October 2017 - Alsa-devel - mailman.alsa-project.org

[alsa-devel] Error patching file pcm_native.c
by Delusion's Master 20 Nov '24

20 Nov '24

Hi, first of all I apologize in case you receive multiple copies of this message, but I forgot to specify a subject in the original mail. I have a problem compiling the latest alsa driver snapshots. Specifically, it's a problem when patching pcm_native.c: [..] copying file alsa-kernel/core/info.c patching file info.c copying file alsa-kernel/core/pcm.c patching file pcm.c Hunk #3 succeeded at 739 (offset 2 lines). Hunk #5 succeeded at 954 (offset 2 lines). Hunk #7 succeeded at 1045 (offset 2 lines). copying file alsa-kernel/core/pcm_native.c patching file pcm_native.c Hunk #4 succeeded at 1544 (offset -25 lines). Hunk #5 succeeded at 2827 (offset 12 lines). Hunk #6 succeeded at 2878 (offset 10 lines). Hunk #7 FAILED at 2906. Hunk #8 FAILED at 2922. Hunk #9 succeeded at 2915 (offset -23 lines). Hunk #10 succeeded at 3036 (offset 10 lines). Hunk #11 succeeded at 3027 (offset -23 lines). Hunk #12 succeeded at 3074 (offset 10 lines). Hunk #13 succeeded at 3098 (offset -23 lines). Hunk #14 succeeded at 3143 (offset 10 lines). Hunk #15 succeeded at 3124 (offset -23 lines). Hunk #16 succeeded at 3214 (offset 10 lines). Hunk #17 succeeded at 3208 (offset -23 lines). Hunk #18 succeeded at 3271 (offset 10 lines). Hunk #19 succeeded at 3304 (offset -23 lines). Hunk #20 succeeded at 3386 (offset 10 lines). Hunk #21 succeeded at 3411 (offset -23 lines). Hunk #22 succeeded at 3456 (offset 10 lines). Hunk #23 succeeded at 3450 (offset -23 lines). Hunk #24 succeeded at 3525 (offset 10 lines). Hunk #25 succeeded at 3644 (offset -23 lines). 2 out of 25 hunks FAILED -- saving rejects to file pcm_native.c.rej make[2]: *** [pcm_native.c] Error 1 It's some days I get this, but I waited to report this since I thought it could be a temporary problem. My OS is Fedora 11 x86_64. Previous snapshots didn't give me this problem, but I can't say which one broke the compilation, since I usually try updated snapshots only at kernel updates. I searched for this error and the only relevant report I could find is this one: http://mailman.alsa-project.org/pipermail/alsa-devel/2009-April/016420.html a report from last April. What could the problem be? Thanks _________________________________________________________________ Invite your mail contacts to join your friends list with Windows Live Spaces. It's easy! http://spaces.live.com/spacesapi.aspx?wx_action=create&wx_url=/friends.aspx…

3 10

[alsa-devel] unsupported variant of snd_hda_intel
by Alexandre 04 Jul '24

04 Jul '24

Hello, I have an Acer Aspire 4736G laptop. The soundcard that is shipped in this laptop is intel-hda based as shown (actually it's a realtek): cat /proc/asound/cards 0 [Intel ]: HDA-Intel - HDA Intel HDA Intel at 0x9b300000 irq 46 Basic sound output is correct but few feature are missing: - internal mic seems to be found but do not works. - sound output stop working when trying to send it using HDMI output - heads do not work when plugged neither when forcing pulseaudio to use it. I have an ubuntu 10.10 which is shipped with alsa 1.0.23. I have googled around and tried loading the snd-hda-intel with several parameters () as found in HD-Audio-Models.txt to no avail. cat /proc/asound/pcm 00-00: ALC888 Analog : ALC888 Analog : playback 1 : capture 1 00-01: ALC888 Digital : ALC888 Digital : playback 1 00-03: NVIDIA HDMI : NVIDIA HDMI : playback 1 Where can I begin to help implement the support for this sound card? I have very skills in developement in general but am ready to try if it is not to hard (I supose I can use previous code for similar models).. I just need a shepherd. Thanks

