Skip to content

Commit

Permalink
ARM: 7983/1: atomics: implement a better __atomic_add_unless for v6+
Browse files Browse the repository at this point in the history
Looking at perf profiles of multi-threaded hackbench runs, a significant
performance hit appears to manifest from the cmpxchg loop used to
implement the 32-bit atomic_add_unless function. This can be mitigated
by writing a direct implementation of __atomic_add_unless which doesn't
require iteration outside of the atomic operation.

Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
  • Loading branch information
Will Deacon authored and Russell King committed Feb 25, 2014
1 parent d98b90e commit db38ee8
Showing 1 changed file with 31 additions and 4 deletions.
35 changes: 31 additions & 4 deletions arch/arm/include/asm/atomic.h
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,33 @@ static inline int atomic_cmpxchg(atomic_t *ptr, int old, int new)
return oldval;
}

static inline int __atomic_add_unless(atomic_t *v, int a, int u)
{
int oldval, newval;
unsigned long tmp;

smp_mb();
prefetchw(&v->counter);

__asm__ __volatile__ ("@ atomic_add_unless\n"
"1: ldrex %0, [%4]\n"
" teq %0, %5\n"
" beq 2f\n"
" add %1, %0, %6\n"
" strex %2, %1, [%4]\n"
" teq %2, #0\n"
" bne 1b\n"
"2:"
: "=&r" (oldval), "=&r" (newval), "=&r" (tmp), "+Qo" (v->counter)
: "r" (&v->counter), "r" (u), "r" (a)
: "cc");

if (oldval != u)
smp_mb();

return oldval;
}

#else /* ARM_ARCH_6 */

#ifdef CONFIG_SMP
Expand Down Expand Up @@ -189,10 +216,6 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
return ret;
}

#endif /* __LINUX_ARM_ARCH__ */

#define atomic_xchg(v, new) (xchg(&((v)->counter), new))

static inline int __atomic_add_unless(atomic_t *v, int a, int u)
{
int c, old;
Expand All @@ -203,6 +226,10 @@ static inline int __atomic_add_unless(atomic_t *v, int a, int u)
return c;
}

#endif /* __LINUX_ARM_ARCH__ */

#define atomic_xchg(v, new) (xchg(&((v)->counter), new))

#define atomic_inc(v) atomic_add(1, v)
#define atomic_dec(v) atomic_sub(1, v)

Expand Down

0 comments on commit db38ee8

Please sign in to comment.