if CONFIG_HIGH_RES_TIMERS is not configured, still the hrtimer-softirq
is started, so to prevent the name 'softirq-NULL' in 'ps' output we
should still define its name here.
Preventing the softirq from starting at all would touch much more code, so
I chose the easy way here, it is a cosmetic fix anyway.
Signed-off-by: Remy Bohmer <[email protected]>
---
kernel/softirq.c | 2 --
1 files changed, 0 insertions(+), 2 deletions(-)
diff --git a/kernel/softirq.c b/kernel/softirq.c
index aae8d45..590049c 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -1139,9 +1139,7 @@ static const char *softirq_names [] =
[NET_RX_SOFTIRQ] = "net-rx",
[BLOCK_SOFTIRQ] = "block",
[TASKLET_SOFTIRQ] = "tasklet",
-#ifdef CONFIG_HIGH_RES_TIMERS
[HRTIMER_SOFTIRQ] = "hrtimer",
-#endif
[RCU_SOFTIRQ] = "rcu",
};
--
1.6.0.4
2009/9/24 Remy Bohmer <[email protected]>:
> if CONFIG_HIGH_RES_TIMERS is not configured, still the hrtimer-softirq
> is started, so to prevent the name 'softirq-NULL' in 'ps' output we
> should still define its name here.
> Preventing the softirq from starting at all would touch much more code, so
> I chose the easy way here, it is a cosmetic fix anyway.
>
> Signed-off-by: Remy Bohmer <[email protected]>
oops... Forgot to mention that this patch applies to 2.6.31-rt11 (preempt-RT)
> ---
> ?kernel/softirq.c | ? ?2 --
> ?1 files changed, 0 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/softirq.c b/kernel/softirq.c
> index aae8d45..590049c 100644
> --- a/kernel/softirq.c
> +++ b/kernel/softirq.c
> @@ -1139,9 +1139,7 @@ static const char *softirq_names [] =
> ? [NET_RX_SOFTIRQ] ? ? = "net-rx",
> ? [BLOCK_SOFTIRQ] ? ? ?= "block",
> ? [TASKLET_SOFTIRQ] ? ?= "tasklet",
> -#ifdef CONFIG_HIGH_RES_TIMERS
> ? [HRTIMER_SOFTIRQ] ? ?= "hrtimer",
> -#endif
> ? [RCU_SOFTIRQ] ? ? ? ? ? ? ? ?= "rcu",
> ?};
>
> --
> 1.6.0.4
>
>
On Thu, Sep 24, 2009 at 10:43 PM, Remy Bohmer <[email protected]> wrote:
> if CONFIG_HIGH_RES_TIMERS is not configured, still the hrtimer-softirq
> is started, so to prevent the name 'softirq-NULL' in 'ps' output we
> should still define its name here.
> Preventing the softirq from starting at all would touch much more code, so
> I chose the easy way here, it is a cosmetic fix anyway.
Anyway, it would be the right thing to do just that.
Just un-define the enum HRTIMER_SOFTIRQ and the code that needs
modification should pop up. Removing it from the enum also disables
the kthread. Most of the code that raises the softirq should be
guarded by CONFIG_HIGH_RES_TIMERS anyway.
HtH,
Jan