4 4

[alsa-devel] echo-cancellation plugin advice?
by Max Klein 27 Oct '23

27 Oct '23

> Well, asym wouldn't be the answer. Basically you'll need to open two > slave PCMs in one plugin. So, it's no simple filter-type plugin. > It's doable, but not easy as it sounds, I guess. Takashi, can you please further explain why asym is not the solution and what you see as a problem? I am also trying to figure out if it would be possible to create such a plugin for ALSA? I am still searching for some good starting points on how to write an ALSA plugin... Jean-Marc Valin (http://lists.linuxaudio.org/pipermail/linux-audio-dev/2005-June/012454.html) already released his work as part of speex but as far I know there is no plugin that has been made out of it? Thanks for any hint! -- NEU: FreePhone 3-fach-Flat mit kostenlosem Smartphone! Jetzt informieren: http://mobile.1und1.de/?ac=OM.PW.PW003K20328T7073a

3 2

[alsa-devel] Attempting to understand odd snd-usb-audio code and behavior
by Keith A. Milner 21 Aug '20

21 Aug '20

Over the last few weeks I've been trying to dig through the snd-usb-audio code to try to get an understanding of it. I've used ftrace and also a local version which I have heavily polluted with kprintf(KERN_INFO statements so I can track what it's doing in practice. I've been playing with this with some USB audio interfaces I have some which work, and some which don't, and trying to understand the differences and, ultimately, what I can do to make the non-functioning ones work. I've got to the point where I have a fairly good understanding of some parts, but I'm struggling to understand why the code is behaving as it does in practice. My observation when plugging the USB device in is that, as expected, it tries to build the audio device, and add components to it including attaching capture and playback streams. It seems to go through this phase without too many surprises, and we get to the standard dmesg output: usbcore: registered new interface driver snd-usb-audio After this, it gets confusing. It seems to open and close the PCMs (e.g, snd_usb_capture_open) several times in a row. Each time it sets up the hardware info, checks and sets the format and then closes the PCM. I don't see anything obviously different between most of these cycles. Here's an example my logging output (hopefully this will make sense): Aug 22 20:04:16 kerneldev kernel: [ 923.140789] Called: pcm.c - snd_usb_capture_open for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.140789] Called: pcm.c - snd_usb_pcm_open for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.140789] Setting up hw info Aug 22 20:04:16 kerneldev kernel: [ 923.140789] Checking format: Aug 22 20:04:16 kerneldev kernel: [ 923.140790] Interface number = 2 Aug 22 20:04:16 kerneldev kernel: [ 923.140790] alt setting = 1 Aug 22 20:04:16 kerneldev kernel: [ 923.140790] alt setting index = 1 Aug 22 20:04:16 kerneldev kernel: [ 923.140790] format type = 1 Aug 22 20:04:16 kerneldev kernel: [ 923.140790] frame size = 0 Aug 22 20:04:16 kerneldev kernel: [ 923.140791] Setting pt=125 ptmin=125 Aug 22 20:04:16 kerneldev kernel: [ 923.140792] Checking for unusual bit rates Aug 22 20:04:16 kerneldev kernel: [ 923.140794] Called: init.c - snd_card_file_remove Aug 22 20:04:16 kerneldev kernel: [ 923.140921] Called: pcm.c - snd_usb_hw_params for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.140945] Setting format: Aug 22 20:04:16 kerneldev kernel: [ 923.140945] Bytes per period = 8816 Aug 22 20:04:16 kerneldev kernel: [ 923.140945] Frames per period = 1102 Aug 22 20:04:16 kerneldev kernel: [ 923.140946] Periods per buffer = 4 Aug 22 20:04:16 kerneldev kernel: [ 923.140946] Channels = 2 Aug 22 20:04:16 kerneldev kernel: [ 923.140946] Rate = 44100 Aug 22 20:04:16 kerneldev kernel: [ 923.141279] Called: endpoint.c - snd_usb_add_endpoint of type 0 with direction 1 Aug 22 20:04:16 kerneldev kernel: [ 923.141292] Called: pcm.c - snd_usb_pcm_prepare for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.141293] Called: pcm.c - configure_endpoint Aug 22 20:04:16 kerneldev kernel: [ 923.141294] Called: endpoint.c - snd_usb_endpoint_set_params for endpoint 142, type 0 Aug 22 20:04:16 kerneldev kernel: [ 923.141294] Setting endpoint params for endpoint num 142 Aug 22 20:04:16 kerneldev kernel: [ 923.141294] channels=2 Aug 22 20:04:16 kerneldev kernel: [ 923.141294] period bytes=8816 Aug 22 20:04:16 kerneldev kernel: [ 923.141295] frames per period=1102 Aug 22 20:04:16 kerneldev kernel: [ 923.141295] periods per buffer=4 Aug 22 20:04:16 kerneldev kernel: [ 923.141296] Called: endpoint.c - data_ep_set_params for endpoint 142, type 0 Aug 22 20:04:16 kerneldev kernel: [ 923.141301] Called: pcm.c - snd_usb_pcm_prepare for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.142768] Called: init.c - snd_card_file_add Aug 22 20:04:16 kerneldev kernel: [ 923.142772] Called: init.c - snd_card_file_remove Aug 22 20:04:16 kerneldev kernel: [ 