2006-01-17 19:16:31

by john stultz

Subject: [PATCH] Time: Delay clocksource selection until later in boot

Delay installing new clocksources until later in boot. This avoids some
of the clocksource churn that can occur at boot, possibly allowing the
system to run for a brief time with a bad clocksource.

This patch resolves the boot time stalls seen by Mattia Dongili.

thanks
-john

Signed-off-by: John Stultz <[email protected]>


diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c
--- a/kernel/time/clocksource.c
+++ b/kernel/time/clocksource.c
@@ -51,13 +51,27 @@ static LIST_HEAD(clocksource_list);
static DEFINE_SPINLOCK(clocksource_lock);
static char override_name[32];

+static int finished_booting;
+
+/* clocksource_done_booting - Called near the end of bootup
+ *
+ * Hack to avoid lots of clocksource churn at boot time
+ */
+static int clocksource_done_booting(void)
+{
+ finished_booting = 1;
+ return 0;
+}
+
+late_initcall(clocksource_done_booting);
+
/**
* get_next_clocksource - Returns the selected clocksource
*/
struct clocksource *get_next_clocksource(void)
{
spin_lock(&clocksource_lock);
- if (next_clocksource) {
+ if (next_clocksource && finished_booting) {
curr_clocksource = next_clocksource;
next_clocksource = NULL;
}

2006-01-17 23:41:18

by Andrew Morton

Subject: Re: [PATCH] Time: Delay clocksource selection until later in boot

john stultz <[email protected]> wrote:
>
> Delay installing new clocksources until later in boot. This avoids some
> of the clocksource churn that can occur at boot, possibly allowing the
> system to run for a brief time with a bad clocksource.
>
> This patch resolves the boot time stalls seen by Mattia Dongili.
>

Except he now thinks that it doesn't. Do you think the patch is good anyway?

2006-01-17 23:48:04

by john stultz

Subject: Re: [PATCH] Time: Delay clocksource selection until later in boot

On Tue, 2006-01-17 at 15:43 -0800, Andrew Morton wrote:
> john stultz <[email protected]> wrote:
> >
> > Delay installing new clocksources until later in boot. This avoids some
> > of the clocksource churn that can occur at boot, possibly allowing the
> > system to run for a brief time with a bad clocksource.
> >
> > This patch resolves the boot time stalls seen by Mattia Dongili.
> >
>
> Except he now thinks that it doesn't. Do you think the patch is good anyway?

I actually do still think its a good idea to avoid the bootup
clocksource churn. I've still got a few ideas for further improving the
TSC registration so I'll probably have another patch or two to try later
this week.

thanks
-john