[PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[Richard Henderson]

Version 8 incorporates quite a bit of review (thanks, PMM):

 * Include HCR_DCT in the HCR_EL2 update.
 * Use a separate decode function for ADDG/SUBG.
 * Use the system arm_cpu_do_unaligned_access for user-only as well.
 * Fix alignment for LDG.
 * Fix no-fault for mte_probe1.
 * Rearrage how tagged pages are recorded.

and perhaps most visibly, for those tracking the kernel patches:

 * Add arm,armv8.5-memtag to the dtb.


r~


In need of review:
0010-target-arm-Revise-decoding-for-disas_add_sub_imm.patch
0011-target-arm-Implement-the-ADDG-SUBG-instructions.patch
0014-target-arm-Define-arm_cpu_do_unaligned_access-for.patch
0015-target-arm-Implement-LDG-STG-ST2G-instructions.patch
0025-target-arm-Implement-helper_mte_check1.patch
0028-target-arm-Use-mte_checkN-for-sve-unpredicated-lo.patch
0029-target-arm-Use-mte_checkN-for-sve-unpredicated-st.patch
0030-target-arm-Use-mte_check1-for-sve-LD1R.patch
0031-target-arm-Tidy-trans_LD1R_zpri.patch
0032-target-arm-Add-arm_tlb_bti_gp.patch
0033-target-arm-Add-mte-helpers-for-sve-scalar-int-loa.patch
0034-target-arm-Add-mte-helpers-for-sve-scalar-int-sto.patch
0035-target-arm-Add-mte-helpers-for-sve-scalar-int-ff-.patch
0036-target-arm-Handle-TBI-for-sve-scalar-int-memory-o.patch
0037-target-arm-Add-mte-helpers-for-sve-scatter-gather.patch
0038-target-arm-Complete-TBI-clearing-for-user-only-fo.patch
0041-target-arm-Always-pass-cacheattr-to-get_phys_addr.patch
0042-target-arm-Cache-the-Tagged-bit-for-a-page-in-Mem.patch
0043-target-arm-Create-tagged-ram-when-MTE-is-enabled.patch
0044-target-arm-Add-allocation-tag-storage-for-system-.patch
0045-target-arm-Enable-MTE.patch


Richard Henderson (45):
  target/arm: Add isar tests for mte
  target/arm: Improve masking of SCR RES0 bits
  target/arm: Add support for MTE to SCTLR_ELx
  target/arm: Add support for MTE to HCR_EL2 and SCR_EL3
  target/arm: Rename DISAS_UPDATE to DISAS_UPDATE_EXIT
  target/arm: Add DISAS_UPDATE_NOCHAIN
  target/arm: Add MTE system registers
  target/arm: Add MTE bits to tb_flags
  target/arm: Implement the IRG instruction
  target/arm: Revise decoding for disas_add_sub_imm
  target/arm: Implement the ADDG, SUBG instructions
  target/arm: Implement the GMI instruction
  target/arm: Implement the SUBP instruction
  target/arm: Define arm_cpu_do_unaligned_access for user-only
  target/arm: Implement LDG, STG, ST2G instructions
  target/arm: Implement the STGP instruction
  target/arm: Restrict the values of DCZID.BS under TCG
  target/arm: Simplify DC_ZVA
  target/arm: Implement the LDGM, STGM, STZGM instructions
  target/arm: Implement the access tag cache flushes
  target/arm: Move regime_el to internals.h
  target/arm: Move regime_tcr to internals.h
  target/arm: Add gen_mte_check1
  target/arm: Add gen_mte_checkN
  target/arm: Implement helper_mte_check1
  target/arm: Implement helper_mte_checkN
  target/arm: Add helper_mte_check_zva
  target/arm: Use mte_checkN for sve unpredicated loads
  target/arm: Use mte_checkN for sve unpredicated stores
  target/arm: Use mte_check1 for sve LD1R
  target/arm: Tidy trans_LD1R_zpri
  target/arm: Add arm_tlb_bti_gp
  target/arm: Add mte helpers for sve scalar + int loads
  target/arm: Add mte helpers for sve scalar + int stores
  target/arm: Add mte helpers for sve scalar + int ff/nf loads
  target/arm: Handle TBI for sve scalar + int memory ops
  target/arm: Add mte helpers for sve scatter/gather memory ops
  target/arm: Complete TBI clearing for user-only for SVE
  target/arm: Implement data cache set allocation tags
  target/arm: Set PSTATE.TCO on exception entry
  target/arm: Always pass cacheattr to get_phys_addr
  target/arm: Cache the Tagged bit for a page in MemTxAttrs
  target/arm: Create tagged ram when MTE is enabled
  target/arm: Add allocation tag storage for system mode
  target/arm: Enable MTE

 include/hw/arm/boot.h          |    3 +
 target/arm/cpu.h               |   50 +-
 target/arm/helper-a64.h        |   16 +
 target/arm/helper-sve.h        |  488 ++++++++++++
 target/arm/helper.h            |    2 +
 target/arm/internals.h         |  153 +++-
 target/arm/translate-a64.h     |    5 +
 target/arm/translate.h         |   23 +-
 hw/arm/boot.c                  |   12 +-
 hw/arm/virt.c                  |   57 +-
 target/arm/cpu.c               |   77 +-
 target/arm/cpu64.c             |    1 +
 target/arm/helper-a64.c        |   94 +--
 target/arm/helper.c            |  423 +++++++---
 target/arm/m_helper.c          |   11 +-
 target/arm/mte_helper.c        |  901 +++++++++++++++++++++
 target/arm/op_helper.c         |   16 +
 target/arm/sve_helper.c        |  616 +++++++++++---
 target/arm/tlb_helper.c        |   13 +-
 target/arm/translate-a64.c     |  652 +++++++++++++--
 target/arm/translate-sve.c     | 1366 +++++++++++++++++++++-----------
 target/arm/translate-vfp.inc.c |    2 +-
 target/arm/translate.c         |   16 +-
 target/arm/Makefile.objs       |    1 +
 24 files changed, 4164 insertions(+), 834 deletions(-)
 create mode 100644 target/arm/mte_helper.c

-- 
2.25.1

[PATCH v8 01/45] target/arm: Add isar tests for mte

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/cpu.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 677584e5da..f8ac11e73b 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3814,6 +3814,16 @@ static inline bool isar_feature_aa64_bti(const ARMISARegisters *id)
     return FIELD_EX64(id->id_aa64pfr1, ID_AA64PFR1, BT) != 0;
 }
 
+static inline bool isar_feature_aa64_mte_insn_reg(const ARMISARegisters *id)
+{
+    return FIELD_EX64(id->id_aa64pfr1, ID_AA64PFR1, MTE) != 0;
+}
+
+static inline bool isar_feature_aa64_mte(const ARMISARegisters *id)
+{
+    return FIELD_EX64(id->id_aa64pfr1, ID_AA64PFR1, MTE) >= 2;
+}
+
 static inline bool isar_feature_aa64_pmu_8_1(const ARMISARegisters *id)
 {
     return FIELD_EX64(id->id_aa64dfr0, ID_AA64DFR0, PMUVER) >= 4 &&
-- 
2.25.1

[PATCH v8 02/45] target/arm: Improve masking of SCR RES0 bits

[Richard Henderson]

Protect reads of aa64 id registers with ARM_CP_STATE_AA64.
Use this as a simpler test than arm_el_is_aa64, since EL3
cannot change mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 972a766730..a29f0a28d8 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -2011,9 +2011,16 @@ static void scr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
     uint32_t valid_mask = 0x3fff;
     ARMCPU *cpu = env_archcpu(env);
 
-    if (arm_el_is_aa64(env, 3)) {
+    if (ri->state == ARM_CP_STATE_AA64) {
         value |= SCR_FW | SCR_AW;   /* these two bits are RES1.  */
         valid_mask &= ~SCR_NET;
+
+        if (cpu_isar_feature(aa64_lor, cpu)) {
+            valid_mask |= SCR_TLOR;
+        }
+        if (cpu_isar_feature(aa64_pauth, cpu)) {
+            valid_mask |= SCR_API | SCR_APK;
+        }
     } else {
         valid_mask &= ~(SCR_RW | SCR_ST);
     }
@@ -2032,12 +2039,6 @@ static void scr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
             valid_mask &= ~SCR_SMD;
         }
     }
-    if (cpu_isar_feature(aa64_lor, cpu)) {
-        valid_mask |= SCR_TLOR;
-    }
-    if (cpu_isar_feature(aa64_pauth, cpu)) {
-        valid_mask |= SCR_API | SCR_APK;
-    }
 
     /* Clear all-context RES0 bits.  */
     value &= valid_mask;
-- 
2.25.1

[PATCH v8 03/45] target/arm: Add support for MTE to SCTLR_ELx

[Richard Henderson]

This does not attempt to rectify all of the res0 bits, but does
clear the mte bits when not enabled.  Since there is no high-part
mapping of SCTLR, aa32 mode cannot write to these bits.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index a29f0a28d8..8a0fb01581 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -4698,6 +4698,22 @@ static void sctlr_write(CPUARMState *env, const ARMCPRegInfo *ri,
 {
     ARMCPU *cpu = env_archcpu(env);
 
