Re: [PATCH 3/3] WorkStruct: Use direct assignment rather than cmpxchg()

[Russell King]

On Thu, Dec 07, 2006 at 11:58:23PM +0000, David Howells wrote:
> Russell King <rmk+lkml@xxxxxxxxxxxxxxxx> wrote:
> 
> > Incorrect.  pre-v6 ARM bitops for test_and_xxx_bit() all do:
> > 
> > 	save and disable irqs
> > 	load value
> > 	test bit
> > 	if not in desired state, alter bit and write it back
> > 	restore irqs
> 
> Hmmm...  ARM has two implementations.  One in the header files which is what I
> consulted when writing that email:
> 
> static inline void ____atomic_set_bit(unsigned int bit, volatile unsigned long *p)
> {
> 	unsigned long flags;
> 	unsigned long mask = 1UL << (bit & 31);
> 
> 	p += bit >> 5;
> 
> 	raw_local_irq_save(flags);
> 	*p |= mask;
> 	raw_local_irq_restore(flags);
> }
> 
> And the other in the libs which does as you say.  Why the one in the header
> file at all?

These are the constant versions, where the compiler can optimise the
mask and word offset itself.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/