Skip to contentSkip to conversation

itimers: fix racy writes to cpu_itimer fields

PATCH In review Stanislaw Gruszka 3 participants 8 messages Sep 24, 2009 – Nov 18, 2009

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 itimers: fix racy writes to cpu_itimer fields

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

---
 kernel/itimer.c |    7 +++++--
 1 files changed, 5 insertions(+), 2 deletions(-)
kernel/itimer.c +5 -2
index b03451e..d802883 100644
@@ -146,6 +146,7 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 {
 	cputime_t cval, nval, cinterval, ninterval;
 	s64 ns_ninterval, ns_nval;
+	u32 error, incr_error;
 	struct cpu_itimer *it = &tsk->signal->it[clock_id];
 
 	nval = timeval_to_cputime(&value->it_value);
@@ -153,8 +154,8 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 	ninterval = timeval_to_cputime(&value->it_interval);
 	ns_ninterval = timeval_to_ns(&value->it_interval);
 
-	it->incr_error = cputime_sub_ns(ninterval, ns_ninterval);
-	it->error = cputime_sub_ns(nval, ns_nval);
+	error = cputime_sub_ns(nval, ns_nval);
+	incr_error = cputime_sub_ns(ninterval, ns_ninterval);
 
 	spin_lock_irq(&tsk->sighand->siglock);
 
@@ -168,6 +169,8 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 	}
 	it->expires = nval;
 	it->incr = ninterval;
+	it->error = error;
+	it->incr_error = incr_error;
 	trace_itimer_state(clock_id == CPUCLOCK_VIRT ?
 			   ITIMER_VIRTUAL : ITIMER_PROF, value, nval);
 
Signature · 1 line
1.6.2.5
#1 lore raw

PATCH 2/2 posix-cpu-timers: initialize new_itimer->it.cpu.firing

In case when posix_cpu_timer_create() is called from sys_timer_create() it.cpu.firing field of struct k_itimer is uninitialized.

---
 kernel/posix-cpu-timers.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
kernel/posix-cpu-timers.c +1 -0
index 5c9dc22..e73a386 100644
@@ -398,6 +398,7 @@ int posix_cpu_timer_create(struct k_itimer *new_timer)
 	INIT_LIST_HEAD(&new_timer->it.cpu.entry);
 	new_timer->it.cpu.incr.sched = 0;
 	new_timer->it.cpu.expires.sched = 0;
+	new_timer->it.cpu.firing = 0;
 
 	read_lock(&tasklist_lock);
 	if (CPUCLOCK_PERTHREAD(new_timer->it_clock)) {
Signature · 1 line
1.6.2.5
#2 lore raw
In reply to Stanislaw Gruszka
Quoted 3 lines · Stanislaw Gruszka

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

Won't it be all-round much better to convert these things to hrtimers instead of adding more and more fuzz on top to make them deal with jiffies?

#3 lore raw
Quoted 8 lines · On Thu, 24 Sep 2009 16:48:07 +0200 Peter Zijlstra <[email protected]>…
Quoted 3 lines · Stanislaw Gruszka

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

Won't it be all-round much better to convert these things to hrtimers instead of adding more and more fuzz on top to make them deal with jiffies?

Perhaps it would, but I don't know how to do it :{ . Especially how to precisely account user time. The only idea I have is make something like microstate accounting (http://lwn.net/Articles/127296/), but this patch and whole idea was rejected long time ago.

Stanislaw

#4 lore raw
Quoted 8 lines · Stanislaw Gruszka
Quoted 5 lines · On Thu, 24 Sep 2009 16:48:07 +0200 Peter Zijlstra <[email protected]>…
Quoted 3 lines · Stanislaw Gruszka

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

Quoted 8 lines
Quoted 3 lines

Won't it be all-round much better to convert these things to hrtimers instead of adding more and more fuzz on top to make them deal with jiffies?

Perhaps it would, but I don't know how to do it :{ . Especially how to precisely account user time. The only idea I have is make something like microstate accounting (http://lwn.net/Articles/127296/), but this patch and whole idea was rejected long time ago.

That patch does look a little painful indeed.

I was more thinking about about looking if an itimer was to expire less than 1 tick away on either sched-in or the tick.

When we find it is indeed less than 1 tick away, program an hrtimer for that cpu to expire at the required moment, see hrtick_start().

If we happen to de-schedule the task before the timer fires, we clear the hrtimer again (or let it pend and ignore the fire), see hrtick_clear().

