Yamaha 754 question...

Yamaha 754 question...

[Adam K Kirchhoff]

A number of months ago, the possibility of having the rear speakers on the
yamaha 754 duplicate the front speakers by default (instead of having to
setup .asoundrc to handle the duplication).  

I have a couple quick questions :-)  Would this duplication carry over
when using the OSS emualation from Alsa?  And has this development
happened yet? 

Adam




_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Yamaha 754 question...

[Adam K Kirchhoff]

No comment on this one? :-(

Adam

On Wed, 8 May 2002, Adam K Kirchhoff wrote:

> 
> A number of months ago, the possibility of having the rear speakers on the
> yamaha 754 duplicate the front speakers by default (instead of having to
> setup .asoundrc to handle the duplication).  
> 
> I have a couple quick questions :-)  Would this duplication carry over
> when using the OSS emualation from Alsa?  And has this development
> happened yet? 
> 
> Adam




_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Andy Wingo]

I'm not an expert, but this sounds like something out of a default asoundrc
that involves ttables. To me that means that you would need to use
LD_PRELOAD, etc. For what it's worth, anyway...

regards,

wingo.

On Fri, 10 May 2002, Adam K Kirchhoff wrote:

> 
> No comment on this one? :-(
> 
> Adam
> 
> On Wed, 8 May 2002, Adam K Kirchhoff wrote:
> 
> > 
> > A number of months ago, the possibility of having the rear speakers on the
> > yamaha 754 duplicate the front speakers by default (instead of having to
> > setup .asoundrc to handle the duplication).  
> > 
> > I have a couple quick questions :-)  Would this duplication carry over
> > when using the OSS emualation from Alsa?  And has this development
> > happened yet? 
> > 
> > Adam
> 
> 
> 
> 
> _______________________________________________________________
> 
> Have big pipes? SourceForge.net is looking for download mirrors. We supply
> the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/alsa-devel

_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

Re: Re: Yamaha 754 question...

[Takashi Iwai]

[-- Attachment #1: Type: text/plain, Size: 303 bytes --]

At Fri, 10 May 2002 12:46:07 -0400 (EDT),
Adam K Kirchhoff wrote:
> 
> 
> No comment on this one? :-(

oh, you had bad luck ;)

now i did a very quick-n-dirty hack, the recipe was finished in 15
minutes.  can you try the attached patch?

btw i've not tested at all (nor have i the hardware)..


Takashi

[-- Attachment #2: ymfpci-4ch.dif --]
[-- Type: application/octet-stream, Size: 13417 bytes --]

Index: alsa-kernel/include/ymfpci.h
===================================================================
RCS file: /suse/tiwai/cvs/alsa/alsa-kernel/include/ymfpci.h,v
retrieving revision 1.5
diff -u -r1.5 ymfpci.h
--- alsa-kernel/include/ymfpci.h	18 Feb 2002 09:25:29 -0000	1.5
+++ alsa-kernel/include/ymfpci.h	13 May 2002 11:30:46 -0000
@@ -264,9 +264,9 @@
 	snd_ymfpci_pcm_type_t type;
 	snd_pcm_substream_t *substream;
 	ymfpci_voice_t *voices[2];	/* playback only */
-	int running: 1,
-	    spdif: 1,
-	    mode4ch : 1;	
+	int running: 1;
+	int output_front: 1;
+	int output_rear: 1;
 	u32 period_size;		/* cached from runtime->period_size */
 	u32 buffer_size;		/* cached from runtime->buffer_size */
 	u32 period_pos;
@@ -304,6 +304,9 @@
 	dma_addr_t bank_base_capture_addr;
 	dma_addr_t bank_base_effect_addr;
 	dma_addr_t work_base_addr;
+
+	struct semaphore ac3_mutex;
+	int ac3_inited;
 	void *ac3_tmp_base;
 	dma_addr_t ac3_tmp_base_addr;
 
@@ -333,6 +336,9 @@
 	snd_kcontrol_t *ctl_vol_spdifrec;
 	unsigned short spdif_bits, spdif_pcm_bits;
 	snd_kcontrol_t *spdif_pcm_ctl;
+	int mode_dup4ch;
+	int rear_opened;
+	int spdif_opened;
 
 	spinlock_t reg_lock;
 	spinlock_t voice_lock;
Index: alsa-kernel/pci/ymfpci/ymfpci_main.c
===================================================================
RCS file: /suse/tiwai/cvs/alsa/alsa-kernel/pci/ymfpci/ymfpci_main.c,v
retrieving revision 1.16
diff -u -r1.16 ymfpci_main.c
--- alsa-kernel/pci/ymfpci/ymfpci_main.c	26 Apr 2002 09:32:48 -0000	1.16
+++ alsa-kernel/pci/ymfpci/ymfpci_main.c	13 May 2002 11:33:53 -0000
@@ -453,7 +453,8 @@
 
 static void snd_ymfpci_pcm_init_voice(ymfpci_voice_t *voice, int stereo,
 				      int rate, int w_16, unsigned long addr,
-				      unsigned int end, int eff2)
+				      unsigned int end,
+				      int output_front, int output_rear)
 {
 	u32 format;
 	u32 delta = snd_ymfpci_calc_delta(rate);
@@ -499,19 +500,20 @@
 		bank->eff3_gain_end = 0;
 
 		if (!stereo) {
-			if (!eff2) {
+			if (output_front) {
 				bank->left_gain = 
 				bank->right_gain =
 				bank->left_gain_end =
 				bank->right_gain_end = cpu_to_le32(0x40000000);
-			} else {
+			}
+			if (output_rear) {
 				bank->eff2_gain =
 				bank->eff2_gain_end =
 				bank->eff3_gain =
 				bank->eff3_gain_end = cpu_to_le32(0x40000000);
 			}
 		} else {
-			if (!eff2) {
+			if (output_front) {
 				if ((voice->number & 1) == 0) {
 					bank->left_gain =
 					bank->left_gain_end = cpu_to_le32(0x40000000);
@@ -520,7 +522,8 @@
 					bank->right_gain =
 					bank->right_gain_end = cpu_to_le32(0x40000000);
 				}
-			} else {
+			}
+			if (output_rear) {
 				if ((voice->number & 1) == 0) {
 					bank->eff3_gain =
 					bank->eff3_gain_end = cpu_to_le32(0x40000000);
@@ -539,10 +542,19 @@
 	unsigned char *ptr;
 	dma_addr_t ptr_addr;
 
