July 2018 - Alsa-devel - mailman.alsa-project.org

Re: [alsa-devel] 答复: A bug about cache inconsistency report
by Takashi Iwai 31 Jul '18

31 Jul '18

On Tue, 31 Jul 2018 15:51:06 +0200, Hans Hu(SH-RD) wrote: > > I did the test and Oops report when install driver( both snoop and non-snoop mode) I don't understand. Did my one-liner change the behavior or not? thanks, Takashi > > -----邮件原件----- > 发件人: Takashi Iwai [mailto:tiwai@suse.de] > 发送时间: 31 July 2018 19:59 > 收件人: Hans Hu(SH-RD) <HansHu(a)zhaoxin.com> > 抄送: 'alsa-devel(a)alsa-project.org' <alsa-devel(a)alsa-project.org> > 主题: Re: [alsa-devel] A bug about cache inconsistency report > > On Tue, 31 Jul 2018 12:52:04 +0200, > Hans Hu(SH-RD) wrote: > > > > Hi All, > > > > Try send an email. > > > > Appearance: > > when use non-snoop + ioctl path combination, an issue occurs: > > usage: aplay -D plughw:0,0 44100_S16_LE_2c.wav Ubuntu, it performance > > intermittently noise; > > > > Root case: > > Function set_pages_array_wc marked the PTE of the original address; Function set_memory_wc marked the PTE both original and after vmaped address. > > [cid:image002.png@01D428FF.92621520] > > However, cache or not in transmission process is based on the PTE of the later, and the set_pages_array_wc did't mark it, so a cache inconsistency issue occurs. > > > > Related bug: > > SHA1 ID: 9ddf1aeb2134e72275c97a2c6ff2e3eb04f2f27a > > OK, so did you confirm that dropping the SG buffer support "fixes" the noise problem on your system? The easiest test would be like: > > --- a/sound/pci/hda/hda_controller.c > +++ b/sound/pci/hda/hda_controller.c > @@ -770,7 +770,7 @@ int snd_hda_attach_pcm_stream(struct hda_bus *_bus, struct hda_codec *codec, > size = CONFIG_SND_HDA_PREALLOC_SIZE * 1024; > if (size > MAX_PREALLOC_SIZE) > size = MAX_PREALLOC_SIZE; > -snd_pcm_lib_preallocate_pages_for_all(pcm, SNDRV_DMA_TYPE_DEV_SG, > +snd_pcm_lib_preallocate_pages_for_all(pcm, SNDRV_DMA_TYPE_DEV, > chip->card->dev, > size, MAX_PREALLOC_SIZE); > return 0; > > > thanks, > > Takashi > > > 保密声明： > 本邮件含有保密或专有信息，仅供指定收件人使用。严禁对本邮件或其内容做任何未经授权的查阅、使用、复制或转发。 > CONFIDENTIAL NOTE: > This email contains confidential or legally privileged information and is for the sole use of its intended recipient. Any unauthorized review, use, copying or forwarding of this email or the content of this email is strictly prohibited.

1 0

[alsa-devel] [PATCH 0/4] usb-audio: Add UAC3 Power Domains
by Jorge Sanjuan 31 Jul '18

