Skip to contentSkip to conversation

Fix emu10k1 breakages in kernel 2.6.12

PATCH Lee Revell 5 participants 19 messages Jul 2, 2005 – Jul 5, 2005 Sound

lore
Share & tools
lore

Lee Revell

author
#0 lore raw

This set of critical fixes did not make it into the 2.6.12 kernel.

To better handle the large variety of devices it supports, the emu10k1 driver was recently cleaned up by moving all hardware details into a table keyed by PCI ID; unfortunately some IDs were missing from the table.

The typical effect is loss of mixer settings, mic not working, or no sound at all when upgrading to 2.6.12.

Here are the broken out patches for reference:

http://cvs.sourceforge.net/viewcvs.py/alsa/alsa-kernel/pci/emu10k1/emu10k1_main.c?rev=1.58&view=log

I hope I did not omit any of the authors.

From: James Courtier-Dutton <[email protected]>
From: Arnaud Patard <[email protected]>
From: Mikael Magnusson <[email protected]>
From: Peter Zubaj <[email protected]>
From: Lee Revell <[email protected]>
From: Takashi Iwai <[email protected]>

home/rlrevell/cvs/alsa/alsa-kernel/include/emu10k1.h renamed from sound/emu10k1.h +7 -8
@@ -1039,28 +1057,28 @@
 	u32 vendor;
 	u32 device;
 	u32 subsystem;
+	unsigned char revision;
 	unsigned char emu10k1_chip; /* Original SB Live. Not SB Live 24bit. */
 	unsigned char emu10k2_chip; /* Audigy 1 or Audigy 2. */
 	unsigned char ca0102_chip;  /* Audigy 1 or Audigy 2. Not SB Audigy 2 Value. */
 	unsigned char ca0108_chip;  /* Audigy 2 Value */
 	unsigned char ca0151_chip;  /* P16V */
 	unsigned char spk71;        /* Has 7.1 speakers */
+	unsigned char sblive51;	    /* SBLive! 5.1 - extout 0x11 -> center, 0x12 -> lfe */
 	unsigned char spdif_bug;    /* Has Spdif phasing bug */
 	unsigned char ac97_chip;    /* Has an AC97 chip */
 	unsigned char ecard;        /* APS EEPROM */
-	char * driver;
-	char * name;
+	const char *driver;
+	const char *name;
+	const char *id;		/* for backward compatibility - can be NULL if not needed */
 } emu_chip_details_t;
 
 struct _snd_emu10k1 {
 	int irq;
 
 	unsigned long port;			/* I/O port number */
-	unsigned int APS: 1,			/* APS flag */
-	    no_ac97: 1,				/* no AC'97 */
-	    tos_link: 1,			/* tos link detected */
-	    rear_ac97: 1,			/* rear channels are on AC'97 */
-	    spk71:1;				/* 7.1 configuration (Audigy 2 ZS) */
+	unsigned int tos_link: 1,		/* tos link detected */
+	    rear_ac97: 1;			/* rear channels are on AC'97 */
 	const emu_chip_details_t *card_capabilities;	/* Contains profile of card capabilities */
 	unsigned int audigy;			/* is Audigy? */
 	unsigned int revision;			/* chip revision */
@@ -1453,7 +1474,6 @@
 #endif
 
 typedef struct {
-	unsigned int card;			/* card type */
 	unsigned int internal_tram_size;	/* in samples */
 	unsigned int external_tram_size;	/* in samples */
 	char fxbus_names[16][32];		/* names of FXBUSes */
home/rlrevell/cvs/alsa/alsa-kernel/pci/emu10k1/emuproc.c renamed from pci/emu10k1/emuproc.c +1 -1
@@ -182,7 +189,7 @@
 	
 	snd_iprintf(buffer, "EMU10K1\n\n");
 	snd_iprintf(buffer, "Card                  : %s\n",
-		    emu->audigy ? "Audigy" : (emu->APS ? "EMU APS" : "Creative"));
+		    emu->audigy ? "Audigy" : (emu->card_capabilities->ecard ? "EMU APS" : "Creative"));
 	snd_iprintf(buffer, "Internal TRAM (words) : 0x%x\n", emu->fx8010.itram_size);
 	snd_iprintf(buffer, "External TRAM (words) : 0x%x\n", (int)emu->fx8010.etram_pages.bytes / 2);
 	snd_iprintf(buffer, "\n");
pci/emu10k1/emumixer.c renamed from pci/emu10k1/emumixer.c.orig +5 -5
@@ -791,7 +791,7 @@
 		NULL
 	};
 