-	if (chip->ac3_tmp_base != NULL)
+	down(&chip->ac3_mutex);
+	if (chip->ac3_inited) {
+		up(&chip->ac3_mutex);
+		return 0;
+	}
+	if (chip->ac3_tmp_base != NULL) {
+		up(&chip->ac3_mutex);
 		return -EBUSY;
-	if ((ptr = snd_malloc_pci_pages(chip->pci, 4096, &ptr_addr)) == NULL)
+	}
+	if ((ptr = snd_malloc_pci_pages(chip->pci, 4096, &ptr_addr)) == NULL) {
+		up(&chip->ac3_mutex);
 		return -ENOMEM;
+	}
 
 	chip->ac3_tmp_base = ptr;
 	chip->ac3_tmp_base_addr = ptr_addr;
@@ -559,20 +571,31 @@
 	snd_ymfpci_writel(chip, YDSXGR_MAPOFEFFECT,
 			  snd_ymfpci_readl(chip, YDSXGR_MAPOFEFFECT) | 3 << 3);
 	spin_unlock_irq(&chip->reg_lock);
+	chip->ac3_inited++;
+	up(&chip->ac3_mutex);
 	return 0;
 }
 
 static int snd_ymfpci_ac3_done(ymfpci_t *chip)
 {
-	spin_lock_irq(&chip->reg_lock);
-	snd_ymfpci_writel(chip, YDSXGR_MAPOFEFFECT,
-			  snd_ymfpci_readl(chip, YDSXGR_MAPOFEFFECT) & ~(3 << 3));
-	spin_unlock_irq(&chip->reg_lock);
-	snd_ymfpci_irq_wait(chip);
-	if (chip->ac3_tmp_base) {
-		snd_free_pci_pages(chip->pci, 4096, chip->ac3_tmp_base, chip->ac3_tmp_base_addr);
-		chip->ac3_tmp_base = NULL;
+	down(&chip->ac3_mutex);
+	if (! chip->ac3_inited) {
+		up(&chip->ac3_mutex);
+		return 0;
 	}
+	chip->ac3_inited--;
+	if (! chip->ac3_inited) {
+		spin_lock_irq(&chip->reg_lock);
+		snd_ymfpci_writel(chip, YDSXGR_MAPOFEFFECT,
+				  snd_ymfpci_readl(chip, YDSXGR_MAPOFEFFECT) & ~(3 << 3));
+		spin_unlock_irq(&chip->reg_lock);
+		snd_ymfpci_irq_wait(chip);
+		if (chip->ac3_tmp_base) {
+			snd_free_pci_pages(chip->pci, 4096, chip->ac3_tmp_base, chip->ac3_tmp_base_addr);
+			chip->ac3_tmp_base = NULL;
+		}
+	}
+	up(&chip->ac3_mutex);
 	return 0;
 }
 
@@ -588,7 +611,7 @@
 		return err;
 	if ((err = snd_ymfpci_pcm_voice_alloc(ypcm, params_channels(hw_params))) < 0)
 		return err;
-	if (ypcm->spdif || ypcm->mode4ch)
+	if (ypcm->output_rear)
 		if ((err = snd_ymfpci_ac3_init(chip)) < 0)
 			return err;
 	return 0;
@@ -616,7 +639,7 @@
 		snd_ymfpci_voice_free(chip, ypcm->voices[0]);
 		ypcm->voices[0] = NULL;
 	}
-	if (ypcm->spdif || ypcm->mode4ch)
+	if (ypcm->output_rear)
 		snd_ymfpci_ac3_done(chip);
 	return 0;
 }
@@ -639,7 +662,8 @@
 					  snd_pcm_format_width(runtime->format) == 16,
 					  runtime->dma_addr,
 					  ypcm->buffer_size,
-					  ypcm->spdif || ypcm->mode4ch);
+					  ypcm->output_front,
+					  ypcm->output_rear);
 	return 0;
 }
 
@@ -842,7 +866,7 @@
 		snd_magic_kfree(ypcm);
 }
 
-static int snd_ymfpci_playback_open(snd_pcm_substream_t * substream)
+static int snd_ymfpci_playback_open_1(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
 	snd_pcm_runtime_t *runtime = substream->runtime;
@@ -862,6 +886,31 @@
 	return 0;
 }
 
+static int snd_ymfpci_playback_open(snd_pcm_substream_t * substream)
+{
+	ymfpci_t *chip = snd_pcm_substream_chip(substream);
+	snd_pcm_runtime_t *runtime = substream->runtime;
+	ymfpci_pcm_t *ypcm;
+	unsigned long flags;
+	int err;
+	
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
+		return err;
+	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
+	ypcm->output_front = 1;
+	ypcm->output_rear = chip->mode_dup4ch ? 1 : 0;
+	spin_lock_irqsave(&chip->reg_lock, flags);
+	if (! chip->rear_opened)
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0030) | 0x0010);
+	if (! chip->spdif_opened && ! chip->rear_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	chip->rear_opened++;
+	spin_unlock_irqrestore(&chip->reg_lock, flags);
+	return 0;
+}
+
 static int snd_ymfpci_playback_spdif_open(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
@@ -870,25 +919,25 @@
 	unsigned long flags;
 	int err;
 	