31 Jul '18

This patchset add support for UAC3 Power Domains. This feature of the USB audio class 3 allows the host to notify the device what it is making use of so power comsumption can be optimized. This proposal implements this feature for Power Domains that include an Input/Output Terminal associated to an audio Streaming interface. This is the main usage of this feature according to the spec. For that reason, the logic for the Power Domain state change has been implemented within the ALSA PCMs logic and the suspend/resume callbacks of the usb_driver. The behaviour would be as follows: * Power Domain State D0: A Power Domain will reach this state only when the audio substream associated to that domain is being used (i,e. Audio playback/capture is happening). * Power Domain State D1: This is the Idle state where the driver is going to always want to be in order to reduce power consumption. * Power Domain State D2: This state is only set when the usb driver asumes the device is not going to be used anymore and hence, it wont care about getting any interrupts from the device. This will only happen when power level is set to "auto" in sysfs so the usb driver gets suspended when the interfaces are not in use. NOTE: The way this has been implemented will always try to put the Power Domain in state D1 if the Power Domain exists so there is not a way a user could disable this feature. It may be worth getting a control exposed to userland that enables/disables this feature (?). Power Domains affecting other units independently are required to be bypassed via a Selector Unit first before the host can change the power state. This sceneario is not covered in this patchset. based on next-20180719 Jorge Sanjuan (4): ALSA: usb-audio: Initial Power Domain support ALSA: usb-audio: AudioStreaming Power Domain parsing ALSA: usb-audio: Operate UAC3 Power Domains in PCM callbacks ALSA: usb-audio: Add UAC3 Power Domains to suspend/resume include/linux/usb/audio-v3.h | 4 ++ sound/usb/Makefile | 1 + sound/usb/card.c | 9 ++++ sound/usb/card.h | 2 + sound/usb/pcm.c | 64 +++++++++++++++++++++-- sound/usb/pcm.h | 2 + sound/usb/power.c | 117 +++++++++++++++++++++++++++++++++++++++++++ sound/usb/power.h | 19 +++++++ sound/usb/stream.c | 70 +++++++++++++++++++++++--- 9 files changed, 277 insertions(+), 11 deletions(-) create mode 100644 sound/usb/power.c -- 2.11.0

5 32

[alsa-devel] A bug about cache inconsistency report
by Hans Hu(SH-RD) 31 Jul '18

31 Jul '18

Hi All, Try send an email. Appearance: when use non-snoop + ioctl path combination, an issue occurs: usage: aplay -D plughw:0,0 44100_S16_LE_2c.wav Ubuntu, it performance intermittently noise; Root case: Function set_pages_array_wc marked the PTE of the original address; Function set_memory_wc marked the PTE both original and after vmaped address. [cid:image002.png@01D428FF.92621520] However, cache or not in transmission process is based on the PTE of the later, and the set_pages_array_wc did't mark it, so a cache inconsistency issue occurs. Related bug: SHA1 ID: 9ddf1aeb2134e72275c97a2c6ff2e3eb04f2f27a Best Regards! Hans Hu ????? ????????????????????????????????????????????????????? CONFIDENTIAL NOTE: This email contains confidential or legally privileged information and is for the sole use of its intended recipient. Any unauthorized review, use, copying or forwarding of this email or the content of this email is strictly prohibited.

2 1

[alsa-devel] More snd_pcm_ioplug_avail_update() questions
by Rob Duncan 31 Jul '18

31 Jul '18

Is it required that all available data has been committed between calls to snd_pcm_ioplug_avail_update() on an IO plugin capture stream? If this is not a requirement then there can still be uncommitted data in the mmap after the current appl_ptr. This data will get overwritten by the transfer call. If it is a requirement then snd_pcm_rate_avail_update() seems to be doing the wrong thing, because it only commits in units of the period size, which means that there could be a partial period left in the mmap. Can someone clarify what the contract is for these calls? Thanks, Rob.

2 7

[alsa-devel] [asoc:for-4.19 199/244] sound/soc/meson/axg-tdm-interface.c:225:8: error: implicit declaration of function 'clk_set_duty_cycle'
by kbuild test robot 31 Jul '18

