mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Mel Gorman <mgorman@suse.de>
To: H Peter Anvin <hpa@zytor.com>, Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Mike Galbraith <mgalbraith@suse.de>, Linux-X86 <x86@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>, Mel Gorman <mgorman@suse.de>
Subject: [PATCH 2/3] x86, fpu: Split FPU save state preparation into eagerfpu and !eagerfpu parts
Date: Wed,  6 Aug 2014 13:55:47 +0100	[thread overview]
Message-ID: <1407329748-3928-3-git-send-email-mgorman@suse.de> (raw)
In-Reply-To: <1407329748-3928-1-git-send-email-mgorman@suse.de>

If the CPU supports non-lazy FPU support using xsave then a different save
path is used for the FPU during context switch. The xsave path is heavier
than it needs to be so this patch splits the two cases in preparation.

Signed-off-by: Mel Gorman <mgorman@suse.de>
---
 arch/x86/include/asm/fpu-internal.h | 40 ++++++++++++++++++++++++++++++++-----
 1 file changed, 35 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/fpu-internal.h b/arch/x86/include/asm/fpu-internal.h
index b8771c4a..8d92807 100644
--- a/arch/x86/include/asm/fpu-internal.h
+++ b/arch/x86/include/asm/fpu-internal.h
@@ -416,15 +416,36 @@ static inline int fpu_lazy_restore(struct task_struct *new, unsigned int cpu)
 		cpu == new->thread.fpu.last_cpu;
 }
 
-static inline void switch_fpu_prepare(struct task_struct *old, struct task_struct *new, int cpu,
+static inline void switch_eagerfpu_prepare(struct task_struct *old, struct task_struct *new, int cpu,
+						fpu_switch_t *fpu)
+{
+	fpu->preload = tsk_used_math(new);
+
+	if (__thread_has_fpu(old)) {
+		if (!__save_init_fpu(old))
+			cpu = ~0;
+		old->thread.fpu.last_cpu = cpu;
+		old->thread.fpu.has_fpu = 0;	/* But leave fpu_owner_task! */
+	} else {
+		old->thread.fpu_counter = 0;
+		old->thread.fpu.last_cpu = ~0;
+	}
+
+	if (fpu->preload) {
+		new->thread.fpu_counter++;
+		__thread_set_has_fpu(new);
+		prefetch(new->thread.fpu.state);
+	}
+}
+
+static inline void switch_preloadfpu_prepare(struct task_struct *old, struct task_struct *new, int cpu,
 						fpu_switch_t *fpu)
 {
 	/*
 	 * If the task has used the math, pre-load the FPU on xsave processors
 	 * or if the past 5 consecutive context-switches used math.
 	 */
-	fpu->preload = tsk_used_math(new) && (use_eager_fpu() ||
-					     new->thread.fpu_counter > 5);
+	fpu->preload = tsk_used_math(new) && (new->thread.fpu_counter > 5);
 	if (__thread_has_fpu(old)) {
 		if (!__save_init_fpu(old))
 			cpu = ~0;
@@ -436,14 +457,14 @@ static inline void switch_fpu_prepare(struct task_struct *old, struct task_struc
 			new->thread.fpu_counter++;
 			__thread_set_has_fpu(new);
 			prefetch(new->thread.fpu.state);
-		} else if (!use_eager_fpu())
+		} else
 			stts();
 	} else {
 		old->thread.fpu_counter = 0;
 		old->thread.fpu.last_cpu = ~0;
 		if (fpu->preload) {
 			new->thread.fpu_counter++;
-			if (!use_eager_fpu() && fpu_lazy_restore(new, cpu))
+			if (fpu_lazy_restore(new, cpu))
 				fpu->preload = 0;
 			else
 				prefetch(new->thread.fpu.state);
@@ -452,6 +473,15 @@ static inline void switch_fpu_prepare(struct task_struct *old, struct task_struc
 	}
 }
 
+static inline void switch_fpu_prepare(struct task_struct *old, struct task_struct *new, int cpu,
+						fpu_switch_t *fpu)
+{
+	if (use_eager_fpu())
+		switch_eagerfpu_prepare(old, new, cpu, fpu);
+	else
+		switch_preloadfpu_prepare(old, new, cpu, fpu);
+}
+
 /*
  * By the time this gets called, we've already cleared CR0.TS and
  * given the process the FPU if we are going to preload the FPU
-- 
1.8.4.5


  parent reply	other threads:[~2014-08-06 12:55 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-08-06 12:55 [PATCH 0/3] Reduce length of the eagerfpu path during x86 context switches Mel Gorman
2014-08-06 12:55 ` [PATCH 1/3] x86, fpu: Do not copy fpu preload state Mel Gorman
2014-08-06 12:55 ` Mel Gorman [this message]
2014-08-06 12:55 ` [PATCH 3/3] x86, fpu: Do not update fpu_counter in the eagerfpu case Mel Gorman
2014-08-27 16:03 ` [PATCH 0/3] Reduce length of the eagerfpu path during x86 context switches Mel Gorman
2014-08-27 17:03   ` H. Peter Anvin

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=1407329748-3928-3-git-send-email-mgorman@suse.de \
    --to=mgorman@suse.de \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mgalbraith@suse.de \
    --cc=suresh.b.siddha@intel.com \
    --cc=x86@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®