-	if (!emu->no_ac97) {
+	if (emu->card_capabilities->ac97_chip) {
 		ac97_bus_t *pbus;
 		ac97_template_t ac97;
 		static ac97_bus_ops_t ops = {
@@ -833,7 +833,7 @@
 		for (; *c; c++)
 			remove_ctl(card, *c);
 	} else {
-		if (emu->APS)
+		if (emu->card_capabilities->ecard)
 			strcpy(emu->card->mixername, "EMU APS");
 		else if (emu->audigy)
 			strcpy(emu->card->mixername, "SB Audigy");
@@ -918,7 +918,7 @@
 		mix->attn[0] = 0xffff;
 	}
 	
-	if (! emu->APS) { /* FIXME: APS has these controls? */
+	if (! emu->card_capabilities->ecard) { /* FIXME: APS has these controls? */
 		/* sb live! and audigy */
 		if ((kctl = snd_ctl_new1(&snd_emu10k1_spdif_mask_control, emu)) == NULL)
 			return -ENOMEM;
@@ -939,14 +939,14 @@
 			return -ENOMEM;
 		if ((err = snd_ctl_add(card, kctl)))
 			return err;
-	} else if (! emu->APS) {
+	} else if (! emu->card_capabilities->ecard) {
 		/* sb live! */
 		if ((kctl = snd_ctl_new1(&snd_emu10k1_shared_spdif, emu)) == NULL)
 			return -ENOMEM;
 		if ((err = snd_ctl_add(card, kctl)))
 			return err;
 	}
-	if (emu->audigy && emu->revision == 4) { /* P16V */
+	if (emu->card_capabilities->ca0151_chip) { /* P16V */
 		if ((err = snd_p16v_mixer(emu)))
 			return err;
 	}
home/rlrevell/cvs/alsa/alsa-kernel/pci/emu10k1/emufx.c renamed from pci/emu10k1/emufx.c +10 -11
@@ -1077,7 +1077,7 @@
 	gpr += 2;
 	
 	/* PCM Side Playback (independent from stereo mix) */
-	if (emu->spk71) {
+	if (emu->card_capabilities->spk71) {
 		A_OP(icode, &ptr, iMAC0, A_GPR(playback+6), A_C_00000000, A_GPR(gpr), A_FXBUS(FXBUS_PCM_LEFT_SIDE));
 		A_OP(icode, &ptr, iMAC0, A_GPR(playback+7), A_C_00000000, A_GPR(gpr+1), A_FXBUS(FXBUS_PCM_RIGHT_SIDE));
 		snd_emu10k1_init_stereo_control(&controls[nctl++], "PCM Side Playback Volume", gpr, 100);
@@ -1145,14 +1145,14 @@
 	A_ADD_VOLUME_IN(stereo_mix, gpr, A_EXTIN_SPDIF_CD_L);
 	A_ADD_VOLUME_IN(stereo_mix+1, gpr+1, A_EXTIN_SPDIF_CD_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "CD Playback Volume" : "Audigy CD Playback Volume",
+					emu->card_capabilities->ac97_chip ? "Audigy CD Playback Volume" : "CD Playback Volume",
 					gpr, 0);
 	gpr += 2;
 	/* Audigy CD Capture Volume */
 	A_ADD_VOLUME_IN(capture, gpr, A_EXTIN_SPDIF_CD_L);
 	A_ADD_VOLUME_IN(capture+1, gpr+1, A_EXTIN_SPDIF_CD_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "CD Capture Volume" : "Audigy CD Capture Volume",
+					emu->card_capabilities->ac97_chip ? "Audigy CD Capture Volume" : "CD Capture Volume",
 					gpr, 0);
 	gpr += 2;
 