-	if ((err = snd_ymfpci_playback_open(substream)) < 0)
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
 		return err;
 	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
-	ypcm->spdif = 1;
+	ypcm->output_front = 0;
+	ypcm->output_rear = 1;
 	spin_lock_irqsave(&chip->reg_lock, flags);
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTCTRL,
 			  snd_ymfpci_readw(chip, YDSXGR_SPDIFOUTCTRL) | 2);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	if (! chip->rear_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
 	chip->spdif_pcm_bits = chip->spdif_bits;
 	snd_ymfpci_writel(chip, YDSXGR_SPDIFOUTSTATUS, chip->spdif_pcm_bits);
+	chip->spdif_opened++;
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
 
 	chip->spdif_pcm_ctl->access &= ~SNDRV_CTL_ELEM_ACCESS_INACTIVE;
 	snd_ctl_notify(chip->card, SNDRV_CTL_EVENT_MASK_VALUE |
 		       SNDRV_CTL_EVENT_MASK_INFO, &chip->spdif_pcm_ctl->id);
-
-	/* FIXME? True value is 256/48 = 5.33333 ms */
-	snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_PERIOD_TIME, 5333, UINT_MAX);
 	return 0;
 }
 
@@ -900,19 +949,20 @@
 	unsigned long flags;
 	int err;
 	
-	if ((err = snd_ymfpci_playback_open(substream)) < 0)
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
 		return err;
 	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
-	ypcm->mode4ch = 1;
+	ypcm->output_front = 0;
+	ypcm->output_rear = 1;
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
-			  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0030) | 0x0010);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	if (! chip->rear_opened)
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0030) | 0x0010);
+	if (! chip->spdif_opened && ! chip->rear_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	chip->rear_opened++;
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
-
-	/* FIXME? True value is 256/48 = 5.33333 ms */
-	snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_PERIOD_TIME, 5333, UINT_MAX);
 	return 0;
 }
 
@@ -950,19 +1000,38 @@
 	return snd_ymfpci_capture_open(substream, 1);
 }
 
-static int snd_ymfpci_playback_close(snd_pcm_substream_t * substream)
+static int snd_ymfpci_playback_close_1(snd_pcm_substream_t * substream)
 {
 	return 0;
 }
 
+static int snd_ymfpci_playback_close(snd_pcm_substream_t * substream)
+{
+	ymfpci_t *chip = snd_pcm_substream_chip(substream);
+	unsigned long flags;
+
+	spin_lock_irqsave(&chip->reg_lock, flags);
+	chip->rear_opened--;
+	if (! chip->rear_opened && ! chip->spdif_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
+	if (! chip->rear_opened)
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) | 0x0010);
+	spin_unlock_irqrestore(&chip->reg_lock, flags);
+	return snd_ymfpci_playback_close_1(substream);
+}
+
 static int snd_ymfpci_playback_spdif_close(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
 	unsigned long flags;
 
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
+	chip->spdif_opened = 0;
+	if (! chip->rear_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTCTRL,
 			  snd_ymfpci_readw(chip, YDSXGR_SPDIFOUTCTRL) & ~2);
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTSTATUS, chip->spdif_bits);
@@ -970,7 +1039,7 @@
 	chip->spdif_pcm_ctl->access |= SNDRV_CTL_ELEM_ACCESS_INACTIVE;
 	snd_ctl_notify(chip->card, SNDRV_CTL_EVENT_MASK_VALUE |
 		       SNDRV_CTL_EVENT_MASK_INFO, &chip->spdif_pcm_ctl->id);
-	return snd_ymfpci_playback_close(substream);
+	return snd_ymfpci_playback_close_1(substream);
 }
 
 static int snd_ymfpci_playback_4ch_close(snd_pcm_substream_t * substream)
@@ -979,12 +1048,15 @@
 	unsigned long flags;
 
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
-	snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
-			  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) | 0x0010);
+	chip->rear_opened--;
+	if (! chip->rear_opened && ! chip->spdif_opened)
+		snd_ymfpci_writel(chip, YDSXGR_MODE,
+				  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
+	if (! chip->rear_opened)
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) | 0x0010);
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
-	return snd_ymfpci_playback_close(substream);
+	return snd_ymfpci_playback_close_1(substream);
 }
 
 static int snd_ymfpci_capture_close(snd_pcm_substream_t * substream)
@@ -1446,6 +1518,36 @@
 	return change;
 }
 
+/*
+ * 4ch duplication
+ */
+static int snd_ymfpci_info_dup4ch(snd_kcontrol_t *kcontrol, snd_ctl_elem_info_t * uinfo)
+{
+	uinfo->type = SNDRV_CTL_ELEM_TYPE_BOOLEAN;
+	uinfo->count = 1;
+	uinfo->value.integer.min = 0;
+	uinfo->value.integer.max = 1;
+	return 0;
+}
+
+static int snd_ymfpci_get_dup4ch(snd_kcontrol_t * kcontrol, snd_ctl_elem_value_t * ucontrol)
+{
+	ymfpci_t *chip = snd_kcontrol_chip(kcontrol);
+	ucontrol->value.integer.value[0] = chip->mode_dup4ch;
+	return 0;
+}
+
+static int snd_ymfpci_put_dup4ch(snd_kcontrol_t * kcontrol, snd_ctl_elem_value_t * ucontrol)
+{
+	ymfpci_t *chip = snd_kcontrol_chip(kcontrol);
+	int change;
+	change = (ucontrol->value.integer.value[0] != chip->mode_dup4ch);
+	if (change)
+		chip->mode_dup4ch = !!ucontrol->value.integer.value[0];
+	return change;
+}
+
+
 #define YMFPCI_CONTROLS (sizeof(snd_ymfpci_controls)/sizeof(snd_kcontrol_new_t))
 
 static snd_kcontrol_new_t snd_ymfpci_controls[] __devinitdata = {
@@ -1462,7 +1564,14 @@
 YMFPCI_DOUBLE(SNDRV_CTL_NAME_IEC958("AC97 ",PLAYBACK,VOLUME), 1, YDSXGR_SPDIFOUTVOL),
 YMFPCI_DOUBLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,VOLUME), 1, YDSXGR_SPDIFLOOPVOL),
 YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",PLAYBACK,SWITCH), 0, YDSXGR_SPDIFOUTCTRL),
-YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,SWITCH), 0, YDSXGR_SPDIFINCTRL)
+YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,SWITCH), 0, YDSXGR_SPDIFINCTRL),
+{
+	iface: SNDRV_CTL_ELEM_IFACE_MIXER,
+	name: "4ch Duplication",
+	info: snd_ymfpci_info_dup4ch,
+	get: snd_ymfpci_get_dup4ch,
+	put: snd_ymfpci_put_dup4ch,
+},
 };
 
 
@@ -2114,6 +2223,7 @@
 	spin_lock_init(&chip->voice_lock);
 	init_waitqueue_head(&chip->interrupt_sleep);
 	atomic_set(&chip->interrupt_sleep_count, 0);
+	init_MUTEX(&chip->ac3_mutex);
 	chip->card = card;
 	chip->pci = pci;
 	chip->irq = -1;

Re: Re: Yamaha 754 question...

