", ret); if (count >= PAGE_SIZE - 1) break; -- 1.7.3.5

3 4

[alsa-devel] [PATCH] ASoC: soc-core: Support debugfs entries larger than PAGE_SIZE bytes
by Dimitris Papastamos 02 Feb '11

02 Feb '11

For some codecs with large register maps, it was not possible to dump all registers via the codec_reg file but only up to PAGE_SIZE bytes. This patch fixes this problem. Signed-off-by: Dimitris Papastamos <dp(a)opensource.wolfsonmicro.com> --- sound/soc/soc-core.c | 114 +++++++++++++++++++++++++++++++++---------------- 1 files changed, 77 insertions(+), 37 deletions(-) diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index 205cbd7..a7710d2 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -87,15 +87,55 @@ static int min_bytes_needed(unsigned long val) return c; } +/* fill buf which is 'len' bytes with a formatted + * string of the form 'reg: value\n' */ +static int format_register_str(struct snd_soc_codec *codec, + unsigned int regidx, char *buf, size_t len) +{ + int wordsize = codec->driver->reg_word_size * 2; + int regsize = min_bytes_needed(codec->driver->reg_cache_size) * 2; + int ret; + char tmpbuf[len + 1]; + + /* since tmpbuf is allocated on the stack, warn the callers if they + * try to abuse this function */ + WARN_ON(len > 63); + + /* +2 for ': ' and + 1 for '\n' */ + if (wordsize + regsize + 2 + 1 != len) + return -EINVAL; + + /* we don't currently handle failed reads */ + ret = snd_soc_read(codec , regidx); + if (ret < 0) { + dev_err(codec->dev, "Failed to read from %#x\n", regidx); + return ret; + } + + /* prepare the buffer */ + snprintf(tmpbuf, len + 1, "%.*x: %.*x\n", wordsize, + regidx, regsize, ret); + /* copy it back to the caller without the '\0' */ + memcpy(buf, tmpbuf, len); + + return 0; +} + /* codec register dump */ -static ssize_t soc_codec_reg_show(struct snd_soc_codec *codec, char *buf) +static ssize_t soc_codec_reg_show(struct snd_soc_codec *codec, char *buf, + size_t count, loff_t pos) { - int ret, i, step = 1, count = 0; + int i, step = 1; int wordsize, regsize; + int len; + size_t total = 0; + loff_t p = 0; wordsize = codec->driver->reg_word_size * 2; regsize = min_bytes_needed(codec->driver->reg_cache_size) * 2; + len = wordsize + regsize + 2 + 1; + if (!codec->driver->reg_cache_size) return 0; @@ -105,51 +145,34 @@ static ssize_t soc_codec_reg_show(struct snd_soc_codec *codec, char *buf) for (i = 0; i < codec->driver->reg_cache_size; i += step) { if (codec->readable_register && !codec->readable_register(codec, i)) continue; - - count += sprintf(buf + count, "%.*x: ", regsize, i); - if (count >= PAGE_SIZE - 1) - break; - if (codec->driver->display_register) { count += codec->driver->display_register(codec, buf + count, PAGE_SIZE - count, i); } else { - /* If the read fails it's almost certainly due to - * the register being volatile and the device being - * powered off. - */ - ret = snd_soc_read(codec, i); - if (ret >= 0) - count += snprintf(buf + count, - PAGE_SIZE - count, - "%.*x", wordsize, ret); - else - count += snprintf(buf + count, - PAGE_SIZE - count, - "<no data: %d>", ret); + /* only support larger than PAGE_SIZE bytes debugfs + * entries for the default case */ + if (p >= pos) { + if (total + len >= count - 1) + break; + format_register_str(codec, i, buf + total, len); + total += len; + } + p += len; } - - if (count >= PAGE_SIZE - 1) - break; - - count += snprintf(buf + count, PAGE_SIZE - count, "\n"); - if (count >= PAGE_SIZE - 1) - break; } - /* Truncate count; min() would cause a warning */ - if (count >= PAGE_SIZE) - count = PAGE_SIZE - 1; + total = min(total, count - 1); - return count; + return total; } + static ssize_t codec_reg_show(struct device *dev, struct device_attribute *attr, char *buf) { struct snd_soc_pcm_runtime *rtd = container_of(dev, struct snd_soc_pcm_runtime, dev); - return soc_codec_reg_show(rtd->codec, buf); + return soc_codec_reg_show(rtd->codec, buf, PAGE_SIZE, 0); } static DEVICE_ATTR(codec_reg, 0444, codec_reg_show, NULL); @@ -188,16 +211,33 @@ static int codec_reg_open_file(struct inode *inode, struct file *file) } static ssize_t codec_reg_read_file(struct file *file, char __user *user_buf, - size_t count, loff_t *ppos) + size_t count, loff_t *ppos) { ssize_t ret; struct snd_soc_codec *codec = file->private_data; - char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL); + char *buf; + + if (*ppos < 0 || !count) + return -EINVAL; + + buf = kmalloc(count, GFP_KERNEL); if (!buf) return -ENOMEM; - ret = soc_codec_reg_show(codec, buf); - if (ret >= 0) - ret = simple_read_from_buffer(user_buf, count, ppos, buf, ret); + + ret = soc_codec_reg_show(codec, buf, count, *ppos); + if (ret >= 0) { + if (!access_ok(VERIFY_WRITE, user_buf, ret)) { + ret = -EFAULT; + goto out; + } + if (copy_to_user(user_buf, buf, ret)) { + ret = -EFAULT; + goto out; + } + *ppos += ret; + } + +out: kfree(buf); return ret; } -- 1.7.3.5