@@ -1171,14 +1171,14 @@
 	A_ADD_VOLUME_IN(stereo_mix, gpr, A_EXTIN_LINE2_L);
 	A_ADD_VOLUME_IN(stereo_mix+1, gpr+1, A_EXTIN_LINE2_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "Line Playback Volume" : "Line2 Playback Volume",
+					emu->card_capabilities->ac97_chip ? "Line2 Playback Volume" : "Line Playback Volume",
 					gpr, 0);
 	gpr += 2;
 	/* Line2 Capture Volume */
 	A_ADD_VOLUME_IN(capture, gpr, A_EXTIN_LINE2_L);
 	A_ADD_VOLUME_IN(capture+1, gpr+1, A_EXTIN_LINE2_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "Line Capture Volume" : "Line2 Capture Volume",
+					emu->card_capabilities->ac97_chip ? "Line2 Capture Volume" : "Line Capture Volume",
 					gpr, 0);
 	gpr += 2;
         
@@ -1197,14 +1197,14 @@
 	A_ADD_VOLUME_IN(stereo_mix, gpr, A_EXTIN_AUX2_L);
 	A_ADD_VOLUME_IN(stereo_mix+1, gpr+1, A_EXTIN_AUX2_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "Aux Playback Volume" : "Aux2 Playback Volume",
+					emu->card_capabilities->ac97_chip ? "Aux2 Playback Volume" : "Aux Playback Volume",
 					gpr, 0);
 	gpr += 2;
 	/* Aux2 Capture Volume */
 	A_ADD_VOLUME_IN(capture, gpr, A_EXTIN_AUX2_L);
 	A_ADD_VOLUME_IN(capture+1, gpr+1, A_EXTIN_AUX2_R);
 	snd_emu10k1_init_stereo_control(&controls[nctl++],
-					emu->no_ac97 ? "Aux Capture Volume" : "Aux2 Capture Volume",
+					emu->card_capabilities->ac97_chip ? "Aux2 Capture Volume" : "Aux Capture Volume",
 					gpr, 0);
 	gpr += 2;
 	
@@ -1232,7 +1232,7 @@
 	snd_emu10k1_init_mono_control(&controls[nctl++], "LFE Playback Volume", gpr, 0);
 	gpr++;
 	
-	if (emu->spk71) {
+	if (emu->card_capabilities->spk71) {
 		/* Stereo Mix Side Playback */
 		A_OP(icode, &ptr, iMAC0, A_GPR(playback+6), A_GPR(playback+6), A_GPR(gpr), A_GPR(stereo_mix));
 		A_OP(icode, &ptr, iMAC0, A_GPR(playback+7), A_GPR(playback+7), A_GPR(gpr+1), A_GPR(stereo_mix+1));
@@ -1266,7 +1266,7 @@
 	A_OP(icode, &ptr, iACC3, A_GPR(playback + SND_EMU10K1_PLAYBACK_CHANNELS + 3), A_GPR(playback + 3), A_C_00000000, A_C_00000000); /* rear right */
 	A_OP(icode, &ptr, iACC3, A_GPR(playback + SND_EMU10K1_PLAYBACK_CHANNELS + 4), A_GPR(playback + 4), A_C_00000000, A_C_00000000); /* center */
 	A_OP(icode, &ptr, iACC3, A_GPR(playback + SND_EMU10K1_PLAYBACK_CHANNELS + 5), A_GPR(playback + 5), A_C_00000000, A_C_00000000); /* LFE */
-	if (emu->spk71) {
+	if (emu->card_capabilities->spk71) {
 		A_OP(icode, &ptr, iACC3, A_GPR(playback + SND_EMU10K1_PLAYBACK_CHANNELS + 6), A_GPR(playback + 6), A_C_00000000, A_C_00000000); /* side left */
 		A_OP(icode, &ptr, iACC3, A_GPR(playback + SND_EMU10K1_PLAYBACK_CHANNELS + 7), A_GPR(playback + 7), A_C_00000000, A_C_00000000); /* side right */
 	}
@@ -1359,7 +1359,7 @@
 	A_PUT_STEREO_OUTPUT(A_EXTOUT_AREAR_L, A_EXTOUT_AREAR_R, playback+2 + SND_EMU10K1_PLAYBACK_CHANNELS);
 	A_PUT_OUTPUT(A_EXTOUT_ACENTER, playback+4 + SND_EMU10K1_PLAYBACK_CHANNELS);
 	A_PUT_OUTPUT(A_EXTOUT_ALFE, playback+5 + SND_EMU10K1_PLAYBACK_CHANNELS);
-	if (emu->spk71)
+	if (emu->card_capabilities->spk71)
 		A_PUT_STEREO_OUTPUT(A_EXTOUT_ASIDE_L, A_EXTOUT_ASIDE_R, playback+6 + SND_EMU10K1_PLAYBACK_CHANNELS);
 
