Re: linux-next: Tree for Feb 18 (sound/soc/codecs/wm5110.c)
On 2/17/20 8:28 PM, Stephen Rothwell wrote:
Hi all,
Changes since 20200217:
on x86_64:
WARNING: unmet direct dependencies detected for SND_SOC_WM5110 Depends on [n]: SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && MFD_WM5110 [=n] Selected by [y]: - SND_SOC_BELLS [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (MACH_WLF_CRAGG_6410 || COMPILE_TEST [=y]) - SND_SOC_SAMSUNG_TM2_WM5110 [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (GPIOLIB [=y] || COMPILE_TEST [=y])
Full randconfig file is attached.
[resending to broonie@kernel.org; other email address got message too big for mailbox]
On 2/17/20 11:04 PM, Randy Dunlap wrote:
On 2/17/20 8:28 PM, Stephen Rothwell wrote:
Hi all,
Changes since 20200217:
on x86_64:
WARNING: unmet direct dependencies detected for SND_SOC_WM5110 Depends on [n]: SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && MFD_WM5110 [=n] Selected by [y]:
- SND_SOC_BELLS [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (MACH_WLF_CRAGG_6410 || COMPILE_TEST [=y])
- SND_SOC_SAMSUNG_TM2_WM5110 [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (GPIOLIB [=y] || COMPILE_TEST [=y])
Full randconfig file is attached.
On Mon, Feb 17, 2020 at 11:05:49PM -0800, Randy Dunlap wrote:
[resending to broonie@kernel.org; other email address got message too big for mailbox]
Will have a word with our IT guys, Mark's opensource.wolfsonmicro.com account certainly shouldn't still be accepting email.
On 2/17/20 11:04 PM, Randy Dunlap wrote:
On 2/17/20 8:28 PM, Stephen Rothwell wrote:
Hi all,
Changes since 20200217:
on x86_64:
WARNING: unmet direct dependencies detected for SND_SOC_WM5110 Depends on [n]: SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && MFD_WM5110 [=n] Selected by [y]:
- SND_SOC_BELLS [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (MACH_WLF_CRAGG_6410 || COMPILE_TEST [=y])
- SND_SOC_SAMSUNG_TM2_WM5110 [=y] && SOUND [=y] && !UML && SND [=y] && SND_SOC [=y] && SND_SOC_SAMSUNG [=y] && MFD_ARIZONA [=y] && I2C [=y] && SPI_MASTER [=y] && (GPIOLIB [=y] || COMPILE_TEST [=y])
Full randconfig file is attached.
This looks like it comes from these two patches:
ea00d95200d0 ("ASoC: Use imply for SND_SOC_ALL_CODECS") d8dd3f92a6ba ("ASoC: Fix SND_SOC_ALL_CODECS imply misc fallout")
What is happening is the machine drivers still select these symbols but this doesn't take account of the dependencies, which are now added by the second of those patches.
I guess there are a couple of options:
1) We could update the machine drivers to do an imply as well, like the changes to SND_SOC_ALL_CODECS. I think this will generally work but doesn't really feel right as you can end up with the machine driver built without the CODEC.
2) We could update the machine drivers to a depends on and make the CODEC symbols manually selectable. I guess this also has the advantages that it would be necessary for the simple card stuff that seems to be everyones first choice for machine drivers these days.
Mark do you have any strong feelings on this I am leaning towards 2?
Thanks, Charles
On Tue, Feb 18, 2020 at 10:09:59AM +0000, Charles Keepax wrote:
- We could update the machine drivers to do an imply as well,
like the changes to SND_SOC_ALL_CODECS. I think this will generally work but doesn't really feel right as you can end up with the machine driver built without the CODEC.
Yes, that's not helpful.
- We could update the machine drivers to a depends on and make
the CODEC symbols manually selectable. I guess this also has the advantages that it would be necessary for the simple card stuff that seems to be everyones first choice for machine drivers these days.
That's not good, it means people have to know exactly what CODECs and DAIs are on their board to even see the config option.
Mark do you have any strong feelings on this I am leaning towards 2?
Why not just continue adding the relevant dependencies to the machine drivers like we've always done?
On Tue, Feb 18, 2020 at 11:31:16AM +0000, Mark Brown wrote:
On Tue, Feb 18, 2020 at 10:09:59AM +0000, Charles Keepax wrote:
- We could update the machine drivers to a depends on and make
the CODEC symbols manually selectable. I guess this also has the advantages that it would be necessary for the simple card stuff that seems to be everyones first choice for machine drivers these days.
That's not good, it means people have to know exactly what CODECs and DAIs are on their board to even see the config option.
Mark do you have any strong feelings on this I am leaning towards 2?
Why not just continue adding the relevant dependencies to the machine drivers like we've always done?
Hmm... a good point. The machine drivers are already depending on MFD_ARIZONA, if we swapped that for the actual CODECs involved I guess that should fix the situation as well.
I will fire out a patch shortly.
Thanks, Charles
participants (3)
-
Charles Keepax -
Mark Brown -
Randy Dunlap