Indan Zupancic
Indan Zupancic
For non-hyp user space would get a `seL4_Fault_UserException` with no extra info at all, that doesn't seem much better overall. Ideally ESR is always passed to user space, also for...
> > For non-hyp user space would get a seL4_Fault_UserException with no extra info at all, that doesn't seem much better overall. > > That's not true, you usually get...
I just ran into this issue, it's so lovely to waste an hour of your life debugging what the hell is going on. :-(
Depends on the compiler, I've seen both those and IP relative loads in generated ASM. Slow is all relative here, it will never be "extra SDRAM read" slow, at worst...
I see I have been too slow. Please fix issues in additional PR.
Implemented by #1183.
Well, I thought we didn't plan on supporting this for GICv2: https://github.com/seL4/seL4/pull/1183#issuecomment-1922355618. From your post, it doesn't seem worth the hassle.
It's x86, you can get an NMI at any time which will take unknown amount of time to be handled by the system firmware. Still, 2 ms deviation seems like...
This might have been fixed with my sel4bench bugfixes recently, but either way haven't seen this for a while. Should we close this?
> I have also not seen this in a very long time. Would it be possible to run a stress test to confirm? (E.g. that one test a few hundred...