target/i386: kvm: Re-inject #DB to guest with updated DR6
authorLiran Alon <liran.alon@oracle.com>
Wed, 19 Jun 2019 16:21:34 +0000 (19:21 +0300)
committerPaolo Bonzini <pbonzini@redhat.com>
Fri, 21 Jun 2019 00:29:39 +0000 (02:29 +0200)
If userspace (QEMU) debug guest, when #DB is raised in guest and
intercepted by KVM, KVM forwards information on #DB to userspace
instead of injecting #DB to guest.
While doing so, KVM don't update vCPU DR6 but instead report the #DB DR6
value to userspace for further handling.
See KVM's handle_exception() DB_VECTOR handler.

QEMU handler for this case is kvm_handle_debug(). This handler basically
checks if #DB is related to one of user set hardware breakpoints and if
not, it re-inject #DB into guest.
The re-injection is done by setting env->exception_injected to #DB which
will later be passed as events.exception.nr to KVM_SET_VCPU_EVENTS ioctl
by kvm_put_vcpu_events().

However, in case userspace re-injects #DB, KVM expects userspace to set
vCPU DR6 as reported to userspace when #DB was intercepted! Otherwise,
KVM_REQ_EVENT handler will inject #DB with wrong DR6 to guest.

Fix this issue by updating vCPU DR6 appropriately when re-inject #DB to
guest.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20190619162140.133674-5-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
target/i386/kvm.c

index c763643847ca7cda6bb6dbb9aeed178a81824989..9864aa09e67f26a2024db7ef2ee53b8a3a6bbb03 100644 (file)
@@ -3897,6 +3897,9 @@ static int kvm_handle_debug(X86CPU *cpu,
         /* pass to guest */
         env->exception_injected = arch_info->exception;
         env->has_error_code = 0;
+        if (arch_info->exception == EXCP01_DB) {
+            env->dr[6] = arch_info->dr6;
+        }
     }
 
     return ret;