[Adam K Kirchhoff]

It's looking good so far :-)

I applied the patch and recompiled.  I haven't had much chance to test
since I had to leave for work (I'll test more tonight), but alsaplayer
worked without using the ch4dup bit from .asoundrc...  xmms worked fine,
with both the Alsa output plugin and the OSS plugin.  In all three cases,
I was getting four speaker output.

In addition, alsamixer shows a new option "4ch Duplication" so, in theory,
I can turn it off and have two alsaplayers going, one stream on the front
channel and a different one on the rear channel.  I haven't tested that
yet, though.

I'll have to take a close look at your patch now to see exactly what
you're doing.  I tried my handa t it some time last month, but ran into
some difficulties :-)

Adam

On Mon, 13 May 2002, Takashi Iwai wrote:

> At Fri, 10 May 2002 12:46:07 -0400 (EDT),
> Adam K Kirchhoff wrote:
> > 
> > 
> > No comment on this one? :-(
> 
> oh, you had bad luck ;)
> 
> now i did a very quick-n-dirty hack, the recipe was finished in 15
> minutes.  can you try the attached patch?
> 
> btw i've not tested at all (nor have i the hardware)..
> 
> 
> Takashi
> 




_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Takashi Iwai]

At Mon, 13 May 2002 08:54:57 -0400 (EDT),
Adam K Kirchhoff wrote:
> 
> 
> It's looking good so far :-)
> 
> I applied the patch and recompiled.  I haven't had much chance to test
> since I had to leave for work (I'll test more tonight), but alsaplayer
> worked without using the ch4dup bit from .asoundrc...  xmms worked fine,
> with both the Alsa output plugin and the OSS plugin.  In all three cases,
> I was getting four speaker output.
 
good.
the only concern is whether snd_ymfpci_ac3_init/_done is properly
handled.  could you check (by adding printk in these routines, for
example) whether these functions are called only once per open/close
devices?
also, could you check multiple playbacks?


> In addition, alsamixer shows a new option "4ch Duplication" so, in theory,
> I can turn it off and have two alsaplayers going, one stream on the front
> channel and a different one on the rear channel.  I haven't tested that
> yet, though.
 
this is even possible without my patch.
the pcm device #2 (hw:0,2) to play from rear speaker.  this is aliased
as "rear".  so in the case of aplay,
	aplay foo.wav
		=> front (or 4ch if "4ch Dup" switch is on)
	aplay -Drear foo.wav
		=> rear only


Takashi

_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Adam K Kirchhoff]

On Mon, 13 May 2002, Takashi Iwai wrote:

> At Mon, 13 May 2002 08:54:57 -0400 (EDT),
> Adam K Kirchhoff wrote:
> > 
> > 
> > It's looking good so far :-)
> > 
> > I applied the patch and recompiled.  I haven't had much chance to test
> > since I had to leave for work (I'll test more tonight), but alsaplayer
> > worked without using the ch4dup bit from .asoundrc...  xmms worked fine,
> > with both the Alsa output plugin and the OSS plugin.  In all three cases,
> > I was getting four speaker output.
>  
> good.
> the only concern is whether snd_ymfpci_ac3_init/_done is properly
> handled.  could you check (by adding printk in these routines, for
> example) whether these functions are called only once per open/close
> devices?

I'll try that out tonight.

> also, could you check multiple playbacks?

I'm getting some interesting results with this :-)

I'm booted into KDE and I have artsd running...  As KDE starts up artsd
plays it's audio file fine, with all four speakers.

If I set xmms to use the OSS output plugin, the mp3 I'm playing starts up
fine with all four speakers...  Somewhere along the way, the rear channel
drops out completely.  It seems to be hit or miss.  Sometimes I can
actually get through the file without any problems.  Sometimes it drops
out nearly immediately.

A sure fire way to duplicate the problem, though, is to start up
alsaplayer with another song..  Both songs will now play on all four
speakers.  So now I have artsd running, xmms running with the OSS plugin,
and alsaplayer is running :-)  If I quit alsaplayer, the rear speakers
will cut out for the song playing on xmms.  If I then hit "x" to restart
the song, it begins again with all four speakers :-)

> > In addition, alsamixer shows a new option "4ch Duplication" so, in theory,
> > I can turn it off and have two alsaplayers going, one stream on the front
> > channel and a different one on the rear channel.  I haven't tested that
> > yet, though.
>  
> this is even possible without my patch.

Yeah, I did notice that too.  The only reason I mentioned it this time
around is because I thought it was nice that it was still possible to do
this by muting that switch :-)

Adam


_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Adam K Kirchhoff]

OK...  So I got around to testing how snd_ymfpci_ac3_init and
snd_ymfpci_ac3_done are being handled.  You'll have to tell me if the
results sound good, though.

Playing back one small wav file (3930 bytes) with "play" (so it's not
using any sound daemons) results in entering _init one and entering _done
once.

Playing back a slightly larger wav file (12250 bytes) results in entering
_init 3 times, and _done once.

An even larger wav file: _init 4 times, _done once...

The largest wav file I can find will still only enter _init 4 times...
_done only ever seems to get called once (as you'd expect).

The results are similar for using xmms to play mp3s...  esd will also
enter _init four times.