[ there is no reason to rely on the tick though, we can program the hrtimer on sched in to expire on at the right moment, and do so on each schedule for as long as an itimer is active - re-setting whatever pending timer the cpu still had. ]

#5 lore raw
Quoted 26 lines · Peter Zijlstra
Quoted 8 lines · Stanislaw Gruszka
Quoted 5 lines · On Thu, 24 Sep 2009 16:48:07 +0200 Peter Zijlstra <[email protected]>…
Quoted 3 lines · Stanislaw Gruszka

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

Quoted 8 lines
Quoted 3 lines

Won't it be all-round much better to convert these things to hrtimers instead of adding more and more fuzz on top to make them deal with jiffies?

Perhaps it would, but I don't know how to do it :{ . Especially how to precisely account user time. The only idea I have is make something like microstate accounting (http://lwn.net/Articles/127296/), but this patch and whole idea was rejected long time ago.

That patch does look a little painful indeed.

I was more thinking about about looking if an itimer was to expire less than 1 tick away on either sched-in or the tick.

When we find it is indeed less than 1 tick away, program an hrtimer for that cpu to expire at the required moment, see hrtick_start().

Quoted 8 lines

If we happen to de-schedule the task before the timer fires, we clear the hrtimer again (or let it pend and ignore the fire), see hrtick_clear().

[ there is no reason to rely on the tick though, we can program the hrtimer on sched in to expire on at the right moment, and do so on each schedule for as long as an itimer is active - re-setting whatever pending timer the cpu still had. ]

we should think about the simplest approach: switching itimers to hrtimers.

Ingo

#6 lore raw
Quoted 7 lines · Ingo Molnar
Quoted 4 lines

[ there is no reason to rely on the tick though, we can program the hrtimer on sched in to expire on at the right moment, and do so on each schedule for as long as an itimer is active - re-setting whatever pending timer the cpu still had. ]

we should think about the simplest approach: switching itimers to hrtimers.

I agree we need better implementation of posix-cpu-timers, however as far we have this not perfect code and two patches fix bugs on it. Not critical bugs, but I think they should be applied. Have I resend them ?

Stanislaw

#7 lore raw

[tip:timers/urgent] itimers: Fix racy writes to cpu_itimer fields

In reply to Stanislaw Gruszka
Commit-ID:  8747d793fc5c4d3e4decd41d55f6dc24498dd5f5
Gitweb:     http://git.kernel.org/tip/8747d793fc5c4d3e4decd41d55f6dc24498dd5f5
Author:     Stanislaw Gruszka <[email protected]>
AuthorDate: Tue, 17 Nov 2009 14:14:12 -0800
Committer:  Thomas Gleixner <[email protected]>
CommitDate: Wed, 18 Nov 2009 16:32:12 +0100

itimers: Fix racy writes to cpu_itimer fields

incr_error and error fields of struct cpu_itimer are used when calculating next timer tick in check_cpu_itimers() and should not be modified without tsk->sighand->siglock taken.

LKML-Reference: <[email protected]>

Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Oleg Nesterov <[email protected]>
Cc Peter Zijlstra <[email protected]>
---
 kernel/itimer.c |    7 +++++--
 1 files changed, 5 insertions(+), 2 deletions(-)
kernel/itimer.c +5 -2
index b03451e..d802883 100644
@@ -146,6 +146,7 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 {
 	cputime_t cval, nval, cinterval, ninterval;
 	s64 ns_ninterval, ns_nval;
+	u32 error, incr_error;
 	struct cpu_itimer *it = &tsk->signal->it[clock_id];
 
 	nval = timeval_to_cputime(&value->it_value);
@@ -153,8 +154,8 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 	ninterval = timeval_to_cputime(&value->it_interval);
 	ns_ninterval = timeval_to_ns(&value->it_interval);
 
-	it->incr_error = cputime_sub_ns(ninterval, ns_ninterval);
-	it->error = cputime_sub_ns(nval, ns_nval);
+	error = cputime_sub_ns(nval, ns_nval);
+	incr_error = cputime_sub_ns(ninterval, ns_ninterval);
 
 	spin_lock_irq(&tsk->sighand->siglock);
 
@@ -168,6 +169,8 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 	}
 	it->expires = nval;
 	it->incr = ninterval;
+	it->error = error;
+	it->incr_error = incr_error;
 	trace_itimer_state(clock_id == CPUCLOCK_VIRT ?
 			   ITIMER_VIRTUAL : ITIMER_PROF, value, nval);