 	/* headphone */
@@ -2128,7 +2133,6 @@
 	int res;
 
 	memset(info, 0, sizeof(info));
-	info->card = emu->card_type;
 	info->internal_tram_size = emu->fx8010.itram_size;
 	info->external_tram_size = emu->fx8010.etram_pages.bytes / 2;
 	fxbus = fxbuses;
home/rlrevell/cvs/alsa/alsa-kernel/pci/emu10k1/emu10k1_main.c renamed from pci/emu10k1/emu10k1_main.c +169 -23

305 lines not on this page · raw message

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#1 lore raw
Quoted 1 line · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

Is this patch in mainline already?

A few minor complaints that make me worried about it for the -stable kernel:
- it's bigger than expected for the -stable tree to take in just one patch. Can you split it up into smaller pieces (one with the new table values, the other with the logic changes?)
- wrong patch level, will not apply cleanly to a kernel tree.

thanks,

greg k-h

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#2 lore raw
Quoted 5 lines · Greg KH
Quoted 1 line · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

Is this patch in mainline already?

Yes, these fixes have been in ALSA CVS for 2-6 weeks.

Quoted 5 lines

A few minor complaints that make me worried about it for the -stable kernel:
- it's bigger than expected for the -stable tree to take in just one patch. Can you split it up into smaller pieces (one with the new table values, the other with the logic changes?)

The logic does have to change somewhat, because we were not handling unknown devices in a backward compatible way.

Here's the updated patch.

This is all only compile tested, as a "real" test would require access to a *lot* of hardware...

From: James Courtier-Dutton <[email protected]>
From: Arnaud Patard <[email protected]>
From: Mikael Magnusson <[email protected]>
From: Peter Zubaj <[email protected]>
From: Lee Revell <[email protected]>
From: Takashi Iwai <[email protected]>

include/sound/emu10k1.h +5 -2
@@ -1039,17 +1039,20 @@
 	u32 vendor;
 	u32 device;
 	u32 subsystem;
+	unsigned char revision;
 	unsigned char emu10k1_chip; /* Original SB Live. Not SB Live 24bit. */
 	unsigned char emu10k2_chip; /* Audigy 1 or Audigy 2. */
 	unsigned char ca0102_chip;  /* Audigy 1 or Audigy 2. Not SB Audigy 2 Value. */
 	unsigned char ca0108_chip;  /* Audigy 2 Value */
 	unsigned char ca0151_chip;  /* P16V */
+	unsigned char sblive51;	    /* SBLive! 5.1 - extout 0x11 -> center, 0x12 -> lfe */
 	unsigned char spk71;        /* Has 7.1 speakers */
 	unsigned char spdif_bug;    /* Has Spdif phasing bug */
 	unsigned char ac97_chip;    /* Has an AC97 chip */
 	unsigned char ecard;        /* APS EEPROM */
-	char * driver;
-	char * name;
+	const char *driver;
+	const char *name;
+	const char *id;		/* for backward compatibility - can be NULL if not needed */
 } emu_chip_details_t;
 
 struct _snd_emu10k1 {
sound/pci/emu10k1/emu10k1_main.c +165 -10

258 lines not on this page · raw message

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#3 lore raw
Quoted 8 lines · Lee Revell
Quoted 5 lines · Greg KH
Quoted 1 line · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

Is this patch in mainline already?

Yes, these fixes have been in ALSA CVS for 2-6 weeks.

Not my question. Are they in Linus's tree yet?

Quoted 8 lines
Quoted 5 lines

A few minor complaints that make me worried about it for the -stable kernel:
- it's bigger than expected for the -stable tree to take in just one patch. Can you split it up into smaller pieces (one with the new table values, the other with the logic changes?)

