[lkml]   [2009]   [Dec]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [patch 04/23] locking: Convert raw_spinlock to arch_spinlock

On Sun, 6 Dec 2009, Thomas Gleixner wrote:
> Linus suggested to convert the raw_ to arch_ locks and cleanup the
> name space instead of using an artifical name like core_spin,
> atomic_spin or whatever
> No functional change.

Ok, I'm obviously biased, since I suggested this as a possible solution to
the naming wars, but I have to say that I like this patch regardless of
any of the other patches in the series. IOW, even without any issue of
then re-using 'raw_spinlock' for the non-preemptable one, I like how this
kind of change JustMakesSense(tm):

> --- linux-2.6-tip.orig/arch/alpha/include/asm/spinlock_types.h
> +++ linux-2.6-tip/arch/alpha/include/asm/spinlock_types.h
> @@ -7,7 +7,7 @@
> typedef struct {
> volatile unsigned int lock;
> -} raw_spinlock_t;
> +} arch_spinlock_t;
> #define __RAW_SPIN_LOCK_UNLOCKED { 0 }

ie I just think that even just looking at the patch, this kind of change
simply makes sense. The architectures declare their own per-architecture
"arch_spinlock", and we can then do whatever we want at a higher level
around that notion.

That said, it _is_

51 files changed, 164 insertions(+), 164 deletions(-)

but it seems to really be mainly the obvious arch header files and the
tracing infrastructure, so while it's 51 files, the impact seems to be
reasonably well-contained.

So I like it, but maybe the arch people hate it?


 \ /
  Last update: 2009-12-06 19:13    [W:0.118 / U:23.888 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site