linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tejun Heo <tj@kernel.org>
To: jeff@garzik.org, mingo@elte.hu, linux-kernel@vger.kernel.org,
	akpm@linux-foundation.org, jens.axboe@oracle.com,
	rusty@rustcorp.com.au, cl@linux-foundation.org,
	dhowells@redhat.com, arjan@linux.intel.com
Cc: Tejun Heo <tj@kernel.org>
Subject: [PATCH 10/19] workqueue: update cwq alignement and make one more flag bit available
Date: Thu,  1 Oct 2009 17:09:09 +0900	[thread overview]
Message-ID: <1254384558-1018-11-git-send-email-tj@kernel.org> (raw)
In-Reply-To: <1254384558-1018-1-git-send-email-tj@kernel.org>

Currently cwqs are aligned to cacheline and lower two bits of
work_struct->data are considered to be available and used as flags.
Make the alignement requirement official by defining
WORK_STRUCT_FLAG_BITS and aligning cwqs to two's power of it.

This is in preparation of concurrency managed workqueue and cwqs being
aligned to cacheline wouldn't matter as much.  While at it, this patch
reserves one more bit for work flags and make sure the resulting
alignment is at least equal to or larger than that of long long.

NOT_SIGNED_OFF_YET
---
 include/linux/workqueue.h |   17 ++++++++++++++---
 kernel/workqueue.c        |   15 +++++++++++++--
 2 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h
index d3334a0..3d11ce3 100644
--- a/include/linux/workqueue.h
+++ b/include/linux/workqueue.h
@@ -22,11 +22,22 @@ typedef void (*work_func_t)(struct work_struct *work);
  */
 #define work_data_bits(work) ((unsigned long *)(&(work)->data))
 
+enum {
+	WORK_STRUCT_PENDING	= 0,	/* work item is pending execution */
+
+	/*
+	 * Reserve 3bits off of cwq pointer.  This is enough and
+	 * provides acceptable alignment on both 32 and 64bit
+	 * machines.
+	 */
+	WORK_STRUCT_FLAG_BITS	= 3,
+
+	WORK_STRUCT_FLAG_MASK	= (1UL << WORK_STRUCT_FLAG_BITS) - 1,
+	WORK_STRUCT_WQ_DATA_MASK = ~WORK_STRUCT_FLAG_MASK,
+};
+
 struct work_struct {
 	atomic_long_t data;
-#define WORK_STRUCT_PENDING 0		/* T if work item pending execution */
-#define WORK_STRUCT_FLAG_MASK (3UL)
-#define WORK_STRUCT_WQ_DATA_MASK (~WORK_STRUCT_FLAG_MASK)
 	struct list_head entry;
 	work_func_t func;
 #ifdef CONFIG_LOCKDEP
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index befda6c..28de966 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -50,7 +50,9 @@
 
 /*
  * The per-CPU workqueue (if single thread, we always use the first
- * possible cpu).
+ * possible cpu).  The lower WORK_STRUCT_FLAG_BITS of
+ * work_struct->data are used for flags and thus cwqs need to be
+ * aligned at two's power of the number of flag bits.
  */
 struct cpu_workqueue_struct {
 
@@ -62,7 +64,7 @@ struct cpu_workqueue_struct {
 
 	struct workqueue_struct *wq;		/* I: the owning workqueue */
 	struct task_struct	*thread;
-} ____cacheline_aligned;
+} __attribute__((aligned(1 << WORK_STRUCT_FLAG_BITS)));
 
 /*
  * The externally visible workqueue abstraction is an array of
@@ -1049,6 +1051,15 @@ EXPORT_SYMBOL_GPL(work_on_cpu);
 
 void __init init_workqueues(void)
 {
+	/*
+	 * cwqs are forced aligned according to WORK_STRUCT_FLAG_BITS.
+	 * Make sure that the alignment isn't lower than that of
+	 * unsigned long long in case this code survives for longer
+	 * than twenty years.  :-P
+	 */
+	BUILD_BUG_ON(__alignof__(struct cpu_workqueue_struct) <
+		     __alignof__(unsigned long long));
+
 	alloc_cpumask_var(&cpu_populated_map, GFP_KERNEL);
 
 	cpumask_copy(cpu_populated_map, cpu_online_mask);
-- 
1.6.4.2


  parent reply	other threads:[~2009-10-01  8:10 UTC|newest]