Adam

On Mon, 13 May 2002, Adam K Kirchhoff wrote:

> 
> On Mon, 13 May 2002, Takashi Iwai wrote:
> 
> > At Mon, 13 May 2002 08:54:57 -0400 (EDT),
> > Adam K Kirchhoff wrote:
> > > 
> > > 
> > > It's looking good so far :-)
> > > 
> > > I applied the patch and recompiled.  I haven't had much chance to test
> > > since I had to leave for work (I'll test more tonight), but alsaplayer
> > > worked without using the ch4dup bit from .asoundrc...  xmms worked fine,
> > > with both the Alsa output plugin and the OSS plugin.  In all three cases,
> > > I was getting four speaker output.
> >  
> > good.
> > the only concern is whether snd_ymfpci_ac3_init/_done is properly
> > handled.  could you check (by adding printk in these routines, for
> > example) whether these functions are called only once per open/close
> > devices?
> 
> I'll try that out tonight.
> 
> > also, could you check multiple playbacks?
> 
> I'm getting some interesting results with this :-)
> 
> I'm booted into KDE and I have artsd running...  As KDE starts up artsd
> plays it's audio file fine, with all four speakers.
> 
> If I set xmms to use the OSS output plugin, the mp3 I'm playing starts up
> fine with all four speakers...  Somewhere along the way, the rear channel
> drops out completely.  It seems to be hit or miss.  Sometimes I can
> actually get through the file without any problems.  Sometimes it drops
> out nearly immediately.
> 
> A sure fire way to duplicate the problem, though, is to start up
> alsaplayer with another song..  Both songs will now play on all four
> speakers.  So now I have artsd running, xmms running with the OSS plugin,
> and alsaplayer is running :-)  If I quit alsaplayer, the rear speakers
> will cut out for the song playing on xmms.  If I then hit "x" to restart
> the song, it begins again with all four speakers :-)
> 
> > > In addition, alsamixer shows a new option "4ch Duplication" so, in theory,
> > > I can turn it off and have two alsaplayers going, one stream on the front
> > > channel and a different one on the rear channel.  I haven't tested that
> > > yet, though.
> >  
> > this is even possible without my patch.
> 
> Yeah, I did notice that too.  The only reason I mentioned it this time
> around is because I thought it was nice that it was still possible to do
> this by muting that switch :-)
> 
> Adam
> 
> 
> _______________________________________________________________
> 
> Have big pipes? SourceForge.net is looking for download mirrors. We supply
> the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/alsa-devel
> 




_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Takashi Iwai]

[-- Attachment #1: Type: text/plain, Size: 960 bytes --]

At Tue, 14 May 2002 11:13:56 -0400 (EDT),
Adam K Kirchhoff wrote:
> 
> 
> OK...  So I got around to testing how snd_ymfpci_ac3_init and
> snd_ymfpci_ac3_done are being handled.  You'll have to tell me if the
> results sound good, though.
> 
> Playing back one small wav file (3930 bytes) with "play" (so it's not
> using any sound daemons) results in entering _init one and entering _done
> once.
> 
> Playing back a slightly larger wav file (12250 bytes) results in entering
> _init 3 times, and _done once.
> 
> An even larger wav file: _init 4 times, _done once...
> 
> The largest wav file I can find will still only enter _init 4 times...
> _done only ever seems to get called once (as you'd expect).

hmm, init can be called multiple times.  i forgot it.

the following is the revised patch.
now ac3_init and _done are called at module init/end stage, so there
will be no problems any more to switch 2ch/4ch accidentally.
could you give a try?


Takashi

[-- Attachment #2: ymfpci-4ch-2.dif --]
[-- Type: application/octet-stream, Size: 13131 bytes --]

Index: alsa-kernel/include/ymfpci.h
===================================================================
RCS file: /suse/tiwai/cvs/alsa/alsa-kernel/include/ymfpci.h,v
retrieving revision 1.5
diff -u -r1.5 ymfpci.h
--- alsa-kernel/include/ymfpci.h	18 Feb 2002 09:25:29 -0000	1.5
+++ alsa-kernel/include/ymfpci.h	16 May 2002 10:19:59 -0000
@@ -264,9 +264,9 @@
 	snd_ymfpci_pcm_type_t type;
 	snd_pcm_substream_t *substream;
 	ymfpci_voice_t *voices[2];	/* playback only */
-	int running: 1,
-	    spdif: 1,
-	    mode4ch : 1;	
+	int running: 1;
+	int output_front: 1;
+	int output_rear: 1;
 	u32 period_size;		/* cached from runtime->period_size */
 	u32 buffer_size;		/* cached from runtime->buffer_size */
 	u32 period_pos;
@@ -333,6 +333,9 @@
 	snd_kcontrol_t *ctl_vol_spdifrec;
 	unsigned short spdif_bits, spdif_pcm_bits;
 	snd_kcontrol_t *spdif_pcm_ctl;
+	int mode_dup4ch;
+	int rear_opened;
+	int spdif_opened;
 
 	spinlock_t reg_lock;
 	spinlock_t voice_lock;
Index: alsa-kernel/pci/ymfpci/ymfpci_main.c
===================================================================
RCS file: /suse/tiwai/cvs/alsa/alsa-kernel/pci/ymfpci/ymfpci_main.c,v
retrieving revision 1.16
diff -u -r1.16 ymfpci_main.c
--- alsa-kernel/pci/ymfpci/ymfpci_main.c	26 Apr 2002 09:32:48 -0000	1.16
+++ alsa-kernel/pci/ymfpci/ymfpci_main.c	16 May 2002 10:12:41 -0000
@@ -453,7 +453,8 @@
 
 static void snd_ymfpci_pcm_init_voice(ymfpci_voice_t *voice, int stereo,
 				      int rate, int w_16, unsigned long addr,
-				      unsigned int end, int eff2)
+				      unsigned int end,
+				      int output_front, int output_rear)
 {
 	u32 format;
 	u32 delta = snd_ymfpci_calc_delta(rate);
@@ -499,19 +500,20 @@
 		bank->eff3_gain_end = 0;
 
 		if (!stereo) {
-			if (!eff2) {
+			if (output_front) {
 				bank->left_gain = 
 				bank->right_gain =
 				bank->left_gain_end =
 				bank->right_gain_end = cpu_to_le32(0x40000000);
-			} else {
+			}
+			if (output_rear) {
 				bank->eff2_gain =
 				bank->eff2_gain_end =
 				bank->eff3_gain =
 				bank->eff3_gain_end = cpu_to_le32(0x40000000);
 			}
 		} else {
-			if (!eff2) {
+			if (output_front) {
 				if ((voice->number & 1) == 0) {
 					bank->left_gain =
 					bank->left_gain_end = cpu_to_le32(0x40000000);
@@ -520,7 +522,8 @@
 					bank->right_gain =
 					bank->right_gain_end = cpu_to_le32(0x40000000);
 				}
-			} else {
+			}
+			if (output_rear) {
 				if ((voice->number & 1) == 0) {
 					bank->eff3_gain =
 					bank->eff3_gain_end = cpu_to_le32(0x40000000);
@@ -534,13 +537,11 @@
 	}
 }
 
