[PATCH] wss_lib: snd_wss_calibrate_mute improvement

[PATCH] wss_lib: snd_wss_calibrate_mute improvement

[Krzysztof Helt]

From: Krzysztof Helt <krzysztof.h1@wp.pl>

Mute sound by setting mute bit without
setting volume to 0. It makes both source code
and binary shorter.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
---


diff -urp alsa-ref/sound/isa/wss/wss_lib.c alsa-driver/sound/isa/wss/wss_lib.c
--- alsa-ref/sound/isa/wss/wss_lib.c	2008-08-14 00:05:30.000000000 +0200
+++ alsa-driver/sound/isa/wss/wss_lib.c	2008-08-17 22:33:24.620085406 +0200
@@ -574,7 +574,7 @@ static void snd_wss_calibrate_mute(struc
 {
 	unsigned long flags;
 
-	mute = mute ? 1 : 0;
+	mute = mute ? 0x80 : 0;
 	spin_lock_irqsave(&chip->reg_lock, flags);
 	if (chip->calibrate_mute == mute) {
 		spin_unlock_irqrestore(&chip->reg_lock, flags);
@@ -589,34 +589,34 @@ static void snd_wss_calibrate_mute(struc
 			     chip->image[CS4231_LOOPBACK]);
 	}
 	snd_wss_dout(chip, CS4231_AUX1_LEFT_INPUT,
-		     mute ? 0x80 : chip->image[CS4231_AUX1_LEFT_INPUT]);
+		     mute | chip->image[CS4231_AUX1_LEFT_INPUT]);
 	snd_wss_dout(chip, CS4231_AUX1_RIGHT_INPUT,
-		     mute ? 0x80 : chip->image[CS4231_AUX1_RIGHT_INPUT]);
+		     mute | chip->image[CS4231_AUX1_RIGHT_INPUT]);
 	snd_wss_dout(chip, CS4231_AUX2_LEFT_INPUT,
-		     mute ? 0x80 : chip->image[CS4231_AUX2_LEFT_INPUT]);
+		     mute | chip->image[CS4231_AUX2_LEFT_INPUT]);
 	snd_wss_dout(chip, CS4231_AUX2_RIGHT_INPUT,
-		     mute ? 0x80 : chip->image[CS4231_AUX2_RIGHT_INPUT]);
+		     mute | chip->image[CS4231_AUX2_RIGHT_INPUT]);
 	snd_wss_dout(chip, CS4231_LEFT_OUTPUT,
-		     mute ? 0x80 : chip->image[CS4231_LEFT_OUTPUT]);
+		     mute | chip->image[CS4231_LEFT_OUTPUT]);
 	snd_wss_dout(chip, CS4231_RIGHT_OUTPUT,
-		     mute ? 0x80 : chip->image[CS4231_RIGHT_OUTPUT]);
+		     mute | chip->image[CS4231_RIGHT_OUTPUT]);
 	if (!(chip->hardware & WSS_HW_AD1848_MASK)) {
 		snd_wss_dout(chip, CS4231_LEFT_LINE_IN,
-			     mute ? 0x80 : chip->image[CS4231_LEFT_LINE_IN]);
+			     mute | chip->image[CS4231_LEFT_LINE_IN]);
 		snd_wss_dout(chip, CS4231_RIGHT_LINE_IN,
-			     mute ? 0x80 : chip->image[CS4231_RIGHT_LINE_IN]);
+			     mute | chip->image[CS4231_RIGHT_LINE_IN]);
 		snd_wss_dout(chip, CS4231_MONO_CTRL,
 			     mute ? 0xc0 : chip->image[CS4231_MONO_CTRL]);
 	}
 	if (chip->hardware == WSS_HW_INTERWAVE) {
 		snd_wss_dout(chip, CS4231_LEFT_MIC_INPUT,
-			     mute ? 0x80 : chip->image[CS4231_LEFT_MIC_INPUT]);
+			     mute | chip->image[CS4231_LEFT_MIC_INPUT]);
 		snd_wss_dout(chip, CS4231_RIGHT_MIC_INPUT,
-			     mute ? 0x80 : chip->image[CS4231_RIGHT_MIC_INPUT]);
+			     mute | chip->image[CS4231_RIGHT_MIC_INPUT]);
 		snd_wss_dout(chip, CS4231_LINE_LEFT_OUTPUT,
-			mute ? 0x80 : chip->image[CS4231_LINE_LEFT_OUTPUT]);
+			     mute | chip->image[CS4231_LINE_LEFT_OUTPUT]);
 		snd_wss_dout(chip, CS4231_LINE_RIGHT_OUTPUT,
-			mute ? 0x80 : chip->image[CS4231_LINE_RIGHT_OUTPUT]);
+			     mute | chip->image[CS4231_LINE_RIGHT_OUTPUT]);
 	}
 	chip->calibrate_mute = mute;
 	spin_unlock_irqrestore(&chip->reg_lock, flags);

