Re: [kbuild-all] Re: [ebiederm-user-namespace:signal-for-v5.17 4/12] arch/x86/kernel/dumpstack.o: warning: objtool: oops_end() falls through to next function show_opcodes()

[Rong Chen <rong.a.chen@intel.com>]

On 12/15/21 12:48, Eric W. Biederman wrote:
> Rong Chen <rong.a.chen@intel.com> writes:
>
>> On 12/15/21 01:29, Eric W. Biederman wrote:
>>> kernel test robot <lkp@intel.com> writes:
>>>
>>>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git signal-for-v5.17
>>>> head:   6b1248798eb6f6d5285db214299996ecc5dc1e6b
>>>> commit: 0e25498f8cd43c1b5aa327f373dd094e9a006da7 [4/12] exit: Add and use make_task_dead.
>>>> config: x86_64-randconfig-a011-20211213 (https://download.01.org/0day-ci/archive/20211214/202112140949.Uq5sFKR1-lkp@intel.com/config)
>>>> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
>>>> reproduce (this is a W=1 build):
>>>>           # https://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git/commit/?id=0e25498f8cd43c1b5aa327f373dd094e9a006da7
>>>>           git remote add ebiederm-user-namespace https://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git
>>>>           git fetch --no-tags ebiederm-user-namespace signal-for-v5.17
>>>>           git checkout 0e25498f8cd43c1b5aa327f373dd094e9a006da7
>>>>           # save the config file to linux build tree
>>>>           mkdir build_dir
>>>>           make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash arch/x86/kernel/ kernel/ lib/kunit/
>>>>
>>>> If you fix the issue, kindly add following tag as appropriate
>>>> Reported-by: kernel test robot <lkp@intel.com>
>>>>
>>>> All warnings (new ones prefixed by >>):
>>>>
>>>>>> lib/kunit/kunit-example-test.o: warning: objtool: .text.unlikely: unexpected end of section
>>>> --
>>>>>> arch/x86/kernel/dumpstack.o: warning: objtool: oops_end() falls through to next function show_opcodes()
>>> I am confused.  This change patches objtool to rename
>>> rewind_stack_and_do_exit to rewind_stack_and_make_dead in the list
>>> global_noreturns in tools/objtool/check.c
>>>
>>> There is no other change to oops_end other than that renaming.
>>>
>>> Did the robot somehow mange to run an old version of objtool while
>>> building and this get this error?
>>>
>>> I tried and I am not currently able to reproduce this error.  Do you
>>> have any additional pointers on how I might reproduce this?
>> Hi Eric,
>>
>> The reproduce step in report based on a full build, I can reproduce
>> the warnings with the below command:
>>
>>    make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash
> Any chance you could do "make arch/x86/kernel/dumpstack.s" and send it
> to me?

$ make O=build_dir ARCH=x86_64 SHELL=/bin/bash arch/x86/kernel/dumpstack.s
make[1]: Entering directory '/mnt/memdrive/linux/build_dir'
   GEN     Makefile
   CALL    ../scripts/checksyscalls.sh
   CALL    ../scripts/atomic/check-atomics.sh
   DESCEND objtool
   CC      arch/x86/kernel/dumpstack.s
make[1]: Leaving directory '/mnt/memdrive/linux/build_dir'

$ make O=build_dir ARCH=x86_64 SHELL=/bin/bash arch/x86/kernel/
make[1]: Entering directory '/mnt/memdrive/linux/build_dir'
   GEN     Makefile
   CALL    ../scripts/checksyscalls.sh
   CALL    ../scripts/atomic/check-atomics.sh
   DESCEND objtool
   CC      arch/x86/kernel/dumpstack.o
arch/x86/kernel/dumpstack.o: warning: objtool: oops_end() falls through 
to next function show_opcodes()
   AR      arch/x86/kernel/built-in.a
make[1]: Leaving directory '/mnt/memdrive/linux/build_dir'

Best Regards,
Rong Chen

>
> I want to see what code is in your oops_end.
>
> It really does not make sense that simply renaming a function would
> trigger this error.
>
> I did a full build (on debian 10 with gcc-8.3.0) and I did not manage to
> reproduce this.  In a bit I will try with a newer debian and see if I
> can get the issue to reproduce.  Still it does not make sense to me
> that a simple rename would trigger this error.
>
> Eric
>