The logic does have to change somewhat, because we were not handling unknown devices in a backward compatible way.

I'm not asking you to change the logic of the change, just break it up into smaller pieces that are easier for you to look at and say, "of course that is correct!" :)

thanks,

greg k-h

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#4 lore raw
Quoted 12 lines · Greg KH
Quoted 8 lines · Lee Revell
Quoted 5 lines · Greg KH
Quoted 1 line · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

Is this patch in mainline already?

Yes, these fixes have been in ALSA CVS for 2-6 weeks.

Not my question. Are they in Linus's tree yet?

Yes, they are in 2.6.13-rc1.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#5 lore raw
In reply to Lee Revell
Quoted 10 lines · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

To better handle the large variety of devices it supports, the emu10k1 driver was recently cleaned up by moving all hardware details into a table keyed by PCI ID; unfortunately some IDs were missing from the table.

The typical effect is loss of mixer settings, mic not working, or no sound at all when upgrading to 2.6.12.

Hold on. It is supposed to be done in a backwards compatible way already. Before all card varients we handled as being the same card, except for a few special cases. The first version of the patch took the general case and named it "Audigy Unknown" and gave it the correct parameters as were used before in the general case of all card varients being the same. Then, I looked through the code, and extracted all the special cases, and created specific recognition of those cards and set the correct feature variables, so they were also handled. The result of this should have been that there were no performance change between either versions. Then, as more card types were identified, I added special cases for them, but in their case, they used all the default parameters for the general case, the only difference was that the correct Model name was recorded and displayed in /proc/asound/cards. This is mainly just to help users identify their particular card, without having to remove the case of their PC. So this helps them report bugs better.

If there are loss of features during the upgrade, I would be very suprised. I suspect that the same problems one sees now with some cards, has been there all along. If that is not the case, please report them in the bug tracker in the normal way.

James

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#6 lore raw
Quoted 12 lines · James Courtier-Dutton
Quoted 10 lines · Lee Revell

This set of critical fixes did not make it into the 2.6.12 kernel.

To better handle the large variety of devices it supports, the emu10k1 driver was recently cleaned up by moving all hardware details into a table keyed by PCI ID; unfortunately some IDs were missing from the table.

The typical effect is loss of mixer settings, mic not working, or no sound at all when upgrading to 2.6.12.

Quoted 17 lines

Hold on. It is supposed to be done in a backwards compatible way already. Before all card varients we handled as being the same card, except for a few special cases. The first version of the patch took the general case and named it "Audigy Unknown" and gave it the correct parameters as were used before in the general case of all card varients being the same. Then, I looked through the code, and extracted all the special cases, and created specific recognition of those cards and set the correct feature variables, so they were also handled. The result of this should have been that there were no performance change between either versions. Then, as more card types were identified, I added special cases for them, but in their case, they used all the default parameters for the general case, the only difference was that the correct Model name was recorded and displayed in /proc/asound/cards. This is mainly just to help users identify their particular card, without having to remove the case of their PC. So this helps them report bugs better.

Right, but if the name does not match, alsactl cannot restore the previous mixer settings, so the mixer is reset to the default state, ie no sound.

Quoted 5 lines

If there are loss of features during the upgrade, I would be very suprised. I suspect that the same problems one sees now with some cards, has been there all along. If that is not the case, please report them in the bug tracker in the normal way.

The big one is that several emu10k2 based devices like "Audigy 1 or 2 [Unknown]", "Audigy 2 Value [Unknown]", and "Audigy 1 [SB0090]" failed to set ac97_chip, resulting in non-working mic/line in. Here's pzad's fix:

http://cvs.sourceforge.net/viewcvs.py/alsa/alsa-kernel/pci/emu10k1/emu10k1_main.c?r1=1.52&r2=1.53

Users are already getting bitten, here's a bug report:

https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1218

In addition to 2.6.12, it looks like FC4 shipped with this bug.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#7 lore raw
Quoted 5 lines · Lee Revell

Right, but if the name does not match, alsactl cannot restore the previous mixer settings, so the mixer is reset to the default state, ie no sound.

