Handle armv8m unpriv_access with SAU

Previously when we got to unpriv_access we verfied the operation with the TT
assembly command, we now switch to using the SAU with uVisor vmpu functions.

This could slow down the operation since we switch an assembly command
with a C function which scans the virtual regions.

core/vmpu/src/mpu_armv8m/vmpu_armv8m_unpriv_access.c

@@ -15,112 +15,38 @@
  * limitations under the License.
  */
 #include <uvisor.h>
+#include <stdbool.h>
 #include "context.h"
-#include "debug.h"
-#include "vmpu.h"
+#include "halt.h"
+#include "vmpu_mpu.h"
 #include "vmpu_unpriv_access.h"
-#include <stdbool.h>
-
-#define TT_RESP_IREGION_Pos     24U
-#define TT_RESP_IREGION_Msk     (0xFFUL << TT_RESP_IREGION_Pos)
-
-#define TT_RESP_IRVALID_Pos     23U
-#define TT_RESP_IRVALID_Msk     (1UL << TT_RESP_IRVALID_Pos)
-
-#define TT_RESP_S_Pos           22U
-#define TT_RESP_S_Msk           (1UL << TT_RESP_S_Pos)
-
-#define TT_RESP_NSRW_Pos        21U
-#define TT_RESP_NSRW_Msk        (1UL << TT_RESP_NSRW_Pos)
-
-#define TT_RESP_NSR_Pos         20U
-#define TT_RESP_NSR_Msk         (1UL << TT_RESP_NSR_Pos)
-
-#define TT_RESP_RW_Pos          19U
-#define TT_RESP_RW_Msk          (1UL << TT_RESP_RW_Pos)
-
-#define TT_RESP_R_Pos           18U
-#define TT_RESP_R_Msk           (1UL << TT_RESP_R_Pos)
-
-#define TT_RESP_SRVALID_Pos     17U
-#define TT_RESP_SRVALID_Msk     (1UL << TT_RESP_SRVALID_Pos)
-
-#define TT_RESP_MRVALID_Pos     16U
-#define TT_RESP_MRVALID_Msk     (1UL << TT_RESP_MRVALID_Pos)
-
-#define TT_RESP_SREGION_Pos     8U
-#define TT_RESP_SREGION_Msk     (0xFFUL << TT_RESP_SREGION_Pos)
-
-#define TT_RESP_MREGION_Pos     0U
-#define TT_RESP_MREGION_Msk     (0xFFUL << TT_RESP_MREGION_Pos)
-
-
-static uint32_t vmpu_unpriv_test_range(uint32_t addr, uint32_t size)
-{
-    if (!size) size = 1;
-    uint32_t response_lower, response_upper;
-    uint32_t test_addr_lower = addr & ~31UL;
-    uint32_t test_addr_upper = (addr + size - 1) & ~31UL;
-
-    /* Test lower address. */
-    asm volatile (
-        "tta %[response], %[addr]"
-        : [response] "=r" (response_lower)
-        : [addr] "r" (test_addr_lower)
-    );
-    if (test_addr_lower != test_addr_upper) {
-        /* Test upper address. */
-        asm volatile (
-            "tta %[response], %[addr]"
-            : [response] "=r" (response_upper)
-            : [addr] "r" (test_addr_upper)
-        );
-        /* If lower and upper do not have the same S|SRVALID|SREGION, then it's definitely not the same region. */
-        if (((response_lower ^ response_upper) & (TT_RESP_S_Msk | TT_RESP_SRVALID_Msk | TT_RESP_SREGION_Msk))) {
-            /* Upper memory region has different SAU region than lower memory region! */
-            return 0;
-        }
-        /* Both memory locations have the same non-secure SAU region and therefore same properties.
-         * No Secure SAU region can be inbetween due to SAU region overlap rules. */
-        response_lower &= response_upper;
-    }
-
-    return response_lower & (TT_RESP_NSRW_Msk | TT_RESP_NSR_Msk | TT_RESP_RW_Msk | TT_RESP_R_Msk |
-                             TT_RESP_S_Msk | TT_RESP_SRVALID_Msk | TT_RESP_SREGION_Msk);
-}
-
-extern int vmpu_fault_recovery_mpu(uint32_t pc, uint32_t sp, uint32_t fault_addr, uint32_t fault_status);
 
 uint32_t vmpu_unpriv_access(uint32_t addr, uint32_t size, uint32_t data)
 {
-    unsigned int tries = 0;
-    while(1) {
-        if ((vmpu_unpriv_test_range(addr, UVISOR_UNPRIV_ACCESS_SIZE(size)) & (TT_RESP_NSRW_Msk | TT_RESP_SRVALID_Msk)) == (TT_RESP_NSRW_Msk | TT_RESP_SRVALID_Msk)) {
-            switch(size) {
-                case UVISOR_UNPRIV_ACCESS_READ(1):
-                    return *((uint8_t *) addr);
-                case UVISOR_UNPRIV_ACCESS_READ(2):
-                    return *((uint16_t *) addr);
-                case UVISOR_UNPRIV_ACCESS_READ(4):
-                    return *((uint32_t *) addr);
-                case UVISOR_UNPRIV_ACCESS_WRITE(1):
-                    *((uint8_t *) addr) = (uint8_t) data;
-                    return 0;
-                case UVISOR_UNPRIV_ACCESS_WRITE(2):
-                    *((uint16_t *) addr) = (uint16_t) data;
-                    return 0;
-                case UVISOR_UNPRIV_ACCESS_WRITE(4):
-                    *((uint32_t *) addr) = data;
-                    return 0;
-                default:
-                    break;
-            }
-            break;
-        }
-        if (++tries > 1 || !vmpu_fault_recovery_mpu(0, 0, addr, 0)) {
-            break;
+    // This operation could be slow since we are scanning all the regions defined for the box
+    // an implementation using the TTA command is faster but causes misses since it works with the MPU
+    if (vmpu_buffer_access_is_ok(g_active_box, (const void *) addr, UVISOR_UNPRIV_ACCESS_SIZE(size))){
+        switch(size) {
+            case UVISOR_UNPRIV_ACCESS_READ(1):
+                return *((uint8_t *) addr);
+            case UVISOR_UNPRIV_ACCESS_READ(2):
+                return *((uint16_t *) addr);
+            case UVISOR_UNPRIV_ACCESS_READ(4):
+                return *((uint32_t *) addr);
+            case UVISOR_UNPRIV_ACCESS_WRITE(1):
+                *((uint8_t *) addr) = (uint8_t) data;
+                return 0;
+            case UVISOR_UNPRIV_ACCESS_WRITE(2):
+                *((uint16_t *) addr) = (uint16_t) data;
+                return 0;
+            case UVISOR_UNPRIV_ACCESS_WRITE(4):
+                *((uint32_t *) addr) = data;
+                return 0;
+            default:
+                break;
         }
     }
+
     HALT_ERROR(PERMISSION_DENIED, "Access to restricted resource denied");
     return 0;
 }