March 2010 - Alsa-devel - mailman.alsa-project.org

[alsa-devel] [PATCH - ALSA: usbaudio: Add basic support for M-AudioAdd basic support for M-Audio Fast Track Ultra series
by Felix Homann 25 Mar '10

25 Mar '10

This adds basic support for M-Audio's Fast Track Ultra series of USB audio interfaces. It is a refactored version of the patch Clemens Ladisch posted some time ago. Neither playback nor capturing work properly at 44100 Hz (don't know why). The other sampling rates work properly. There's no support for the DSP mixer, yet. Kind regards, Felix Signed-off-by: Felix Homann <fexpop(a)web.de> diff --git a/usb/endpoint.c b/usb/endpoint.c index b1309cd..7b84b61 100644 --- a/usb/endpoint.c +++ b/usb/endpoint.c @@ -317,6 +317,8 @@ int snd_usb_parse_audio_endpoints(struct snd_usb_audio *chip, int iface_no) break; case USB_ID(0x041e, 0x3020): /* Creative SB Audigy 2 NX */ case USB_ID(0x0763, 0x2003): /* M-Audio Audiophile USB */ + case USB_ID(0x0763, 0x2080): /* M-Audio Fast Track Ultra 8 */ + case USB_ID(0x0763, 0x2081): /* M-Audio Fast Track Ultra 8R */ /* doesn't set the sample rate attribute, but supports it */ fp->attributes |= UAC_EP_CS_ATTR_SAMPLE_RATE; break; diff --git a/usb/quirks-table.h b/usb/quirks-table.h index 81c5f8a..91ddef3 100644 --- a/usb/quirks-table.h +++ b/usb/quirks-table.h @@ -1826,6 +1826,60 @@ YAMAHA_DEVICE(0x7010, "UB99"), } } }, +{ + USB_DEVICE(0x0763, 0x2080), + .driver_info = (unsigned long) & (const struct snd_usb_audio_quirk) { + /* .vendor_name = "M-Audio", */ + /* .product_name = "Fast Track Ultra 8", */ + .ifnum = QUIRK_ANY_INTERFACE, + .type = QUIRK_COMPOSITE, + .data = & (const struct snd_usb_audio_quirk[]) { + { + .ifnum = 0, + .type = QUIRK_IGNORE_INTERFACE + }, + { + .ifnum = 1, + .type = QUIRK_AUDIO_STANDARD_INTERFACE + }, + { + .ifnum = 2, + .type = QUIRK_AUDIO_STANDARD_INTERFACE + }, + /* interface 3 (MIDI) is standard compliant */ + { + .ifnum = -1 + } + } + } +}, +{ + USB_DEVICE(0x0763, 0x2081), + .driver_info = (unsigned long) & (const struct snd_usb_audio_quirk) { + /* .vendor_name = "M-Audio", */ + /* .product_name = "Fast Track Ultra 8R", */ + .ifnum = QUIRK_ANY_INTERFACE, + .type = QUIRK_COMPOSITE, + .data = & (const struct snd_usb_audio_quirk[]) { + { + .ifnum = 0, + .type = QUIRK_IGNORE_INTERFACE + }, + { + .ifnum = 1, + .type = QUIRK_AUDIO_STANDARD_INTERFACE + }, + { + .ifnum = 2, + .type = QUIRK_AUDIO_STANDARD_INTERFACE + }, + /* interface 3 (MIDI) is standard compliant */ + { + .ifnum = -1 + } + } + } +}, /* Casio devices */ { diff --git a/usb/urb.c b/usb/urb.c index ad50d43..e2b7c47 100644 --- a/usb/urb.c +++ b/usb/urb.c @@ -918,6 +918,11 @@ void snd_usb_init_substream(struct snd_usb_stream *as, case USB_ID(0x041e, 0x3f0a): /* E-Mu Tracker Pre */ subs->ops.retire_sync = retire_playback_sync_urb_hs_emu; break; + case USB_ID(0x0763, 0x2080): /* M-Audio Fast Track Ultra 8 */ + case USB_ID(0x0763, 0x2081): /* M-Audio Fast Track Ultra 8R */ + subs->ops.prepare_sync = prepare_playback_sync_urb; + subs->ops.retire_sync = retire_playback_sync_urb; + break; } } -- 1.7.0.3

