KVM: x86: fix failure of injecting exceptions in x86_emulate_instruction
diff mbox series

Message ID 1536220964-8412-1-git-send-email-wang.yi59@zte.com.cn
State New
Headers show
Series
  • KVM: x86: fix failure of injecting exceptions in x86_emulate_instruction
Related show

Commit Message

Yi Wang Sept. 6, 2018, 8:02 a.m. UTC
In order to fix a page table walk issue, commit 6ea6e84309ca
("KVM: x86: inject exceptions produced by x86_decode_insn") check
if variable ctxt->have_exception true and inject the exception.
Unfortunately, ctxt->have_exception is set to true only in function
x86_emulate_insn(), which won't be called before invoking
inject_emulated_exception() in the 6ea6e84309ca.
This patch fix this issue.

Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
Reviewed-by: Xi Xu <xu.xi8@zte.com.cn>
---
 arch/x86/kvm/emulate.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Comments

Paolo Bonzini Sept. 20, 2018, 4:30 p.m. UTC | #1
On 06/09/2018 10:02, Yi Wang wrote:
> In order to fix a page table walk issue, commit 6ea6e84309ca
> ("KVM: x86: inject exceptions produced by x86_decode_insn") check
> if variable ctxt->have_exception true and inject the exception.
> Unfortunately, ctxt->have_exception is set to true only in function
> x86_emulate_insn(), which won't be called before invoking
> inject_emulated_exception() in the 6ea6e84309ca.
> This patch fix this issue.
> 
> Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
> Reviewed-by: Xi Xu <xu.xi8@zte.com.cn>

Queued, thanks.  Do you have a testcase for this?

> ---
>  arch/x86/kvm/emulate.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
> index 106482d..aecf9a72 100644
> --- a/arch/x86/kvm/emulate.c
> +++ b/arch/x86/kvm/emulate.c
> @@ -5105,8 +5105,11 @@ int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len)
>  		memcpy(ctxt->fetch.data, insn, insn_len);
>  	else {
>  		rc = __do_insn_fetch_bytes(ctxt, 1);
> -		if (rc != X86EMUL_CONTINUE)
> +		if (rc != X86EMUL_CONTINUE) {
> +			if (rc == X86EMUL_PROPAGATE_FAULT)
> +				ctxt->have_exception = true;
>  			return rc;
> +		}
>  	}
>  
>  	switch (mode) {
>
Paolo Bonzini Sept. 20, 2018, 4:53 p.m. UTC | #2
On 20/09/2018 18:30, Paolo Bonzini wrote:
> On 06/09/2018 10:02, Yi Wang wrote:
>> In order to fix a page table walk issue, commit 6ea6e84309ca
>> ("KVM: x86: inject exceptions produced by x86_decode_insn") check
>> if variable ctxt->have_exception true and inject the exception.
>> Unfortunately, ctxt->have_exception is set to true only in function
>> x86_emulate_insn(), which won't be called before invoking
>> inject_emulated_exception() in the 6ea6e84309ca.
>> This patch fix this issue.
>>
>> Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
>> Reviewed-by: Xi Xu <xu.xi8@zte.com.cn>
> 
> Queued, thanks.  Do you have a testcase for this?

Unqueued, sorry.  The hypercall test from kvm-unit-tests fails.  A
VMCALL on the "edge" of canonical address space, i.e. at 0x7ffffffffffd,
raises a general protection fault before this patch and a double fault
afterwards.

Paolo
Sean Christopherson Sept. 20, 2018, 5:50 p.m. UTC | #3
+cc Peng Hao and Jiang Biao

On Thu, Sep 20, 2018 at 06:53:55PM +0200, Paolo Bonzini wrote:
> On 20/09/2018 18:30, Paolo Bonzini wrote:
> > On 06/09/2018 10:02, Yi Wang wrote:
> >> In order to fix a page table walk issue, commit 6ea6e84309ca
> >> ("KVM: x86: inject exceptions produced by x86_decode_insn") check
> >> if variable ctxt->have_exception true and inject the exception.
> >> Unfortunately, ctxt->have_exception is set to true only in function
> >> x86_emulate_insn(), which won't be called before invoking
> >> inject_emulated_exception() in the 6ea6e84309ca.
> >> This patch fix this issue.
> >>
> >> Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
> >> Reviewed-by: Xi Xu <xu.xi8@zte.com.cn>
> > 
> > Queued, thanks.  Do you have a testcase for this?
> 
> Unqueued, sorry.  The hypercall test from kvm-unit-tests fails.  A
> VMCALL on the "edge" of canonical address space, i.e. at 0x7ffffffffffd,
> raises a general protection fault before this patch and a double fault
> afterwards.

Peng Hao submitted a very similar patch[1], the difference being that
it also modified x86_emulate_instruction() to ignore the result of
inject_emulated_exception().  I have no idea if that change is correct
but it may be related to the aforementioned unit test failing.

[1] https://lkml.org/lkml/2018/9/19/38

Patch
diff mbox series

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 106482d..aecf9a72 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -5105,8 +5105,11 @@  int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len)
 		memcpy(ctxt->fetch.data, insn, insn_len);
 	else {
 		rc = __do_insn_fetch_bytes(ctxt, 1);
-		if (rc != X86EMUL_CONTINUE)
+		if (rc != X86EMUL_CONTINUE) {
+			if (rc == X86EMUL_PROPAGATE_FAULT)
+				ctxt->have_exception = true;
 			return rc;
+		}
 	}
 
 	switch (mode) {