public inbox for linux-arch@vger.kernel.org
 help / color / mirror / Atom feed
* atomic ops for a VM fix
@ 2004-05-13 20:35 Andrew Morton
  2004-05-13 20:37 ` Andrew Morton
  2004-05-13 21:52 ` Ivan Kokshaysky
  0 siblings, 2 replies; 6+ messages in thread
From: Andrew Morton @ 2004-05-13 20:35 UTC (permalink / raw)
  To: linux-arch

Guys, I have four patches here which I'd like to merge into Linus's tree
within a week.

They require that the architecture implement atomic_add_negative() and
atomic_inc_and_test().  I managed to cobble these together for many
architectures, but there are a few gaps.  Please review the third and
fourth patches.

Dave, sparc64 is missing atomic_add_negative(), sorry.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: atomic ops for a VM fix
  2004-05-13 20:35 atomic ops for a VM fix Andrew Morton
@ 2004-05-13 20:37 ` Andrew Morton
  2004-05-13 21:52 ` Ivan Kokshaysky
  1 sibling, 0 replies; 6+ messages in thread
From: Andrew Morton @ 2004-05-13 20:37 UTC (permalink / raw)
  To: linux-arch

oh, and open-coded references to page->count are deliberately
broken at compile time.  Please convert any leftovers to get_page(),
set_page_count(), page_count() and __put_page(), thanks.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: atomic ops for a VM fix
  2004-05-13 20:35 atomic ops for a VM fix Andrew Morton
  2004-05-13 20:37 ` Andrew Morton
@ 2004-05-13 21:52 ` Ivan Kokshaysky
  2004-05-13 22:33   ` David Mosberger
  1 sibling, 1 reply; 6+ messages in thread
From: Ivan Kokshaysky @ 2004-05-13 21:52 UTC (permalink / raw)
  To: Andrew Morton; +Cc: linux-arch

On Thu, May 13, 2004 at 01:35:26PM -0700, Andrew Morton wrote:
> They require that the architecture implement atomic_add_negative() and
> atomic_inc_and_test().  I managed to cobble these together for many
> architectures, but there are a few gaps.  Please review the third and
> fourth patches.

It seems atomic_inc_and_test() is missing on alpha.

Ivan.

--- 2.6/include/asm-alpha/atomic.h	Mon May 10 06:32:52 2004
+++ linux/include/asm-alpha/atomic.h	Fri May 14 01:41:57 2004
@@ -178,6 +178,7 @@ static __inline__ long atomic64_sub_retu
 #define atomic_sub_and_test(i,v) (atomic_sub_return((i), (v)) == 0)
 #define atomic64_sub_and_test(i,v) (atomic64_sub_return((i), (v)) == 0)
 
+#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)
 #define atomic_dec_and_test(v) (atomic_sub_return(1, (v)) == 0)
 #define atomic64_dec_and_test(v) (atomic64_sub_return(1, (v)) == 0)
 

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: atomic ops for a VM fix
  2004-05-13 21:52 ` Ivan Kokshaysky
@ 2004-05-13 22:33   ` David Mosberger
  2004-05-13 22:52     ` Andrew Morton
  0 siblings, 1 reply; 6+ messages in thread
From: David Mosberger @ 2004-05-13 22:33 UTC (permalink / raw)
  To: Ivan Kokshaysky; +Cc: Andrew Morton, linux-arch

>>>>> On Fri, 14 May 2004 01:52:22 +0400, Ivan Kokshaysky <ink@jurassic.park.msu.ru> said:

  Ivan> It seems atomic_inc_and_test() is missing on alpha.

  Ivan> +#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)

And I seem to have gotten the ia64-version wrong (it tests for != 0).
I'll fix that now.

Can someone say with certainty whether the s390 version is also
wrong?

