NoWrite instructions (e.g. cmp or test) never set the "write access"
bit in the error code, even if one of the operands is treated as a
destination.
Fixes: c205fb7d7d4f81e46fc577b707ceb9e356af1456
Cc: Nadav Amit <[email protected]>
Signed-off-by: Paolo Bonzini <[email protected]>
---
arch/x86/kvm/emulate.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 81dcf7964701..a943bf0c06d0 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -4954,7 +4954,8 @@ int x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
rc = segmented_read(ctxt, ctxt->dst.addr.mem,
&ctxt->dst.val, ctxt->dst.bytes);
if (rc != X86EMUL_CONTINUE) {
- if (rc == X86EMUL_PROPAGATE_FAULT &&
+ if (!(ctxt->d & NoWrite) &&
+ rc == X86EMUL_PROPAGATE_FAULT &&
ctxt->exception.vector == PF_VECTOR)
ctxt->exception.error_code |= PFERR_WRITE_MASK;
goto done;
--
1.8.3.1
Sorry for that. I’ll try to run some few more tests next week.
Nadav
Paolo Bonzini <[email protected]> wrote:
> NoWrite instructions (e.g. cmp or test) never set the "write access"
> bit in the error code, even if one of the operands is treated as a
> destination.
>
> Fixes: c205fb7d7d4f81e46fc577b707ceb9e356af1456
> Cc: Nadav Amit <[email protected]>
> Signed-off-by: Paolo Bonzini <[email protected]>
> ---
> arch/x86/kvm/emulate.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
> index 81dcf7964701..a943bf0c06d0 100644
> --- a/arch/x86/kvm/emulate.c
> +++ b/arch/x86/kvm/emulate.c
> @@ -4954,7 +4954,8 @@ int x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
> rc = segmented_read(ctxt, ctxt->dst.addr.mem,
> &ctxt->dst.val, ctxt->dst.bytes);
> if (rc != X86EMUL_CONTINUE) {
> - if (rc == X86EMUL_PROPAGATE_FAULT &&
> + if (!(ctxt->d & NoWrite) &&
> + rc == X86EMUL_PROPAGATE_FAULT &&
> ctxt->exception.vector == PF_VECTOR)
> ctxt->exception.error_code |= PFERR_WRITE_MASK;
> goto done;
> --
> 1.8.3.1
>
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Paolo,
2015-02-09 17:03 GMT+08:00 Paolo Bonzini <[email protected]>:
> NoWrite instructions (e.g. cmp or test) never set the "write access"
> bit in the error code, even if one of the operands is treated as a
> destination.
Sorry to reply to an old commit, btw, could you point out where in SDM
describe above?
Regards,
Wanpeng Li
2015-11-20 10:52 GMT+08:00 Wanpeng Li <[email protected]>:
> Hi Paolo,
> 2015-02-09 17:03 GMT+08:00 Paolo Bonzini <[email protected]>:
>> NoWrite instructions (e.g. cmp or test) never set the "write access"
>> bit in the error code, even if one of the operands is treated as a
>> destination.
>
> Sorry to reply to an old commit, btw, could you point out where in SDM
> describe above?
I see.
Wanpeng Li <[email protected]> wrote:
> 2015-11-20 10:52 GMT+08:00 Wanpeng Li <[email protected]>:
>> Hi Paolo,
>> 2015-02-09 17:03 GMT+08:00 Paolo Bonzini <[email protected]>:
>>> NoWrite instructions (e.g. cmp or test) never set the "write access"
>>> bit in the error code, even if one of the operands is treated as a
>>> destination.
>>
>> Sorry to reply to an old commit, btw, could you point out where in SDM
>> describe above?
>
> I see.
I don’t understand whether you still need my help, so to clarify: on a
page-fault the error code should indicate whether the access was due to a
write access. Previously KVM marked it as “write access” for instructions
such as test and cmp that do not perform write.
Nadav-