2 1

[alsa-devel] aconnect patch
by lieven moors 25 Mar '10

25 Mar '10

Hi everybody, I'm a musician working intensively with Linux, and I like writing programs. Now I am working on a little command line utility to store and restore all ALSA (midi) and JACK connections to/from an xml file, and I've been using aconnect as an example for writing the ALSA part. I also needed a function to remove all connections, similar to the remove_connections function in aconnect. After using aconnect -x for some time, I noticed that it didn't work as it should. Very often a number of connections were left untouched after issuing this command. After digging a little bit deeper, I think I found the reason for the bug, and wrote a fix for it. This is what I think happens in remove_connections: Every time the for loop is executed, the index of the snd_seq_query_subscribe_t is updated (+ 1). If the connection has the right capabilities, it is removed. When the connection is removed, I think the index gets reassigned, and the next connection gets the position at index 0. The result is that when one connection is removed, the removal of the next connection is skipped (because the for loop updates the index every time). In the patch, I only update the index when unsubscribing fails, or if it is unwanted because the capabilities or wrong. This seems to solve the problems I had with aconnect -x, and now it disconnects all connections as it should. In the original function, connections are removed both ways. First all ports with SND_SEQ_PORT_CAP_SUBS_READ are checked, and associated connections removed, and then all ports with SND_SEQ_PORT_CAP_SUBS_WRITE. In the patch, I removed the second part, because I think it is doing the same thing twice. Please let me know what you think about this patch, and if you think my reasoning is sane... Greetings, Lieven Moors diff --git a/aconnect.c b/aconnect.c index 1a50666..6092748 100644 --- a/aconnect.c +++ b/aconnect.c @@ -192,52 +192,33 @@ static void remove_connection(snd_seq_t *seq, snd_seq_client_info_t *cinfo, snd_seq_port_info_t *pinfo, int count) { snd_seq_query_subscribe_t *query; - snd_seq_query_subscribe_alloca(&query); snd_seq_query_subscribe_set_root(query, snd_seq_port_info_get_addr(pinfo)); - snd_seq_query_subscribe_set_type(query, SND_SEQ_QUERY_SUBS_READ); snd_seq_query_subscribe_set_index(query, 0); - for (; snd_seq_query_port_subscribers(seq, query) >= 0; - snd_seq_query_subscribe_set_index(query, snd_seq_query_subscribe_get_index(query) + 1)) { + + while (snd_seq_query_port_subscribers(seq, query) >= 0) + { snd_seq_port_info_t *port; snd_seq_port_subscribe_t *subs; const snd_seq_addr_t *sender = snd_seq_query_subscribe_get_root(query); const snd_seq_addr_t *dest = snd_seq_query_subscribe_get_addr(query); snd_seq_port_info_alloca(&port); - if (snd_seq_get_any_port_info(seq, dest->client, dest->port, port) < 0) - continue; - if (!(snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_SUBS_WRITE)) - continue; - if (snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_NO_EXPORT) - continue; - snd_seq_port_subscribe_alloca(&subs); - snd_seq_port_subscribe_set_queue(subs, snd_seq_query_subscribe_get_queue(query)); - snd_seq_port_subscribe_set_sender(subs, sender); - snd_seq_port_subscribe_set_dest(subs, dest); - snd_seq_unsubscribe_port(seq, subs); - } - snd_seq_query_subscribe_set_type(query, SND_SEQ_QUERY_SUBS_WRITE); - snd_seq_query_subscribe_set_index(query, 0); - for (; snd_seq_query_port_subscribers(seq, query) >= 0; - snd_seq_query_subscribe_set_index(query, snd_seq_query_subscribe_get_index(query) + 1)) { - snd_seq_port_info_t *port; - snd_seq_port_subscribe_t *subs; - const snd_seq_addr_t *dest = snd_seq_query_subscribe_get_root(query); - const snd_seq_addr_t *sender = snd_seq_query_subscribe_get_addr(query); - snd_seq_port_info_alloca(&port); - if (snd_seq_get_any_port_info(seq, sender->client, sender->port, port) < 0) - continue; - if (!(snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_SUBS_READ)) - continue; - if (snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_NO_EXPORT) + if ((snd_seq_get_any_port_info(seq, dest->client, dest->port, port) < 0) || + !(snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_SUBS_WRITE) || + (snd_seq_port_info_get_capability(port) & SND_SEQ_PORT_CAP_NO_EXPORT)) + { + snd_seq_query_subscribe_set_index(query, snd_seq_query_subscribe_get_index(query) + 1); continue; + } snd_seq_port_subscribe_alloca(&subs); snd_seq_port_subscribe_set_queue(subs, snd_seq_query_subscribe_get_queue(query)); snd_seq_port_subscribe_set_sender(subs, sender); snd_seq_port_subscribe_set_dest(subs, dest); - snd_seq_unsubscribe_port(seq, subs); + if(snd_seq_unsubscribe_port(seq, subs) < 0){ + snd_seq_query_subscribe_set_index(query, snd_seq_query_subscribe_get_index(query) + 1); + } } }