-static int snd_ymfpci_ac3_init(ymfpci_t *chip)
+static int __devinit snd_ymfpci_ac3_init(ymfpci_t *chip)
 {
 	unsigned char *ptr;
 	dma_addr_t ptr_addr;
 
-	if (chip->ac3_tmp_base != NULL)
-		return -EBUSY;
 	if ((ptr = snd_malloc_pci_pages(chip->pci, 4096, &ptr_addr)) == NULL)
 		return -ENOMEM;
 
@@ -568,7 +569,7 @@
 	snd_ymfpci_writel(chip, YDSXGR_MAPOFEFFECT,
 			  snd_ymfpci_readl(chip, YDSXGR_MAPOFEFFECT) & ~(3 << 3));
 	spin_unlock_irq(&chip->reg_lock);
-	snd_ymfpci_irq_wait(chip);
+	// snd_ymfpci_irq_wait(chip);
 	if (chip->ac3_tmp_base) {
 		snd_free_pci_pages(chip->pci, 4096, chip->ac3_tmp_base, chip->ac3_tmp_base_addr);
 		chip->ac3_tmp_base = NULL;
@@ -579,7 +580,6 @@
 static int snd_ymfpci_playback_hw_params(snd_pcm_substream_t * substream,
 					 snd_pcm_hw_params_t * hw_params)
 {
-	ymfpci_t *chip = snd_pcm_substream_chip(substream);
 	snd_pcm_runtime_t *runtime = substream->runtime;
 	ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return -ENXIO);
 	int err;
@@ -588,9 +588,6 @@
 		return err;
 	if ((err = snd_ymfpci_pcm_voice_alloc(ypcm, params_channels(hw_params))) < 0)
 		return err;
-	if (ypcm->spdif || ypcm->mode4ch)
-		if ((err = snd_ymfpci_ac3_init(chip)) < 0)
-			return err;
 	return 0;
 }
 
@@ -600,7 +597,6 @@
 	snd_pcm_runtime_t *runtime = substream->runtime;
 	ymfpci_pcm_t *ypcm;
 	
-	
 	if (runtime->private_data == NULL)
 		return 0;
 	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return -ENXIO);
@@ -616,8 +612,6 @@
 		snd_ymfpci_voice_free(chip, ypcm->voices[0]);
 		ypcm->voices[0] = NULL;
 	}
-	if (ypcm->spdif || ypcm->mode4ch)
-		snd_ymfpci_ac3_done(chip);
 	return 0;
 }
 
@@ -639,7 +633,8 @@
 					  snd_pcm_format_width(runtime->format) == 16,
 					  runtime->dma_addr,
 					  ypcm->buffer_size,
-					  ypcm->spdif || ypcm->mode4ch);
+					  ypcm->output_front,
+					  ypcm->output_rear);
 	return 0;
 }
 
@@ -842,7 +837,7 @@
 		snd_magic_kfree(ypcm);
 }
 
-static int snd_ymfpci_playback_open(snd_pcm_substream_t * substream)
+static int snd_ymfpci_playback_open_1(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
 	snd_pcm_runtime_t *runtime = substream->runtime;
@@ -862,6 +857,53 @@
 	return 0;
 }
 
+/* call with spinlock held */
+static void ymfpci_open_extension(ymfpci_t *chip)
+{
+	if (! chip->rear_opened) {
+		if (! chip->spdif_opened) /* set AC3 */
+			snd_ymfpci_writel(chip, YDSXGR_MODE,
+					  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+		/* enable second codec (4CHEN) */
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) | 0x0010);
+	}
+}
+
+/* call with spinlock held */
+static void ymfpci_close_extension(ymfpci_t *chip)
+{
+	if (! chip->rear_opened) {
+		if (! chip->spdif_opened)
+			snd_ymfpci_writel(chip, YDSXGR_MODE,
+					  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
+		snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
+				  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) & ~0x0010);
+	}
+}
+
+static int snd_ymfpci_playback_open(snd_pcm_substream_t * substream)
+{
+	ymfpci_t *chip = snd_pcm_substream_chip(substream);
+	snd_pcm_runtime_t *runtime = substream->runtime;
+	ymfpci_pcm_t *ypcm;
+	unsigned long flags;
+	int err;
+	
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
+		return err;
+	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
+	ypcm->output_front = 1;
+	ypcm->output_rear = chip->mode_dup4ch ? 1 : 0;
+	spin_lock_irqsave(&chip->reg_lock, flags);
+	if (ypcm->output_rear) {
+		ymfpci_open_extension(chip);
+		chip->rear_opened++;
+	}
+	spin_unlock_irqrestore(&chip->reg_lock, flags);
+	return 0;
+}
+
 static int snd_ymfpci_playback_spdif_open(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
@@ -870,25 +912,23 @@
 	unsigned long flags;
 	int err;
 	
