Re: [PATCH] s390/irqflags: add out-of-line definitions of arch_local_irq_*() for KMSAN

[Ilya Leoshkevich <iii@linux.ibm.com>]

On 8/6/26 12:17, Heiko Carstens wrote:
> On Thu, Aug 06, 2026 at 12:00:19PM +0200, Ilya Leoshkevich wrote:
>> Inline KMSAN arch_local_irq_*() definitions run afoul of
>> -Wstatic-in-inline. Move them out-of-line. Decompressor is not
>> instrumented, so make sure it keeps using inline definitions.
>> Make sure non-GPL modules see the out-of-line definitions.
>>
>> Cc: Boqun Feng <boqun@kernel.org>
>> Reported-by: kernel test robot <lkp@intel.com>
>> Closes: https://lore.kernel.org/oe-kbuild-all/202607131219.euJHPSJ5-lkp@intel.com/
>> Suggested-by: Heiko Carstens <hca@linux.ibm.com>
>> Fixes: 1b301f5f28ba ("s390/irqflags: do not instrument arch_local_irq_*() with KMSAN")
>> Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
>> ---
>>   arch/s390/include/asm/irqflags.h | 20 +++++++++++++-------
>>   arch/s390/kernel/Makefile        |  1 +
>>   arch/s390/kernel/irqflags.c      | 27 +++++++++++++++++++++++++++
>>   3 files changed, 41 insertions(+), 7 deletions(-)
>>   create mode 100644 arch/s390/kernel/irqflags.c
> 
> Don't we have the very same problem in arch/s390/include/asm/uaccess.h
> with uaccess_kmsan_or_inline too?

Apparently yes. I will send a separate patch.


And I'll fix the rest of your findings in v2.