3 7

[alsa-devel] [rfc patch] wm8994: range checking issue
by Dan Carpenter 25 Mar '10

25 Mar '10

Smatch complained about BUG_ON(reg > WM8994_MAX_REGISTER) because the actual number of elements in the array was WM8994_REG_CACHE_SIZE + 1. I changed the BUG_ON() to return -EINVAL. I was confused why WM8994_REG_CACHE_SIZE was different from the actual size of ->reg_cache and I was concerned because some places used ARRAY_SIZE() to find the end of the array and other places used WM8994_REG_CACHE_SIZE. In my patch, I made them the same. I don't have the hardware to test this and some of this patch is just guess work. For example, I left it in, but why is there a -1 here? 3711 /* Fill the cache with physical values we inherited; don't reset */ 3712 ret = wm8994_bulk_read(codec->control_data, 0, 3713 ARRAY_SIZE(wm8994->reg_cache) - 1, 3714 codec->reg_cache); I didn't sign this off because I figured I'd probably need to send a second version after I hear the feed back. regards, dan carpenter diff --git a/sound/soc/codecs/wm8994.c b/sound/soc/codecs/wm8994.c index 29f3771..d9c179a 100644 --- a/sound/soc/codecs/wm8994.c +++ b/sound/soc/codecs/wm8994.c @@ -59,13 +59,13 @@ static int wm8994_retune_mobile_base[] = { WM8994_AIF2_EQ_GAINS_1, }; -#define WM8994_REG_CACHE_SIZE 0x621 +#define WM8994_REG_CACHE_SIZE 0x622 /* codec private data */ struct wm8994_priv { struct wm_hubs_data hubs; struct snd_soc_codec codec; - u16 reg_cache[WM8994_REG_CACHE_SIZE + 1]; + u16 reg_cache[WM8994_REG_CACHE_SIZE]; int sysclk[2]; int sysclk_rate[2]; int mclk[2]; @@ -1697,7 +1697,8 @@ static int wm8994_write(struct snd_soc_codec *codec, unsigned int reg, { struct wm8994_priv *wm8994 = codec->private_data; - BUG_ON(reg > WM8994_MAX_REGISTER); + if (reg >= WM8994_REG_CACHE_SIZE) + return -EINVAL; if (!wm8994_volatile(reg)) wm8994->reg_cache[reg] = value; @@ -1710,7 +1711,8 @@ static unsigned int wm8994_read(struct snd_soc_codec *codec, { u16 *reg_cache = codec->reg_cache; - BUG_ON(reg > WM8994_MAX_REGISTER); + if (reg >= WM8994_REG_CACHE_SIZE) + return -EINVAL; if (wm8994_volatile(reg)) return wm8994_reg_read(codec->control_data, reg); @@ -3700,7 +3702,7 @@ static int wm8994_codec_probe(struct platform_device *pdev) codec->set_bias_level = wm8994_set_bias_level; codec->dai = &wm8994_dai[0]; codec->num_dai = 3; - codec->reg_cache_size = WM8994_MAX_REGISTER; + codec->reg_cache_size = WM8994_REG_CACHE_SIZE; codec->reg_cache = &wm8994->reg_cache; codec->dev = &pdev->dev;

