Messages in this thread | | | From | Roland McGrath <> | Subject | Re: s390 && user_enable_single_step() (Was: odd utrace testing results on s390x) | Date | Thu, 7 Jan 2010 13:44:29 -0800 (PST) |
| |
> Can't understand why do we need TIF_SINGLE_STEP at all.
I think you mean "why we need to set it in do_signal" here, not "why do we need it to exist at all".
> Just pass current->thread.per_info.single_step to > tracehook_signal_handler() ?
Yes. I believe this is what Martin meant, and it's what I meant to endorse.
do_signal should not do anything with TIF_SINGLE_STEP at all. Its only purpose should be to communicate from the low-level trap assembly code up to the return-to-user assembly code so it calls do_single_step.
Thanks, Roland
| |