923.142774] Called: init.c - snd_card_file_add Aug 22 20:04:16 kerneldev kernel: [ 923.142830] Called: init.c - snd_card_file_remove Aug 22 20:04:16 kerneldev kernel: [ 923.142836] Called: pcm.c - snd_usb_hw_free for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.142837] Called: endpoint.c - snd_usb_endpoint_deactivate for endpoint 142, type 0 Aug 22 20:04:16 kerneldev kernel: [ 923.142846] Called: pcm.c - snd_usb_capture_close for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.146760] Called: init.c - snd_card_file_remove Aug 22 20:04:16 kerneldev kernel: [ 923.146834] Called: init.c - snd_card_file_add Aug 22 20:04:16 kerneldev kernel: [ 923.146838] Called: init.c - snd_card_file_remove Aug 22 20:04:16 kerneldev kernel: [ 923.146841] Called: init.c - snd_card_file_add Aug 22 20:04:16 kerneldev kernel: [ 923.146845] Called: init.c - snd_card_file_add Aug 22 20:04:16 kerneldev kernel: [ 923.146848] Called: pcm.c - snd_usb_capture_open for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.146849] Called: pcm.c - snd_usb_pcm_open for pcm substream 0 Aug 22 20:04:16 kerneldev kernel: [ 923.146849] Setting up hw info It seems to do this several times, before the behavior changing slightly, as it now starts to try to set the endpoint parameters and seems to leave the stream open. However, it then tries to open the stream and set the endpoint parameters again resulting in errors: Aug 22 20:22:00 kerneldev kernel: [ 1987.452945] Called: endpoint.c - snd_usb_endpoint_set_params for endpoint 142, type 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.452945] Setting endpoint params for endpoint num 142 Aug 22 20:22:00 kerneldev kernel: [ 1987.452945] channels=2 Aug 22 20:22:00 kerneldev kernel: [ 1987.452945] period bytes=1768 Aug 22 20:22:00 kerneldev kernel: [ 1987.452946] frames per period=221 Aug 22 20:22:00 kerneldev kernel: [ 1987.452946] periods per buffer=593 Aug 22 20:22:00 kerneldev kernel: [ 1987.452947] usb 1-1: Unable to change format on ep #8e: already in use Maybe this is driven by hotplug or something, but it seems to be performing a lot of pointless activities. Eventually it seems to try to open the PCM playback or capture stream (with some devices it opens playback, with others capture) and then checks the pointer: Aug 22 20:22:00 kerneldev kernel: [ 1987.470194] Called: pcm.c - snd_usb_substream_playback_trigger for pcm substream 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470198] Called: pcm.c - snd_usb_pcm_pointer for pcm substream 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470198] Pointer is 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470656] Called: pcm.c - snd_usb_pcm_pointer for pcm substream 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470657] Pointer is 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470660] Called: pcm.c - snd_usb_pcm_pointer for pcm substream 0 Aug 22 20:22:00 kerneldev kernel: [ 1987.470660] Pointer is 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.473567] Called: pcm.c - snd_usb_pcm_pointer for pcm substream 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.473569] Pointer is 0 The above is from a non-working device. For a working device the pointer increments. Eventually this activity ends: Aug 22 20:22:05 kerneldev kernel: [ 1992.473575] Called: pcm.c - snd_usb_substream_playback_trigger for pcm substream 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.473576] Called: endpoint.c - snd_usb_endpoint_stop for endpoint 142, type 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.473578] Called: endpoint.c - snd_usb_endpoint_stop for endpoint 13, type 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.473602] Called: pcm.c - snd_usb_hw_free for pcm substream 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.503287] Called: endpoint.c - snd_usb_endpoint_deactivate for endpoint 142, type 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.503288] Called: endpoint.c - snd_usb_endpoint_deactivate for endpoint 13, type 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.503323] Called: pcm.c - snd_usb_playback_close for pcm substream 0 Aug 22 20:22:05 kerneldev kernel: [ 1992.503656] Called: init.c - snd_card_file_remove This seems as if it might be some sort of test of the PCMs, to make sure they are working. Certainly in non-working devices the pcm_pointer stays at zero and in working devices it increments. It also seems to try harder with some devices than others, even though the device descriptors seem very similar. Can anyone explain any of this to me, particularly why it's exhibiting this behaviour. I've uploaded one of the logs from a non-working device to Pastebin if anyone wants to look at a full example: https://pastebin.com/uBZEuQ57 Also, is this the best way to approach this, or are there other tools/techniques that can help? I've experimented with ftrace and that was useful, but didn't give the level of detail I wanted. Cheers, Keith