2 4

[alsa-devel] [PATCH 4/4 v2] ASoC: fsi: Add FSI2 device support
by Kuninori Morimoto 25 Mar '10

25 Mar '10

ARM-SHMOBILE series have FIFO-buffered serial interface 2 (FSI2) device which is advanced version of FSI. This patch add simple support for it. Signed-off-by: Kuninori Morimoto <morimoto.kuninori(a)renesas.com> --- v1 -> v2 o based on new [PATCH 2/4 v2] ASoC: fsi: Add FIFO size calculate sound/soc/sh/Kconfig | 3 +- sound/soc/sh/fsi.c | 56 +++++++++++++++++++++++++++++++++++++++++-------- 2 files changed, 48 insertions(+), 11 deletions(-) diff --git a/sound/soc/sh/Kconfig b/sound/soc/sh/Kconfig index f07f6d8..a1d14bc 100644 --- a/sound/soc/sh/Kconfig +++ b/sound/soc/sh/Kconfig @@ -1,5 +1,5 @@ menu "SoC Audio support for SuperH" - depends on SUPERH + depends on SUPERH || ARCH_SHMOBILE config SND_SOC_PCM_SH7760 tristate "SoC Audio support for Renesas SH7760" @@ -22,7 +22,6 @@ config SND_SOC_SH4_SSI config SND_SOC_SH4_FSI tristate "SH4 FSI support" - depends on CPU_SUBTYPE_SH7724 help This option enables FSI sound support diff --git a/sound/soc/sh/fsi.c b/sound/soc/sh/fsi.c index ae88865..f14bbb0 100644 --- a/sound/soc/sh/fsi.c +++ b/sound/soc/sh/fsi.c @@ -40,6 +40,10 @@ #define MUTE_ST 0x0028 #define REG_END MUTE_ST + +#define CPU_INT_ST 0x01F4 +#define CPU_IEMSK 0x01F8 +#define CPU_IMSK 0x01FC #define INT_ST 0x0200 #define IEMSK 0x0204 #define IMSK 0x0208 @@ -47,7 +51,7 @@ #define CLK_RST 0x0210 #define SOFT_RST 0x0214 #define FIFO_SZ 0x0218 -#define MREG_START INT_ST +#define MREG_START CPU_INT_ST #define MREG_END FIFO_SZ /* DO_FMT */ @@ -116,11 +120,18 @@ struct fsi_priv { int periods; }; +struct fsi_regs { + u32 int_st; + u32 iemsk; + u32 imsk; +}; + struct fsi_master { void __iomem *base; int irq; struct fsi_priv fsia; struct fsi_priv fsib; + struct fsi_regs *regs; struct sh_fsi_platform_info *info; spinlock_t lock; }; @@ -337,8 +348,8 @@ static void fsi_irq_enable(struct fsi_priv *fsi, int is_play) u32 data = fsi_port_ab_io_bit(fsi, is_play); struct fsi_master *master = fsi_get_master(fsi); - fsi_master_mask_set(master, IMSK, data, data); - fsi_master_mask_set(master, IEMSK, data, data); + fsi_master_mask_set(master, master->regs->imsk, data, data); + fsi_master_mask_set(master, master->regs->iemsk, data, data); } static void fsi_irq_disable(struct fsi_priv *fsi, int is_play) @@ -346,18 +357,18 @@ static void fsi_irq_disable(struct fsi_priv *fsi, int is_play) u32 data = fsi_port_ab_io_bit(fsi, is_play); struct fsi_master *master = fsi_get_master(fsi); - fsi_master_mask_set(master, IMSK, data, 0); - fsi_master_mask_set(master, IEMSK, data, 0); + fsi_master_mask_set(master, master->regs->imsk, data, 