2 3

[alsa-devel] [PATCH] ASoC: Amstrad Delta: fix NULL pointer dereference
by Janusz Krzysztofik 02 Feb '11

02 Feb '11

It looks like the .card member of the snd_soc_pcm_runtime structure pointed to by the snd_soc_dai_link.init() argument is no longer initialized before the function being called. As the Amstrad Delta sound card driver was making use of it for locating its snd_soc_card structure, this resulted in actual or potential NULL pointer dereferences. Since the structure in question is local to the driver, its address can be referred directly instead of trying to derive it from the function argument. Fix it. Created and tested against linux-2.6.38-rc2 Signed-off-by: Janusz Krzysztofik <jkrzyszt(a)tis.icnet.pl> --- Please consider queuing this patch for the current rc cycle. sound/soc/omap/ams-delta.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) --- git/sound/soc/omap/ams-delta.c.orig 2011-01-31 19:49:29.000000000 +0100 +++ git/sound/soc/omap/ams-delta.c 2011-02-01 17:54:22.000000000 +0100 @@ -492,12 +492,14 @@ static void ams_delta_shutdown(struct sn * Card initialization */ +static struct snd_soc_card ams_delta_audio_card; + static int ams_delta_cx20442_init(struct snd_soc_pcm_runtime *rtd) { struct snd_soc_codec *codec = rtd->codec; struct snd_soc_dapm_context *dapm = &codec->dapm; struct snd_soc_dai *codec_dai = rtd->codec_dai; - struct snd_soc_card *card = rtd->card; + struct snd_soc_card *card = &ams_delta_audio_card; int ret; /* Codec is ready, now add/activate board specific controls */

2 2

Re: [alsa-devel] problem with mms stream - [AO_ALSA] Trying to reset soundcard
by Raymond Yau 02 Feb '11

02 Feb '11

>Приветствую, Reimar > On Sun, Jan 24, 2010 at 02:36:50PM +0300, Goga777 wrote: > > [AO_ALSA] Write error: Broken pipe4428928.0 (-24.-8) 96.4% 6% > > [AO_ALSA] Trying to reset soundcard. > > Increasing filtered audio buffer size from 65536 to 675361% 9% > > Increasing filtered audio buffer size from 67536 to 675442% 12% > > [AO_ALSA] Write error: Broken pipe4428928.0 (-24.-8) 24.0% 6% > > [AO_ALSA] Trying to reset soundcard. > > I am not sure this is the same issue, but I had this part as well. res = snd_pcm_writei(alsa_handler, data, num_frames); ... if (res < 0) { mp_msg(MSGT_AO,MSGL_ERR,MSGTR_AO_ALSA_WriteError, snd_strerror(res)); mp_msg(MSGT_AO,MSGL_INFO,MSGTR_AO_ALSA_TryingToResetSoundcard); if ((res = snd_pcm_prepare(alsa_handler)) < 0) { mp_msg(MSGT_AO,MSGL_ERR,MSGTR_AO_ALSA_PcmPrepareError, snd_strerror(res)); return 0; break; } } The "pulse" plugin return -EPIPE "Broken pipe" at four conditions after mplayer call snd_pcm_writei() 1) io->state == SND_PCM_STATE_XRUN 2) pcm->underrun 3) update_ptr(pcm) return err mplayer call snd_pcm_prepare() to recover from "Broken Pipe". but why mplayer increasing filtered audio buffer ? static snd_pcm_sframes_t pulse_pointer(snd_pcm_ioplug_t * io) { ... if (io->state == SND_PCM_STATE_XRUN) return -EPIPE; if (pcm->underrun) { ret = -EPIPE; goto finish; } if (io->state != SND_PCM_STATE_RUNNING) return 0; pa_threaded_mainloop_lock(pcm->p->mainloop); ret = check_stream(pcm); if (ret < 0) goto finish; if (pcm->underrun) { ret = -EPIPE; goto finish; } ret = update_ptr(pcm); if (ret < 0) { ret = -EPIPE; goto finish; } if (pcm->underrun) ret = -EPIPE; else ret = snd_pcm_bytes_to_frames(io->pcm, pcm->ptr); finish:

1 0

Re: [alsa-devel] [PATCH] ASoC: EDB93xx machine sound driver with CS4271
by Alexander Sverdlin 02 Feb '11

02 Feb '11

Dear Todor, > Zdrastvuj Aleks, > I'd like to test the sound module. Which is the latest patch versions > and > coresponding kernel version? In which official kernel release do you > plan to > merge it? As I can conclude from list, you should see it in 2.6.39 and possibly -next already. 75% of the driver has been merged already. All you have to do to start testing is to add patch that I've just sent to the list ([PATCH] EDB93xx: Add support for CS4271 CODEC on EDB93xx boards). Best regards, Alexander.

1 0

[alsa-devel] snd_soc_jack_add_gpios vs. SOC_DAPM_PIN_SWITCH
by Stephen Warren 01 Feb '11

01 Feb '11

I've set up headphone hack detection in my machine driver, using snd_soc_jack_new, snd_soc_jack_add_pins, and snd_soc_jack_add_gpios. I've also created a control to allow the user to enable/disable the headphone jack: static const struct snd_kcontrol_new harmony_controls[] = { SOC_DAPM_PIN_SWITCH("Headphone Jack"), }; On their own, both of these work fine. However, if I use the control to disable the headphones, then unplug/replug the headphones, the jack detect seems to override the control, and I hear audio from the headphones. Is this something that's known not yet implemented or was never intended to be used this way, or should I look further to find the issue? FYI, sound/soc/samsung/smartq_wm8987.c has this same code structure, and I imagine will have the same issue. Thanks. -- nvpublic

2 3