https://bugs.kde.org/show_bug.cgi?id=126348

--- Comment #4 from Julian Seward <[email protected]> ---
Vince, thanks for the patch.  I have to say I am a little concerned
about landing it as-is, due to capability-check issues.

IIUC (and correct me if I am wrong): rdpmc support is something that
is either enabled or disabled at a process granularity (or larger), but
doesn't change whilst a process runs.

My concern here is that if we execute (as a guest insn, that is,
simulate) a rdpmc, then with your patch we'll simply pass that through
to the host.  If the host doesn't support that then we'll get a SIGILL
on the host, which isn't what we want -- we need to deliver a SIGILL in
the guest context.

What this amounts to is that V needs to test at startup whether the 
host can do rdpmc, and if it can't, then we simply decline to decode
the insn in the normal way and synthesise a SIGILL for the guest.

Is the above a correct understanding?  If so I can try to hack up the
relevant hwcaps checking stuff.  A framework for hwcaps is already in
place, so adding this one feature shouldn't be too much hassle.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to