linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Christoph Lameter <cl@linux.com>
To: Tejun Heo <tj@kernel.org>
Cc: akpm@linuxfoundation.org, rostedt@goodmis.org,
	linux-kernel@vger.kernel.org, Ingo Molnar <mingo@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
	Matt Turner <mattst88@gmail.com>,
	Richard Henderson <rth@twiddle.net>
Subject: [PATCH 24/31] alpha: Replace __get_cpu_var
Date: Fri, 20 Jun 2014 14:31:39 -0500	[thread overview]
Message-ID: <20140620193127.753042628@linux.com> (raw)
In-Reply-To: 20140620193115.547427118@linux.com

[-- Attachment #1: this_alpha --]
[-- Type: text/plain, Size: 6521 bytes --]

__get_cpu_var() is used for multiple purposes in the kernel source. One of
them is address calculation via the form &__get_cpu_var(x).  This calculates
the address for the instance of the percpu variable of the current processor
based on an offset.

Other use cases are for storing and retrieving data from the current
processors percpu area.  __get_cpu_var() can be used as an lvalue when
writing data or on the right side of an assignment.

__get_cpu_var() is defined as :


#define __get_cpu_var(var) (*this_cpu_ptr(&(var)))



__get_cpu_var() always only does an address determination. However, store
and retrieve operations could use a segment prefix (or global register on
other platforms) to avoid the address calculation.

this_cpu_write() and this_cpu_read() can directly take an offset into a
percpu area and use optimized assembly code to read and write per cpu
variables.


This patch converts __get_cpu_var into either an explicit address
calculation using this_cpu_ptr() or into a use of this_cpu operations that
use the offset.  Thereby address calculations are avoided and less registers
are used when code is generated.

At the end of the patch set all uses of __get_cpu_var have been removed so
the macro is removed too.

The patch set includes passes over all arches as well. Once these operations
are used throughout then specialized macros can be defined in non -x86
arches as well in order to optimize per cpu access by f.e.  using a global
register that may be set to the per cpu base.




Transformations done to __get_cpu_var()


1. Determine the address of the percpu instance of the current processor.

	DEFINE_PER_CPU(int, y);
	int *x = &__get_cpu_var(y);

    Converts to

	int *x = this_cpu_ptr(&y);


2. Same as #1 but this time an array structure is involved.

	DEFINE_PER_CPU(int, y[20]);
	int *x = __get_cpu_var(y);

    Converts to

	int *x = this_cpu_ptr(y);


3. Retrieve the content of the current processors instance of a per cpu
variable.

	DEFINE_PER_CPU(int, y);
	int x = __get_cpu_var(y)

   Converts to

	int x = __this_cpu_read(y);


4. Retrieve the content of a percpu struct

	DEFINE_PER_CPU(struct mystruct, y);
	struct mystruct x = __get_cpu_var(y);

   Converts to

	memcpy(&x, this_cpu_ptr(&y), sizeof(x));


5. Assignment to a per cpu variable

	DEFINE_PER_CPU(int, y)
	__get_cpu_var(y) = x;

   Converts to

	__this_cpu_write(y, x);


6. Increment/Decrement etc of a per cpu variable

	DEFINE_PER_CPU(int, y);
	__get_cpu_var(y)++

   Converts to

	__this_cpu_inc(y)

CC: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Acked-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Christoph Lameter <cl@linux.com>