Ok, which name should we be using here?
cat /proc/asound/cards
0 [Audigy2 ]: Audigy2 - Audigy 4 PRO [SB0380]

                     Audigy 4 PRO [SB0380] (rev.4, serial:0x20071102) at
0x9000, irq 23
1 [Audigy2_1      ]: Audigy2 - Audigy 2 Value [SB0400]
                     Audigy 2 Value [SB0400] (rev.0, serial:0x10011102)
at 0x9400, irq 17
2 [CA0106         ]: CA0106 - CA0106
                     Live! 7.1 24bit [SB0410] at 0x8400 irq 21
3 [CA0106_1       ]: CA0106 - CA0106
                     AudigyLS [SB0310] at 0x8800 irq 22

But alsactl asound.state uses the following names:
state.SB0400 {
state.SB0240 {
state.SB0380 {
state.CA0106 {
state.Audigy2 {

If I clear asound.state, I now get:
state.Audigy2 {
state.Audigy2_1 {
state.CA0106 {
state.CA0106_1 {

This must have been a change to alsactl. I think that using Audigy2_1 etc. is a much better idea than using something like SB0400.

I don't believe my changes to emu10k1 caused this to change. It was something independent that changed.

James

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#8 lore raw
Quoted 2 lines · James Courtier-Dutton

I don't believe my changes to emu10k1 caused this to change. It was something independent that changed.

Ugh, you are right, this is alsactl's problem. So it looks like users who upgrade to alsa-utils 1.0.9+ will lose their mixer settings...

I prefer the SB*** names, because they uniquely identify the hardware, and users seem to know them. AFAICT it's possible for 'Audigy2' to have an ac97 codec and 'Audigy2_1' not, resulting in different & incompatible mixer settings.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#9 lore raw
Quoted 6 lines · Lee Revell
Quoted 2 lines · James Courtier-Dutton

I don't believe my changes to emu10k1 caused this to change. It was something independent that changed.

Ugh, you are right, this is alsactl's problem. So it looks like users who upgrade to alsa-utils 1.0.9+ will lose their mixer settings...

Never mind, I just tested and it seems to restore the settings correctly, whether the card is called Audigy2 or SB0350. It looks like alsactl is much smarter than it used to be.

So please disregard the part of my message about mixer settings, for -stable we just have to handle the ac97/noac97 type issues.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#10 lore raw
In reply to Lee Revell
Quoted 15 lines · Lee Revell
Quoted 6 lines

If there are loss of features during the upgrade, I would be very suprised. I suspect that the same problems one sees now with some cards, has been there all along. If that is not the case, please report them in the bug tracker in the normal way.

The big one is that several emu10k2 based devices like "Audigy 1 or 2 [Unknown]", "Audigy 2 Value [Unknown]", and "Audigy 1 [SB0090]" failed to set ac97_chip, resulting in non-working mic/line in. Here's pzad's fix:

http://cvs.sourceforge.net/viewcvs.py/alsa/alsa-kernel/pci/emu10k1/emu10k1_main.c?r1=1.52&r2=1.53

Quoted 8 lines

Users are already getting bitten, here's a bug report:

https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1218

In addition to 2.6.12, it looks like FC4 shipped with this bug.

Lee

I suppose it depends what one wants to happen when the driver comes across a unrecognised sound card variant.
For example, we have ac97 for the Audigy 2 LS (a ca0106 chip), but newer cards with the ca0106 chip have no ac97 codec. I thought that if we come across an unrecognised card, we should assume that it does not have external chips, and then add specific support into the driver if it does. So, now, if a new Audigy 2 arrives without any ac97 chip, the snd-emu10k1 module will fail to load at all. At least if the default unknown entry excluded the ac97 chip, the user would at least have sound playback, just no capture.

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

James

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#11 lore raw
Quoted 11 lines · James Courtier-Dutton

I suppose it depends what one wants to happen when the driver comes across a unrecognised sound card variant.
For example, we have ac97 for the Audigy 2 LS (a ca0106 chip), but newer cards with the ca0106 chip have no ac97 codec. I thought that if we come across an unrecognised card, we should assume that it does not have external chips, and then add specific support into the driver if it does. So, now, if a new Audigy 2 arrives without any ac97 chip, the snd-emu10k1 module will fail to load at all. At least if the default unknown entry excluded the ac97 chip, the user would at least have sound playback, just no capture.

