Re: {standard input}:11828: Error: pcrel offset for branch to .LS005A too far (0x3e)

[Juri Lelli <juri.lelli@redhat.com>]

Hello,

On 21/01/26 21:14, kernel test robot wrote:
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> head:   6c790212c588fddeb0d852f2790840753bb604b1
> commit: bb4700adc3abec34c0a38b64f66258e4e233fc16 sched/deadline: Always stop dl-server before changing parameters
> date:   5 months ago
> config: csky-randconfig-r122-20260121 (https://download.01.org/0day-ci/archive/20260121/202601212126.OCGOmVW7-lkp@intel.com/config)
> compiler: csky-linux-gcc (GCC) 15.2.0
> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260121/202601212126.OCGOmVW7-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@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202601212126.OCGOmVW7-lkp@intel.com/
> 
> All errors (new ones prefixed by >>):
> 
>    {standard input}: Assembler messages:
> >> {standard input}:11828: Error: pcrel offset for branch to .LS005A too far (0x3e)

Thanks for the report. I was able to reproduce the issue following the
instructions provided:

  {standard input}:11828: Error: pcrel offset for branch to .LS005A too far (0x3e)
  make[5]: *** [../scripts/Makefile.build:287: kernel/sched/build_utility.o] Error 1

The failure occurs when compiling kernel/sched/build_utility.c.

I tested with several GCC versions using the same config and got
different results:

  GCC 14.2.0: builds successfully (build_utility.o created, 133KB)
  GCC 14.3.0: builds successfully (build_utility.o created, 133KB)
  GCC 15.1.0: fails with pcrel offset error at line 11797
  GCC 15.2.0: fails with pcrel offset error at line 11828 (as reported)

The error appears consistently with GCC 15.x but not with GCC 14.x when
building the same commit (bb4700adc3abec) with the provided randconfig
for csky architecture.

Interestingly, comparing the generated assembly output, GCC 15.2
produces approximately ~5% more code than GCC 14.3 for build_utility.c
(912KB vs 868KB assembly, 25479 vs 24356 lines).

I'm not sure whether this indicates a GCC 15 code generation issue for
csky, something specific about this randconfig that triggers a boundary
condition, or if there's an actual code issue I'm not seeing. The code
changes in the commit are minimal (removing a condition check before
calling dl_server_stop) and seem straightforward.

Is this maybe a known compiler issue?

Thanks,
Juri