[PATCH 6/9] asm-generic/bitops/atomic.h: Rewrite using atomic_fetch_*

On Thu, May 24, 2018 at 01:47:39PM +0100, Will Deacon wrote:

On Thu, May 24, 2018 at 02:44:10PM +0200, Peter Zijlstra wrote:

quoted

On Thu, May 24, 2018 at 11:59:43AM +0100, Will Deacon wrote:

quoted

+static inline void set_bit(unsigned int nr, volatile unsigned long *p)
 {
+	p += BIT_WORD(nr);
+	atomic_long_fetch_or_relaxed(BIT_MASK(nr), (atomic_long_t *)p);
 }
 
+static inline void clear_bit(unsigned int nr, volatile unsigned long *p)
 {
+	p += BIT_WORD(nr);
+	atomic_long_fetch_andnot_relaxed(BIT_MASK(nr), (atomic_long_t *)p);
 }
 
+static inline void change_bit(unsigned int nr, volatile unsigned long *p)
 {
+	p += BIT_WORD(nr);
+	atomic_long_fetch_xor_relaxed(BIT_MASK(nr), (atomic_long_t *)p);
 }

Why use the fetch variants here?

I noticed the same thing just now; I'll drop that and just use the
non-value-returning variants. It's shame that I can't do the same for
the lock.h unlock code, but we don't have non-returning release variants.

As an aside, If I complete the autogeneration stuff, it'll be possible
to generate those. I split out the necessary barriers in [1], but I
still have a lot of other preparatory cleanup to do.

IIUC, the void-returning atomic ops are relaxed, so trying to unify that
with the usual rule that no suffix means fence will slow things down
unless we want to do a treewide substitition to fixup for that.

Thanks,
Mark.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/commit/?h=atomics/api-unification&id=c6b9ff2627d06776e427a7f1a7f83caeff3db536