Re: riscv gcc-13 allyesconfig error the frame size of 2064 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]

On Fri, May 23, 2025, at 15:19, Naresh Kamboju wrote:
> On Thu, 22 May 2025 at 22:18, Kent Overstreet <[email protected]> wrote:
>>
>> On Thu, May 22, 2025 at 06:59:53PM +0530, Naresh Kamboju wrote:
>> > Regressions on riscv allyesconfig build failed with gcc-13 on the Linux next
>> > tag next-20250516 and next-20250522.
>> >
>> > First seen on the next-20250516
>> >  Good: next-20250515
>> >  Bad:  next-20250516
>> >
>> > Regressions found on riscv:
>> >  - build/gcc-13-allyesconfig
>> >
>> > Regression Analysis:
>> >  - New regression? Yes
>> >  - Reproducible? Yes
>> >
>> > Build regression: riscv gcc-13 allyesconfig error the frame size of
>> > 2064 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
>>
>> Is this a kmsan build? kmsan seems to inflate stack usage by quite a
>> lot.

KMSAN is currently a clang-only feature.

> This is allyesconfig build which has KASAN builds.
>
> CONFIG_HAVE_ARCH_KASAN=y
> CONFIG_HAVE_ARCH_KASAN_VMALLOC=y
> CONFIG_CC_HAS_KASAN_GENERIC=y
> CONFIG_CC_HAS_WORKING_NOSANITIZE_ADDRESS=y
> CONFIG_KASAN=y
> CONFIG_CC_HAS_KASAN_MEMINTRINSIC_PREFIX=y
> CONFIG_KASAN_GENERIC=y
> # CONFIG_KASAN_OUTLINE is not set
> CONFIG_KASAN_INLINE=y
> CONFIG_KASAN_STACK=y

I reproduced the problem locally and found this to go down to
1440 bytes after I turn off KASAN_STACK. next-20250523 has
some changes that take the number down further to 1136 with
KASAN_STACK and or 1552 with KASAN_STACK.

I've turned bcachefs with kasan-stack on for my randconfig
builds again to see if there are any remaining corner cases.

     Arnd