3 10

[alsa-devel] ALSA 1.0.29 release
by Jaroslav Kysela 19 Feb '20

19 Feb '20

Hello all, all userspace packages (except alsa-oss) were released today. You may download them from the ALSA website http://www.alsa-project.org or directly from the FTP server ftp://ftp.alsa-project.org/pub . Full list of changes: http://www.alsa-project.org/main/index.php/Changes_v1.0.28_v1.0.29 Have fun, Jaroslav -- Jaroslav Kysela <perex(a)perex.cz> Linux Kernel Sound Maintainer ALSA Project; Red Hat, Inc.

2 16

[alsa-devel] [PATCH] ALSA: usb-audio: fix PCM device order
by Alexander Tsoy 27 Jul '19

27 Jul '19

Some cards have alternate setting with non-PCM format as the first altsetting in the interface descriptors. This confuses userspace, since alsa-lib uses device 0 by default. So lets parse interfaces in two steps: 1. Parse altsettings with PCM formats. 2. Parse altsettings with non-PCM formats. This fixes at least following cards: - Audinst HUD-mx2 - Audinst HUD-mini Signed-off-by: Alexander Tsoy <alexander(a)tsoy.me> --- sound/usb/stream.c | 27 ++++++++++++++++++++++++++- 1 file changed, 26 insertions(+), 1 deletion(-) diff --git a/sound/usb/stream.c b/sound/usb/stream.c index 8e9548bc1f1a..b2ebeae91396 100644 --- a/sound/usb/stream.c +++ b/sound/usb/stream.c @@ -477,7 +477,8 @@ static struct uac2_output_terminal_descriptor * return NULL; } -int snd_usb_parse_audio_interface(struct snd_usb_audio *chip, int iface_no) +static int __snd_usb_parse_audio_interface(struct snd_usb_audio *chip, int iface_no, + bool *has_non_pcm, bool non_pcm) { struct usb_device *dev; struct usb_interface *iface; @@ -634,6 +635,10 @@ int snd_usb_parse_audio_interface(struct snd_usb_audio *chip, int iface_no) iface_no, altno); continue; } + if (fmt->bFormatType != UAC_FORMAT_TYPE_I) + *has_non_pcm = true; + if ((fmt->bFormatType == UAC_FORMAT_TYPE_I) == non_pcm) + continue; if (((protocol == UAC_VERSION_1) && (fmt->bLength < 8)) || ((protocol == UAC_VERSION_2) && (fmt->bLength < 6))) { dev_err(&dev->dev, @@ -740,3 +745,23 @@ int snd_usb_parse_audio_interface(struct snd_usb_audio *chip, int iface_no) return 0; } +int snd_usb_parse_audio_interface(struct snd_usb_audio *chip, int iface_no) +{ + int err; + bool has_non_pcm = false; + + /* parse PCM formats */ + err = __snd_usb_parse_audio_interface(chip, iface_no, &has_non_pcm, false); + if (err < 0) + return err; + + if (has_non_pcm) { + /* parse non-PCM formats */ + err = __snd_usb_parse_audio_interface(chip, iface_no, &has_non_pcm, true); + if (err < 0) + return err; + } + + return 0; +} + -- 2.13.0

