vbox的更動 61371 路徑 trunk/src/VBox/VMM/VMMAll/IOMAllMMIO.cpp

時間撮記:

2016-6-1 下午12:58:24 (9 年 以前)

作者:

vboxsync

訊息:

iomMmioHandler: Gracefully deal with large MMIO writes and read (FXSAVE and such) in ring-0 and raw-mode by deflecting them to ring-3. These things shouldn't normally happen.

檔案:

• trunk/src/VBox/VMM/VMMAll/IOMAllMMIO.cpp (修改) (1 筆差異)

trunk/src/VBox/VMM/VMMAll/IOMAllMMIO.cpp

@@ -754 +754 @@
     STAM_COUNTER_INC(&pVM->iom.s.StatR3MMIOHandler);
 
-    AssertMsg(cbBuf >= 1 && cbBuf <= 16, ("%zu\n", cbBuf));
+    NOREF(pvPhys); NOREF(enmOrigin);
     AssertPtr(pRange);
-    NOREF(pvPhys); NOREF(enmOrigin);
+    AssertMsg(cbBuf >= 1, ("%zu\n", cbBuf));
+
+
+#ifndef IN_RING3
+    /*
+     * If someone is doing FXSAVE, FXRSTOR, XSAVE, XRSTOR or other stuff dealing with
+     * large amounts of data, just go to ring-3 where we don't need to deal with partial
+     * successes.  No chance any of these will be problematic read-modify-write stuff.
+     */
+    if (cbBuf > sizeof(pVCpu->iom.s.PendingMmioWrite.abValue))
+        return enmAccessType == PGMACCESSTYPE_WRITE ? VINF_IOM_R3_MMIO_WRITE : VINF_IOM_R3_MMIO_READ;
+#endif
 
     /*