+    if (arm_feature(env, ARM_FEATURE_PMSA) && !cpu->has_mpu) {
+        /* M bit is RAZ/WI for PMSA with no MPU implemented */
+        value &= ~SCTLR_M;
+    }
+
+    /* ??? Lots of these bits are not implemented.  */
+
+    if (ri->state == ARM_CP_STATE_AA64 && !cpu_isar_feature(aa64_mte, cpu)) {
+        if (ri->opc1 == 6) { /* SCTLR_EL3 */
+            value &= ~(SCTLR_ITFSB | SCTLR_TCF | SCTLR_ATA);
+        } else {
+            value &= ~(SCTLR_ITFSB | SCTLR_TCF0 | SCTLR_TCF |
+                       SCTLR_ATA0 | SCTLR_ATA);
+        }
+    }
+
     if (raw_read(env, ri) == value) {
         /* Skip the TLB flush if nothing actually changed; Linux likes
          * to do a lot of pointless SCTLR writes.
@@ -4705,13 +4721,8 @@ static void sctlr_write(CPUARMState *env, const ARMCPRegInfo *ri,
         return;
     }
 
-    if (arm_feature(env, ARM_FEATURE_PMSA) && !cpu->has_mpu) {
-        /* M bit is RAZ/WI for PMSA with no MPU implemented */
-        value &= ~SCTLR_M;
-    }
-
     raw_write(env, ri, value);
-    /* ??? Lots of these bits are not implemented.  */
+
     /* This may enable/disable the MMU, so do a TLB flush.  */
     tlb_flush(CPU(cpu));
 
-- 
2.25.1

[PATCH v8 04/45] target/arm: Add support for MTE to HCR_EL2 and SCR_EL3

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Include HCR_DCT.
---
 target/arm/helper.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 8a0fb01581..d6c326b58e 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -2021,6 +2021,9 @@ static void scr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
         if (cpu_isar_feature(aa64_pauth, cpu)) {
             valid_mask |= SCR_API | SCR_APK;
         }
+        if (cpu_isar_feature(aa64_mte, cpu)) {
+            valid_mask |= SCR_ATA;
+        }
     } else {
         valid_mask &= ~(SCR_RW | SCR_ST);
     }
@@ -5248,17 +5251,22 @@ static void do_hcr_write(CPUARMState *env, uint64_t value, uint64_t valid_mask)
         if (cpu_isar_feature(aa64_pauth, cpu)) {
             valid_mask |= HCR_API | HCR_APK;
         }
+        if (cpu_isar_feature(aa64_mte, cpu)) {
+            valid_mask |= HCR_ATA | HCR_DCT | HCR_TID5;
+        }
     }
 
     /* Clear RES0 bits.  */
     value &= valid_mask;
 
-    /* These bits change the MMU setup:
+    /*
+     * These bits change the MMU setup:
      * HCR_VM enables stage 2 translation
      * HCR_PTW forbids certain page-table setups
-     * HCR_DC Disables stage1 and enables stage2 translation
+     * HCR_DC disables stage1 and enables stage2 translation
+     * HCR_DCT enables tagging on (disabled) stage1 translation
      */
-    if ((env->cp15.hcr_el2 ^ value) & (HCR_VM | HCR_PTW | HCR_DC)) {
+    if ((env->cp15.hcr_el2 ^ value) & (HCR_VM | HCR_PTW | HCR_DC | HCR_DCT)) {
         tlb_flush(CPU(cpu));
     }
     env->cp15.hcr_el2 = value;
-- 
2.25.1

[PATCH v8 05/45] target/arm: Rename DISAS_UPDATE to DISAS_UPDATE_EXIT

[Richard Henderson]

Emphasize that the is_jmp option exits to the main loop.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate.h         | 14 ++++++++------
 target/arm/translate-a64.c     |  8 ++++----
 target/arm/translate-vfp.inc.c |  2 +-
 target/arm/translate.c         | 12 ++++++------
 4 files changed, 19 insertions(+), 17 deletions(-)

diff --git a/target/arm/translate.h b/target/arm/translate.h
index 62ed5c4780..6c01f47983 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -148,7 +148,8 @@ static inline void disas_set_insn_syndrome(DisasContext *s, uint32_t syn)
 
 /* is_jmp field values */
 #define DISAS_JUMP      DISAS_TARGET_0 /* only pc was modified dynamically */
-#define DISAS_UPDATE    DISAS_TARGET_1 /* cpu state was modified dynamically */
+/* CPU state was modified dynamically; exit to main loop for interrupts. */
+#define DISAS_UPDATE_EXIT  DISAS_TARGET_1
 /* These instructions trap after executing, so the A32/T32 decoder must
  * defer them until after the conditional execution state has been updated.
  * WFI also needs special handling when single-stepping.
@@ -164,11 +165,12 @@ static inline void disas_set_insn_syndrome(DisasContext *s, uint32_t syn)
  * custom end-of-TB code)
  */
 #define DISAS_BX_EXCRET DISAS_TARGET_8
-/* For instructions which want an immediate exit to the main loop,
- * as opposed to attempting to use lookup_and_goto_ptr. Unlike
- * DISAS_UPDATE this doesn't write the PC on exiting the translation
- * loop so you need to ensure something (gen_a64_set_pc_im or runtime
- * helper) has done so before we reach return from cpu_tb_exec.
+/*
+ * For instructions which want an immediate exit to the main loop, as opposed
+ * to attempting to use lookup_and_goto_ptr.  Unlike DISAS_UPDATE_EXIT, this
+ * doesn't write the PC on exiting the translation loop so you need to ensure
+ * something (gen_a64_set_pc_im or runtime helper) has done so before we reach
+ * return from cpu_tb_exec.
  */
 #define DISAS_EXIT      DISAS_TARGET_9
 