3 2

[alsa-devel] [PATCH RFC 00/26] Kill set_fs() in ALSA codes
by Takashi Iwai 27 Mar '19

27 Mar '19

Hi, this is a slightly lengthy patchset, basically targeted for 4.13, thus it's still in RFC. The main purpose of this patchset is to get rid of get_fs() / set_fs() usages in ALSA codes. The patchset starts from a patch that looks fairly irrelevant with the goal (the conversion of HD-audio bind-mute code), but it's a step for the long trail. The biggest change in the patchset is the conversion of copy & silence PCM ops into the new copy_silence ops. It helped for reducing the code size and for handling the in-kernel buffer copies. The current patches are found in topic/kill-set_fs branch of my sound.git tree. thanks, Takashi === Takashi Iwai (26): ALSA: hda - Simplify bound-beep mute control for ALC268 ALSA: hda - Move bind-mixer switch codes to generic parser ALSA: hda - Remove the generic bind ctl helpers ALSA: hda - Remove the use of set_fs() ALSA: hda - Fix a typo in comment ALSA: hda - Remove superfluous header inclusions ALSA: opl3: Kill unused set_fs() ALSA: emu10k1: Get rid of set_fs() usage ALSA: pcm: Remove set_fs() in PCM core code ALSA: pcm: Introduce copy_silence PCM ops ALSA: Update document about copy_silence PCM ops ALSA: dummy: Convert to copy_silence ops ALSA: es1938: Convert to copy_silence ops ALSA: korg1212: Convert to copy_silence ops ALSA: nm256: Convert to copy_silence ops ALSA: rme32: Convert to copy_silence ops ALSA: rme96: Convert to copy_silence ops ALSA: rme9652: Convert to copy_silence ops ALSA: hdsp: Convert to copy_silence ops ALSA: gus: Convert to copy_silence ops ALSA: sb: Convert to copy_silence ops ALSA: sh: Convert to copy_silence ops ASoC: blackfin: Convert to copy_silence ops [media] solo6x10: Convert to copy_silence ops ALSA: pcm: Drop the old copy and silence ops ALSA: pcm: Kill set_fs() usage in OSS layer and USB gadget .../sound/kernel-api/writing-an-alsa-driver.rst | 110 +++++---- drivers/media/pci/solo6x10/solo6x10-g723.c | 11 +- drivers/usb/gadget/function/u_uac1.c | 8 +- include/sound/pcm.h | 19 +- sound/core/oss/pcm_oss.c | 62 ++--- sound/core/pcm_compat.c | 22 +- sound/core/pcm_lib.c | 174 +++++++++----- sound/core/pcm_native.c | 122 ++++++---- sound/drivers/dummy.c | 13 +- sound/drivers/opl3/opl3_oss.c | 14 -- sound/isa/gus/gus_pcm.c | 43 +--- sound/isa/sb/emu8000_pcm.c | 99 +++----- sound/pci/emu10k1/emufx.c | 127 ++++++----- sound/pci/es1938.c | 11 +- sound/pci/hda/hda_codec.c | 251 +++------------------ sound/pci/hda/hda_generic.c | 46 +++- sound/pci/hda/hda_local.h | 61 ----- sound/pci/hda/patch_realtek.c | 37 ++- sound/pci/korg1212/korg1212.c | 128 +++-------- sound/pci/nm256/nm256.c | 35 ++- sound/pci/rme32.c | 49 ++-- sound/pci/rme96.c | 52 ++--- sound/pci/rme9652/hdsp.c | 44 ++-- sound/pci/rme9652/rme9652.c | 46 ++-- sound/sh/sh_dac_audio.c | 40 +--- sound/soc/blackfin/bf5xx-ac97-pcm.c | 6 +- sound/soc/blackfin/bf5xx-ac97.c | 18 +- sound/soc/blackfin/bf5xx-i2s-pcm.c | 46 ++-- sound/soc/soc-pcm.c | 3 +- 29 files changed, 696 insertions(+), 1001 deletions(-) -- 2.12.2