0); + fsi_master_mask_set(master, master->regs->iemsk, data, 0); } static u32 fsi_irq_get_status(struct fsi_master *master) { - return fsi_master_read(master, INT_ST); + return fsi_master_read(master, master->regs->int_st); } static void fsi_irq_clear_all_status(struct fsi_master *master) { - fsi_master_write(master, INT_ST, 0x0000000); + fsi_master_write(master, master->regs->int_st, 0x0000000); } static void fsi_irq_clear_status(struct fsi_priv *fsi) @@ -369,7 +380,7 @@ static void fsi_irq_clear_status(struct fsi_priv *fsi) data |= fsi_port_ab_io_bit(fsi, 1); /* clear interrupt factor */ - fsi_master_mask_set(master, INT_ST, data, 0); + fsi_master_mask_set(master, master->regs->int_st, data, 0); } /************************************************************************ @@ -953,6 +964,7 @@ EXPORT_SYMBOL_GPL(fsi_soc_platform); static int fsi_probe(struct platform_device *pdev) { struct fsi_master *master; + const struct platform_device_id *id_entry; struct resource *res; unsigned int irq; int ret; @@ -962,6 +974,12 @@ static int fsi_probe(struct platform_device *pdev) return -ENODEV; } + id_entry = pdev->id_entry; + if (!id_entry) { + dev_err(&pdev->dev, "unknown fsi device\n"); + return -ENODEV; + } + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); irq = platform_get_irq(pdev, 0); if (!res || (int)irq <= 0) { @@ -990,6 +1008,7 @@ static int fsi_probe(struct platform_device *pdev) master->fsia.master = master; master->fsib.base = master->base + 0x40; master->fsib.master = master; + master->regs = (struct fsi_regs *)id_entry->driver_data; spin_lock_init(&master->lock); pm_runtime_enable(&pdev->dev); @@ -1002,7 +1021,8 @@ static int fsi_probe(struct platform_device *pdev) fsi_soft_all_reset(master); - ret = request_irq(irq, &fsi_interrupt, IRQF_DISABLED, "fsi", master); + ret = request_irq(irq, &fsi_interrupt, IRQF_DISABLED, + id_entry->name, master); if (ret) { dev_err(&pdev->dev, "irq request err\n"); goto exit_iounmap; @@ -1069,6 +1089,23 @@ static struct dev_pm_ops fsi_pm_ops = { .runtime_resume = fsi_runtime_nop, }; +static struct fsi_regs fsi_regs = { + .int_st = INT_ST, + .iemsk = IEMSK, + .imsk = IMSK, +}; + +static struct fsi_regs fsi2_regs = { + .int_st = CPU_INT_ST, + .iemsk = CPU_IEMSK, + .imsk = CPU_IMSK, +}; + +static struct platform_device_id fsi_id_table[] = { + { "sh_fsi", (kernel_ulong_t)&fsi_regs }, + { "sh_fsi2", (kernel_ulong_t)&fsi2_regs }, +}; + static struct platform_driver fsi_driver = { .driver = { .name = "sh_fsi", @@ -1076,6 +1113,7 @@ static struct platform_driver fsi_driver = { }, .probe = fsi_probe, .remove = fsi_remove, + .id_table = fsi_id_table, }; static int __init fsi_mobile_init(void) -- 1.6.3.3