-	if ((err = snd_ymfpci_playback_open(substream)) < 0)
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
 		return err;
 	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
-	ypcm->spdif = 1;
+	ypcm->output_front = 0;
+	ypcm->output_rear = 1;
 	spin_lock_irqsave(&chip->reg_lock, flags);
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTCTRL,
 			  snd_ymfpci_readw(chip, YDSXGR_SPDIFOUTCTRL) | 2);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	ymfpci_open_extension(chip);
 	chip->spdif_pcm_bits = chip->spdif_bits;
 	snd_ymfpci_writel(chip, YDSXGR_SPDIFOUTSTATUS, chip->spdif_pcm_bits);
+	chip->spdif_opened++;
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
 
 	chip->spdif_pcm_ctl->access &= ~SNDRV_CTL_ELEM_ACCESS_INACTIVE;
 	snd_ctl_notify(chip->card, SNDRV_CTL_EVENT_MASK_VALUE |
 		       SNDRV_CTL_EVENT_MASK_INFO, &chip->spdif_pcm_ctl->id);
-
-	/* FIXME? True value is 256/48 = 5.33333 ms */
-	snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_PERIOD_TIME, 5333, UINT_MAX);
 	return 0;
 }
 
@@ -900,19 +940,15 @@
 	unsigned long flags;
 	int err;
 	
-	if ((err = snd_ymfpci_playback_open(substream)) < 0)
+	if ((err = snd_ymfpci_playback_open_1(substream)) < 0)
 		return err;
 	ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, return 0);
-	ypcm->mode4ch = 1;
+	ypcm->output_front = 0;
+	ypcm->output_rear = 1;
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
-			  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0030) | 0x0010);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) | (1 << 30));
+	ymfpci_open_extension(chip);
+	chip->rear_opened++;
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
-
-	/* FIXME? True value is 256/48 = 5.33333 ms */
-	snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_PERIOD_TIME, 5333, UINT_MAX);
 	return 0;
 }
 
@@ -950,19 +986,34 @@
 	return snd_ymfpci_capture_open(substream, 1);
 }
 
-static int snd_ymfpci_playback_close(snd_pcm_substream_t * substream)
+static int snd_ymfpci_playback_close_1(snd_pcm_substream_t * substream)
 {
 	return 0;
 }
 
+static int snd_ymfpci_playback_close(snd_pcm_substream_t * substream)
+{
+	ymfpci_t *chip = snd_pcm_substream_chip(substream);
+	ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, substream->runtime->private_data, return -ENXIO);
+	unsigned long flags;
+
+	spin_lock_irqsave(&chip->reg_lock, flags);
+	if (ypcm->output_rear && chip->rear_opened > 0) {
+		chip->rear_opened--;
+		ymfpci_close_extension(chip);
+	}
+	spin_unlock_irqrestore(&chip->reg_lock, flags);
+	return snd_ymfpci_playback_close_1(substream);
+}
+
 static int snd_ymfpci_playback_spdif_close(snd_pcm_substream_t * substream)
 {
 	ymfpci_t *chip = snd_pcm_substream_chip(substream);
 	unsigned long flags;
 
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
+	chip->spdif_opened = 0;
+	ymfpci_close_extension(chip);
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTCTRL,
 			  snd_ymfpci_readw(chip, YDSXGR_SPDIFOUTCTRL) & ~2);
 	snd_ymfpci_writew(chip, YDSXGR_SPDIFOUTSTATUS, chip->spdif_bits);
@@ -970,7 +1021,7 @@
 	chip->spdif_pcm_ctl->access |= SNDRV_CTL_ELEM_ACCESS_INACTIVE;
 	snd_ctl_notify(chip->card, SNDRV_CTL_EVENT_MASK_VALUE |
 		       SNDRV_CTL_EVENT_MASK_INFO, &chip->spdif_pcm_ctl->id);
-	return snd_ymfpci_playback_close(substream);
+	return snd_ymfpci_playback_close_1(substream);
 }
 
 static int snd_ymfpci_playback_4ch_close(snd_pcm_substream_t * substream)
@@ -979,12 +1030,12 @@
 	unsigned long flags;
 
 	spin_lock_irqsave(&chip->reg_lock, flags);
-	snd_ymfpci_writel(chip, YDSXGR_MODE,
-			  snd_ymfpci_readl(chip, YDSXGR_MODE) & ~(1 << 30));
-	snd_ymfpci_writew(chip, YDSXGR_SECCONFIG,
-			  (snd_ymfpci_readw(chip, YDSXGR_SECCONFIG) & ~0x0330) | 0x0010);
+	if (chip->rear_opened > 0) {
+		chip->rear_opened--;
+		ymfpci_close_extension(chip);
+	}
 	spin_unlock_irqrestore(&chip->reg_lock, flags);
-	return snd_ymfpci_playback_close(substream);
+	return snd_ymfpci_playback_close_1(substream);
 }
 
 static int snd_ymfpci_capture_close(snd_pcm_substream_t * substream)
@@ -1446,6 +1497,36 @@
 	return change;
 }
 
+/*
+ * 4ch duplication
+ */
+static int snd_ymfpci_info_dup4ch(snd_kcontrol_t *kcontrol, snd_ctl_elem_info_t * uinfo)
+{
+	uinfo->type = SNDRV_CTL_ELEM_TYPE_BOOLEAN;
+	uinfo->count = 1;
+	uinfo->value.integer.min = 0;
+	uinfo->value.integer.max = 1;
+	return 0;
+}
+
+static int snd_ymfpci_get_dup4ch(snd_kcontrol_t * kcontrol, snd_ctl_elem_value_t * ucontrol)
+{
+	ymfpci_t *chip = snd_kcontrol_chip(kcontrol);
+	ucontrol->value.integer.value[0] = chip->mode_dup4ch;
+	return 0;
+}
+
+static int snd_ymfpci_put_dup4ch(snd_kcontrol_t * kcontrol, snd_ctl_elem_value_t * ucontrol)
+{
+	ymfpci_t *chip = snd_kcontrol_chip(kcontrol);
+	int change;
+	change = (ucontrol->value.integer.value[0] != chip->mode_dup4ch);
+	if (change)
+		chip->mode_dup4ch = !!ucontrol->value.integer.value[0];
+	return change;
+}
+
+
 #define YMFPCI_CONTROLS (sizeof(snd_ymfpci_controls)/sizeof(snd_kcontrol_new_t))
 
 static snd_kcontrol_new_t snd_ymfpci_controls[] __devinitdata = {
@@ -1462,7 +1543,14 @@
 YMFPCI_DOUBLE(SNDRV_CTL_NAME_IEC958("AC97 ",PLAYBACK,VOLUME), 1, YDSXGR_SPDIFOUTVOL),
 YMFPCI_DOUBLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,VOLUME), 1, YDSXGR_SPDIFLOOPVOL),
 YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",PLAYBACK,SWITCH), 0, YDSXGR_SPDIFOUTCTRL),
-YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,SWITCH), 0, YDSXGR_SPDIFINCTRL)
+YMFPCI_SINGLE(SNDRV_CTL_NAME_IEC958("",CAPTURE,SWITCH), 0, YDSXGR_SPDIFINCTRL),
+{
+	iface: SNDRV_CTL_ELEM_IFACE_MIXER,
+	name: "4ch Duplication",
+	info: snd_ymfpci_info_dup4ch,
+	get: snd_ymfpci_get_dup4ch,
+	put: snd_ymfpci_put_dup4ch,
+},
 };
 
