target/arm: Add mte helpers for sve scalar + int ff/nf loads

Because the elements are sequential, we can eliminate many tests all
at once when the tag hits TCMA, or if the page(s) are not Tagged.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20200626033144.790098-36-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

3 files changed, 357 insertions, 100 deletions

diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index 1425f33..f48752e 100644
--- a/target/arm/helper-sve.h
+++ b/target/arm/helper-sve.h
@@ -1285,6 +1285,55 @@ DEF_HELPER_FLAGS_4(sve_ldff1sds_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldff1dd_le_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldff1dd_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_4(sve_ldff1bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bhu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bsu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bdu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bhs_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bss_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1bds_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldff1hh_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hsu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hdu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hss_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hds_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldff1hh_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hsu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hdu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hss_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1hds_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldff1ss_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1sdu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1sds_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldff1ss_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1sdu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1sds_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldff1dd_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldff1dd_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_4(sve_ldnf1bb_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldnf1bhu_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldnf1bsu_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
@@ -1316,6 +1365,55 @@ DEF_HELPER_FLAGS_4(sve_ldnf1sds_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldnf1dd_le_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldnf1dd_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_4(sve_ldnf1bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bhu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bsu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bdu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bhs_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bss_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1bds_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldnf1hh_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hsu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hdu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hss_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hds_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldnf1hh_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hsu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hdu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hss_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1hds_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldnf1ss_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1sdu_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1sds_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldnf1ss_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1sdu_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1sds_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ldnf1dd_le_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ldnf1dd_be_r_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_4(sve_st1bb_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_st2bb_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_st3bb_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index ded9ced..7aca4ad 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -4794,7 +4794,7 @@ static void record_fault(CPUARMState *env, uintptr_t i, uintptr_t oprsz)
  */
 static inline QEMU_ALWAYS_INLINE
 void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
-                   uint32_t desc, const uintptr_t retaddr,
+                   uint32_t desc, const uintptr_t retaddr, uint32_t mtedesc,
                    const int esz, const int msz, const SVEContFault fault,
                    sve_ldst1_host_fn *host_fn,
                    sve_ldst1_tlb_fn *tlb_fn)
@@ -4826,13 +4826,25 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
     mem_off = info.mem_off_first[0];
     flags = info.page[0].flags;
 
+    /*
+     * Disable MTE checking if the Tagged bit is not set.  Since TBI must
+     * be set within MTEDESC for MTE, !mtedesc => !mte_active.
+     */
+    if (arm_tlb_mte_tagged(&info.page[0].attrs)) {
+        mtedesc = 0;
+    }
+
     if (fault == FAULT_FIRST) {
+        /* Trapping mte check for the first-fault element.  */
+        if (mtedesc) {
+            mte_check1(env, mtedesc, addr + mem_off, retaddr);
+        }
+
         /*
          * Special handling of the first active element,
          * if it crosses a page boundary or is MMIO.
          */
         bool is_split = mem_off == info.mem_off_split;
-        /* TODO: MTE check. */
         if (unlikely(flags != 0) || unlikely(is_split)) {
             /*
              * Use the slow path for cross-page handling.
@@ -4868,7 +4880,9 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
                 /* Watchpoint hit, see below. */
                 goto do_fault;
             }
-            /* TODO: MTE check. */
+            if (mtedesc && !mte_probe1(env, mtedesc, addr + mem_off)) {
+                goto do_fault;
+            }
             /*
              * Use the slow path for cross-page handling.
              * This is RAM, without a watchpoint, and will not trap.
@@ -4916,7 +4930,9 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
                      & BP_MEM_READ)) {
                     goto do_fault;
                 }
-                /* TODO: MTE check. */
+                if (mtedesc && !mte_probe1(env, mtedesc, addr + mem_off)) {
+                    goto do_fault;
+                }
                 host_fn(vd, reg_off, host + mem_off);
             }
             reg_off += 1 << esz;
@@ -4954,44 +4970,103 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
     record_fault(env, reg_off, reg_max);
 }
 
-#define DO_LDFF1_LDNF1_1(PART, ESZ) \
+static inline QEMU_ALWAYS_INLINE
+void sve_ldnfff1_r_mte(CPUARMState *env, void *vg, target_ulong addr,
+                       uint32_t desc, const uintptr_t retaddr,
+                       const int esz, const int msz, const SVEContFault fault,
+                       sve_ldst1_host_fn *host_fn,
+                       sve_ldst1_tlb_fn *tlb_fn)
+{
+    uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+    int bit55 = extract64(addr, 55, 1);
+
+    /* Remove mtedesc from the normal sve descriptor. */
+    desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+
+    /* Perform gross MTE suppression early. */
+    if (!tbi_check(desc, bit55) ||
+        tcma_check(desc, bit55, allocation_tag_from_addr(addr))) {
+        mtedesc = 0;
+    }
+
+    sve_ldnfff1_r(env, vg, addr, desc, retaddr, mtedesc,
+                  esz, msz, fault, host_fn, tlb_fn);
+}
+
+#define DO_LDFF1_LDNF1_1(PART, ESZ)                                     \
 void HELPER(sve_ldff1##PART##_r)(CPUARMState *env, void *vg,            \
                                  target_ulong addr, uint32_t desc)      \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_FIRST, \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MO_8, FAULT_FIRST, \
                   sve_ld1##PART##_host, sve_ld1##PART##_tlb);           \
 }                                                                       \
 void HELPER(sve_ldnf1##PART##_r)(CPUARMState *env, void *vg,            \
                                  target_ulong addr, uint32_t desc)      \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_NO,    \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MO_8, FAULT_NO, \
+                  sve_ld1##PART##_host, sve_ld1##PART##_tlb);           \
+}                                                                       \
+void HELPER(sve_ldff1##PART##_r_mte)(CPUARMState *env, void *vg,        \
+                                     target_ulong addr, uint32_t desc)  \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_FIRST, \
+                      sve_ld1##PART##_host, sve_ld1##PART##_tlb);       \
+}                                                                       \
+void HELPER(sve_ldnf1##PART##_r_mte)(CPUARMState *env, void *vg,        \
+                                     target_ulong addr, uint32_t desc)  \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_NO, \
                   sve_ld1##PART##_host, sve_ld1##PART##_tlb);           \
 }
 
-#define DO_LDFF1_LDNF1_2(PART, ESZ, MSZ) \
+#define DO_LDFF1_LDNF1_2(PART, ESZ, MSZ)                                \
 void HELPER(sve_ldff1##PART##_le_r)(CPUARMState *env, void *vg,         \
                                     target_ulong addr, uint32_t desc)   \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST,  \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_FIRST, \
                   sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb);     \
 }                                                                       \
 void HELPER(sve_ldnf1##PART##_le_r)(CPUARMState *env, void *vg,         \
                                     target_ulong addr, uint32_t desc)   \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO,     \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_NO,  \
                   sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb);     \
 }                                                                       \
 void HELPER(sve_ldff1##PART##_be_r)(CPUARMState *env, void *vg,         \
                                     target_ulong addr, uint32_t desc)   \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST,  \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_FIRST, \
                   sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb);     \
 }                                                                       \
 void HELPER(sve_ldnf1##PART##_be_r)(CPUARMState *env, void *vg,         \
                                     target_ulong addr, uint32_t desc)   \
 {                                                                       \
-    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO,     \
+    sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_NO,  \
                   sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb);     \
+}                                                                       \
+void HELPER(sve_ldff1##PART##_le_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST, \
+                      sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \
+}                                                                       \
+void HELPER(sve_ldnf1##PART##_le_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO, \
+                      sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \
+}                                                                       \
+void HELPER(sve_ldff1##PART##_be_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST, \
+                      sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \
+}                                                                       \
+void HELPER(sve_ldnf1##PART##_be_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO, \
+                      sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \
 }
 
 DO_LDFF1_LDNF1_1(bb,  MO_8)
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index daac858..e4fbe48 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4773,104 +4773,188 @@ static bool trans_LD_zpri(DisasContext *s, arg_rpri_load *a)
 
 static bool trans_LDFF1_zprr(DisasContext *s, arg_rprr_load *a)
 {
-    static gen_helper_gvec_mem * const fns[2][16] = {
-        /* Little-endian */
-        { gen_helper_sve_ldff1bb_r,
-          gen_helper_sve_ldff1bhu_r,
-          gen_helper_sve_ldff1bsu_r,
-          gen_helper_sve_ldff1bdu_r,
-
-          gen_helper_sve_ldff1sds_le_r,
-          gen_helper_sve_ldff1hh_le_r,
-          gen_helper_sve_ldff1hsu_le_r,
-          gen_helper_sve_ldff1hdu_le_r,
-
-          gen_helper_sve_ldff1hds_le_r,
-          gen_helper_sve_ldff1hss_le_r,
-          gen_helper_sve_ldff1ss_le_r,
-          gen_helper_sve_ldff1sdu_le_r,
-
-          gen_helper_sve_ldff1bds_r,
-          gen_helper_sve_ldff1bss_r,
-          gen_helper_sve_ldff1bhs_r,
-          gen_helper_sve_ldff1dd_le_r },
-
-        /* Big-endian */
-        { gen_helper_sve_ldff1bb_r,
-          gen_helper_sve_ldff1bhu_r,
-          gen_helper_sve_ldff1bsu_r,
-          gen_helper_sve_ldff1bdu_r,
-
-          gen_helper_sve_ldff1sds_be_r,
-          gen_helper_sve_ldff1hh_be_r,
-          gen_helper_sve_ldff1hsu_be_r,
-          gen_helper_sve_ldff1hdu_be_r,
-
-          gen_helper_sve_ldff1hds_be_r,
-          gen_helper_sve_ldff1hss_be_r,
-          gen_helper_sve_ldff1ss_be_r,
-          gen_helper_sve_ldff1sdu_be_r,
-
-          gen_helper_sve_ldff1bds_r,
-          gen_helper_sve_ldff1bss_r,
-          gen_helper_sve_ldff1bhs_r,
-          gen_helper_sve_ldff1dd_be_r },
+    static gen_helper_gvec_mem * const fns[2][2][16] = {
+        { /* mte inactive, little-endian */
+          { gen_helper_sve_ldff1bb_r,
+            gen_helper_sve_ldff1bhu_r,
+            gen_helper_sve_ldff1bsu_r,
+            gen_helper_sve_ldff1bdu_r,
+
+            gen_helper_sve_ldff1sds_le_r,
+            gen_helper_sve_ldff1hh_le_r,
+            gen_helper_sve_ldff1hsu_le_r,
+            gen_helper_sve_ldff1hdu_le_r,
+
+            gen_helper_sve_ldff1hds_le_r,
+            gen_helper_sve_ldff1hss_le_r,
+            gen_helper_sve_ldff1ss_le_r,
+            gen_helper_sve_ldff1sdu_le_r,
+
+            gen_helper_sve_ldff1bds_r,
+            gen_helper_sve_ldff1bss_r,
+            gen_helper_sve_ldff1bhs_r,
+            gen_helper_sve_ldff1dd_le_r },
+
+          /* mte inactive, big-endian */
+          { gen_helper_sve_ldff1bb_r,
+            gen_helper_sve_ldff1bhu_r,
+            gen_helper_sve_ldff1bsu_r,
+            gen_helper_sve_ldff1bdu_r,
+
+            gen_helper_sve_ldff1sds_be_r,
+            gen_helper_sve_ldff1hh_be_r,
+            gen_helper_sve_ldff1hsu_be_r,
+            gen_helper_sve_ldff1hdu_be_r,
+
+            gen_helper_sve_ldff1hds_be_r,
+            gen_helper_sve_ldff1hss_be_r,
+            gen_helper_sve_ldff1ss_be_r,
+            gen_helper_sve_ldff1sdu_be_r,
+
+            gen_helper_sve_ldff1bds_r,
+            gen_helper_sve_ldff1bss_r,
+            gen_helper_sve_ldff1bhs_r,
+            gen_helper_sve_ldff1dd_be_r } },
+
+        { /* mte active, little-endian */
+          { gen_helper_sve_ldff1bb_r_mte,
+            gen_helper_sve_ldff1bhu_r_mte,
+            gen_helper_sve_ldff1bsu_r_mte,
+            gen_helper_sve_ldff1bdu_r_mte,
+
+            gen_helper_sve_ldff1sds_le_r_mte,
+            gen_helper_sve_ldff1hh_le_r_mte,
+            gen_helper_sve_ldff1hsu_le_r_mte,
+            gen_helper_sve_ldff1hdu_le_r_mte,
+
+            gen_helper_sve_ldff1hds_le_r_mte,
+            gen_helper_sve_ldff1hss_le_r_mte,
+            gen_helper_sve_ldff1ss_le_r_mte,
+            gen_helper_sve_ldff1sdu_le_r_mte,
+
+            gen_helper_sve_ldff1bds_r_mte,
+            gen_helper_sve_ldff1bss_r_mte,
+            gen_helper_sve_ldff1bhs_r_mte,
+            gen_helper_sve_ldff1dd_le_r_mte },
+
+          /* mte active, big-endian */
+          { gen_helper_sve_ldff1bb_r_mte,
+            gen_helper_sve_ldff1bhu_r_mte,
+            gen_helper_sve_ldff1bsu_r_mte,
+            gen_helper_sve_ldff1bdu_r_mte,
+
+            gen_helper_sve_ldff1sds_be_r_mte,
+            gen_helper_sve_ldff1hh_be_r_mte,
+            gen_helper_sve_ldff1hsu_be_r_mte,
+            gen_helper_sve_ldff1hdu_be_r_mte,
+
+            gen_helper_sve_ldff1hds_be_r_mte,
+            gen_helper_sve_ldff1hss_be_r_mte,
+            gen_helper_sve_ldff1ss_be_r_mte,
+            gen_helper_sve_ldff1sdu_be_r_mte,
+
+            gen_helper_sve_ldff1bds_r_mte,
+            gen_helper_sve_ldff1bss_r_mte,
+            gen_helper_sve_ldff1bhs_r_mte,
+            gen_helper_sve_ldff1dd_be_r_mte } },
     };
 
     if (sve_access_check(s)) {
         TCGv_i64 addr = new_tmp_a64(s);
         tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype));
         tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, a->rn));
-        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 0, false,
-                   fns[s->be_data == MO_BE][a->dtype]);
+        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 1, false,
+                   fns[s->mte_active[0]][s->be_data == MO_BE][a->dtype]);
     }
     return true;
 }
 
 static bool trans_LDNF1_zpri(DisasContext *s, arg_rpri_load *a)
 {
-    static gen_helper_gvec_mem * const fns[2][16] = {
-        /* Little-endian */
-        { gen_helper_sve_ldnf1bb_r,
-          gen_helper_sve_ldnf1bhu_r,
-          gen_helper_sve_ldnf1bsu_r,
-          gen_helper_sve_ldnf1bdu_r,
-
-          gen_helper_sve_ldnf1sds_le_r,
-          gen_helper_sve_ldnf1hh_le_r,
-          gen_helper_sve_ldnf1hsu_le_r,
-          gen_helper_sve_ldnf1hdu_le_r,
-
-          gen_helper_sve_ldnf1hds_le_r,
-          gen_helper_sve_ldnf1hss_le_r,
-          gen_helper_sve_ldnf1ss_le_r,
-          gen_helper_sve_ldnf1sdu_le_r,
-
-          gen_helper_sve_ldnf1bds_r,
-          gen_helper_sve_ldnf1bss_r,
-          gen_helper_sve_ldnf1bhs_r,
-          gen_helper_sve_ldnf1dd_le_r },
-
-        /* Big-endian */
-        { gen_helper_sve_ldnf1bb_r,
-          gen_helper_sve_ldnf1bhu_r,
-          gen_helper_sve_ldnf1bsu_r,
-          gen_helper_sve_ldnf1bdu_r,
-
-          gen_helper_sve_ldnf1sds_be_r,
-          gen_helper_sve_ldnf1hh_be_r,
-          gen_helper_sve_ldnf1hsu_be_r,
-          gen_helper_sve_ldnf1hdu_be_r,
-
-          gen_helper_sve_ldnf1hds_be_r,
-          gen_helper_sve_ldnf1hss_be_r,
-          gen_helper_sve_ldnf1ss_be_r,
-          gen_helper_sve_ldnf1sdu_be_r,
-
-          gen_helper_sve_ldnf1bds_r,
-          gen_helper_sve_ldnf1bss_r,
-          gen_helper_sve_ldnf1bhs_r,
-          gen_helper_sve_ldnf1dd_be_r },
+    static gen_helper_gvec_mem * const fns[2][2][16] = {
+        { /* mte inactive, little-endian */
+          { gen_helper_sve_ldnf1bb_r,
+            gen_helper_sve_ldnf1bhu_r,
+            gen_helper_sve_ldnf1bsu_r,
+            gen_helper_sve_ldnf1bdu_r,
+
+            gen_helper_sve_ldnf1sds_le_r,
+            gen_helper_sve_ldnf1hh_le_r,
+            gen_helper_sve_ldnf1hsu_le_r,
+            gen_helper_sve_ldnf1hdu_le_r,
+
+            gen_helper_sve_ldnf1hds_le_r,
+            gen_helper_sve_ldnf1hss_le_r,
+            gen_helper_sve_ldnf1ss_le_r,
+            gen_helper_sve_ldnf1sdu_le_r,
+
+            gen_helper_sve_ldnf1bds_r,
+            gen_helper_sve_ldnf1bss_r,
+            gen_helper_sve_ldnf1bhs_r,
+            gen_helper_sve_ldnf1dd_le_r },
+
+          /* mte inactive, big-endian */
+          { gen_helper_sve_ldnf1bb_r,
+            gen_helper_sve_ldnf1bhu_r,
+            gen_helper_sve_ldnf1bsu_r,
+            gen_helper_sve_ldnf1bdu_r,
+
+            gen_helper_sve_ldnf1sds_be_r,
+            gen_helper_sve_ldnf1hh_be_r,
+            gen_helper_sve_ldnf1hsu_be_r,
+            gen_helper_sve_ldnf1hdu_be_r,
+
+            gen_helper_sve_ldnf1hds_be_r,
+            gen_helper_sve_ldnf1hss_be_r,
+            gen_helper_sve_ldnf1ss_be_r,
+            gen_helper_sve_ldnf1sdu_be_r,
+
+            gen_helper_sve_ldnf1bds_r,
+            gen_helper_sve_ldnf1bss_r,
+            gen_helper_sve_ldnf1bhs_r,
+            gen_helper_sve_ldnf1dd_be_r } },
+
+        { /* mte inactive, little-endian */
+          { gen_helper_sve_ldnf1bb_r_mte,
+            gen_helper_sve_ldnf1bhu_r_mte,
+            gen_helper_sve_ldnf1bsu_r_mte,
+            gen_helper_sve_ldnf1bdu_r_mte,
+
+            gen_helper_sve_ldnf1sds_le_r_mte,
+            gen_helper_sve_ldnf1hh_le_r_mte,
+            gen_helper_sve_ldnf1hsu_le_r_mte,
+            gen_helper_sve_ldnf1hdu_le_r_mte,
+
+            gen_helper_sve_ldnf1hds_le_r_mte,
+            gen_helper_sve_ldnf1hss_le_r_mte,
+            gen_helper_sve_ldnf1ss_le_r_mte,
+            gen_helper_sve_ldnf1sdu_le_r_mte,
+
+            gen_helper_sve_ldnf1bds_r_mte,
+            gen_helper_sve_ldnf1bss_r_mte,
+            gen_helper_sve_ldnf1bhs_r_mte,
+            gen_helper_sve_ldnf1dd_le_r_mte },
+
+          /* mte inactive, big-endian */
+          { gen_helper_sve_ldnf1bb_r_mte,
+            gen_helper_sve_ldnf1bhu_r_mte,
+            gen_helper_sve_ldnf1bsu_r_mte,
+            gen_helper_sve_ldnf1bdu_r_mte,
+
+            gen_helper_sve_ldnf1sds_be_r_mte,
+            gen_helper_sve_ldnf1hh_be_r_mte,
+            gen_helper_sve_ldnf1hsu_be_r_mte,
+            gen_helper_sve_ldnf1hdu_be_r_mte,
+
+            gen_helper_sve_ldnf1hds_be_r_mte,
+            gen_helper_sve_ldnf1hss_be_r_mte,
+            gen_helper_sve_ldnf1ss_be_r_mte,
+            gen_helper_sve_ldnf1sdu_be_r_mte,
+
+            gen_helper_sve_ldnf1bds_r_mte,
+            gen_helper_sve_ldnf1bss_r_mte,
+            gen_helper_sve_ldnf1bhs_r_mte,
+            gen_helper_sve_ldnf1dd_be_r_mte } },
     };
 
     if (sve_access_check(s)) {
@@ -4880,8 +4964,8 @@ static bool trans_LDNF1_zpri(DisasContext *s, arg_rpri_load *a)
         TCGv_i64 addr = new_tmp_a64(s);
 
         tcg_gen_addi_i64(addr, cpu_reg_sp(s, a->rn), off);
-        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 0, false,
-                   fns[s->be_data == MO_BE][a->dtype]);
+        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 1, false,
+                   fns[s->mte_active[0]][s->be_data == MO_BE][a->dtype]);
     }
     return true;
 }