Thank you Heinrich,
partially the scenario (like lack of image to boot) is an artifact of me 
simplifying the testcase (as one can see going through the case above). But it 
hit that issue in riscv_pmu_icount_update_priv even when I did provide content 
to boot into - so I think all of your analysis is still valid.

So it comes down to the """riscv_cpu_set_mode(env, PRV_M, virt) is executed in
riscv_cpu_do_interrupt(). When trapping to M-mode virt should always be false 
as M-mode is not
virtualized. virt should always be false if the hypervisor extension (h) is 
disabled.""" part of your analysis.
Are you pondering about what a patch for your finding could look like or 
consider to report that upstream?

But independent to that I wonder:
- why can I show it working fine with qemu 10.0?
- why only on ppc64?
There seems to be a variable in this equation we just have not seen yet.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2120835

Title:
  EDK2 tests for riscv emulation fail against qemu 10.1 on ppc64

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/edk2/+bug/2120835/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to