1 0

[alsa-devel] [PATCH 0/4] ASoC: fsi: FSI2 device support
by Kuninori Morimoto 25 Mar '10

25 Mar '10

Dear Mark These patches are FSI2 support Kuninori Morimoto (4): ASoC: fsi: ensures process inside master lock ASoC: fsi: Add FIFO size calculate ASoC: fsi: IRQ related process had be united ASoC: fsi: Add FSI2 device support ARM SH-MOBILE series have FSI2/FMSI device which are advance version of FSI. Above patches add simple FSI2 support. These patches are tested on AP4 board Best regards -- Kuninori Morimoto

3 12

[alsa-devel] [PATCH 0/7] snd-asihpi playback improvement + others
by linux＠audioscience.com 24 Mar '10

24 Mar '10

From: Eliot Blennerhassett <eblennerhassett(a)audioscience.com> Important: alsa-firmware/asihpi should also be updated with contents of http://www.audioscience.com/internet/download/firmware/dspbins40313.zip asihpi.c + hpi6205.c fix choppy audio playback on busmastering adapters the rest are minor fixes or cosmetic Eliot Blennerhassett (7): Remove obsolete comment - const correct pointer parameters. - add a new stream state wait and function Reinit response size for every msg/response transaction. Minor fix const ptr Bump lib version due to added and removed APIs Use adapter properties for stream buffer constraints. Default to less logging, remove VPRINTK3 Log buffer info in decimal. Keep HPI buffer pointers in sync with ALSA after prewrite. Support mic control caching. Move an enum out of public api. pci/asihpi/asihpi.c | 102 +++++---- pci/asihpi/hpi.h | 513 +++++++++++++++++++++++---------------------- pci/asihpi/hpi6205.c | 31 ++-- pci/asihpi/hpi_internal.h | 33 ++- pci/asihpi/hpicmn.c | 10 + pci/asihpi/hpidebug.h | 6 +- pci/asihpi/hpifunc.c | 469 ++++++++++++++++++++++------------------- pci/asihpi/hpimsginit.c | 2 +- pci/asihpi/hpimsgx.c | 26 ++- pci/asihpi/hpioctl.c | 3 - 10 files changed, 641 insertions(+), 554 deletions(-)

3 11

Re: [alsa-devel] 2.6.34-rc1: regression: ^Z no longer stops sound
by Pavel Machek 24 Mar '10

24 Mar '10

On Sun 2010-03-14 17:20:13, Bill Davidsen wrote: > Pavel Machek wrote: >> Hi! >> >> In 2.6.33, hitting ^Z stopped mpg123 in a way you'd expect (both on >> thinkpad and geode notebook). >> >> On 2.6.34-rc1, hit ^Z and sound starts looping. >> Pavel > That's not new, I have seen it back as far as 2.6.27.25 doing a suspend > in a VM (KVM, no libvirt). But it only happens a few times a year, so I > never chased it. I hate to report an occasional problem in an obsolete > kernel, and I can't reproduce it. > > Does it happen repeatably with the new kernel? It happened reliably so far, yes. -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

3 2

[alsa-devel] Would it be possible to Alias current versions??
by Wolf Halton 24 Mar '10

24 Mar '10

I am working on fixing a script that was supposed to update alsa drivers on ubuntu. I am trying to figure out how to get the script to wget the most recent versions of everything without having to rewrite the script every time. How difficult would it be to alias the most recent versions? for instance ftp://ftp.alsa-project.org/pub/driver/alsa-driver-current.tar.bz2 I could work up a script for you that would check the current stable versions and update the aliases. -Wolf Halton -- This Apt Has Super Cow Powers - http://sourcefreedom.com

2 1

[alsa-devel] [patch] wm8994: playback => capture
by Dan Carpenter 24 Mar '10

24 Mar '10

Sparse caught that initialize "playback" two times instead of initializing "capture". Signed-off-by: Dan Carpenter <error27(a)gmail.com> diff --git a/sound/soc/codecs/wm8994.c b/sound/soc/codecs/wm8994.c index 29f3771..d10d651 100644 --- a/sound/soc/codecs/wm8994.c +++ b/sound/soc/codecs/wm8994.c @@ -3401,7 +3401,7 @@ struct snd_soc_dai wm8994_dai[] = { .rates = WM8994_RATES, .formats = WM8994_FORMATS, }, - .playback = { + .capture = { .stream_name = "AIF3 Capture", .channels_min = 2, .channels_max = 2,

3 2

Re: [alsa-devel] cmipci and sound skipping ?
by Clemens Ladisch 24 Mar '10

24 Mar '10

Matija Nalis wrote: > On Tue, Mar 23, 2010 at 09:36:01AM +0100, Clemens Ladisch wrote: > > Are you okay with your email address (which one?) to be published with > > the patch: > > Reported-and-tested-by: Matija Nalis <...> > > Sure, mnalis-alsadev @ voyager.hr would be ok for e-mail. Applied. Thanks! Clemens

1 0