Re: [PATCH v2 1/3] locking/rwsem/x86: Add stack frame dependency for __up_read()

[Ingo Molnar]

* Miguel Bernal Marin <miguel.bernal.marin@xxxxxxxxxxxxxxx> wrote:

> kernel/locking/rwsem.o: warning: objtool: up_read()+0x11: call without frame pointer save/setup
> 
> The warning means gcc 7.2.0 placed the __up_read() inline asm (and its
> call instruction) before the frame pointer setup in up_read(),
> which breaks frame pointer convention and can result in incorrect
> stack traces.
> 
> Force a stack frame to be created before the call instruction by listing
> the stack pointer as an output operand in the inline asm statement.
> 
> Signed-off-by: Miguel Bernal Marin <miguel.bernal.marin@xxxxxxxxxxxxxxx>
> ---
>  arch/x86/include/asm/rwsem.h | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/include/asm/rwsem.h b/arch/x86/include/asm/rwsem.h
> index a34e0d4b957d..762167afaec0 100644
> --- a/arch/x86/include/asm/rwsem.h
> +++ b/arch/x86/include/asm/rwsem.h
> @@ -166,14 +166,16 @@ static inline bool __down_write_trylock(struct rw_semaphore *sem)
>  static inline void __up_read(struct rw_semaphore *sem)
>  {
>  	long tmp;
> +	register void *__sp asm(_ASM_SP);
> +
>  	asm volatile("# beginning __up_read\n\t"
> -		     LOCK_PREFIX "  xadd      %1,(%2)\n\t"
> +		     LOCK_PREFIX "  xadd      %1,(%3)\n\t"
>  		     /* subtracts 1, returns the old value */
>  		     "  jns        1f\n\t"
>  		     "  call call_rwsem_wake\n" /* expects old value in %edx */
>  		     "1:\n"
>  		     "# ending __up_read\n"
> -		     : "+m" (sem->count), "=d" (tmp)
> +		     : "+m" (sem->count), "=d" (tmp), "+r" (__sp)
>  		     : "a" (sem), "1" (-RWSEM_ACTIVE_READ_BIAS)
>  		     : "memory", "cc");

Please also convert it to named operands (in separate patches!) - this apparently 
BASIC inspired labeling syntax of GCC is utterly confusing, counterproductive and 
somewhat embarrasing as well, considering that we write 2017.

Thanks,

	Ingo