|
From: | Paolo Bonzini |
Subject: | Re: [Qemu-devel] [PATCH qom-cpu v2 09/40] cpu: Turn cpu_handle_mmu_fault() into a CPUClass hook |
Date: | Wed, 12 Mar 2014 18:58:59 +0100 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.2.0 |
Il 12/03/2014 00:47, Andreas Färber ha scritto:
> I'm not sure I understand why you omit the hook for *-softmmu. It > doesn't hurt to have it, does it?Actually I wondered the same thing while preparing v2 :) and prepared an alternate version, but it did not really help cut down on complexity.
No, I agree it doesn't. Just in case common non-user-only code will need it in the future.
The hook is needed for dispatching from common code to target code, and that part is user-only. Some targets but not all reuse the user function from softmmu code, others have two different implementations, some don't reuse at all. It's a mess.
Ok, that explains it. :) Paolo
[Prev in Thread] | Current Thread | [Next in Thread] |