	--david

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: atomic ops for a VM fix
  2004-05-13 22:33   ` David Mosberger
@ 2004-05-13 22:52     ` Andrew Morton
  0 siblings, 0 replies; 6+ messages in thread
From: Andrew Morton @ 2004-05-13 22:52 UTC (permalink / raw)
  To: davidm; +Cc: davidm, ink, linux-arch

David Mosberger <davidm@napali.hpl.hp.com> wrote:
>
> >>>>> On Fri, 14 May 2004 01:52:22 +0400, Ivan Kokshaysky <ink@jurassic.park.msu.ru> said:
> 
>   Ivan> It seems atomic_inc_and_test() is missing on alpha.
> 
>   Ivan> +#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)
> 
> And I seem to have gotten the ia64-version wrong (it tests for != 0).
> I'll fix that now.

Neato.


From: David Mosberger <davidm@napali.hpl.hp.com>


---

 25-akpm/include/asm-ia64/atomic.h |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

diff -puN include/asm-ia64/atomic.h~ia64-atomic_inc_and_test-fix include/asm-ia64/atomic.h
--- 25/include/asm-ia64/atomic.h~ia64-atomic_inc_and_test-fix	Thu May 13 15:49:32 2004
+++ 25-akpm/include/asm-ia64/atomic.h	Thu May 13 15:49:32 2004
@@ -159,7 +159,7 @@ atomic64_add_negative (__s64 i, atomic64
 
 #define atomic_sub_and_test(i,v)	(atomic_sub_return((i), (v)) == 0)
 #define atomic_dec_and_test(v)		(atomic_sub_return(1, (v)) == 0)
-#define atomic_inc_and_test(v)		(atomic_add_return(1, (v)) != 0)
+#define atomic_inc_and_test(v)		(atomic_add_return(1, (v)) == 0)
 #define atomic64_sub_and_test(i,v)	(atomic64_sub_return((i), (v)) == 0)
 #define atomic64_dec_and_test(v)	(atomic64_sub_return(1, (v)) == 0)
 #define atomic64_inc_and_test(v)	(atomic64_add_return(1, (v)) != 0)

_

> Can someone say with certainty whether the s390 version is also
> wrong?
> 

Either s390's atomic_inc_and_test() is wrong, or their atomic_inc_return()
is wrong.  I think the former.


David Mosberger <davidm@napali.hpl.hp.com>


---

 25-akpm/include/asm-s390/atomic.h |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

diff -puN include/asm-s390/atomic.h~s390-atomic_inc_and_test-fix include/asm-s390/atomic.h
--- 25/include/asm-s390/atomic.h~s390-atomic_inc_and_test-fix	Thu May 13 15:50:46 2004
+++ 25-akpm/include/asm-s390/atomic.h	Thu May 13 15:50:50 2004
@@ -72,7 +72,7 @@ static __inline__ int atomic_inc_return(
 
 static __inline__ int atomic_inc_and_test(volatile atomic_t * v)
 {
-	return __CS_LOOP(v, 1, "ar") != 0;
+	return __CS_LOOP(v, 1, "ar") == 0;
 }
 static __inline__ void atomic_dec(volatile atomic_t * v)
 {

_

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: atomic ops for a VM fix
@ 2004-05-14 11:36 Martin Schwidefsky
  0 siblings, 0 replies; 6+ messages in thread
From: Martin Schwidefsky @ 2004-05-14 11:36 UTC (permalink / raw)
  To: Andrew Morton; +Cc: davidm, davidm, ink, linux-arch





> Either s390's atomic_inc_and_test() is wrong, or their
atomic_inc_return()
> is wrong.  I think the former.

atomic_inc_return is fine, atomic_inc_and_test is broken. The __CS_LOOP always
returns the new value.

blue skies,
   Martin

Linux/390 Design & Development, IBM Deutschland Entwicklung GmbH
Schönaicherstr. 220, D-71032 Böblingen, Telefon: 49 - (0)7031 - 16-2247
E-Mail: schwidefsky@de.ibm.com

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2004-05-14 11:40 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-05-13 20:35 atomic ops for a VM fix Andrew Morton
2004-05-13 20:37 ` Andrew Morton
2004-05-13 21:52 ` Ivan Kokshaysky
2004-05-13 22:33   ` David Mosberger
2004-05-13 22:52     ` Andrew Morton
  -- strict thread matches above, loose matches on Subject: below --
2004-05-14 11:36 Martin Schwidefsky

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox