[Xen-devel] [PATCH] x86emul: correct compat mode system descriptor handling

["Jan Beulich" <JBeulich@xxxxxxxx>]

There are some oddities to take care of here - see the code comment.

Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>

--- a/xen/arch/x86/x86_emulate/x86_emulate.c
+++ b/xen/arch/x86/x86_emulate/x86_emulate.c
@@ -1818,11 +1818,24 @@ protmode_load_seg(
 
     if ( !is_x86_user_segment(seg) )
     {
-        int lm = (desc.b & (1u << 12)) ? 0 : in_longmode(ctxt, ops);
+        /*
+         * Whether to use an 8- or 16-byte descriptor in long mode depends
+         * on sub-mode, descriptor type, and vendor:
+         * - non-system descriptors are always 8-byte ones,
+         * - system descriptors are always 16-byte ones in 64-bit mode,
+         * - (call) gates are always 16-byte ones,
+         * - other system descriptors in compatibility mode have
+         *   - only their low 8-byte bytes read on Intel,
+         *   - all 16 bytes read with the high 8 bytes ignored on AMD.
+         */
+        int wide = desc.b & 0x1000
+                   ? 0 : (desc.b & 0xf00) != 0xc00 &&
+                         ctxt->vendor != X86_VENDOR_AMD
+                         ? mode_64bit() : in_longmode(ctxt, ops);
 
-        if ( lm < 0 )
+        if ( wide < 0 )
             return X86EMUL_UNHANDLEABLE;
-        if ( lm )
+        if ( wide )
         {
             switch ( rc = ops->read(sel_seg, (sel & 0xfff8) + 8,
                                     &desc_hi, sizeof(desc_hi), ctxt) )
@@ -1837,6 +1850,9 @@ protmode_load_seg(
             default:
                 return rc;
             }
+            if ( !mode_64bit() && ctxt->vendor == X86_VENDOR_AMD &&
+                 (desc.b & 0xf00) != 0xc00 )
+                desc_hi.b = desc_hi.a = 0;
             if ( (desc_hi.b & 0x00001f00) ||
                  (seg != x86_seg_none &&
                   !is_canonical_address((uint64_t)desc_hi.a << 32)) )

Attachment: x86emul-compat-mode-sysdesc.patch
Description: Text document

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel