Messages in this thread |  | | | Date | Sun, 2 Aug 2009 22:46:47 +0200 | | From | Ingo Molnar <> | | Subject | Re: [tip:core/debug] debug lockups: Improve lockup detection |
| |
* Andrew Morton <akpm@linux-foundation.org> wrote:
> #ifndef trigger_all_cpu_backtrace > +#ifdef CONFIG_SMP > +#warning This architecture is missing a trigger_all_cpu_backtrace() implementation > +#endif > #define trigger_all_cpu_backtrace() do { } while (0) > #endif
I think a better solution will be to make this function return 1 if it generated a backtrace.
That way we can fall back to the generic schedule_work based codepath.
Albeit the best and cleanest solution would be to implement a workqueue based trigger_all_cpu_backtrace() __weak variant and throw away the workqueue bits from sysrq.c.
Ingo
|  |