Index: linux/arch/alpha/kernel/perf_event.c
===================================================================
--- linux.orig/arch/alpha/kernel/perf_event.c	2014-06-16 09:35:01.360630259 -0500
+++ linux/arch/alpha/kernel/perf_event.c	2014-06-16 09:35:13.880385571 -0500
@@ -431,7 +431,7 @@
  */
 static int alpha_pmu_add(struct perf_event *event, int flags)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
 	int n0;
 	int ret;
@@ -483,7 +483,7 @@
  */
 static void alpha_pmu_del(struct perf_event *event, int flags)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
 	unsigned long irq_flags;
 	int j;
@@ -531,7 +531,7 @@
 static void alpha_pmu_stop(struct perf_event *event, int flags)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 
 	if (!(hwc->state & PERF_HES_STOPPED)) {
 		cpuc->idx_mask &= ~(1UL<<hwc->idx);
@@ -551,7 +551,7 @@
 static void alpha_pmu_start(struct perf_event *event, int flags)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 
 	if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
 		return;
@@ -724,7 +724,7 @@
  */
 static void alpha_pmu_enable(struct pmu *pmu)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 
 	if (cpuc->enabled)
 		return;
@@ -750,7 +750,7 @@
 
 static void alpha_pmu_disable(struct pmu *pmu)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 
 	if (!cpuc->enabled)
 		return;
@@ -814,8 +814,8 @@
 	struct hw_perf_event *hwc;
 	int idx, j;
 
-	__get_cpu_var(irq_pmi_count)++;
-	cpuc = &__get_cpu_var(cpu_hw_events);
+	__this_cpu_inc(irq_pmi_count);
+	cpuc = this_cpu_ptr(&cpu_hw_events);
 
 	/* Completely counting through the PMC's period to trigger a new PMC
 	 * overflow interrupt while in this interrupt routine is utterly
Index: linux/arch/alpha/kernel/time.c
===================================================================
--- linux.orig/arch/alpha/kernel/time.c	2014-06-16 09:35:01.360630259 -0500
+++ linux/arch/alpha/kernel/time.c	2014-06-16 09:35:01.356630339 -0500
@@ -56,9 +56,9 @@
 
 DEFINE_PER_CPU(u8, irq_work_pending);
 
-#define set_irq_work_pending_flag()  __get_cpu_var(irq_work_pending) = 1
-#define test_irq_work_pending()      __get_cpu_var(irq_work_pending)
-#define clear_irq_work_pending()     __get_cpu_var(irq_work_pending) = 0
+#define set_irq_work_pending_flag()  __this_cpu_write(irq_work_pending, 1)
+#define test_irq_work_pending()      __this_cpu_read(irq_work_pending)
+#define clear_irq_work_pending()     __this_cpu_write(irq_work_pending, 0)
 
 void arch_irq_work_raise(void)
 {
Index: linux/arch/powerpc/include/asm/cputime.h
===================================================================
--- linux.orig/arch/powerpc/include/asm/cputime.h	2014-06-16 09:34:54.000000000 -0500
+++ linux/arch/powerpc/include/asm/cputime.h	2014-06-16 09:36:08.787312456 -0500
@@ -56,10 +56,10 @@
 static inline cputime_t cputime_to_scaled(const cputime_t ct)
 {
 	if (cpu_has_feature(CPU_FTR_SPURR) &&
-	    __get_cpu_var(cputime_last_delta))
+	    __this_cpu_read(cputime_last_delta))
 		return (__force u64) ct *
-			__get_cpu_var(cputime_scaled_last_delta) /
-			__get_cpu_var(cputime_last_delta);
+			__this_cpu_read(cputime_scaled_last_delta) /
+			__this_cpu_read(cputime_last_delta);
 	return ct;
 }
 