---------------------------------------------------------------------- 
Igrzyska na Dzikim Zachodzie!
Sprawdz >> http://link.interia.pl/f1edc 
	

Re: [PATCH] wss_lib: snd_wss_calibrate_mute improvement

[Rene Herman]

On 17-08-08 22:38, Krzysztof Helt wrote:

> From: Krzysztof Helt <krzysztof.h1@wp.pl>
> 
> Mute sound by setting mute bit without
> setting volume to 0. It makes both source code
> and binary shorter.
> 
> Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>

Grmmbbled-over-by: Rene Herman <rene.herman@gmail.com>

This should be safe and the current function indeed generates crap code 
but in the sense of always expecting the worst broken hardware variants 
to be out there I must say I'd rather have seen that GCC were beter. But 
yes, sure.


Rene.

Re: [PATCH] wss_lib: snd_wss_calibrate_mute improvement

[Takashi Iwai]

At Mon, 18 Aug 2008 10:50:15 +0200,
Rene Herman wrote:
> 
> On 17-08-08 22:38, Krzysztof Helt wrote:
> 
> > From: Krzysztof Helt <krzysztof.h1@wp.pl>
> > 
> > Mute sound by setting mute bit without
> > setting volume to 0. It makes both source code
> > and binary shorter.
> > 
> > Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
> 
> Grmmbbled-over-by: Rene Herman <rene.herman@gmail.com>
> 
> This should be safe and the current function indeed generates crap code 
> but in the sense of always expecting the worst broken hardware variants 
> to be out there I must say I'd rather have seen that GCC were beter. But 
> yes, sure.

Yeah, the slightest concern is the intention of using 0x80 there.
If this patch was already tested on some real hardwares (and I guess
it should work), I'm willing to apply it.  Krzystof, was this tested
on your machines?


Takashi

Re: [PATCH] wss_lib: snd_wss_calibrate_mute improvement

[Rene Herman]

On 18-08-08 12:16, Takashi Iwai wrote:

> At Mon, 18 Aug 2008 10:50:15 +0200,
> Rene Herman wrote:
>> On 17-08-08 22:38, Krzysztof Helt wrote:
>>
>>> From: Krzysztof Helt <krzysztof.h1@wp.pl>
>>>
>>> Mute sound by setting mute bit without
>>> setting volume to 0. It makes both source code
>>> and binary shorter.
>>>
>>> Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
>> Grmmbbled-over-by: Rene Herman <rene.herman@gmail.com>
>>
>> This should be safe and the current function indeed generates crap code 
>> but in the sense of always expecting the worst broken hardware variants 
>> to be out there I must say I'd rather have seen that GCC were beter. But 
>> yes, sure.
> 
> Yeah, the slightest concern is the intention of using 0x80 there.

Yes, shared that same concern but it's very likely not significant. The 
original thought may just have been that you have to write _something_ 
so you might as well simply write 0x80 instead or or-ing in a bit and 
with volume at 0, your're even "doubly safe"...

> If this patch was already tested on some real hardwares (and I guess
> it should work), I'm willing to apply it.  Krzystof, was this tested
> on your machines?