Quoted 3 lines

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Right, if the user has hardware that we've never seen, I guess the behavior is always undefined. IMHO for the -stable series we should default to whatever the driver from 2.6.11 would have done (if possible).

My main point was not to handle any new devices that come along, but to fix the regressions that users with known hardware are seeing. The simplest way seems to be to add the missing entries to the list but default to assuming no ac97 etc. as you suggested.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#12 lore raw
In reply to James Courtier-Dutton
Quoted 3 lines · James Courtier-Dutton

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#13 lore raw

At Tue, 05 Jul 2005 11:33:37 -0400,

Quoted 8 lines · Lee Revell
Quoted 3 lines · James Courtier-Dutton

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

I agree with James.

Takashi

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

author
#14 lore raw
Quoted 4 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:33:37 -0400,

Quoted 2 lines · Lee Revell

On Sat, 2005-07-02 at 23:46 +0100, James Courtier-Dutton wrote:

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

I agree with James.

OK, James, do you want to generate a new patch for -stable? As the
developer of this feature, I think you understand the behavior better than me...

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#15 lore raw

At Tue, 05 Jul 2005 11:41:44 -0400,

Quoted 10 lines · Lee Revell
Quoted 8 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:33:37 -0400,

Quoted 6 lines · Lee Revell
Quoted 3 lines · James Courtier-Dutton

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

I agree with James.

OK, James, do you want to generate a new patch for -stable? As the
developer of this feature, I think you understand the behavior better than me...

BTW, it would be nice, too, if we have a "model" module option like other drivers. It would give users a chance to choose another card model without compiling drivers if the card is not detected correctly.

This isn't necessarily put to -stable tree, though.

Takashi

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#16 lore raw
Quoted 26 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:41:44 -0400,

Quoted 10 lines · Lee Revell
Quoted 8 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:33:37 -0400,

Quoted 6 lines · Lee Revell
Quoted 3 lines · James Courtier-Dutton

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

I agree with James.

OK, James, do you want to generate a new patch for -stable? As the
developer of this feature, I think you understand the behavior better than me...

BTW, it would be nice, too, if we have a "model" module option like other drivers. It would give users a chance to choose another card model without compiling drivers if the card is not detected correctly.

Quoted 5 lines

This isn't necessarily put to -stable tree, though.

Takashi

I don't know about stable stuff.
I currently only have access to the alsa cvs, so I put fixes in their and let other people migrate those into kernel xyz.
I will look at the "model" option when I get a chance next weekend.

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id492&op=click

author
#17 lore raw
Quoted 5 lines · James Courtier-Dutton

I don't know about stable stuff.
I currently only have access to the alsa cvs, so I put fixes in their and let other people migrate those into kernel xyz.
I will look at the "model" option when I get a chance next weekend.

I think for -stable we need to just apply the original patch that I sent. We just need the minimal change to restore the mic and line in on Audigy 1 devices that resulted from the incorrect setting of ac97.

Lee

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

#18 lore raw
In reply to James Courtier-Dutton

At Tue, 5 Jul 2005 17:16:05 +0100,

Quoted 28 lines · James Courtier-Dutton
Quoted 26 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:41:44 -0400,

Quoted 10 lines · Lee Revell
Quoted 8 lines · Takashi Iwai

At Tue, 05 Jul 2005 11:33:37 -0400,

Quoted 6 lines · Lee Revell
Quoted 3 lines · James Courtier-Dutton

So, my general point is that people are going to get bitten no matter what we do for the "unknown" case, but my original method was going to have less of an impact that potentially yours will.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Any more feedback on this? We absolutely need to address the problem in the next -stable release.

I agree with James.

OK, James, do you want to generate a new patch for -stable? As the
developer of this feature, I think you understand the behavior better than me...

BTW, it would be nice, too, if we have a "model" module option like other drivers. It would give users a chance to choose another card model without compiling drivers if the card is not detected correctly.

Quoted 7 lines
Quoted 5 lines

This isn't necessarily put to -stable tree, though.

Takashi

I don't know about stable stuff.

It's 2.6.12.x kernel. Just make a diff for that.

Takashi

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click