 
@@ -1952,12 +2040,19 @@
 		snd_ymfpci_writew(chip, YDSXGR_GLOBALCTRL, ctrl & ~0x0007);
 	}
 
+	snd_ymfpci_ac3_done(chip);
+
 	/* Set PCI device to D3 state */
-	// pci_set_power_state(chip->pci, 3);
+#if 0
+	/* FIXME: temporarily disabled, otherwise we cannot fire up
+	 * the chip again unless reboot.  ACPI bug?
+	 */
+	pci_set_power_state(chip->pci, 3);
+#endif
 
 #ifdef CONFIG_PM
 	if (chip->saved_regs)
-		kfree(chip->saved_regs);
+		vfree(chip->saved_regs);
 #endif
 	if (chip->reg_area_virt)
 		iounmap((void *)chip->reg_area_virt);
@@ -2150,8 +2245,11 @@
 		return -EIO;
 	}
 
+	if ((err = snd_ymfpci_ac3_init(chip)) < 0)
+		return err;
+
 #ifdef CONFIG_PM
-	chip->saved_regs = kmalloc(YDSXGR_NUM_SAVED_REGS * sizeof(u32), GFP_KERNEL);
+	chip->saved_regs = vmalloc(YDSXGR_NUM_SAVED_REGS * sizeof(u32));
 	if (chip->saved_regs == NULL) {
 		snd_ymfpci_free(chip);
 		return -ENOMEM;

Re: Re: Yamaha 754 question...

[Adam K Kirchhoff]

On Thu, 16 May 2002, Takashi Iwai wrote:

> At Tue, 14 May 2002 11:13:56 -0400 (EDT),
> Adam K Kirchhoff wrote:
> > 
> > 
> > OK...  So I got around to testing how snd_ymfpci_ac3_init and
> > snd_ymfpci_ac3_done are being handled.  You'll have to tell me if the
> > results sound good, though.
> > 
> > Playing back one small wav file (3930 bytes) with "play" (so it's not
> > using any sound daemons) results in entering _init one and entering _done
> > once.
> > 
> > Playing back a slightly larger wav file (12250 bytes) results in entering
> > _init 3 times, and _done once.
> > 
> > An even larger wav file: _init 4 times, _done once...
> > 
> > The largest wav file I can find will still only enter _init 4 times...
> > _done only ever seems to get called once (as you'd expect).
> 
> hmm, init can be called multiple times.  i forgot it.
> 
> the following is the revised patch.
> now ac3_init and _done are called at module init/end stage, so there
> will be no problems any more to switch 2ch/4ch accidentally.
> could you give a try?

I haven't had a chance to thoroughly test it, but I gave it a quick shot
this morning.  Using the same method that I was able to replicate the
problem last time (playing a file with xmms and alsaplayer and then
killing alsaplayer), I was unable to make it happen again.  

I'll try some more after I get home from work tonight.

Adam



_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net

Re: Re: Yamaha 754 question...

[Adam K Kirchhoff]

On Thu, 16 May 2002, Adam K Kirchhoff wrote:

> >
> > hmm, init can be called multiple times.  i forgot it.
> >
> > the following is the revised patch.
> > now ac3_init and _done are called at module init/end stage, so there
> > will be no problems any more to switch 2ch/4ch accidentally.
> > could you give a try?
> 
> I haven't had a chance to thoroughly test it, but I gave it a quick shot
> this morning.  Using the same method that I was able to replicate the
> problem last time (playing a file with xmms and alsaplayer and then
> killing alsaplayer), I was unable to make it happen again.  
> 
> I'll try some more after I get home from work tonight.

Sorry for the delay in getting back to you...

I have tested this patch as rigorously as I think possible :-)  I have
been unable to get it to break and switch from 4ch to 2ch (or vice versa)
after a number of hours of playing games, mp3s, videos, etc.  Seems to
work great here.

Adam



_______________________________________________________________
Hundreds of nodes, one monster rendering program.
Now that's a super model! Visit http://clustering.foundries.sf.net/

Re: Re: Yamaha 754 question...

[Takashi Iwai]

At Mon, 20 May 2002 12:36:53 -0400 (EDT),
Adam K Kirchhoff wrote:
> 
> 
> On Thu, 16 May 2002, Adam K Kirchhoff wrote:
> 
> > >
> > > hmm, init can be called multiple times.  i forgot it.
> > >
> > > the following is the revised patch.
> > > now ac3_init and _done are called at module init/end stage, so there
> > > will be no problems any more to switch 2ch/4ch accidentally.
> > > could you give a try?
> > 
> > I haven't had a chance to thoroughly test it, but I gave it a quick shot
> > this morning.  Using the same method that I was able to replicate the
> > problem last time (playing a file with xmms and alsaplayer and then
> > killing alsaplayer), I was unable to make it happen again.  
> > 
> > I'll try some more after I get home from work tonight.
> 
> Sorry for the delay in getting back to you...
> 
> I have tested this patch as rigorously as I think possible :-)  I have
> been unable to get it to break and switch from 4ch to 2ch (or vice versa)
> after a number of hours of playing games, mp3s, videos, etc.  Seems to
> work great here.

ok, now the patch is on cvs.
thanks for your report!


Takashi

_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm