x86/HVM: grow MMIO cache data size to 64 bytes

This is needed before enabling any AVX512 insns in the emulator. Change
the way alignment is enforced at the same time.

Add a check that the buffer won't actually overflow, and while at it
also convert the check for accesses to not cross page boundaries.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

xen/arch/x86/hvm/emulate.c

@@ -876,7 +876,18 @@ static int hvmemul_phys_mmio_access(
     int rc = X86EMUL_OKAY;
 
     /* Accesses must fall within a page. */
-    BUG_ON((gpa & ~PAGE_MASK) + size > PAGE_SIZE);
+    if ( (gpa & ~PAGE_MASK) + size > PAGE_SIZE )
+    {
+        ASSERT_UNREACHABLE();
+        return X86EMUL_UNHANDLEABLE;
+    }
+
+    /* Accesses must not overflow the cache's buffer. */
+    if ( size > sizeof(cache->buffer) )
+    {
+        ASSERT_UNREACHABLE();
+        return X86EMUL_UNHANDLEABLE;
+    }
 
     /*
      * hvmemul_do_io() cannot handle non-power-of-2 accesses or

xen/include/asm-x86/hvm/vcpu.h

@@ -42,15 +42,14 @@ struct hvm_vcpu_asid {
 };
 
 /*
- * We may read or write up to m256 as a number of device-model
+ * We may read or write up to m512 as a number of device-model
  * transactions.
  */
 struct hvm_mmio_cache {
     unsigned long gla;
     unsigned int size;
     uint8_t dir;
-    uint8_t pad[3]; /* make buffer[] long-aligned */
-    uint8_t buffer[32];
+    uint8_t buffer[64] __aligned(sizeof(long));
 };
 
 struct hvm_vcpu_io {