linux-kernel - Re: [RFC][PATCH 0/2] localmodconfig / ALSA: hda - Have codecs be disabled by localmodconfig

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] [thread-next>] [day] [month] [year] [list]
Date:	Wed, 18 Dec 2013 20:05:11 +0100
From:	Takashi Iwai <tiwai@...e.de>
To:	Steven Rostedt <rostedt@...dmis.org>
Cc:	linux-kernel@...r.kernel.org,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Michal Marek <mmarek@...e.cz>,
	Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [RFC][PATCH 0/2] localmodconfig / ALSA: hda - Have codecs be disabled by localmodconfig

At Wed, 18 Dec 2013 12:35:19 -0500,
Steven Rostedt wrote:
> 
> Linus Torvalds reported that 'make localmodconfig' would never work on
> his ALSA codec modules. That is, although his box only had realtek and
> hdmi codec modules, all of the codec modules would still be built after
> running localmodconfig.
> 
> When he showed me his config file which had this:
> 
>   CONFIG_SND_HDA_CODEC_REALTEK=y
>   CONFIG_SND_HDA_CODEC_ANALOG=y
>   CONFIG_SND_HDA_CODEC_SIGMATEL=y
>   CONFIG_SND_HDA_CODEC_VIA=y
>   CONFIG_SND_HDA_CODEC_HDMI=y
>   CONFIG_SND_HDA_CODEC_CIRRUS=y
>   CONFIG_SND_HDA_CODEC_CONEXANT=y
>   CONFIG_SND_HDA_CODEC_CA0110=y
>   CONFIG_SND_HDA_CODEC_CA0132=y
>   CONFIG_SND_HDA_CODEC_CA0132_DSP=y
>   CONFIG_SND_HDA_CODEC_CMEDIA=y
>   CONFIG_SND_HDA_CODEC_SI3054=y
>   ..
> 
> I was confused to why they were '=y' if they were to build modules
> and not '=m'. Localmodconfig will not disable anything with '=y' because
> there's not enough information in lsmod to safely disable a built-in
> config.
> 
> Investigating as to why these were '=y' I found this in the Makefile:
> 
> ifdef CONFIG_SND_HDA_CODEC_REALTEK
> obj-$(CONFIG_SND_HDA_INTEL) += snd-hda-codec-realtek.o
> endif
> ifdef CONFIG_SND_HDA_CODEC_CMEDIA
> obj-$(CONFIG_SND_HDA_INTEL) += snd-hda-codec-cmedia.o
> endif
> ifdef CONFIG_SND_HDA_CODEC_ANALOG
> obj-$(CONFIG_SND_HDA_INTEL) += snd-hda-codec-analog.o
> endif
> ifdef CONFIG_SND_HDA_CODEC_SIGMATEL
> obj-$(CONFIG_SND_HDA_INTEL) += snd-hda-codec-idt.o
> endif
> ifdef CONFIG_SND_HDA_CODEC_SI3054
> obj-$(CONFIG_SND_HDA_INTEL) += snd-hda-codec-si3054.o
> endif
> [...]
> 
> And the reason for this is because if the code for SND_HDA_INTEL
> is a built-in, then all the codecs that are configured must also be
> built-in. If it is a module, then all the codecs must also be a module.
> 
> To put this another way; if SND_HDA_INTEL is 'y' then all the codecs
> must either be 'y' or 'n'. If SND_HDA_INTEL is 'm', then all the
> codecs must be 'm' or 'n'. The latter is already supported by the
> kconfig code, but the former is not. That is, if a tristate config
> is dependent on a config that is built-in 'y', then that config can
> still become a module 'm'. This trick that is done above prevents that.
> But unfortunately, it also confuses localmodconfig, and the user
> does not get the correct result.
> 
> Ideally, having a "match_config" option in the kconfig subsystem would
> be ideal. But anyone that ever took a look at the kconfig code knows that
> is much more difficult to implement than one would expect, as that
> code can cause severe eye strain.
> 
> But luckily for us, we can do another trick to keep the behavior needed
> by the ALSA hda code, and still allow for localmodconfig to disable the
> modules.
> 
> Now to fix this so that the codecs match the tristate of SND_HDA_INTEL
> we can add a config that states that SND_HDA_INTEL is 'y', and then add
> another config for each codec for it to build the module. We then switch
> all the codecs to be tristates too.
> 
> The new config for HDA_INTEL set as 'y' is:
> 
> config SND_HDA_YES
> 	default y if SND_HDA_INTEL=y
> 
> The new codec configs will have the following format:
> 
> config SND_HDA_CODEC_FOO_BUILD
> 	def_tristate (SND_HDA_CODEC_FOO=m && SND_HDA_YES) || SND_HDA_CODEC_FOO
> 
> What the above does is to set SND_HDA_CODEC_FOO_BUILD to 'y' if
> SND_HDA_CODEC_FOO is a module and SND_HDA_YES is set. This forces
> the codec to be built-in. Otherwise, it just sets it to the state of
> SND_HDA_CODEC_FOO.
> 
> This change lets localmodconfig turn off codecs that are not being used.
> 
> The first patch fixes a bug in localmodconfig that this ordeal uncovered,
> and that was the fact that localmodconfig did not honor default dependencies
> like "def_tristate (SND_HDA_CODEC_FOO=m && SND_HDA_YES) || SND_HDA_CODEC_FOO"

Well, as mentioned in the mail thread, I'm afraid that people don't
like to see yet doubled kconfig items just for workarounds by this
approach.  So it'd be likely better not to restrict but just let users
choose tristate.  I already submitted the patches (the same ones as
attached before) to alsa-devel ML.

Another thing we may add is a warning like:

================================================================
config SND_HDA_CODEC_REALTEK
	tristate "Build Realtek HD-audio codec support"
	select SND_HDA_GENERIC
	help
	  Say Y or M here to include Realtek HD-audio codec support in
	  snd-hda-intel driver, such as ALC880.

comment "Set to Y if you want auto-loading the codec driver"
	depends on SND_HDA_CODEC_REALTEK && SND_HDA_INTEL != SND_HDA_CODEC_REALTEK
================================================================

The comment won't break make localmodconfig, right?
This should be intuitive enough for avoiding pitfalls, I hope.


thanks,

Takashi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/