Skip to contentSkip to conversation

sound/soc/codecs: Mark snd-soc-wm2200 and snd-soc-wm-adsp as BROKEN

PATCH In review Peter Hurley 3 participants 6 messages Mar 29, 2013 – Mar 30, 2013

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sound/soc/codecs: Mark snd-soc-wm2200 and snd-soc-wm-adsp as BROKEN

The linux-next tree has failed to build for last 7 days with these errors:
MODPOST 2490 modules
ERROR: "arizona_rate_text" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
ERROR: "arizona_rate_val" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
WARNING: modpost: Found 5 section mismatch(es).

Patches submitted by Xiong Zhou <[email protected]> to fix this have been rejected.

Until the author(s) resolve the situation, mark the offenders as BROKEN.

Cc Xiong Zhou <[email protected]>
Cc Mark Brown <[email protected]>
---
 sound/soc/codecs/Kconfig | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
sound/soc/codecs/Kconfig +2 -1
index 18fea10..470b66b 100644
@@ -79,7 +79,7 @@ config SND_SOC_ALL_CODECS
 	select SND_SOC_WM0010 if SPI_MASTER
 	select SND_SOC_WM1250_EV1 if I2C
 	select SND_SOC_WM2000 if I2C
-	select SND_SOC_WM2200 if I2C
+	select SND_SOC_WM2200 if I2C && BROKEN
 	select SND_SOC_WM5100 if I2C
 	select SND_SOC_WM5102 if MFD_WM5102
 	select SND_SOC_WM5110 if MFD_WM5110
@@ -153,6 +153,7 @@ config SND_SOC_WM_HUBS
 
 config SND_SOC_WM_ADSP
 	tristate
+	depends on BROKEN
 	default y if SND_SOC_WM5102=y
 	default y if SND_SOC_WM2200=y
 	default m if SND_SOC_WM5102=m
Signature · 1 line
1.8.1.2
#1 lore raw
Quoted 2 lines · Peter Hurley

The linux-next tree has failed to build for last 7 days with these errors:

In some pretty specific and basically unrealistic configurations rather than anything a human is likely to want to use or the core automatic ones.

Quoted 9 lines

MODPOST 2490 modules
ERROR: "arizona_rate_text" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
ERROR: "arizona_rate_val" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
WARNING: modpost: Found 5 section mismatch(es).

Patches submitted by Xiong Zhou <[email protected]> to fix this have been rejected.

Until the author(s) resolve the situation, mark the offenders as BROKEN.

I'm not going to apply this, it's clearly not a constructive or helpful effort to address the issue - it's not engaging with any of the discussion about how to resolve the issue (there's no mention of the suggested fix for example).

If you're going to run randconfigs that's great and the testing is useful but if you're going to send patches as well as reports it would be really helpful to look at what the code is doing so that the change makes sense rather than just sending the first thing that springs to mind. The smarter the stuff that gets sent is the more likely it is to either get merged or encourage people to work on the problem.

In this case it seems like Xiong is likely to fix things and it'd be much better overall to get a fix from Xiong merged (which we seem to be progressing towards).

#2 lore raw
Quoted 31 lines · Mark Brown
Quoted 2 lines · Peter Hurley

The linux-next tree has failed to build for last 7 days with these errors:

In some pretty specific and basically unrealistic configurations rather than anything a human is likely to want to use or the core automatic ones.

Quoted 9 lines

MODPOST 2490 modules
ERROR: "arizona_rate_text" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
ERROR: "arizona_rate_val" [sound/soc/codecs/snd-soc-wm-adsp.ko] undefined!
WARNING: modpost: Found 5 section mismatch(es).

Patches submitted by Xiong Zhou <[email protected]> to fix this have been rejected.

Until the author(s) resolve the situation, mark the offenders as BROKEN.

I'm not going to apply this, it's clearly not a constructive or helpful effort to address the issue - it's not engaging with any of the discussion about how to resolve the issue (there's no mention of the suggested fix for example).

If you're going to run randconfigs that's great and the testing is useful but if you're going to send patches as well as reports it would be really helpful to look at what the code is doing so that the change makes sense rather than just sending the first thing that springs to mind. The smarter the stuff that gets sent is the more likely it is to either get merged or encourage people to work on the problem.

Quoted 3 lines

In this case it seems like Xiong is likely to fix things and it'd be much better overall to get a fix from Xiong merged (which we seem to be progressing towards).

This is stock Ubuntu configs.

#3 lore raw
Quoted 1 line · Peter Hurley

This is stock Ubuntu configs.

Then the stock Ubuntu configuration is not a sensible one, enabling SND_SOC_ALL_CODECS for any purpose other than build coverage testing is just a waste of time - that's all it does.

#4 lore raw
Quoted 8 lines · Mark Brown
Quoted 1 line · Peter Hurley

This is stock Ubuntu configs.

Then the stock Ubuntu configuration is not a sensible one, enabling SND_SOC_ALL_CODECS for any purpose other than build coverage testing is just a waste of time - that's all it does.

Take a look at this:

MFD_ARIZONA_I2C select

               > -------- > MFD_ARIZONA <-|
MDF_ARIZONA_SPI                           |
                                          |
					  |depends on
					  |
					  |

SND_SOC_ALL_CODECS-\ / \
I2C ---------/| MFD_WM5110 MFD_WM5102
| | |
| |select | select
| | |
v v v
SND_SOC_WM2200 SND_SOC_WM5102 SND_SOC_WM5110
| | | |
|____________________| |___________|
| |
| if M/Y | if M/Y
v v
SND_SOC_WM_ADSP SND_SOC_ARIZONA

If I2C is enabled normally, and SND_SOC_ALL_CODECS is keeped in the sound card configure options.

Then, disable ALL options in:

	Device drivers  ---->
		Multifunction device drivers --->

which includes MFD_ARIZONA_I2C and MDF_ARIZONA_SPI.

So the failure comes. Yes, this config is kind of specific, not much.

The definition of one array which is cited by one module comes in another module, and this dependence does not appear in Kconfig architecture. Indeed, the help text of ALL_CODECS reminds the separated surpporting bus options. Still I think this is unreasonable. Even though build coverage test is the only proper purpose. It breaks the whole building procedure.

I think copying the same array definition across c file is ugly, and Mark said that this is not a Kconfig issue. So far, I have not got any other idea to fix this. Code cruising.

#5 lore raw
Quoted 1 line · Xiong Zhou

So the failure comes. Yes, this config is kind of specific, not much.

Right, but look at what the options are doing - ALL_CODECS is purely for build coverage so someone seeing the issue in this form must be running a partial build coverage test which happens to cover a relatively obscure subsystem. This is why none of the all*configs that are run on -next as standard fail.

Quoted 6 lines

The definition of one array which is cited by one module comes in another module, and this dependence does not appear in Kconfig architecture. Indeed, the help text of ALL_CODECS reminds the separated surpporting bus options. Still I think this is unreasonable. Even though build coverage test is the only proper purpose. It breaks the whole building procedure.

I'm not saying this doesn't need to be fixed, I'm saying we need a sane fix.

Quoted 3 lines

I think copying the same array definition across c file is ugly, and Mark said that this is not a Kconfig issue. So far, I have not got any other idea to fix this. Code cruising.

I actually looked at this earlier and sent a fix for it (didn't get round to mailing this thread yet, sorry), the code wouldn't have worked at all for WM2200 - this is the sort of thing that should be apparent as soon as people start asking questions like "why was this dependency added, is it sensible?" which should be among the first questions asked when turning up something like this, especially for recent changes.