From: Jan Kiszka <jan.kiszka@domain.hid>
To: adeos-main <adeos-main@gna.org>
Subject: [Adeos-main] [PATCH] avoid suboptimal use of ipipe_lock_cpu
Date: Wed, 13 Dec 2006 18:58:12 +0100 [thread overview]
Message-ID: <45803F34.9000003@domain.hid> (raw)
[-- Attachment #1.1: Type: text/plain, Size: 386 bytes --]
Hi,
while disassembling that bug in start_kernel, I noticed that
ipipe_lock_cpu(flags) is used in a suboptimal way in some
ipipe_XXX_head() services, i.e. when the flags are actually of no
interest. The compiler still generates the unneeded pushf/pop <reg>, and
this in a hot-path.
Attached patch should perform better. Lightly tested, but QEMU SMP still
works here.
Jan
[-- Attachment #1.2: avoid-ipipe_lock_cpu.patch --]
[-- Type: text/plain, Size: 1861 bytes --]
---
include/linux/ipipe.h | 8 ++++----
kernel/ipipe/core.c | 8 ++++----
2 files changed, 8 insertions(+), 8 deletions(-)
Index: linux-2.6.19/include/linux/ipipe.h
===================================================================
--- linux-2.6.19.orig/include/linux/ipipe.h
+++ linux-2.6.19/include/linux/ipipe.h
@@ -471,18 +471,18 @@ static inline void ipipe_restore_pipelin
static inline void ipipe_stall_pipeline_head(void)
{
ipipe_declare_cpuid;
- unsigned long flags;
- ipipe_lock_cpu(flags);
+ local_irq_disable_hw();
+ ipipe_load_cpuid();
__set_bit(IPIPE_STALL_FLAG, &__ipipe_pipeline_head()->cpudata[cpuid].status);
}
static inline unsigned long ipipe_test_and_stall_pipeline_head(void)
{
- unsigned long flags;
ipipe_declare_cpuid;
- ipipe_lock_cpu(flags);
+ local_irq_disable_hw();
+ ipipe_load_cpuid();
return __test_and_set_bit(IPIPE_STALL_FLAG, &__ipipe_pipeline_head()->cpudata[cpuid].status);
}
Index: linux-2.6.19/kernel/ipipe/core.c
===================================================================
--- linux-2.6.19.orig/kernel/ipipe/core.c
+++ linux-2.6.19/kernel/ipipe/core.c
@@ -303,10 +303,10 @@ void fastcall ipipe_restore_pipeline_fro
void ipipe_unstall_pipeline_head(void)
{
struct ipipe_domain *head;
- unsigned long flags;
ipipe_declare_cpuid;
- ipipe_lock_cpu(flags);
+ local_irq_disable_hw();
+ ipipe_load_cpuid();
head = __ipipe_pipeline_head();
__clear_bit(IPIPE_STALL_FLAG, &head->cpudata[cpuid].status);
@@ -323,9 +323,9 @@ void ipipe_unstall_pipeline_head(void)
void fastcall __ipipe_restore_pipeline_head(struct ipipe_domain *head, unsigned long x)
{
ipipe_declare_cpuid;
- unsigned long flags;
- ipipe_lock_cpu(flags);
+ local_irq_disable_hw();
+ ipipe_load_cpuid();
if (x) {
#ifdef CONFIG_DEBUG_KERNEL
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]
next reply other threads:[~2006-12-13 17:58 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-12-13 17:58 Jan Kiszka [this message]
2006-12-14 20:47 ` [Adeos-main] [PATCH] avoid suboptimal use of ipipe_lock_cpu Philippe Gerum
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=45803F34.9000003@domain.hid \
--to=jan.kiszka@domain.hid \
--cc=adeos-main@gna.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.