From: Martin Schwidefsky <schwidefsky@de.ibm.com>
To: linux-kernel@vger.kernel.org, linux-s390@vger.kernel.org
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>,
Matthew Wilcox <matthew@wil.cx>,
Martin Schwidefsky <schwidefsky@de.ibm.com>
Subject: [patch 07/52] [PATCH] Improve code generated by atomic operations.
Date: Fri, 13 Nov 2009 16:08:31 +0100 [thread overview]
Message-ID: <20091113150910.350086040@de.ibm.com> (raw)
In-Reply-To: 20091113150824.351347652@de.ibm.com
[-- Attachment #1: 106-atomic-volatile.diff --]
[-- Type: text/plain, Size: 2395 bytes --]
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
Git commit ea435467500612636f8f4fb639ff6e76b2496e4b changed the
definition of atomic_t and atomic64_t for s390 by adding the volatile
modifier to the counter field. This has an unfortunate side effect
with newer versions of the gcc. The typeof operator now picks up the
volatile modifier from the expression. This causes the compiler to
think that it has to store the two temporary variable old_val and
new_val in the __CS_LOOP for the different atomic operations to the
stack as the variables are now volatile. Both stores are superfluous.
The hack to replace typeof(ptr->counter) with int in __CS_LOOP and
and long long in __CSG_LOOP avoids the two stores. A better solution
would be to drop the volatile from the counter field of the atomic_t
and atomic64_t definition. But that is a touchy subject ..
Cc: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/include/asm/atomic.h | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
Index: quilt-2.6/arch/s390/include/asm/atomic.h
===================================================================
--- quilt-2.6.orig/arch/s390/include/asm/atomic.h 2009-11-13 15:48:32.000000000 +0100
+++ quilt-2.6/arch/s390/include/asm/atomic.h 2009-11-13 16:08:13.000000000 +0100
@@ -21,7 +21,7 @@
#if __GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ > 2)
#define __CS_LOOP(ptr, op_val, op_string) ({ \
- typeof(ptr->counter) old_val, new_val; \
+ int old_val, new_val; \
asm volatile( \
" l %0,%2\n" \
"0: lr %1,%0\n" \
@@ -38,7 +38,7 @@
#else /* __GNUC__ */
#define __CS_LOOP(ptr, op_val, op_string) ({ \
- typeof(ptr->counter) old_val, new_val; \
+ int old_val, new_val; \
asm volatile( \
" l %0,0(%3)\n" \
"0: lr %1,%0\n" \
@@ -143,7 +143,7 @@
#if __GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ > 2)
#define __CSG_LOOP(ptr, op_val, op_string) ({ \
- typeof(ptr->counter) old_val, new_val; \
+ long long old_val, new_val; \
asm volatile( \
" lg %0,%2\n" \
"0: lgr %1,%0\n" \
@@ -160,7 +160,7 @@
#else /* __GNUC__ */
#define __CSG_LOOP(ptr, op_val, op_string) ({ \
- typeof(ptr->counter) old_val, new_val; \
+ long long old_val, new_val; \
asm volatile( \
" lg %0,0(%3)\n" \
"0: lgr %1,%0\n" \
next prev parent reply other threads:[~2009-11-13 15:08 UTC|newest]
Thread overview: 56+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-13 15:08 [patch 00/52] s390 patches for the next merge window (2.6.33) Martin Schwidefsky
2009-11-13 15:08 ` [patch 01/52] [PATCH] Improve address space check Martin Schwidefsky
2009-11-13 15:08 ` [patch 02/52] [PATCH] Improve address space mode selection Martin Schwidefsky
2009-11-13 15:08 ` [patch 03/52] [PATCH] Improve notify_page_fault implementation Martin Schwidefsky
2009-11-13 15:08 ` [patch 04/52] [PATCH] fault handler performance optimization Martin Schwidefsky
2009-11-13 15:08 ` [patch 05/52] [PATCH] fault handler access flags check Martin Schwidefsky
2009-11-13 15:08 ` [patch 06/52] [PATCH] Use do_exception() in pagetable walk usercopy functions Martin Schwidefsky
2009-11-13 15:08 ` Martin Schwidefsky [this message]
2009-11-13 15:08 ` [patch 08/52] [PATCH] dasd: support DIAG access for read-only devices Martin Schwidefsky
2009-11-13 15:08 ` [patch 09/52] [PATCH] cmm: free pages on hibernate Martin Schwidefsky
2009-11-13 15:08 ` [patch 10/52] [PATCH] smp: remove unused typedef and defines Martin Schwidefsky
2009-11-13 15:08 ` [patch 11/52] [PATCH] dasd: remove dead code Martin Schwidefsky
2009-11-13 15:08 ` [patch 12/52] [PATCH] use generic termbits.h header file Martin Schwidefsky
2009-11-13 15:08 ` [patch 13/52] [PATCH] use generic sockios.h " Martin Schwidefsky
2009-11-13 15:08 ` [patch 14/52] [PATCH] MAINTAINERS: Add s390 drivers block Martin Schwidefsky
2009-11-13 15:08 ` [patch 15/52] [PATCH] zcrypt: initialize ap_messages for cex3 exploitation Martin Schwidefsky
2009-11-13 15:08 ` [patch 16/52] [PATCH] zcrypt: special command support " Martin Schwidefsky
2009-11-13 15:08 ` [patch 17/52] [PATCH] zcrypt: add support for cex3 device types Martin Schwidefsky
2009-11-13 15:08 ` [patch 18/52] [PATCH] zcrypt: use definitions for cex3 Martin Schwidefsky
2009-11-13 15:08 ` [patch 19/52] [PATCH] zcrypt: adjust speed rating between cex2 and pcixcc Martin Schwidefsky
2009-11-13 15:08 ` [patch 20/52] [PATCH] zcrypt: adjust speed rating of cex3 adapters Martin Schwidefsky
2009-11-13 15:08 ` [patch 21/52] [PATCH] dasd: enable prefix independent of pav support Martin Schwidefsky
2009-11-13 15:08 ` [patch 22/52] [PATCH] dasd: improve error recovery for internal I/O Martin Schwidefsky
2009-11-13 15:08 ` [patch 23/52] [PATCH] dasd: remove strings from s390dbf Martin Schwidefsky
2009-11-13 15:08 ` [patch 24/52] [PATCH] s390: use change recording override for kernel mapping Martin Schwidefsky
2009-11-13 15:08 ` [patch 25/52] [PATCH] sclp: improve servicability setting Martin Schwidefsky
2009-11-13 15:08 ` [patch 26/52] [PATCH] cio: fix double free in case of probe failure Martin Schwidefsky
2009-11-13 15:08 ` [patch 27/52] [PATCH] cio: fix repeat setting of cdev parent association Martin Schwidefsky
2009-11-13 15:08 ` [patch 28/52] [PATCH] cio: introduce parent-initiated device move Martin Schwidefsky
2009-11-13 15:08 ` [patch 29/52] [PATCH] cio: introduce subchannel todos Martin Schwidefsky
2009-11-13 15:08 ` [patch 30/52] [PATCH] cio: introduce ccw device todos Martin Schwidefsky
2009-11-13 15:08 ` [patch 31/52] [PATCH] cio: inform user when online/offline processing fails Martin Schwidefsky
2009-11-13 15:08 ` [patch 32/52] [PATCH] cio: handle error during device recognition consistently Martin Schwidefsky
2009-11-13 15:08 ` [patch 33/52] [PATCH] cio: handle error during path verification consistently Martin Schwidefsky
2009-11-13 15:08 ` [patch 34/52] [PATCH] cio: ensure proper locking during device recognition Martin Schwidefsky
2009-11-13 15:08 ` [patch 35/52] [PATCH] cio: dont panic in non-fatal conditions Martin Schwidefsky
2009-11-13 15:09 ` [patch 36/52] [PATCH] cio: consistent infrastructure for internal I/O requests Martin Schwidefsky
2009-11-13 15:09 ` [patch 37/52] [PATCH] cio: use ccw request infrastructure for sense id Martin Schwidefsky
2009-11-13 15:09 ` [patch 38/52] [PATCH] cio: use ccw request infrastructure for pgid Martin Schwidefsky
2009-11-13 15:09 ` [patch 39/52] [PATCH] cio: allow setting not-operational devices offline Martin Schwidefsky
2009-11-13 15:09 ` [patch 40/52] [PATCH] cio: remove intretry flag Martin Schwidefsky
2009-11-13 15:09 ` [patch 41/52] [PATCH] cio: split PGID settings and status Martin Schwidefsky
2009-11-13 15:09 ` [patch 42/52] [PATCH] cio: use sense-pgid operation for path verification Martin Schwidefsky
2009-11-13 15:09 ` [patch 43/52] [PATCH] cio: make steal lock procedure more robust Martin Schwidefsky
2009-11-13 15:09 ` [patch 44/52] [PATCH] cio: remove registered flag from ccw_device_private Martin Schwidefsky
2009-11-13 15:09 ` [patch 45/52] [PATCH] cio: add per device initialization status flag Martin Schwidefsky
2009-11-13 15:09 ` [patch 46/52] [PATCH] cio: fix quiesce state Martin Schwidefsky
2009-11-13 15:09 ` [patch 47/52] [PATCH] cio: handle failed disable_subchannel after device recognition Martin Schwidefsky
2009-11-13 15:09 ` [patch 48/52] [PATCH] cio: handle busy subchannel in ccw_device_move_to_sch Martin Schwidefsky
2009-11-13 15:09 ` [patch 49/52] [PATCH] cio: quiesce subchannel in io_subchannel_remove Martin Schwidefsky
2009-11-13 15:09 ` [patch 50/52] [PATCH] cio: change locking " Martin Schwidefsky
2009-11-13 15:09 ` [patch 51/52] [PATCH] cio: improve error recovery for internal I/Os Martin Schwidefsky
2009-11-13 15:09 ` [patch 52/52] [PATCH] cio: dont unregister a busy device in ccw_device_set_offline Martin Schwidefsky
2009-11-13 15:28 ` [patch 00/52] s390 patches for the next merge window (2.6.33) Arnd Bergmann
2009-11-13 15:31 ` s390: move keyboard compat ioctls into tty3270 driver Arnd Bergmann
2009-11-16 8:33 ` Martin Schwidefsky
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20091113150910.350086040@de.ibm.com \
--to=schwidefsky@de.ibm.com \
--cc=heiko.carstens@de.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-s390@vger.kernel.org \
--cc=matthew@wil.cx \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).