Re: [PATCH] local_t : powerpc extension - use long for powerpc32

[Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>]

* Kumar Gala (galak@kernel.crashing.org) wrote:
> 
> On Feb 15, 2007, at 2:02 PM, Mathieu Desnoyers wrote:
> 
> >* Kumar Gala (galak@kernel.crashing.org) wrote:
> >>is there a reason this isn't local_add_return(long a, local_t *l) on
> >>ppc32?
> >>
> >>(same comment for other functions)
> >>
> >>- k
> >
> >no, except that we use the code is taken from atomic.h and used an
> >int parameter. However, due to the semantics of local_t, we should
> >manipulate longs instead of ints, even if they are the same size on a
> >given architecture (ppc32 and mips32 are affectec by this).
> >
> >Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
> 
> Right, this means we can simply for ppc version down to:
> 
> +static __inline__ int local_add_return(int a, local_t *l)
> +{
> +	int t;
> +
> +	__asm__ __volatile__(
> +"1:	PPC_LLARX	%0,0,%2		# local_add_return\n\
> +	add	%0,%1,%0\n"
> +	PPC405_ERR77(0,%2)
> +"	PPC_STLCX	%0,0,%2 \n\
what about the dot after stwcx ? (stwcx.) ? Is it no longer required ?

> +	bne-	1b"
> +	: "=&r" (t)
> +	: "r" (a), "r" (&(l->a.counter))
> +	: "cc", "memory");
> +
> +	return t;
> +}
> 
> and that should work for both ppc32 & ppc64 and removes the  
> duplication between the two.
> 
> - k
> 

-- 
Mathieu Desnoyers
Computer Engineering Ph.D. Candidate, École Polytechnique de Montréal
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68