Jan Kiszka wrote:
> just ran into an ugly warning of I-pipe as I happened to have
> CONFIG_PERF_EVENTS enabled in 2.6.32 kernel. That pulls some plain Linux
> spinlock into the finish_task_switch path.
> We should either disable this feature at I-pipe level or patch it
> properly. Haven't checked yet if the latter is feasible with moderate
> effort, though.
At least this one is now explainable: There is no problem related to
CONFIG_PERF_EVENTS. finish_task_switch is not part of the critical
shadow task switch, and the oops I saw finally turned out to be caused
by my buggy virtual hardware (IOW: an unrelated hypervisor bug). Nothing
to worry about for Xenomai or I-pipe.
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux
Xenomai-core mailing list