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index a0e72ad694..63029bbc59 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1616,7 +1616,7 @@ static void handle_msr_i(DisasContext *s, uint32_t insn,
         gen_helper_msr_i_daifclear(cpu_env, t1);
         tcg_temp_free_i32(t1);
         /* For DAIFClear, exit the cpu loop to re-evaluate pending IRQs.  */
-        s->base.is_jmp = DISAS_UPDATE;
+        s->base.is_jmp = DISAS_UPDATE_EXIT;
         break;
 
     default:
@@ -1795,7 +1795,7 @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread,
 
     if ((tb_cflags(s->base.tb) & CF_USE_ICOUNT) && (ri->type & ARM_CP_IO)) {
         /* I/O operations must end the TB here (whether read or write) */
-        s->base.is_jmp = DISAS_UPDATE;
+        s->base.is_jmp = DISAS_UPDATE_EXIT;
     }
     if (!isread && !(ri->type & ARM_CP_SUPPRESS_TB_END)) {
         /*
@@ -1810,7 +1810,7 @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread,
          * but allow this to be suppressed by the register definition
          * (usually only necessary to work around guest bugs).
          */
-        s->base.is_jmp = DISAS_UPDATE;
+        s->base.is_jmp = DISAS_UPDATE_EXIT;
     }
 }
 
@@ -14304,7 +14304,7 @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
             gen_goto_tb(dc, 1, dc->base.pc_next);
             break;
         default:
-        case DISAS_UPDATE:
+        case DISAS_UPDATE_EXIT:
             gen_a64_set_pc_im(dc->base.pc_next);
             /* fall through */
         case DISAS_EXIT:
diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c
index e1a9017598..4e1fbe0dc0 100644
--- a/target/arm/translate-vfp.inc.c
+++ b/target/arm/translate-vfp.inc.c
@@ -2861,6 +2861,6 @@ static bool trans_VLLDM_VLSTM(DisasContext *s, arg_VLLDM_VLSTM *a)
     tcg_temp_free_i32(fptr);
 
     /* End the TB, because we have updated FP control bits */
-    s->base.is_jmp = DISAS_UPDATE;
+    s->base.is_jmp = DISAS_UPDATE_EXIT;
     return true;
 }
diff --git a/target/arm/translate.c b/target/arm/translate.c
index 6d18892ade..45ea788370 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -2876,7 +2876,7 @@ static void gen_msr_banked(DisasContext *s, int r, int sysm, int rn)
     tcg_temp_free_i32(tcg_tgtmode);
     tcg_temp_free_i32(tcg_regno);
     tcg_temp_free_i32(tcg_reg);
-    s->base.is_jmp = DISAS_UPDATE;
+    s->base.is_jmp = DISAS_UPDATE_EXIT;
 }
 
 static void gen_mrs_banked(DisasContext *s, int r, int sysm, int rn)
@@ -2898,7 +2898,7 @@ static void gen_mrs_banked(DisasContext *s, int r, int sysm, int rn)
     tcg_temp_free_i32(tcg_tgtmode);
     tcg_temp_free_i32(tcg_regno);
     store_reg(s, rn, tcg_reg);
-    s->base.is_jmp = DISAS_UPDATE;
+    s->base.is_jmp = DISAS_UPDATE_EXIT;
 }
 
 /* Store value to PC as for an exception return (ie don't
@@ -6153,7 +6153,7 @@ static void gen_srs(DisasContext *s,
         tcg_temp_free_i32(tmp);
     }
     tcg_temp_free_i32(addr);
-    s->base.is_jmp = DISAS_UPDATE;
+    s->base.is_jmp = DISAS_UPDATE_EXIT;
 }
 
 /* Generate a label used for skipping this instruction */
@@ -9202,7 +9202,7 @@ static bool trans_SETEND(DisasContext *s, arg_SETEND *a)
     }
     if (a->E != (s->be_data == MO_BE)) {
         gen_helper_setend(cpu_env);
-        s->base.is_jmp = DISAS_UPDATE;
+        s->base.is_jmp = DISAS_UPDATE_EXIT;
     }
     return true;
 }
@@ -9925,7 +9925,7 @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
             break;
         case DISAS_NEXT:
         case DISAS_TOO_MANY:
-        case DISAS_UPDATE:
+        case DISAS_UPDATE_EXIT:
             gen_set_pc_im(dc, dc->base.pc_next);
             /* fall through */
         default:
@@ -9952,7 +9952,7 @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
         case DISAS_JUMP:
             gen_goto_ptr();
             break;
-        case DISAS_UPDATE:
+        case DISAS_UPDATE_EXIT:
             gen_set_pc_im(dc, dc->base.pc_next);
             /* fall through */
         default:
-- 
2.25.1

[PATCH v8 06/45] target/arm: Add DISAS_UPDATE_NOCHAIN

[Richard Henderson]

Add an option that writes back the PC, like DISAS_UPDATE_EXIT,
but does not exit back to the main loop.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate.h     | 2 ++
 target/arm/translate-a64.c | 3 +++
 target/arm/translate.c     | 4 ++++
 3 files changed, 9 insertions(+)

diff --git a/target/arm/translate.h b/target/arm/translate.h
index 6c01f47983..c6f9376000 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -173,6 +173,8 @@ static inline void disas_set_insn_syndrome(DisasContext *s, uint32_t syn)
  * return from cpu_tb_exec.
  */
 #define DISAS_EXIT      DISAS_TARGET_9
+/* CPU state was modified dynamically; no need to exit, but do not chain. */
+#define DISAS_UPDATE_NOCHAIN  DISAS_TARGET_10
 
 #ifdef TARGET_AARCH64
 void a64_translate_init(void);
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 63029bbc59..b4bf4cce18 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14310,6 +14310,9 @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
         case DISAS_EXIT:
             tcg_gen_exit_tb(NULL, 0);
             break;
+        case DISAS_UPDATE_NOCHAIN:
+            gen_a64_set_pc_im(dc->base.pc_next);
+            /* fall through */
         case DISAS_JUMP:
             tcg_gen_lookup_and_goto_ptr();
             break;
diff --git a/target/arm/translate.c b/target/arm/translate.c
index 45ea788370..00f94371a4 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -9926,6 +9926,7 @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
         case DISAS_NEXT:
         case DISAS_TOO_MANY:
         case DISAS_UPDATE_EXIT:
+        case DISAS_UPDATE_NOCHAIN:
             gen_set_pc_im(dc, dc->base.pc_next);
             /* fall through */
         default:
@@ -9949,6 +9950,9 @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
         case DISAS_TOO_MANY:
             gen_goto_tb(dc, 1, dc->base.pc_next);
             break;
+        case DISAS_UPDATE_NOCHAIN:
+            gen_set_pc_im(dc, dc->base.pc_next);
+            /* fall through */
         case DISAS_JUMP:
             gen_goto_ptr();
             break;
-- 
2.25.1

[PATCH v8 07/45] target/arm: Add MTE system registers

[Richard Henderson]

This is TFSRE0_EL1, TFSR_EL1, TFSR_EL2, TFSR_EL3,
RGSR_EL1, GCR_EL1, GMID_EL1, and PSTATE.TCO.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v3: Add GMID; add access_mte.
v4: Define only TCO at mte_insn_reg.
v6: Define RAZ/WI version of TCO at mte_insn_reg;
    honor TID5 for GMID_EL1; fix TFS crn/crm; recalc hflags after TCO.
---
 target/arm/cpu.h           |  4 ++
 target/arm/internals.h     |  9 ++++
 target/arm/helper.c        | 94 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c | 21 +++++++++
 4 files changed, 128 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index f8ac11e73b..49cf37d43b 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -502,6 +502,9 @@ typedef struct CPUARMState {
         uint64_t pmccfiltr_el0; /* Performance Monitor Filter Register */
         uint64_t vpidr_el2; /* Virtualization Processor ID Register */
         uint64_t vmpidr_el2; /* Virtualization Multiprocessor ID Register */
+        uint64_t tfsr_el[4]; /* tfsre0_el1 is index 0.  */
+        uint64_t gcr_el1;
+        uint64_t rgsr_el1;
     } cp15;
 
     struct {
@@ -1282,6 +1285,7 @@ void pmu_init(ARMCPU *cpu);
 #define PSTATE_SS (1U << 21)
 #define PSTATE_PAN (1U << 22)
 #define PSTATE_UAO (1U << 23)
+#define PSTATE_TCO (1U << 25)
 #define PSTATE_V (1U << 28)
 #define PSTATE_C (1U << 29)
 #define PSTATE_Z (1U << 30)
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 4bdbc3a8ac..56b4672685 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1159,6 +1159,9 @@ static inline uint32_t aarch64_pstate_valid_mask(const ARMISARegisters *id)
     if (isar_feature_aa64_uao(id)) {
         valid |= PSTATE_UAO;
     }
+    if (isar_feature_aa64_mte(id)) {
+        valid |= PSTATE_TCO;
+    }
 
     return valid;
 }
@@ -1234,4 +1237,10 @@ void arm_log_exception(int idx);
 