Not (yet) on any of mine at least but actually having something where it 
does not work would be surprising. I could sort of imagine something 
being out there that ignores the mute bit but I doubt that either of us 
is actually going to have it even if it is.

Avoiding cargo-cult programming isn't unimportant so let's not be _too_ 
cautious I guess. It does currently make for crap code (GCC could be 
using cmov here it seems but it doesn't) which is an excuse...

Rene.

Re: [PATCH] wss_lib: snd_wss_calibrate_mute improvement

[krzysztof.h1]

> At Mon, 18 Aug 2008 10:50:15 +0200,
> Rene Herman wrote:
> > 
> > On 17-08-08 22:38, Krzysztof Helt wrote:
> > 
> > > From: Krzysztof Helt <krzysztof.h1@wp.pl>
> > > 
> > > Mute sound by setting mute bit without
> > > setting volume to 0. It makes both source code
> > > and binary shorter.
> > > 
> > > Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
> > 
> > Grmmbbled-over-by: Rene Herman <rene.herman@gmail.com>
> > 
> > This should be safe and the current function indeed generates crap code
> 
> > but in the sense of always expecting the worst broken hardware variants
> 
> > to be out there I must say I'd rather have seen that GCC were beter. But
> 
> > yes, sure.
> 
> Yeah, the slightest concern is the intention of using 0x80 there.
> If this patch was already tested on some real hardwares (and I guess
> it should work), I'm willing to apply it.  Krzystof, was this tested
> on your machines?
> 

I have tested it only on cs4236 (inside laptop) and opti931 I have in the old PC at the moment. If you want I can retest on more hardware.

The second check I noticed that for some registers the volume is defined as gain (so 0 is the least audible) and for some (CS4231_MONO_CTRL, CS4231_LEFT/RIGHT_OUTPUT) is defined as attentuation (the 0 is the most audible). In two cases, the old code set volume to the loudest and mute it with mute bit, probably the opposite what was intended.

It is not a crucial patch. You may drop it.

Regards,
Krzysztof

----------------------------------------------------------------------
Wygraj laptopa lub iPoda! Wez udzial w igrzyskach!
Kliknij >>> http://link.interia.pl/f1eda 

Re: [PATCH] wss_lib: snd_wss_calibrate_mute improvement

[Takashi Iwai]

At 18 Aug 2008 13:16:56 +0200,
krzysztof.h1@poczta.fm wrote:
> 
> > At Mon, 18 Aug 2008 10:50:15 +0200,
> > Rene Herman wrote:
> > > 
> > > On 17-08-08 22:38, Krzysztof Helt wrote:
> > > 
> > > > From: Krzysztof Helt <krzysztof.h1@wp.pl>
> > > > 
> > > > Mute sound by setting mute bit without
> > > > setting volume to 0. It makes both source code
> > > > and binary shorter.
> > > > 
> > > > Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
> > > 
> > > Grmmbbled-over-by: Rene Herman <rene.herman@gmail.com>
> > > 
> > > This should be safe and the current function indeed generates crap code
> > 
> > > but in the sense of always expecting the worst broken hardware variants
> > 
> > > to be out there I must say I'd rather have seen that GCC were beter. But
> > 
> > > yes, sure.
> > 
> > Yeah, the slightest concern is the intention of using 0x80 there.
> > If this patch was already tested on some real hardwares (and I guess
> > it should work), I'm willing to apply it.  Krzystof, was this tested
> > on your machines?
> > 
> 
> I have tested it only on cs4236 (inside laptop) and opti931 I have in the old PC at the moment. If you want I can retest on more hardware.
> 
> The second check I noticed that for some registers the volume is defined as gain (so 0 is the least audible) and for some (CS4231_MONO_CTRL, CS4231_LEFT/RIGHT_OUTPUT) is defined as attentuation (the 0 is the most audible). In two cases, the old code set volume to the loudest and mute it with mute bit, probably the opposite what was intended.
> 
> It is not a crucial patch. You may drop it.

Oh, no, your patch is actually a correct fix.

The thing is that I wanted to be sure that this doesn't introduce an
obvious breakage.


Takashi