3 40

[alsa-devel] Prepare to covert Codec/Platform into Component
by Kuninori Morimoto 09 May '18

09 May '18

Hi Mark, Lars-Peter I will post all prepare patches to convert Codec/Platform into Component. It has 3 steps. 1) add new necessary functions 2) remove/don't use codec hw_write/read 3) remove/don't use codec reg_cache If these are accepted, all prepares were completed. Next, I can post "convert Codec/Platform into Component" patch bomb. These are too many convert patches, to avoid review issue, I will post few of them as sample. 4) sample of don't use rtd->codec 5) sample of convert Platform into Component 6) sample of convert Codec into Component I will post some of "remove" patches too for each steps. I will re-post all "remove" patches if all are accepted. For your review, please consider below. If these are not good, all covert patch doesn't work. 1. use devm_snd_soc_register_component() and remove snd_soc_unregister_component() My patches is using devm_ version function, and removed unregister function. This is not a big deal 2. New flags. Because "Codec" and "CPU/Platform" are using different features, we need to use some new flags. .idle_bias_off/on .ignore_pmdown_time/pmdown_time These are because "CPU/Platform" and "Codec" are using different default value. "Codec" side convert patches are using these. .endianness .non_legacy_dai_naming These are because "Codec" only needs fixup format, and not use "legacy" DAI name. "Codec" side convert patches are using these I added its flags info on each patch. 3. rtd->platform rtd->platform will be removed. Now we can find all rtd connected component by using rtdcom_lookup. rtd->platform will be replaced to snd_soc_rtdcom_lookup(rtd, DRV_NAME); Here, DRV_NAME is platform's component driver name. 4. rtd->codec rtd->codec will be removed. In soc_bind_dai_link(), It is rtd->codec = rtd->codec_dai->codec; And now, DAI has codec and component pointer. So, rtd->codec will be replaced into rtd->codec_dai->component. Best regards --- Kuninori Morimoto

2 104

[alsa-devel] [PATCH 0/x] ASoC: replace platform to component
by Kuninori Morimoto 12 Feb '18

12 Feb '18

Hi Mark These are for replace "platform" to "component". It needs 3 steps I think. step1. "component" needs _add_ function, like "platform" step2. "component" level pcm_new/free support step3. replace platform to component We can get platform pointer from rtd->platform on current style. But these are very categorized pointer, in other words, non generic pointer. This cleanup is 1st step of merging all feature into component. I think new style wil be CPU + Codec + Platform = Component I think merging CPU into Component is already done. This patch set merges Platform feature into Component, and remove Platform. Merging Codec into Component is next step. This patch set will add new rtd connected component list. All CPU/Codec/Platform are connected to this list. And we can get each component pointer by using rtd and its driver name. Here, We can replace rtd->platform like below - struct device *dev = rtd->platform->dev; + struct snd_soc_component *component = snd_soc_rtdcom_lookup(rtd, DRIVER_NAME); + struct device *dev = component->dev; - if(rtd->platform->pcm_new) - rtd->platform->pcm_new(yyy); + for_each_rtdcom(rtd, rtdcom) { + component = rtdcom->component; /* rtd connected CPU/Codec/Platform */ + if (component->pcm_new) + component->pcm_new(yyy); + } Now, only platform has pcm_new/free, but in new style, all component can have it. I will post step1/2/3 patch-set. step1 is normal patches. step2/3 are now still [RFC]. Best regards --- Kuninori Morimoto

2 23

[alsa-devel] [PATCH 4/6] alsa-lib:pcm: fix the hw_ptr update until the boundary available.
by sutar.mounesh＠gmail.com 19 Jan '18

