Changeset 72639 in vbox for trunk/src/VBox/VMM/VMMR0

Timestamp:

Jun 21, 2018 1:51:44 PM (6 years ago)

Author:

vboxsync

Message:

IPRT,VMM: Try use KeShouldYieldProcessor to implement RTThreadPreemptIsPending on Windows. That'll save us a lot of pain with changing _KPRCB structures. bugref:5102

Location:

trunk/src/VBox/VMM/VMMR0

Files:

• HMSVMR0.cpp (modified) (2 diffs)
• HMVMXR0.cpp (modified) (1 diff)

trunk/src/VBox/VMM/VMMR0/HMSVMR0.cpp

@@ -4167 +4167 @@
 
     /*
-     * We disable interrupts so that we don't miss any interrupts that would flag preemption (IPI/timers etc.)
-     * when thread-context hooks aren't used and we've been running with preemption disabled for a while.
+     * We disable interrupts so that we don't miss any interrupts that would flag
+     * preemption (IPI/timers etc.) when thread-context hooks aren't used and we've
+     * been running with preemption disabled for a while.  Since this is purly to aid
+     * the RTThreadPreemptIsPending code, it doesn't matter that it may temporarily
+     * reenable and disable interrupt on NT.
      *
      * We need to check for force-flags that could've possible been altered since we last checked them (e.g.
@@ -4274 +4277 @@
 
     /*
-     * We disable interrupts so that we don't miss any interrupts that would flag preemption (IPI/timers etc.)
-     * when thread-context hooks aren't used and we've been running with preemption disabled for a while.
+     * We disable interrupts so that we don't miss any interrupts that would flag
+     * preemption (IPI/timers etc.) when thread-context hooks aren't used and we've
+     * been running with preemption disabled for a while.  Since this is purly to aid
+     * the RTThreadPreemptIsPending code, it doesn't matter that it may temporarily
+     * reenable and disable interrupt on NT.
      *
      * We need to check for force-flags that could've possible been altered since we last checked them (e.g.

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

@@ -9035 +9035 @@
 
     /*
-     * We disable interrupts so that we don't miss any interrupts that would flag preemption (IPI/timers etc.)
-     * when thread-context hooks aren't used and we've been running with preemption disabled for a while.
+     * We disable interrupts so that we don't miss any interrupts that would flag
+     * preemption (IPI/timers etc.) when thread-context hooks aren't used and we've
+     * been running with preemption disabled for a while.  Since this is purly to aid
+     * the RTThreadPreemptIsPending code, it doesn't matter that it may temporarily
+     * reenable and disable interrupt on NT.
      *
      * We need to check for force-flags that could've possible been altered since we last checked them (e.g.