All of lore.kernel.org
 help / color / mirror / Atom feed
From: Ingo Molnar <mingo@elte.hu>
To: "Chen, Kenneth W" <kenneth.w.chen@intel.com>
Cc: Keith Owens <kaos@ocs.com.au>,
	David.Mosberger@acm.org, Andrew Morton <akpm@osdl.org>,
	linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org
Subject: Re: Add prefetch switch stack hook in scheduler function
Date: Fri, 29 Jul 2005 07:07:02 +0000	[thread overview]
Message-ID: <20050729070702.GA3327@elte.hu> (raw)
In-Reply-To: <20050729070447.GA3032@elte.hu>


* Ingo Molnar <mingo@elte.hu> wrote:

> > Sorry, this is not enough.  Switch stack on ia64 is 528 bytes.  We 
> > need to prefetch 5 lines.  It probably should use prefetch_range().  
> 
> ok, how about the additional patch below? Does this do the trick on 
> ia64? It makes complete sense on every architecture to prefetch from 
> below the current kernel stack, in the expectation of the next task 
> touching the stack. The only difference is that for ia64 the 'expected 
> minimum stack footprint' is larger, due to the switch_stack.

the patch below unrolls the prefetch_range() loop manually, for up to 5 
cachelines prefetched. This patch, ontop of the 4 previous patches, 
should generate similar code to the assembly code in your original 
patch. The full patch-series is:

 patches/prefetch-next.patch
 patches/prefetch-mm.patch
 patches/prefetch-kstack-size.patch
 patches/prefetch-unroll.patch

	Ingo

---------
unroll prefetch_range() loops manually.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

 include/linux/prefetch.h |   31 +++++++++++++++++++++++++++++--
 1 files changed, 29 insertions(+), 2 deletions(-)

Index: linux/include/linux/prefetch.h
=================================--- linux.orig/include/linux/prefetch.h
+++ linux/include/linux/prefetch.h
@@ -58,11 +58,38 @@ static inline void prefetchw(const void 
 static inline void prefetch_range(void *addr, size_t len)
 {
 #ifdef ARCH_HAS_PREFETCH
-	char *cp;
+	char *cp = addr;
 	char *end = addr + len;
 
-	for (cp = addr; cp < end; cp += PREFETCH_STRIDE)
+	/*
+	 * Unroll agressively:
+	 */
+	if (len <= PREFETCH_STRIDE)
 		prefetch(cp);
+	else if (len <= 2*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+	}
+	else if (len <= 3*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+	}
+	else if (len <= 4*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+		prefetch(cp + 3*PREFETCH_STRIDE);
+	}
+	else if (len <= 5*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+		prefetch(cp + 3*PREFETCH_STRIDE);
+		prefetch(cp + 4*PREFETCH_STRIDE);
+	} else
+		for (; cp < end; cp += PREFETCH_STRIDE)
+			prefetch(cp);
 #endif
 }
 

WARNING: multiple messages have this Message-ID (diff)
From: Ingo Molnar <mingo@elte.hu>
To: "Chen, Kenneth W" <kenneth.w.chen@intel.com>
Cc: Keith Owens <kaos@ocs.com.au>,
	David.Mosberger@acm.org, Andrew Morton <akpm@osdl.org>,
	linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org
Subject: Re: Add prefetch switch stack hook in scheduler function
Date: Fri, 29 Jul 2005 09:07:02 +0200	[thread overview]
Message-ID: <20050729070702.GA3327@elte.hu> (raw)
In-Reply-To: <20050729070447.GA3032@elte.hu>


* Ingo Molnar <mingo@elte.hu> wrote:

> > Sorry, this is not enough.  Switch stack on ia64 is 528 bytes.  We 
> > need to prefetch 5 lines.  It probably should use prefetch_range().  
> 
> ok, how about the additional patch below? Does this do the trick on 
> ia64? It makes complete sense on every architecture to prefetch from 
> below the current kernel stack, in the expectation of the next task 
> touching the stack. The only difference is that for ia64 the 'expected 
> minimum stack footprint' is larger, due to the switch_stack.

the patch below unrolls the prefetch_range() loop manually, for up to 5 
cachelines prefetched. This patch, ontop of the 4 previous patches, 
should generate similar code to the assembly code in your original 
patch. The full patch-series is:

 patches/prefetch-next.patch
 patches/prefetch-mm.patch
 patches/prefetch-kstack-size.patch
 patches/prefetch-unroll.patch

	Ingo

---------
unroll prefetch_range() loops manually.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

 include/linux/prefetch.h |   31 +++++++++++++++++++++++++++++--
 1 files changed, 29 insertions(+), 2 deletions(-)

Index: linux/include/linux/prefetch.h
===================================================================
--- linux.orig/include/linux/prefetch.h
+++ linux/include/linux/prefetch.h
@@ -58,11 +58,38 @@ static inline void prefetchw(const void 
 static inline void prefetch_range(void *addr, size_t len)
 {
 #ifdef ARCH_HAS_PREFETCH
-	char *cp;
+	char *cp = addr;
 	char *end = addr + len;
 