Thread overview: 83+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-10-01  8:08 [RFC PATCHSET] workqueue: implement concurrency managed workqueue Tejun Heo
2009-10-01  8:09 ` [PATCH 01/19] freezer: don't get over-anxious while waiting Tejun Heo
2009-10-01 18:36   ` Pavel Machek
2009-10-01 21:04     ` Rafael J. Wysocki
2009-10-02 10:56       ` Tejun Heo
2009-10-02 19:47       ` Oren Laadan
2009-10-02 21:04         ` Matt Helsley
2009-10-02 21:21           ` Rafael J. Wysocki
2009-10-03  0:43             ` Tejun Heo
2009-10-03 19:36               ` Rafael J. Wysocki
2009-10-01  8:09 ` [PATCH 02/19] scheduler: implement sched_class_equal() Tejun Heo
2009-10-01  8:09 ` [PATCH 03/19] scheduler: implement workqueue scheduler class Tejun Heo
2009-10-01 16:57   ` Linus Torvalds
2009-10-01 18:48     ` Ingo Molnar
2009-10-01 19:00       ` Avi Kivity
2009-10-01 19:11         ` Linus Torvalds
2009-10-01 19:23           ` Ingo Molnar
2009-10-01 20:03             ` Linus Torvalds
2009-10-01 19:15         ` Ingo Molnar
2009-10-01 19:06       ` Linus Torvalds
2009-10-02 12:23     ` Tejun Heo
2009-10-01  8:09 ` [PATCH 04/19] scheduler: implement force_cpus_allowed_ptr() Tejun Heo
2009-10-01  8:09 ` [PATCH 05/19] kthread: implement kthread_data() Tejun Heo
2009-10-01  8:09 ` [PATCH 06/19] acpi: use queue_work_on() instead of binding workqueue worker to cpu0 Tejun Heo
2009-10-01 12:57   ` David Howells
2009-10-01 17:07     ` Tejun Heo
2009-10-01  8:09 ` [PATCH 07/19] stop_machine: reimplement without using workqueue Tejun Heo
2009-10-06  9:36   ` Rusty Russell
2009-10-06 23:42     ` Tejun Heo
2009-10-01  8:09 ` [PATCH 08/19] workqueue: misc/cosmetic updates Tejun Heo
2009-10-01  8:09 ` [PATCH 09/19] workqueue: merge feature parametesr into flags Tejun Heo
2009-10-01  8:09 ` Tejun Heo [this message]
2009-10-01 13:05   ` [PATCH 10/19] workqueue: update cwq alignement and make one more flag bit available David Howells
2009-10-01 16:15     ` Jeff Garzik
2009-10-01 16:20       ` David Howells
2009-10-01 16:30         ` Tejun Heo
2009-10-01 16:39     ` Alan Cox
2009-10-01 18:45     ` Ben Pfaff
2009-10-02 11:56       ` Tejun Heo
2009-10-01  8:09 ` [PATCH 11/19] workqueue: define both bit position and mask for work flags Tejun Heo
2009-10-01  8:09 ` [PATCH 12/19] workqueue: separate out process_one_work() Tejun Heo
2009-10-01  8:09 ` [PATCH 13/19] workqueue: temporarily disable workqueue tracing Tejun Heo
2009-10-01  8:09 ` [PATCH 14/19] workqueue: (TEMPORARY) kill singlethread variant Tejun Heo
2009-10-01  8:09 ` [PATCH 15/19] workqueue: reimplement workqueue flushing using color coded works Tejun Heo
2009-10-01 17:03   ` Linus Torvalds
2009-10-01 17:11     ` Tejun Heo
2009-10-01 17:16       ` Tejun Heo
2009-10-01  8:09 ` [PATCH 16/19] workqueue: introduce worker Tejun Heo
2009-10-01  8:09 ` [PATCH 17/19] workqueue: reimplement work flushing using linked works Tejun Heo
2009-10-01  8:09 ` [PATCH 18/19] workqueue: reimplement workqueue freeze using cwq->frozen_works queue Tejun Heo
2009-10-01  8:09 ` [PATCH 19/19] workqueue: implement concurrency managed workqueue Tejun Heo
2009-10-01 13:15   ` David Howells
2009-10-02 12:03     ` Tejun Heo
2009-10-01 14:49   ` Andrew Morton
2009-10-01 15:12     ` Ingo Molnar
2009-10-01 16:34     ` Tejun Heo
2009-10-04  8:49     ` Peter Zijlstra
2009-10-01 17:12   ` Linus Torvalds
2009-10-01 17:22     ` Tejun Heo
2009-10-01 17:55       ` Linus Torvalds
2009-10-02  0:42   ` Andi Kleen
2009-10-02 12:09     ` Tejun Heo
2009-10-03  2:59       ` Andi Kleen
2009-10-02 14:28   ` Frédéric Weisbecker
2009-10-01  8:24 ` [RFC PATCHSET] " Jens Axboe
2009-10-01 16:36   ` Tejun Heo
2009-10-01  8:24 ` Jens Axboe
2009-10-01 16:25   ` Tejun Heo
2009-10-01  8:40 ` Ingo Molnar
2009-10-01  8:47   ` Jens Axboe
2009-10-01  9:01     ` Ingo Molnar
2009-10-01  9:05       ` Jens Axboe
2009-10-01  9:11   ` Avi Kivity
2009-10-01  9:22     ` Avi Kivity
2009-10-01 16:55     ` Tejun Heo
2009-10-01 17:06       ` Avi Kivity
2009-10-01 16:43   ` Tejun Heo
2009-10-01 12:53 ` David Howells
2009-10-02 11:44   ` Tejun Heo
2009-10-02 12:45     ` Stefan Richter
2009-10-02 15:38     ` David Howells
2009-10-03  5:07       ` Tejun Heo
2009-10-04  8:41 ` Peter Zijlstra

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=1254384558-1018-11-git-send-email-tj@kernel.org \
    --to=tj@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=arjan@linux.intel.com \
    --cc=cl@linux-foundation.org \
    --cc=dhowells@redhat.com \
    --cc=jeff@garzik.org \
    --cc=jens.axboe@oracle.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=rusty@rustcorp.com.au \
    /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).