 #endif /* !CONFIG_USER_ONLY */
 
+/*
+ * The log2 of the words in the tag block, for GMID_EL1.BS.
+ * The is the maximum, 256 bytes, which manipulates 64-bits of tags.
+ */
+#define GMID_EL1_BS  6
+
 #endif
diff --git a/target/arm/helper.c b/target/arm/helper.c
index d6c326b58e..b4842ea23e 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -5881,6 +5881,9 @@ static void define_arm_vh_e2h_redirects_aliases(ARMCPU *cpu)
         { K(3, 0,  1, 2, 0), K(3, 4,  1, 2, 0), K(3, 5, 1, 2, 0),
           "ZCR_EL1", "ZCR_EL2", "ZCR_EL12", isar_feature_aa64_sve },
 
+        { K(3, 0,  5, 6, 0), K(3, 4,  5, 6, 0), K(3, 5, 5, 6, 0),
+          "TFSR_EL1", "TFSR_EL2", "TFSR_EL12", isar_feature_aa64_mte },
+
         /* TODO: ARMv8.2-SPE -- PMSCR_EL2 */
         /* TODO: ARMv8.4-Trace -- TRFCR_EL2 */
     };
@@ -6855,6 +6858,86 @@ static const ARMCPRegInfo dcpodp_reg[] = {
 };
 #endif /*CONFIG_USER_ONLY*/
 
+static CPAccessResult access_aa64_tid5(CPUARMState *env, const ARMCPRegInfo *ri,
+                                       bool isread)
+{
+    if ((arm_current_el(env) < 2) && (arm_hcr_el2_eff(env) & HCR_TID5)) {
+        return CP_ACCESS_TRAP_EL2;
+    }
+
+    return CP_ACCESS_OK;
+}
+
+static CPAccessResult access_mte(CPUARMState *env, const ARMCPRegInfo *ri,
+                                 bool isread)
+{
+    int el = arm_current_el(env);
+
+    if (el < 2 &&
+        arm_feature(env, ARM_FEATURE_EL2) &&
+        !(arm_hcr_el2_eff(env) & HCR_ATA)) {
+        return CP_ACCESS_TRAP_EL2;
+    }
+    if (el < 3 &&
+        arm_feature(env, ARM_FEATURE_EL3) &&
+        !(env->cp15.scr_el3 & SCR_ATA)) {
+        return CP_ACCESS_TRAP_EL3;
+    }
+    return CP_ACCESS_OK;
+}
+
+static uint64_t tco_read(CPUARMState *env, const ARMCPRegInfo *ri)
+{
+    return env->pstate & PSTATE_TCO;
+}
+
+static void tco_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t val)
+{
+    env->pstate = (env->pstate & ~PSTATE_TCO) | (val & PSTATE_TCO);
+}
+
+static const ARMCPRegInfo mte_reginfo[] = {
+    { .name = "TFSRE0_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 5, .crm = 6, .opc2 = 1,
+      .access = PL1_RW, .accessfn = access_mte,
+      .fieldoffset = offsetof(CPUARMState, cp15.tfsr_el[0]) },
+    { .name = "TFSR_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 5, .crm = 6, .opc2 = 0,
+      .access = PL1_RW, .accessfn = access_mte,
+      .fieldoffset = offsetof(CPUARMState, cp15.tfsr_el[1]) },
+    { .name = "TFSR_EL2", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 4, .crn = 5, .crm = 6, .opc2 = 0,
+      .access = PL2_RW, .accessfn = access_mte,
+      .fieldoffset = offsetof(CPUARMState, cp15.tfsr_el[2]) },
+    { .name = "TFSR_EL3", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 6, .crn = 5, .crm = 6, .opc2 = 0,
+      .access = PL3_RW,
+      .fieldoffset = offsetof(CPUARMState, cp15.tfsr_el[3]) },
+    { .name = "RGSR_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 1, .crm = 0, .opc2 = 5,
+      .access = PL1_RW, .accessfn = access_mte,
+      .fieldoffset = offsetof(CPUARMState, cp15.rgsr_el1) },
+    { .name = "GCR_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 1, .crm = 0, .opc2 = 6,
+      .access = PL1_RW, .accessfn = access_mte,
+      .fieldoffset = offsetof(CPUARMState, cp15.gcr_el1) },
+    { .name = "GMID_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 1, .crn = 0, .crm = 0, .opc2 = 4,
+      .access = PL1_R, .accessfn = access_aa64_tid5,
+      .type = ARM_CP_CONST, .resetvalue = GMID_EL1_BS },
+    { .name = "TCO", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 3, .crn = 4, .crm = 2, .opc2 = 7,
+      .type = ARM_CP_NO_RAW,
+      .access = PL0_RW, .readfn = tco_read, .writefn = tco_write },
+    REGINFO_SENTINEL
+};
+
+static const ARMCPRegInfo mte_tco_ro_reginfo[] = {
+    { .name = "TCO", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 3, .crn = 4, .crm = 2, .opc2 = 7,
+      .type = ARM_CP_CONST, .access = PL0_RW, },
+    REGINFO_SENTINEL
+};
 #endif
 
 static CPAccessResult access_predinv(CPUARMState *env, const ARMCPRegInfo *ri,
@@ -7980,6 +8063,17 @@ void register_cp_regs_for_features(ARMCPU *cpu)
         }
     }
 #endif /*CONFIG_USER_ONLY*/
+
+    /*
+     * If full MTE is enabled, add all of the system registers.
+     * If only "instructions available at EL0" are enabled,
+     * then define only a RAZ/WI version of PSTATE.TCO.
+     */
+    if (cpu_isar_feature(aa64_mte, cpu)) {
+        define_arm_cp_regs(cpu, mte_reginfo);
+    } else if (cpu_isar_feature(aa64_mte_insn_reg, cpu)) {
+        define_arm_cp_regs(cpu, mte_tco_ro_reginfo);
+    }
 #endif
 
     if (cpu_isar_feature(any_predinv, cpu)) {
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index b4bf4cce18..efdfd50fb6 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1619,6 +1619,27 @@ static void handle_msr_i(DisasContext *s, uint32_t insn,
         s->base.is_jmp = DISAS_UPDATE_EXIT;
         break;
 
+    case 0x1c: /* TCO */
+        if (dc_isar_feature(aa64_mte, s)) {
+            /* Full MTE is enabled -- set the TCO bit as directed. */
+            if (crm & 1) {
+                set_pstate_bits(PSTATE_TCO);
+            } else {
+                clear_pstate_bits(PSTATE_TCO);
+            }
+            t1 = tcg_const_i32(s->current_el);
+            gen_helper_rebuild_hflags_a64(cpu_env, t1);
+            tcg_temp_free_i32(t1);
+            /* Many factors, including TCO, go into MTE_ACTIVE. */
+            s->base.is_jmp = DISAS_UPDATE_NOCHAIN;
+        } else if (dc_isar_feature(aa64_mte_insn_reg, s)) {
+            /* Only "instructions accessible at EL0" -- PSTATE.TCO is WI.  */
+            s->base.is_jmp = DISAS_NEXT;
+        } else {
+            goto do_unallocated;
+        }
+        break;
+
     default:
     do_unallocated:
         unallocated_encoding(s);
-- 
2.25.1

[PATCH v8 08/45] target/arm: Add MTE bits to tb_flags

[Richard Henderson]

Cache the composite ATA setting.

Cache when MTE is fully enabled, i.e. access to tags are enabled
and tag checks affect the PE.  Do this for both the normal context
and the UNPRIV context.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v3: Remove stub helper_mte_check; moved to a later patch.
v6: Add mte0_active and ata bits; drop reviewed-by.
---
 target/arm/cpu.h           | 12 ++++++++----
 target/arm/internals.h     | 18 +++++++++++++++++
 target/arm/translate.h     |  5 +++++
 target/arm/helper.c        | 40 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c |  4 ++++
 5 files changed, 75 insertions(+), 4 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 49cf37d43b..a5d3b6c9ee 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3187,10 +3187,10 @@ typedef ARMCPU ArchCPU;
  * |              |     |   TBFLAG_A32   |              |
  * |              |     +-----+----------+  TBFLAG_AM32 |
  * |  TBFLAG_ANY  |           |TBFLAG_M32|              |
- * |              |         +-+----------+--------------|
- * |              |         |         TBFLAG_A64        |
- * +--------------+---------+---------------------------+
- *  31          20        15                           0
+ * |              +-----------+----------+--------------|
+ * |              |            TBFLAG_A64               |
+ * +--------------+-------------------------------------+
+ *  31          20                                     0
  *
  * Unless otherwise noted, these bits are cached in env->hflags.
  */
@@ -3257,6 +3257,10 @@ FIELD(TBFLAG_A64, BT, 9, 1)
 FIELD(TBFLAG_A64, BTYPE, 10, 2)         /* Not cached. */
 FIELD(TBFLAG_A64, TBID, 12, 2)
 FIELD(TBFLAG_A64, UNPRIV, 14, 1)
+FIELD(TBFLAG_A64, ATA, 15, 1)
+FIELD(TBFLAG_A64, TCMA, 16, 2)
+FIELD(TBFLAG_A64, MTE_ACTIVE, 18, 1)
+FIELD(TBFLAG_A64, MTE0_ACTIVE, 19, 1)
 
 /**
  * cpu_mmu_index:
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 56b4672685..53e249687b 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1198,6 +1198,24 @@ static inline int exception_target_el(CPUARMState *env)
     return target_el;
 }
 
+/* Determine if allocation tags are available.  */
+static inline bool allocation_tag_access_enabled(CPUARMState *env, int el,
+                                                 uint64_t sctlr)
+{
+    if (el < 3
+        && arm_feature(env, ARM_FEATURE_EL3)
+        && !(env->cp15.scr_el3 & SCR_ATA)) {
+        return false;
+    }
+    if (el < 2
+        && arm_feature(env, ARM_FEATURE_EL2)
+        && !(arm_hcr_el2_eff(env) & HCR_ATA)) {
+        return false;
+    }
+    sctlr &= (el == 0 ? SCTLR_ATA0 : SCTLR_ATA);
+    return sctlr != 0;
+}
+
 #ifndef CONFIG_USER_ONLY
 