-	for (cp = addr; cp < end; cp += PREFETCH_STRIDE)
+	/*
+	 * Unroll agressively:
+	 */
+	if (len <= PREFETCH_STRIDE)
 		prefetch(cp);
+	else if (len <= 2*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+	}
+	else if (len <= 3*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+	}
+	else if (len <= 4*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+		prefetch(cp + 3*PREFETCH_STRIDE);
+	}
+	else if (len <= 5*PREFETCH_STRIDE) {
+		prefetch(cp);
+		prefetch(cp + PREFETCH_STRIDE);
+		prefetch(cp + 2*PREFETCH_STRIDE);
+		prefetch(cp + 3*PREFETCH_STRIDE);
+		prefetch(cp + 4*PREFETCH_STRIDE);
+	} else
+		for (; cp < end; cp += PREFETCH_STRIDE)
+			prefetch(cp);
 #endif
 }
 

  reply	other threads:[~2005-07-29  7:07 UTC|newest]

Thread overview: 73+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-07-27 22:07 Add prefetch switch stack hook in scheduler function Chen, Kenneth W
2005-07-27 22:07 ` Chen, Kenneth W
2005-07-27 23:13 ` Andrew Morton
2005-07-27 23:13   ` Andrew Morton
2005-07-27 23:23   ` david mosberger
2005-07-27 23:23     ` david mosberger
2005-07-28  7:41     ` Ingo Molnar
2005-07-28  7:41       ` Ingo Molnar
2005-07-28  8:09 ` Keith Owens
2005-07-28  8:09   ` Keith Owens
2005-07-28  8:16   ` Ingo Molnar
2005-07-28  8:16     ` Ingo Molnar
2005-07-28  9:09     ` Ingo Molnar
2005-07-28  9:09       ` Ingo Molnar
2005-07-28  8:31   ` Nick Piggin
2005-07-28  8:31     ` Nick Piggin
2005-07-28  8:35     ` Ingo Molnar
2005-07-28  8:35       ` Ingo Molnar
2005-07-28  8:48       ` Nick Piggin
2005-07-28  8:48         ` Nick Piggin
2005-07-28  9:16         ` Ingo Molnar
2005-07-28  9:16           ` Ingo Molnar
2005-07-28  9:19           ` Ingo Molnar
2005-07-28  9:19             ` Ingo Molnar
2005-07-28  9:34           ` Nick Piggin
2005-07-28  9:34             ` Nick Piggin
2005-07-28 10:04             ` Ingo Molnar
2005-07-28 10:04               ` Ingo Molnar
2005-07-28 10:29               ` Nick Piggin
2005-07-28 10:29                 ` Nick Piggin
2005-07-28 19:14 ` Chen, Kenneth W
2005-07-28 19:14   ` Chen, Kenneth W
2005-07-29  7:04   ` Ingo Molnar
2005-07-29  7:04     ` Ingo Molnar
2005-07-29  7:07     ` Ingo Molnar [this message]
2005-07-29  7:07       ` Ingo Molnar
2005-07-29  8:30       ` Eric Dumazet
2005-07-29  8:30         ` Eric Dumazet
2005-07-29  8:44         ` Ingo Molnar
2005-07-29  8:44           ` Ingo Molnar
2005-07-31 16:27         ` hashed spinlocks Daniel Walker
2005-07-31 18:46           ` David S. Miller
2005-07-31 19:06             ` Daniel Walker
2005-07-31 19:11               ` David S. Miller
2005-07-31 19:16                 ` Daniel Walker
2005-07-29  9:17       ` Add prefetch switch stack hook in scheduler function Peter Zijlstra
2005-07-29  9:17         ` Peter Zijlstra
2005-07-29 10:52         ` Ingo Molnar
2005-07-29 10:52           ` Ingo Molnar
2005-07-29  7:22 ` Chen, Kenneth W
2005-07-29  7:22   ` Chen, Kenneth W
2005-07-29  8:28   ` Ingo Molnar
2005-07-29  8:28     ` Ingo Molnar
2005-07-29  9:02     ` Russell King
2005-07-29  9:02       ` Russell King
2005-07-29  9:45       ` Ingo Molnar
2005-07-29  9:45         ` Ingo Molnar
2005-07-29  7:38 ` Keith Owens
2005-07-29  7:38   ` Keith Owens
2005-07-29  7:45 ` Keith Owens
2005-07-29  7:45   ` Keith Owens
2005-07-29  8:02 ` Chen, Kenneth W
2005-07-29  8:02   ` Chen, Kenneth W
2005-07-29  8:08 ` Chen, Kenneth W
2005-07-29  8:08   ` Chen, Kenneth W
2005-07-29  8:30 ` Chen, Kenneth W
2005-07-29  8:30   ` Chen, Kenneth W
2005-07-29  8:35   ` Ingo Molnar
2005-07-29  8:35     ` Ingo Molnar
2005-07-29  8:39 ` Chen, Kenneth W
2005-07-29  8:39   ` Chen, Kenneth W
  -- strict thread matches above, loose matches on Subject: below --
2005-07-29 15:18 linux
2005-07-29 15:49 ` Ingo Molnar

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=20050729070702.GA3327@elte.hu \
    --to=mingo@elte.hu \
    --cc=David.Mosberger@acm.org \
    --cc=akpm@osdl.org \
    --cc=kaos@ocs.com.au \
    --cc=kenneth.w.chen@intel.com \
    --cc=linux-ia64@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.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.