19 Jan '18

From: mahendran.k <mahendran.kuppusamy(a)in.bosch.com> For long time test case, the slave_hw_ptr will exceed the boundary and wraparound the slave_hw_ptr. This slave boundary wraparound will cause the rate->hw_ptr to wraparound irrespective of the rate->boundary availability and due to that the available size goes wrong. Hence, To get the correct available size, - Its necessary to increment the rate->hw_ptr upto the rate->boundary and then wraparound the rate->hw_ptr. - While handling fraction part of slave period, rounded value will be introduced by input_frames(). To eliminate rounding issue on rate->hw_ptr, subtract last rounded value from rate->hw_ptr and add new rounded value of present slave_hw_ptr fraction part to rate->hw_ptr. Signed-off-by: mahendran.k <mahendran.kuppusamy(a)in.bosch.com> Signed-off-by: Mounesh Sutar <mounesh_sutar(a)mentor.com> diff --git a/src/pcm/pcm_rate.c b/src/pcm/pcm_rate.c index 1f830dd..5bee77c 100644 --- a/src/pcm/pcm_rate.c +++ b/src/pcm/pcm_rate.c @@ -50,7 +50,7 @@ typedef struct _snd_pcm_rate snd_pcm_rate_t; struct _snd_pcm_rate { snd_pcm_generic_t gen; - snd_pcm_uframes_t appl_ptr, hw_ptr; + snd_pcm_uframes_t appl_ptr, hw_ptr, last_slave_hw_ptr; snd_pcm_uframes_t last_commit_ptr; snd_pcm_uframes_t orig_avail_min; snd_pcm_sw_params_t sw_params; @@ -563,14 +563,28 @@ static inline void snd_pcm_rate_sync_hwptr0(snd_pcm_t *pcm, snd_pcm_uframes_t sl { snd_pcm_rate_t *rate = pcm->private_data; + snd_pcm_sframes_t slave_hw_ptr_diff = slave_hw_ptr - rate->last_slave_hw_ptr; + snd_pcm_sframes_t last_slave_hw_ptr_frac; + if (pcm->stream != SND_PCM_STREAM_PLAYBACK) return; - /* FIXME: boundary overlap of slave hw_ptr isn't evaluated here! - * e.g. if slave rate is small... - */ - rate->hw_ptr = - (slave_hw_ptr / rate->gen.slave->period_size) * pcm->period_size + - rate->ops.input_frames(rate->obj, slave_hw_ptr % rate->gen.slave->period_size); + + if (slave_hw_ptr_diff < 0) + slave_hw_ptr_diff += rate->gen.slave->boundary; /* slave boundary wraparound */ + else if (slave_hw_ptr_diff == 0) + return; + last_slave_hw_ptr_frac = rate->last_slave_hw_ptr % rate->gen.slave->period_size; + /* While handling fraction part fo slave period, rounded value will be introduced by input_frames(). + * To eliminate rounding issue on rate->hw_ptr, subtract last rounded value from rate->hw_ptr and + * add new rounded value of present slave_hw_ptr fraction part to rate->hw_ptr. Hence, + * rate->hw_ptr += [ (no. of updated slave periods * pcm rate period size) - + * fractional part of last_slave_hw_ptr rounded value + + * fractional part of updated slave hw ptr's rounded value ] */ + rate->hw_ptr += ( + (((last_slave_hw_ptr_frac + slave_hw_ptr_diff) / rate->gen.slave->period_size) * pcm->period_size) - + rate->ops.input_frames(rate->obj, last_slave_hw_ptr_frac) + + rate->ops.input_frames(rate->obj, (last_slave_hw_ptr_frac + slave_hw_ptr_diff) % rate->gen.slave->period_size)); + rate->last_slave_hw_ptr = slave_hw_ptr; rate->hw_ptr %= pcm->boundary; } @@ -635,6 +649,7 @@ static int snd_pcm_rate_prepare(snd_pcm_t *pcm) return err; *pcm->hw.ptr = 0; *pcm->appl.ptr = 0; + rate->last_slave_hw_ptr = 0; err = snd_pcm_rate_init(pcm); if (err < 0) return err; @@ -650,6 +665,7 @@ static int snd_pcm_rate_reset(snd_pcm_t *pcm) return err; *pcm->hw.ptr = 0; *pcm->appl.ptr = 0; + rate->last_slave_hw_ptr = 0; err = snd_pcm_rate_init(pcm); if (err < 0) return err; -- 1.7.9.5

3 2