Messages in this thread | | | Date | Thu, 12 Jan 2012 18:14:36 +0900 | From | Takuya Yoshikawa <> | Subject | Re: [PATCH 1/2] KVM: Exception during emulation decode should propagate |
| |
(2012/01/12 18:07), Nadav Amit wrote: > > On Jan 12, 2012, at 2:26 AM, Takuya Yoshikawa wrote: > >> (2012/01/12 7:11), Takuya Yoshikawa wrote: >>> On Wed, 11 Jan 2012 18:53:30 +0200 >>> Nadav Amit<namit@cs.technion.ac.il> wrote: >>> >>>> An exception might occur during decode (e.g., #PF during fetch). >>>> Currently, the exception is ignored and emulation is performed. >> >> Note that the decode/emulation will not be continued in such a case. >> >> insn_fetch() is a bit tricky macro and it contains "goto done" to outside. >> So if an error happens during fetching the instruction, x86_decode_insn() >> will handle the X86EMUL_* fault value and returns FAIL immediately. > > You got a point. Yet, a problem still exists. > I now notice I was originally working on previous version (3.0.0) > where the return-code of x86_decode_insn is handled differently. > Nonetheless, I think the current implementation might report emulation > error in such a scenario (instead of triggering #PF/#GP in the guest).
It's me who did that fix.
> >>> >>> When I cleaned up insn_fetch(), I thought that fetching the instruction >>> which is being executed by the guest cannot cause #PF. >>> >>> The possibility that a meaningless userspace might similtaneously unmap >>> the page, noted by Avi IIRC, was ignored intentionally, so we just fail >>> in such a case. >>> >>> Did you see any real problem? > > Well, I run some research project for which I emulate instructions quite > often. I do see a real problem with Linux 3.0.0. Please note AFAIK #GP > might occur as well during instruction fetch. I don't think failing is the > right behavior in such case - there is no real reason to fail. > > Please tell me whether you are OK with KVM failing in such a scenario. > If not - I'll send an updated patch (in which x86_decode_insn returns > EMULATION_OK when rc == X86EMUL_PROPAGATE_FAULT).
You need comments from maintainers.
Takuya
| |