31 Jul '18

Hi Jerome, FYI, the error/warning still remains. tree: https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-4.19 head: d8504acca759ae672c6d34d49a33d54ace094cbb commit: c41c2a355b86368608377eaf3df442ec0f342f1e [199/244] ASoC: meson: add tdm output driver config: openrisc-allyesconfig (attached as .config) compiler: or1k-linux-gcc (GCC) 6.0.0 20160327 (experimental) reproduce: wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross git checkout c41c2a355b86368608377eaf3df442ec0f342f1e # save the attached .config to linux build tree make.cross ARCH=openrisc All errors (new ones prefixed by >>): sound/soc/meson/axg-tdm-interface.c: In function 'axg_tdm_iface_set_lrclk': >> sound/soc/meson/axg-tdm-interface.c:225:8: error: implicit declaration of function 'clk_set_duty_cycle' [-Werror=implicit-function-declaration] ret = clk_set_duty_cycle(iface->lrclk, ratio_num, 2); ^~~~~~~~~~~~~~~~~~ cc1: some warnings being treated as errors vim +/clk_set_duty_cycle +225 sound/soc/meson/axg-tdm-interface.c d60e4f1e Jerome Brunet 2018-07-17 188 d60e4f1e Jerome Brunet 2018-07-17 189 static int axg_tdm_iface_set_lrclk(struct snd_soc_dai *dai, d60e4f1e Jerome Brunet 2018-07-17 190 struct snd_pcm_hw_params *params) d60e4f1e Jerome Brunet 2018-07-17 191 { d60e4f1e Jerome Brunet 2018-07-17 192 struct axg_tdm_iface *iface = snd_soc_dai_get_drvdata(dai); d60e4f1e Jerome Brunet 2018-07-17 193 unsigned int ratio_num; d60e4f1e Jerome Brunet 2018-07-17 194 int ret; d60e4f1e Jerome Brunet 2018-07-17 195 d60e4f1e Jerome Brunet 2018-07-17 196 ret = clk_set_rate(iface->lrclk, params_rate(params)); d60e4f1e Jerome Brunet 2018-07-17 197 if (ret) { d60e4f1e Jerome Brunet 2018-07-17 198 dev_err(dai->dev, "setting sample clock failed: %d\n", ret); d60e4f1e Jerome Brunet 2018-07-17 199 return ret; d60e4f1e Jerome Brunet 2018-07-17 200 } d60e4f1e Jerome Brunet 2018-07-17 201 d60e4f1e Jerome Brunet 2018-07-17 202 switch (iface->fmt & SND_SOC_DAIFMT_FORMAT_MASK) { d60e4f1e Jerome Brunet 2018-07-17 203 case SND_SOC_DAIFMT_I2S: d60e4f1e Jerome Brunet 2018-07-17 204 case SND_SOC_DAIFMT_LEFT_J: d60e4f1e Jerome Brunet 2018-07-17 205 case SND_SOC_DAIFMT_RIGHT_J: d60e4f1e Jerome Brunet 2018-07-17 206 /* 50% duty cycle ratio */ d60e4f1e Jerome Brunet 2018-07-17 207 ratio_num = 1; d60e4f1e Jerome Brunet 2018-07-17 208 break; d60e4f1e Jerome Brunet 2018-07-17 209 d60e4f1e Jerome Brunet 2018-07-17 210 case SND_SOC_DAIFMT_DSP_A: d60e4f1e Jerome Brunet 2018-07-17 211 case SND_SOC_DAIFMT_DSP_B: d60e4f1e Jerome Brunet 2018-07-17 212 /* d60e4f1e Jerome Brunet 2018-07-17 213 * A zero duty cycle ratio will result in setting the mininum d60e4f1e Jerome Brunet 2018-07-17 214 * ratio possible which, for this clock, is 1 cycle of the d60e4f1e Jerome Brunet 2018-07-17 215 * parent bclk clock high and the rest low, This is exactly d60e4f1e Jerome Brunet 2018-07-17 216 * what we want here. d60e4f1e Jerome Brunet 2018-07-17 217 */ d60e4f1e Jerome Brunet 2018-07-17 218 ratio_num = 0; d60e4f1e Jerome Brunet 2018-07-17 219 break; d60e4f1e Jerome Brunet 2018-07-17 220 d60e4f1e Jerome Brunet 2018-07-17 221 default: d60e4f1e Jerome Brunet 2018-07-17 222 return -EINVAL; d60e4f1e Jerome Brunet 2018-07-17 223 } d60e4f1e Jerome Brunet 2018-07-17 224 d60e4f1e Jerome Brunet 2018-07-17 @225 ret = clk_set_duty_cycle(iface->lrclk, ratio_num, 2); d60e4f1e Jerome Brunet 2018-07-17 226 if (ret) { d60e4f1e Jerome Brunet 2018-07-17 227 dev_err(dai->dev, d60e4f1e Jerome Brunet 2018-07-17 228 "setting sample clock duty cycle failed: %d\n", ret); d60e4f1e Jerome Brunet 2018-07-17 229 return ret; d60e4f1e Jerome Brunet 2018-07-17 230 } d60e4f1e Jerome Brunet 2018-07-17 231 d60e4f1e Jerome Brunet 2018-07-17 232 /* Set sample clock inversion */ d60e4f1e Jerome Brunet 2018-07-17 233 ret = clk_set_phase(iface->lrclk, d60e4f1e Jerome Brunet 2018-07-17 234 axg_tdm_lrclk_invert(iface->fmt) ? 180 : 0); d60e4f1e Jerome Brunet 2018-07-17 235 if (ret) { d60e4f1e Jerome Brunet 2018-07-17 236 dev_err(dai->dev, d60e4f1e Jerome Brunet 2018-07-17 237 "setting sample clock phase failed: %d\n", ret); d60e4f1e Jerome Brunet 2018-07-17 238 return ret; d60e4f1e Jerome Brunet 2018-07-17 239 } d60e4f1e Jerome Brunet 2018-07-17 240 d60e4f1e Jerome Brunet 2018-07-17 241 return 0; d60e4f1e Jerome Brunet 2018-07-17 242 } d60e4f1e Jerome Brunet 2018-07-17 243 :::::: The code at line 225 was first introduced by commit :::::: d60e4f1e4be5e2dfb55fb084b119aed094227a35 ASoC: meson: add tdm interface driver :::::: TO: Jerome Brunet <jbrunet(a)baylibre.com> :::::: CC: Mark Brown <broonie(a)kernel.org> --- 0-DAY kernel test infrastructure Open Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation

1 0

[alsa-devel] how could unsubscribe ?
by fulgor fulgor 31 Jul '18

31 Jul '18

how could unsubscribe ?

2 1

[alsa-devel] [PATCH 0/4] ASoC: Tegra30 TDM support
by Jorge Sanjuan 30 Jul '18

30 Jul '18

This patchset adds support for TDM audio on Tegra30 hardware. It adds the DAI's `set_tdm_slot` callback and enables a tegra pcm to have up to 8 channels. It also includes support for other audio formats supported by the Tegra30 HW and fixes a broken macro needed for setting the TDM on the registers. Based on Linux 4.18-rc3 tag. Edward Cragg (4): ASoC: tegra: i2s: Fix typo/broken macro ASoC: tegra: Add a TDM configuration callback ASoC: tegra: Allow 32-bit and 24-bit samples ASoC: tegra: i2s: Add support for more than 2 channels sound/soc/tegra/tegra30_i2s.c | 72 ++++++++++++++++++++++++++++++++++++------- sound/soc/tegra/tegra30_i2s.h | 2 +- 2 files changed, 62 insertions(+), 12 deletions(-) -- 2.11.0

6 19

[alsa-devel] [PATCH 1/2] ALSA: seq: virmidi: Offload the output event processing
by Takashi Iwai 30 Jul '18

