From: Andi Kleen <ak@suse.de>
To: linux-kernel@vger.kernel.org, mingo@elte.hu
Subject: [PATCH] [5/6] scheduler: Protect important kernel threads against normalize_rt
Date: Sun, 7 Oct 2007 22:59:58 +0200 (CEST) [thread overview]
Message-ID: <20071007205958.7D8C41474B@wotan.suse.de> (raw)
In-Reply-To: <200710071059.126674000@suse.de>
Not only the migration thread, but also softlockup and stop machine
need to be protected against normalize_rt(). Instead of checking
for them all I added a new process flag for this.
This was the last available bit in 32bit task->flags, sorry for that.
This removes the is_migration_thread macro added in the last patch
again because it is not needed anymore.
Signed-off-by: Andi Kleen <ak@suse.de>
Index: linux-2.6-sched-devel/include/linux/sched.h
===================================================================
--- linux-2.6-sched-devel.orig/include/linux/sched.h
+++ linux-2.6-sched-devel/include/linux/sched.h
@@ -1330,6 +1330,7 @@ static inline void put_task_struct(struc
#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
+#define PF_RT_PROTECTED 0x80000000 /* RT task protected from sysrq */
/*
* Only the _current_ task can read/write to tsk->flags, but other
Index: linux-2.6-sched-devel/kernel/sched.c
===================================================================
--- linux-2.6-sched-devel.orig/kernel/sched.c
+++ linux-2.6-sched-devel/kernel/sched.c
@@ -74,12 +74,6 @@ unsigned long long __attribute__((weak))
return (unsigned long long)jiffies * (1000000000 / HZ);
}
-#ifdef CONFIG_SMP
-#define is_migration_thread(p, rq) ((p) == (rq)->migration_thread)
-#else
-#define is_migration_thread(p, rq) 0
-#endif
-
/*
* Convert user-nice values [ -20 ... 0 ... 19 ]
* to static priority [ MAX_RT_PRIO..MAX_PRIO-1 ],
@@ -5305,6 +5299,7 @@ migration_call(struct notifier_block *nf
kthread_bind(p, cpu);
/* Must be high prio: stop_machine expects to yield to it. */
rq = task_rq_lock(p, &flags);
+ p->flags |= PF_RT_PROTECTED;
__setscheduler(rq, p, SCHED_FIFO, MAX_RT_PRIO-1);
task_rq_unlock(rq, &flags);
cpu_rq(cpu)->migration_thread = p;
@@ -6583,7 +6578,7 @@ void normalize_rt_tasks(void)
spin_lock_irqsave(&p->pi_lock, flags);
rq = __task_rq_lock(p);
- if (!is_migration_thread(p, rq))
+ if (!(p->flags & PF_RT_PROTECTED))
normalize_task(rq, p);
__task_rq_unlock(rq);
Index: linux-2.6-sched-devel/kernel/softlockup.c
===================================================================
--- linux-2.6-sched-devel.orig/kernel/softlockup.c
+++ linux-2.6-sched-devel/kernel/softlockup.c
@@ -116,6 +116,7 @@ static int watchdog(void * __bind_cpu)
{
struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
+ current->flags |= PF_RT_PROTECTED;
sched_setscheduler(current, SCHED_FIFO, ¶m);
/* initialize timestamp */
Index: linux-2.6-sched-devel/kernel/stop_machine.c
===================================================================
--- linux-2.6-sched-devel.orig/kernel/stop_machine.c
+++ linux-2.6-sched-devel/kernel/stop_machine.c
@@ -187,6 +187,7 @@ struct task_struct *__stop_machine_run(i
if (!IS_ERR(p)) {
struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
+ p->flags |= PF_RT_PROTECTED;
/* One high-prio thread per cpu. We'll do this one. */
sched_setscheduler(p, SCHED_FIFO, ¶m);
kthread_bind(p, cpu);
next prev parent reply other threads:[~2007-10-07 21:01 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-10-07 20:59 [PATCH] [0/6] Some scheduler changes for sched-devel Andi Kleen
2007-10-07 20:59 ` [PATCH] [1/6] scheduler: Remove some unnecessary gotos in sched.c Andi Kleen
2007-10-08 11:36 ` Ingo Molnar
2007-10-09 19:17 ` Ingo Molnar
2007-10-10 0:55 ` Andi Kleen
2007-10-10 11:25 ` Ingo Molnar
2007-10-10 11:26 ` Ingo Molnar
2007-10-07 20:59 ` [PATCH] [2/6] scheduler: Refactor common code of sleep_on / wait_for_completion Andi Kleen
2007-10-07 22:22 ` [PATCH] [2/6] scheduler: Refactor common code of sleep_on / wait_for_completion v2 Andi Kleen
2007-10-08 11:39 ` Ingo Molnar
2007-10-08 12:03 ` Ingo Molnar
2007-10-07 20:59 ` [PATCH] [3/6] scheduler: Do devirtualization for sched_fair Andi Kleen
2007-10-08 11:42 ` Ingo Molnar
2007-10-08 12:32 ` Andi Kleen
2007-10-08 12:39 ` Ingo Molnar
2007-10-08 14:33 ` Andi Kleen
2007-10-07 20:59 ` [PATCH] [4/6] scheduler: Refactor normalize_rt_tasks Andi Kleen
2007-10-08 11:44 ` Ingo Molnar
2007-10-07 20:59 ` Andi Kleen [this message]
2007-10-08 11:51 ` [PATCH] [5/6] scheduler: Protect important kernel threads against normalize_rt Ingo Molnar
2007-10-08 12:33 ` Andi Kleen
2007-10-08 12:43 ` Ingo Molnar
2007-10-08 13:08 ` Andi Kleen
2007-10-07 20:59 ` [PATCH] [6/6] scheduler: Remove bogus comment in sched_group_set_shares Andi Kleen
2007-10-08 11:52 ` 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=20071007205958.7D8C41474B@wotan.suse.de \
--to=ak@suse.de \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
/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
Powered by JetHome