linux-kernel - Re: drivers/media/i2c/ccs/ccs.o: error: objtool: ccs_set_selection(): unexpected end of section .text.ccs_set

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: <20251122062946.GA3054484@ax162>
Date: Fri, 21 Nov 2025 23:29:46 -0700
From: Nathan Chancellor <nathan@...nel.org>
To: Josh Poimboeuf <jpoimboe@...nel.org>
Cc: kernel test robot <lkp@...el.com>, oe-kbuild-all@...ts.linux.dev,
	linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...nel.org>,
	Sakari Ailus <sakari.ailus@...ux.intel.com>,
	linux-media@...r.kernel.org
Subject: Re: drivers/media/i2c/ccs/ccs.o: error: objtool:
 ccs_set_selection(): unexpected end of section .text.ccs_set_selection

On Fri, Nov 21, 2025 at 09:51:33PM -0800, Josh Poimboeuf wrote:
> On Fri, Nov 21, 2025 at 06:34:14PM -0700, Nathan Chancellor wrote:
> > On Sat, Nov 22, 2025 at 08:41:37AM +0800, kernel test robot wrote:
> > > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> > > head:   2eba5e05d9bcf4cdea995ed51b0f07ba0275794a
> > > commit: 188d90f817e13b66e03e110eb6f82e8f5f0d654b objtool: Append "()" to function name in "unexpected end of section" warning
> > > date:   8 months ago
> > > :::::: branch date: 4 hours ago
> > > :::::: commit date: 8 months ago
> > > config: x86_64-randconfig-101-20251122 (https://download.01.org/0day-ci/archive/20251122/202511220717.5HHMLUHG-lkp@intel.com/config)
> > > compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
> > > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251122/202511220717.5HHMLUHG-lkp@intel.com/reproduce)
> > > 
> > > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > > the same patch/commit), kindly add following tags
> > > | Reported-by: kernel test robot <lkp@...el.com>
> > > | Closes: https://lore.kernel.org/r/202511220717.5HHMLUHG-lkp@intel.com/
> > > 
> > > All errors (new ones prefixed by >>):
> > > 
> > > >> drivers/media/i2c/ccs/ccs.o: error: objtool: ccs_set_selection(): unexpected end of section .text.ccs_set_selection
> > 
> > That change obviously does not result in this warning/error. This
> > appears to be another divide by zero issue but based on my analysis so
> > far, I do not understand how...
> > 
> > https://github.com/ClangBuiltLinux/linux/issues/2129
> 
> Here ya go :-)  After looking at a gazillion of these I can often spot
> these pretty easily.  I'm not sure what the correct fix is here but this
> made the error go away.
> 
> diff --git a/drivers/media/i2c/ccs/ccs-core.c b/drivers/media/i2c/ccs/ccs-core.c
> index 1c889c878abd..2429c05bffb3 100644
> --- a/drivers/media/i2c/ccs/ccs-core.c
> +++ b/drivers/media/i2c/ccs/ccs-core.c
> @@ -2346,7 +2346,7 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
>  		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
>  	max_m = crops[CCS_PAD_SINK]->width
>  		* CCS_LIM(sensor, SCALER_N_MIN)
> -		/ CCS_LIM(sensor, MIN_X_OUTPUT_SIZE);
> +		/ (CCS_LIM(sensor, MIN_X_OUTPUT_SIZE) ? : 1);
>  
>  	a = clamp(a, CCS_LIM(sensor, SCALER_M_MIN),
>  		  CCS_LIM(sensor, SCALER_M_MAX));

Aha! Thanks a lot :) I had thought it might be something with CCS_LIM()
since ccs_get_limit() returns zero if ccs_limit_ptr() errors and in the
default case of the switch statement. There are a lot of unchecked
divides with the result of CCS_LIM() throughout this driver so I figured
if that was it, there would be other instances of this warning... oh
well.

Something like the following diff also fixes it since LLVM no longer
sees 0 as a possible divisor, which seems a little better to me since it
seems like one of the other uses could turn problematic with other
optimizations.  Given these cases are both errors and have visible WARNs
in case they are hit, it seems like it is better to use a valid divisor
instead of 0.

Just one more warning to tackle in my personal configuration then I can
enable CONFIG_OBJTOOL_WERROR to make new warnings from LLVM uprevs more
obvious :)

  https://github.com/ClangBuiltLinux/linux/issues/2130

diff --git a/drivers/media/i2c/ccs/ccs-core.c b/drivers/media/i2c/ccs/ccs-core.c
index 1c889c878abd..05c3da29f14c 100644
--- a/drivers/media/i2c/ccs/ccs-core.c
+++ b/drivers/media/i2c/ccs/ccs-core.c
@@ -137,7 +137,7 @@ u32 ccs_get_limit(struct ccs_sensor *sensor, unsigned int limit,

        ret = ccs_limit_ptr(sensor, limit, offset, &ptr);
        if (ret)
-               return 0;
+               return 1;

        switch (CCI_REG_WIDTH_BYTES(ccs_limits[ccs_limit_offsets[limit].info].reg)) {
        case sizeof(u8):
@@ -151,7 +151,7 @@ u32 ccs_get_limit(struct ccs_sensor *sensor, unsigned int limit,
                break;
        default:
                WARN_ON(1);
-               return 0;
+               return 1;
        }

        return ccs_reg_conv(sensor, ccs_limits[limit].reg, val);