linux-kernel - Re: [PATCH, RFC] reimplement flush

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 5 Jan 2007 17:07:17 +0300
From:	Oleg Nesterov <oleg@...sign.ru>
To:	Srivatsa Vaddagiri <vatsa@...ibm.com>
Cc:	Andrew Morton <akpm@...l.org>, David Howells <dhowells@...hat.com>,
	Christoph Hellwig <hch@...radead.org>,
	Ingo Molnar <mingo@...e.hu>,
	Linus Torvalds <torvalds@...l.org>,
	linux-kernel@...r.kernel.org, Gautham shenoy <ego@...ibm.com>
Subject: Re: [PATCH, RFC] reimplement flush_workqueue()

On 01/05, Srivatsa Vaddagiri wrote:
>
> On Thu, Jan 04, 2007 at 10:31:07AM -0800, Andrew Morton wrote:
> > But before we do much more of this we should have a wrapper.  Umm
> > 
> > static inline void block_cpu_hotplug(void)
> > {
> > 	preempt_disable();
> > }
> 
> Nack.
> 
> This will only block cpu down, not cpu_up and hence is a misnomer. I would be 
> vary of ignoring cpu_up events totally in writing hotplug safe code.

How about block_cpu_down() ?

These cpu-hotplug races delayed the last workqueue patch I have in my queue.
flush_workqueue() misses an important optimization: we don't need to insert
a barrier and have an extra wake_up + wait_for_completion when cwq has no
pending works. But we need ->current_work (introduced in the next patch) to
implement this correctly.

I'll re-send the patch below later, when we finish with the bug you pointed
out, but it would be nice if you can take a look now.

Oleg.

--- mm-6.20-rc2/kernel/workqueue.c~4_speedup	2006-12-30 18:09:07.000000000 +0300
+++ mm-6.20-rc2/kernel/workqueue.c	2007-01-05 16:32:45.000000000 +0300
@@ -405,12 +405,15 @@ static void wq_barrier_func(struct work_
 	complete(&barr->done);
 }
 
-static inline void init_wq_barrier(struct wq_barrier *barr)
+static void insert_wq_barrier(struct cpu_workqueue_struct *cwq,
+					struct wq_barrier *barr, int tail)
 {
 	INIT_WORK(&barr->work, wq_barrier_func);
 	__set_bit(WORK_STRUCT_PENDING, work_data_bits(&barr->work));
 
 	init_completion(&barr->done);
+
+	insert_work(cwq, &barr->work, tail);
 }
 
 static void flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
@@ -425,13 +428,20 @@ static void flush_cpu_workqueue(struct c
 		mutex_lock(&workqueue_mutex);
 	} else {
 		struct wq_barrier barr;
+		int active = 0;
 
-		init_wq_barrier(&barr);
-		__queue_work(cwq, &barr.work);
+		spin_lock_irq(&cwq->lock);
+		if (!list_empty(&cwq->worklist) || cwq->current_work != NULL) {
+			insert_wq_barrier(cwq, &barr, 1);
+			active = 1;
+		}
+		spin_unlock_irq(&cwq->lock);
 
-		mutex_unlock(&workqueue_mutex);
-		wait_for_completion(&barr.done);
-		mutex_lock(&workqueue_mutex);
+		if (active) {
+			mutex_unlock(&workqueue_mutex);
+			wait_for_completion(&barr.done);
+			mutex_lock(&workqueue_mutex);
+		}
 	}
 }
 
@@ -478,8 +488,7 @@ static void wait_on_work(struct cpu_work
 
 	spin_lock_irq(&cwq->lock);
 	if (unlikely(cwq->current_work == work)) {
-		init_wq_barrier(&barr);
-		insert_work(cwq, &barr.work, 0);
+		insert_wq_barrier(cwq, &barr, 0);
 		running = 1;
 	}
 	spin_unlock_irq(&cwq->lock);

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/