  parent reply	other threads:[~2014-06-20 19:32 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-20 19:31 [PATCH 00/31] percpu: Consistent per cpu operations V5 Christoph Lameter
2014-06-20 19:31 ` [PATCH 01/31] kernel misc: Replace __get_cpu_var uses Christoph Lameter
2014-07-18 23:26   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 02/31] time: " Christoph Lameter
2014-06-20 19:31 ` [PATCH 03/31] scheduler: Replace __get_cpu_var with this_cpu_ptr Christoph Lameter
2014-07-18 23:27   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 04/31] block: Replace __this_cpu_ptr with raw_cpu_ptr Christoph Lameter
2014-07-18 23:28   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 05/31] drivers/char/random: Replace __get_cpu_var uses Christoph Lameter
2014-07-18 23:28   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 06/31] drivers/cpuidle: Replace __get_cpu_var uses for address calculation Christoph Lameter
2014-07-18 23:28   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 07/31] drivers/oprofile: " Christoph Lameter
2014-07-18 23:28   ` Tejun Heo
2014-07-19 11:54     ` Robert Richter
2014-06-20 19:31 ` [PATCH 08/31] drivers/clocksource: Replace __get_cpu_var used " Christoph Lameter
2014-07-18 23:29   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 09/31] drivers/net/ethernet/tile: Replace __get_cpu_var uses " Christoph Lameter
2014-07-18 23:29   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 10/31] watchdog: Replace __raw_get_cpu_var uses Christoph Lameter
2014-07-18 23:29   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 11/31] net: Replace get_cpu_var through this_cpu_ptr Christoph Lameter
2014-07-18 23:29   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 12/31] md: Replace __this_cpu_ptr with raw_cpu_ptr Christoph Lameter
2014-07-18 23:29   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 13/31] metag: Replace __get_cpu_var uses for address calculation Christoph Lameter
2014-07-18 23:30   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 14/31] drivers/net/ethernet/tile: __get_cpu_var call introduced in 3.14 Christoph Lameter
2014-07-18 23:30   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 15/31] irqchips: Replace __this_cpu_ptr uses Christoph Lameter
2014-07-18 23:30   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 16/31] x86: Replace __get_cpu_var uses Christoph Lameter
2014-06-20 19:31 ` [PATCH 17/31] uv: Replace __get_cpu_var Christoph Lameter
2014-07-18 23:30   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 18/31] arm: Replace __this_cpu_ptr with raw_cpu_ptr Christoph Lameter
2014-07-18 23:31   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 19/31] MIPS: Replace __get_cpu_var uses in FPU emulator Christoph Lameter
2014-06-20 20:07   ` David Daney
2014-07-18 23:31   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 20/31] mips: Replace __get_cpu_var uses Christoph Lameter
2014-07-18 23:31   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 21/31] s390: " Christoph Lameter
2014-07-18 23:32   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 22/31] s390: cio driver &__get_cpu_var replacements Christoph Lameter
2014-07-18 23:32   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 23/31] ia64: Replace __get_cpu_var uses Christoph Lameter
2014-07-18 23:32   ` Tejun Heo
2014-06-20 19:31 ` Christoph Lameter [this message]
2014-07-18 23:32   ` [PATCH 24/31] alpha: Replace __get_cpu_var Tejun Heo
2014-06-20 19:31 ` [PATCH 25/31] powerpc: Replace __get_cpu_var uses Christoph Lameter
2014-06-20 19:31 ` [PATCH 26/31] tile: " Christoph Lameter
2014-07-18 23:33   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 27/31] tile: Use this_cpu_ptr() for hardware counters Christoph Lameter
2014-07-18 23:33   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 28/31] blackfin: Replace __get_cpu_var uses Christoph Lameter
2014-07-18 23:33   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 29/31] avr32: Replace __get_cpu_var with __this_cpu_write Christoph Lameter
2014-07-18 23:33   ` Tejun Heo
2014-06-20 19:31 ` [PATCH 30/31] Remove __get_cpu_var and __raw_get_cpu_var macros [only in 3.17] Christoph Lameter
2014-06-20 19:31 ` [PATCH 31/31] percpu: Remove __this_cpu_ptr Christoph Lameter

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=20140620193127.753042628@linux.com \
    --to=cl@linux.com \
    --cc=akpm@linuxfoundation.org \
    --cc=ink@jurassic.park.msu.ru \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mattst88@gmail.com \
    --cc=mingo@kernel.org \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=rth@twiddle.net \
    --cc=tglx@linutronix.de \
    --cc=tj@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;
as well as URLs for NNTP newsgroup(s).