30 Jul '18

The virmidi sequencer stuff tries to translate the rawmidi bytes to sequencer events and deliver the packets at trigger callback. The amount of the whole process of these translations and deliveries depends on the incoming rawmidi bytes, and we have no limit for that; this was the cause of a CPU soft lockup that had been reported and fixed recently. Although we've fixed the soft lockup by putting the temporary unlock and cond_resched(), it's rather a quick band aid. In this patch, meanwhile, the event parsing and delivery process is offloaded to a dedicated work, and the trigger callback just kicks it off. It has three merits, at least: - The processing is always done in a sleepable context, which can assure the event delivery with non-atomic flag without hackish is_atomic() usage. - Other relevant codes can be simplified, reducing the lines - It makes me happier Signed-off-by: Takashi Iwai <tiwai(a)suse.de> --- include/sound/seq_virmidi.h | 1 + sound/core/seq/seq_virmidi.c | 100 ++++++++++++++++------------------- 2 files changed, 47 insertions(+), 54 deletions(-) diff --git a/include/sound/seq_virmidi.h b/include/sound/seq_virmidi.h index 695257ae64ac..d488dcfa3a4e 100644 --- a/include/sound/seq_virmidi.h +++ b/include/sound/seq_virmidi.h @@ -41,6 +41,7 @@ struct snd_virmidi { struct snd_seq_event event; struct snd_virmidi_dev *rdev; struct snd_rawmidi_substream *substream; + struct work_struct output_work; }; #define SNDRV_VIRMIDI_SUBSCRIBE (1<<0) diff --git a/sound/core/seq/seq_virmidi.c b/sound/core/seq/seq_virmidi.c index 8ebbca554e99..67ea5d62cebc 100644 --- a/sound/core/seq/seq_virmidi.c +++ b/sound/core/seq/seq_virmidi.c @@ -154,70 +154,58 @@ static void snd_virmidi_input_trigger(struct snd_rawmidi_substream *substream, i } } -/* - * trigger rawmidi stream for output +/* process rawmidi bytes and send events; + * we need no lock here for vmidi->event since it's handled only in this work */ -static void snd_virmidi_output_trigger(struct snd_rawmidi_substream *substream, int up) +static void snd_vmidi_output_work(struct work_struct *work) { - struct snd_virmidi *vmidi = substream->runtime->private_data; - int count, res; - unsigned char buf[32], *pbuf; - unsigned long flags; - bool check_resched = !in_atomic(); + struct snd_virmidi *vmidi; + struct snd_rawmidi_substream *substream; + unsigned char input; + int ret; + + vmidi = container_of(work, struct snd_virmidi, output_work); + substream = vmidi->substream; + + /* discard the outputs in dispatch mode unless subscribed */ + if (vmidi->seq_mode == SNDRV_VIRMIDI_SEQ_DISPATCH && + !(vmidi->rdev->flags & SNDRV_VIRMIDI_SUBSCRIBE)) { + while (!snd_rawmidi_transmit_empty(substream)) + snd_rawmidi_transmit_ack(substream, 1); + return; + } - if (up) { - vmidi->trigger = 1; - if (vmidi->seq_mode == SNDRV_VIRMIDI_SEQ_DISPATCH && - !(vmidi->rdev->flags & SNDRV_VIRMIDI_SUBSCRIBE)) { - while (snd_rawmidi_transmit(substream, buf, - sizeof(buf)) > 0) { - /* ignored */ - } - return; - } - spin_lock_irqsave(&substream->runtime->lock, flags); + while (vmidi->trigger) { + if (snd_rawmidi_transmit(substream, &input, 1) != 1) + break; + if (snd_midi_event_encode_byte(vmidi->parser, input, + &vmidi->event) <= 0) + continue; if (vmidi->event.type != SNDRV_SEQ_EVENT_NONE) { - if (snd_seq_kernel_client_dispatch(vmidi->client, &vmidi->event, in_atomic(), 0) < 0) - goto out; + ret = snd_seq_kernel_client_dispatch(vmidi->client, + &vmidi->event, + false, 0); vmidi->event.type = SNDRV_SEQ_EVENT_NONE; - } - while (1) { - count = __snd_rawmidi_transmit_peek(substream, buf, sizeof(buf)); - if (count <= 0) + if (ret < 0) break; - pbuf = buf; - while (count > 0) { - res = snd_midi_event_encode(vmidi->parser, pbuf, count, &vmidi->event); - if (res < 0) { - snd_midi_event_reset_encode(vmidi->parser); - continue; - } - __snd_rawmidi_transmit_ack(substream, res); - pbuf += res; - count -= res; - if (vmidi->event.type != SNDRV_SEQ_EVENT_NONE) { - if (snd_seq_kernel_client_dispatch(vmidi->client, &vmidi->event, in_atomic(), 0) < 0) - goto out; - vmidi->event.type = SNDRV_SEQ_EVENT_NONE; - } - } - if (!check_resched) - continue; - /* do temporary unlock & cond_resched() for avoiding - * CPU soft lockup, which may happen via a write from - * a huge rawmidi buffer - */ - spin_unlock_irqrestore(&substream->runtime->lock, flags); - cond_resched(); - spin_lock_irqsave(&substream->runtime->lock, flags); } - out: - spin_unlock_irqrestore(&substream->runtime->lock, flags); - } else { - vmidi->trigger = 0; + /* rawmidi input might be huge, allow to have a break */ + cond_resched(); } } +/* + * trigger rawmidi stream for output + */ +static void snd_virmidi_output_trigger(struct snd_rawmidi_substream *substream, int up) +{ + struct snd_virmidi *vmidi = substream->runtime->private_data; + + vmidi->trigger = !!up; + if (up) + queue_work(system_highpri_wq, &vmidi->output_work); +} + /* * open rawmidi handle for input */ @@ -270,6 +258,7 @@ static int snd_virmidi_output_open(struct snd_rawmidi_substream *substream) vmidi->port = rdev->port; snd_virmidi_init_event(vmidi, &vmidi->event); vmidi->rdev = rdev; + INIT_WORK(&vmidi->output_work, snd_vmidi_output_work); runtime->private_data = vmidi; return 0; } @@ -299,6 +288,9 @@ static int snd_virmidi_input_close(struct snd_rawmidi_substream *substream) static int snd_virmidi_output_close(struct snd_rawmidi_substream *substream) { struct snd_virmidi *vmidi = substream->runtime->private_data; + + vmidi->trigger = 0; /* to be sure */ + cancel_work_sync(&vmidi->output_work); snd_midi_event_free(vmidi->parser); substream->runtime->private_data = NULL; kfree(vmidi); -- 2.18.0

