linux-kernel - Re: linux-next: build warning after merge of the sound-asoc tree

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <b436e795-188a-4b62-95f5-1b11d2d10c77@infradead.org>
Date: Fri, 24 Oct 2025 12:40:07 -0700
From: Randy Dunlap <rdunlap@...radead.org>
To: Troy Mitchell <troy.mitchell@...ux.spacemit.com>,
 Stephen Rothwell <sfr@...b.auug.org.au>, Mark Brown <broonie@...nel.org>,
 Liam Girdwood <lgirdwood@...il.com>
Cc: Jinmei Wei <weijinmei@...ux.spacemit.com>,
 Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
 Linux Next Mailing List <linux-next@...r.kernel.org>
Subject: Re: linux-next: build warning after merge of the sound-asoc tree



On 10/24/25 6:27 AM, Troy Mitchell wrote:
> On Wed, Oct 22, 2025 at 02:17:05PM +1100, Stephen Rothwell wrote:
>> Hi all,
>>
>> After merging the sound-asoc tree, today's linux-next build (powercp
>> allyesconfig) produced this warning:
>>
>> WARNING: unmet direct dependencies detected for DMA_CMA
>>   Depends on [n]: HAVE_DMA_CONTIGUOUS [=n] && CMA [=y]
>>   Selected by [y]:
>>   - SND_SOC_K1_I2S [=y] && SOUND [=y] && SND [=y] && SND_SOC [=y] && (COMPILE_TEST [=y] || ARCH_SPACEMIT) && HAVE_CLK [=y]
>>
>> Probably introduced by commit
>>
>>   fce217449075 ("ASoC: spacemit: add i2s support for K1 SoC")
> Sorry, I didn't realize that this email ended up in my spam folder,
> and I just saw it today.
> 
> I am considering directly selecting `HAVE_DMA_CONTIGUOUS`,
> but another possibility is to make it depend on `DMA_CMA`.
> Can anyone advise which approach would be better?
> If there are no objections, I plan to send a patch on Monday:
> select `HAVE_DMA_CONTIGUOUS`

Hi,

My read of how HAVE_DMA_CONTIGUOUS is used is that an arch or
platform selects it to show that it has this capability.
Drivers don't do that.

Possibly ARCH_SPACEMIT should select HAVE_CONTIGUOUS_DMA but that
wouldn't help here since this is COMPILE_TEST=y, which did
its job here.

I think this driver should
	depend on DMA_CMA
That will take care of being dependent on both CMA and
HAVE_CONTIGUOUS_DMA.

Or IF the driver can operate without DMA_CMA being enabled,
the driver Kconfig could do
	select DMA_CMA if HAVE_DMA_CONTIGUOUS && CMA
and then have alternate code paths for whatever is
enabled.

This driver could
	select CMA if MMU

or it could do
	depends on MMU
	depends on HAVE_CONTIGUOUS_DMA
	select CMA

but
	depends on DMA_CMA
wraps all of driver dependencies into one statement.
But no, not
	select HAVE_CONTIGUOUS_DMA

-- 
~Randy