blob: 57c2b94e59cbbb92aba017f16d8daeb98745e667 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
|
From: Thomas Gleixner <tglx@linutronix.de>
Date: Fri, 7 Dec 2007 18:16:17 +0000 (+0100)
Subject: hrtimers: avoid overflow for large relative timeouts
X-Git-Tag: v2.6.24-rc5~49^2~2
X-Git-Url: http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Ftorvalds%2Flinux-2.6.git;a=commitdiff_plain;h=62f0f61e6673e67151a7c8c0f9a09c7ea43fe2b5;hp=f194d132e4971111f85c18c96067acffb13cee6d
hrtimers: avoid overflow for large relative timeouts
Relative hrtimers with a large timeout value might end up as negative
timer values, when the current time is added in hrtimer_start().
This in turn is causing the clockevents_set_next() function to set an
huge timeout and sleep for quite a long time when we have a clock
source which is capable of long sleeps like HPET. With PIT this almost
goes unnoticed as the maximum delta is ~27ms. The non-hrt/nohz code
sorts this out in the next timer interrupt, so we never noticed that
problem which has been there since the first day of hrtimers.
This bug became more apparent in 2.6.24 which activates HPET on more
hardware.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
Adjusted to apply to Debian's 2.6.18 by dann frazier <dannf@debian.org>
diff -urpN linux-source-2.6.18.orig/kernel/hrtimer.c linux-source-2.6.18/kernel/hrtimer.c
--- linux-source-2.6.18.orig/kernel/hrtimer.c 2006-09-19 21:42:06.000000000 -0600
+++ linux-source-2.6.18/kernel/hrtimer.c 2007-12-16 18:43:03.000000000 -0700
@@ -443,6 +443,14 @@ hrtimer_start(struct hrtimer *timer, kti
#ifdef CONFIG_TIME_LOW_RES
tim = ktime_add(tim, base->resolution);
#endif
+ /*
+ * Careful here: User space might have asked for a
+ * very long sleep, so the add above might result in a
+ * negative number, which enqueues the timer in front
+ * of the queue.
+ */
+ if (tim.tv64 < 0)
+ tim.tv64 = KTIME_MAX;
}
timer->expires = tim;
|