Changeset 48220 in vbox for trunk/src

Timestamp:

Sep 1, 2013 10:46:04 PM (11 years ago)

Author:

vboxsync

Message:

HMVMXR0.cpp: When the guest switches from long mode to one of the legacy modes, we may end up with the wrong host-resume values from what I can tell. Attempting a quick fix to see if that helps with the host tripple faults I'm having.

File:

• trunk/src/VBox/VMM/VMMR0/HMVMXR0.cpp (modified) (2 diffs)

trunk/src/VBox/VMM/VMMR0/HMVMXR0.cpp

@@ -4067 +4067 @@
     {
         /* Guest is not in long mode, use the 32-bit handler. */
+#if HC_ARCH_BITS == 32 && !defined(VBOX_WITH_HYBRID_32BIT_KERNEL)
+        if (pVCpu->hm.s.vmx.pfnStartVM != VMXR0StartVM32)
+        {
+            pVCpu->hm.s.vmx.pfnStartVM = VMXR0StartVM32;
+            /** @todo r=bird: Don't we need to set up the host resume (after
+             *        vmlaunch/vmresume) state here??  I'm forcing a trip to ring-3 now
+             *        in the hope that it will prevent crashing the host.  A better
+             *        fix should be found as the guest may be going back and forth
+             *        between 16/32-bit and long mode frequently at times. */
+            VMCPU_FF_SET(pVCpu, VMCPU_FF_TO_R3);
+        }
+#else
         pVCpu->hm.s.vmx.pfnStartVM = VMXR0StartVM32;
+#endif
     }
     Assert(pVCpu->hm.s.vmx.pfnStartVM);
@@ -7335 +7348 @@
         return VINF_EM_RAW_TO_R3;
     }
-    else if (RTThreadPreemptIsPending(NIL_RTTHREAD))
+    if (RTThreadPreemptIsPending(NIL_RTTHREAD))
     {
         ASMSetFlags(pVmxTransient->uEflags);