1 1

[alsa-devel] [PATCH v2 0/6] arm64: dts: rockchip: add basic dts file for PX30 SoCs
by cl＠rock-chips.com 30 Jul '18

30 Jul '18

From: Liang Chen <cl(a)rock-chips.com> Changes in V2: 1. change the subject to format "dt-bindings: module: ..." for the documents. 2. use new property <num-interpolated-steps> for backlight. 3. remove pinctrl of rk809, because the driver is not ready. Liang Chen (6): dt-bindings: usb: dwc2: add description for px30 dt-bindings: mmc: rockchip-dw-mshc: add description for px30 dt-bindings: sound: rockchip-i2s: add description for px30 dt-bindings: rockchip: grf: add grf and pmugrf description for px30 arm64: dts: rockchip: add core dtsi file for PX30 SoCs arm64: dts: rockchip: add PX30 eavluation board devicetree Documentation/devicetree/bindings/arm/rockchip.txt | 4 + .../devicetree/bindings/mmc/rockchip-dw-mshc.txt | 1 + .../devicetree/bindings/soc/rockchip/grf.txt | 2 + .../devicetree/bindings/sound/rockchip-i2s.txt | 1 + Documentation/devicetree/bindings/usb/dwc2.txt | 1 + arch/arm64/boot/dts/rockchip/Makefile | 1 + arch/arm64/boot/dts/rockchip/px30-evb.dts | 473 +++++ arch/arm64/boot/dts/rockchip/px30.dtsi | 2043 ++++++++++++++++++++ 8 files changed, 2526 insertions(+) create mode 100644 arch/arm64/boot/dts/rockchip/px30-evb.dts create mode 100644 arch/arm64/boot/dts/rockchip/px30.dtsi -- 1.9.1

3 8

[alsa-devel] [PATCH] ASoC: wm8988: fix typo in rate constraints
by Ladislav Michl 30 Jul '18

30 Jul '18

Remove duplicated entry and add missing zero in rate constraints. Signed-off-by: Ladislav Michl <ladis(a)linux-mips.org> --- sound/soc/codecs/wm8988.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/sound/soc/codecs/wm8988.c b/sound/soc/codecs/wm8988.c index 62200117444b..6e52c6a8bab3 100644 --- a/sound/soc/codecs/wm8988.c +++ b/sound/soc/codecs/wm8988.c @@ -522,7 +522,7 @@ static inline int get_coeff(int mclk, int rate) /* The set of rates we can generate from the above for each SYSCLK */ static const unsigned int rates_12288[] = { - 8000, 12000, 16000, 24000, 24000, 32000, 48000, 96000, + 8000, 12000, 16000, 24000, 32000, 48000, 96000, }; static const struct snd_pcm_hw_constraint_list constraints_12288 = { @@ -540,7 +540,7 @@ static const struct snd_pcm_hw_constraint_list constraints_112896 = { }; static const unsigned int rates_12[] = { - 8000, 11025, 12000, 16000, 22050, 2400, 32000, 41100, 48000, + 8000, 11025, 12000, 16000, 22050, 24000, 32000, 41100, 48000, 48000, 88235, 96000, }; -- 2.18.0

2 1