linux-kernel - Re: [PATCH] ASoC: codes: src4xxx: Avoid clang -Wsometimes-uninitialized in src4xxx_hw

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:   Mon, 22 Aug 2022 15:32:47 -0700
From:   Nathan Chancellor <nathan@...nel.org>
To:     Matt Flax <flatmax@...tmax.com>
Cc:     Liam Girdwood <lgirdwood@...il.com>,
        Mark Brown <broonie@...nel.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Tom Rix <trix@...hat.com>, alsa-devel@...a-project.org,
        linux-kernel@...r.kernel.org, llvm@...ts.linux.dev,
        patches@...ts.linux.dev, kernel test robot <lkp@...el.com>,
        "Sudip Mukherjee (Codethink)" <sudipm.mukherjee@...il.com>
Subject: Re: [PATCH] ASoC: codes: src4xxx: Avoid clang
 -Wsometimes-uninitialized in src4xxx_hw_params()

Hi Matt,

On Tue, Aug 23, 2022 at 08:00:07AM +1000, Matt Flax wrote:
> Hi Nathan,
> 
> On 23/8/22 04:31, Nathan Chancellor wrote:
> > Clang warns:
> > 
> >    sound/soc/codecs/src4xxx.c:280:3: error: variable 'd' is used uninitialized whenever switch default is taken [-Werror,-Wsometimes-uninitialized]
> >                    default:
> >                    ^~~~~~~
> >    sound/soc/codecs/src4xxx.c:298:59: note: uninitialized use occurs here
> >                    ret = regmap_write(src4xxx->regmap, SRC4XXX_RCV_PLL_11, d);
> >                                                                            ^
> >    sound/soc/codecs/src4xxx.c:223:20: note: initialize the variable 'd' to silence this warning
> >            int val, pj, jd, d;
> >                              ^
> >                              = 0
> 
> 
> If you really want to get rid of these warnings, you can use this default
> for the variables :
> 
> pj = 0x0;
> jd=0xff;
> 
> d = 0xff;
> 
> It doesn't really make sense why we would choose to initialise these
> variables, but if you want to silence the preprocessor, then perhaps those
> values. Put a message that defaults are not known nor specified in the data
> sheet and these values are chosen to be unlikely matches of real world
> values - which ensures regmap updates later from an initial unknown chip
> state after refresh.

Okay, I could do something like this?

diff --git a/sound/soc/codecs/src4xxx.c b/sound/soc/codecs/src4xxx.c
index a8f143057b41..db4e280dd055 100644
--- a/sound/soc/codecs/src4xxx.c
+++ b/sound/soc/codecs/src4xxx.c
@@ -280,9 +280,14 @@ static int src4xxx_hw_params(struct snd_pcm_substream *substream,
 		default:
 			/* don't error out here,
 			 * other parts of the chip are still functional
+			 * Dummy initialize variables to avoid
+			 * -Wsometimes-uninitialized from clang.
 			 */
 			dev_info(component->dev,
-				"Couldn't set the RCV PLL as this master clock rate is unknown\n");
+				"Couldn't set the RCV PLL as this master clock rate is unknown. Chosen regmap values may not match real world values.\n");
+			pj = 0x0;
+			jd = 0xff;
+			d = 0xff;
 			break;
 		}
 		ret = regmap_write(src4xxx->regmap, SRC4XXX_RCV_PLL_0F, pj);

These warnings need to be eliminated because Linus requires a clean
build as much as possible [1] [2], which has been codified by
CONFIG_WERROR, which is enabled by default with allmodconfig. As a
result, clang allmodconfig builds are broken [3], which you might have
already seen. If there is another way this could be written without
potentially uninitialized variables, I am open to it.

[1]: https://lore.kernel.org/CAHk-=wg-mH-_GYpkhz_psjBWG6ZcjKnPo83fg7YMj_by+-LRTQ@mail.gmail.com/
[2]: https://lore.kernel.org/CAHk-=wi4wyvuXs0SUq-2x=XHxWmJ6jVKRD-WpE0kWgWiqVJNbg@mail.gmail.com/
[3]: https://lore.kernel.org/YwNMUlAmu%2FqCjuva@debian/

> All a bit messy really.

Indeed :/

> > According to the comment in the default case, other parts of the chip
> > are still functional without these values so just return 0 in the
> > default case to avoid using these variables uninitialized.
> > 
> > Link: https://github.com/ClangBuiltLinux/linux/issues/1691
> > Reported-by: kernel test robot <lkp@...el.com>
> > Reported-by: "Sudip Mukherjee (Codethink)" <sudipm.mukherjee@...il.com>
> > Signed-off-by: Nathan Chancellor <nathan@...nel.org>
> > ---
> >   sound/soc/codecs/src4xxx.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/sound/soc/codecs/src4xxx.c b/sound/soc/codecs/src4xxx.c
> > index a8f143057b41..cf45caa4bf7f 100644
> > --- a/sound/soc/codecs/src4xxx.c
> > +++ b/sound/soc/codecs/src4xxx.c
> > @@ -283,7 +283,7 @@ static int src4xxx_hw_params(struct snd_pcm_substream *substream,
> >   			 */
> >   			dev_info(component->dev,
> >   				"Couldn't set the RCV PLL as this master clock rate is unknown\n");
> > -			break;
> > +			return 0;
> 
> 
> Don't return here, the rest of the chip is still functional and probably in
> use. Print the dev_info and continue.

I'll send a v2 tomorrow morning (for me at least) if there are no other
comments.

Cheers,
Nathan