 /* Security attributes for an address, as returned by v8m_security_lookup. */
diff --git a/target/arm/translate.h b/target/arm/translate.h
index c6f9376000..dbbf6145cb 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -30,6 +30,7 @@ typedef struct DisasContext {
     ARMMMUIdx mmu_idx; /* MMU index to use for normal loads/stores */
     uint8_t tbii;      /* TBI1|TBI0 for insns */
     uint8_t tbid;      /* TBI1|TBI0 for data */
+    uint8_t tcma;      /* TCMA1|TCMA0 for MTE */
     bool ns;        /* Use non-secure CPREG bank on access */
     int fp_excp_el; /* FP exception EL or 0 if enabled */
     int sve_excp_el; /* SVE exception EL or 0 if enabled */
@@ -77,6 +78,10 @@ typedef struct DisasContext {
     bool unpriv;
     /* True if v8.3-PAuth is active.  */
     bool pauth_active;
+    /* True if v8.5-MTE access to tags is enabled.  */
+    bool ata;
+    /* True if v8.5-MTE tag checks affect the PE; index with is_unpriv.  */
+    bool mte_active[2];
     /* True with v8.5-BTI and SCTLR_ELx.BT* set.  */
     bool bt;
     /* True if any CP15 access is trapped by HSTR_EL2 */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index b4842ea23e..2c6ec244af 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -10655,6 +10655,16 @@ static int aa64_va_parameter_tbid(uint64_t tcr, ARMMMUIdx mmu_idx)
     }
 }
 
+static int aa64_va_parameter_tcma(uint64_t tcr, ARMMMUIdx mmu_idx)
+{
+    if (regime_has_2_ranges(mmu_idx)) {
+        return extract64(tcr, 57, 2);
+    } else {
+        /* Replicate the single TCMA bit so we always have 2 bits.  */
+        return extract32(tcr, 30, 1) * 3;
+    }
+}
+
 ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va,
                                    ARMMMUIdx mmu_idx, bool data)
 {
@@ -12679,6 +12689,36 @@ static uint32_t rebuild_hflags_a64(CPUARMState *env, int el, int fp_el,
         }
     }
 
+    if (cpu_isar_feature(aa64_mte, env_archcpu(env))) {
+        /*
+         * Set MTE_ACTIVE if any access may be Checked, and leave clear
+         * if all accesses must be Unchecked:
+         * 1) If no TBI, then there are no tags in the address to check,
+         * 2) If Tag Check Override, then all accesses are Unchecked,
+         * 3) If Tag Check Fail == 0, then Checked access have no effect,
+         * 4) If no Allocation Tag Access, then all accesses are Unchecked.
+         */
+        if (allocation_tag_access_enabled(env, el, sctlr)) {
+            flags = FIELD_DP32(flags, TBFLAG_A64, ATA, 1);
+            if (tbid
+                && !(env->pstate & PSTATE_TCO)
+                && (sctlr & (el == 0 ? SCTLR_TCF0 : SCTLR_TCF))) {
+                flags = FIELD_DP32(flags, TBFLAG_A64, MTE_ACTIVE, 1);
+            }
+        }
+        /* And again for unprivileged accesses, if required.  */
+        if (FIELD_EX32(flags, TBFLAG_A64, UNPRIV)
+            && tbid
+            && !(env->pstate & PSTATE_TCO)
+            && (sctlr & SCTLR_TCF0)
+            && allocation_tag_access_enabled(env, 0, sctlr)) {
+            flags = FIELD_DP32(flags, TBFLAG_A64, MTE0_ACTIVE, 1);
+        }
+        /* Cache TCMA as well as TBI. */
+        flags = FIELD_DP32(flags, TBFLAG_A64, TCMA,
+                           aa64_va_parameter_tcma(tcr, mmu_idx));
+    }
+
     return rebuild_hflags_common(env, fp_el, mmu_idx, flags);
 }
 
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index efdfd50fb6..717cb96a40 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14183,6 +14183,7 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->mmu_idx = core_to_aa64_mmu_idx(core_mmu_idx);
     dc->tbii = FIELD_EX32(tb_flags, TBFLAG_A64, TBII);
     dc->tbid = FIELD_EX32(tb_flags, TBFLAG_A64, TBID);
+    dc->tcma = FIELD_EX32(tb_flags, TBFLAG_A64, TCMA);
     dc->current_el = arm_mmu_idx_to_el(dc->mmu_idx);
 #if !defined(CONFIG_USER_ONLY)
     dc->user = (dc->current_el == 0);
@@ -14194,6 +14195,9 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->bt = FIELD_EX32(tb_flags, TBFLAG_A64, BT);
     dc->btype = FIELD_EX32(tb_flags, TBFLAG_A64, BTYPE);
     dc->unpriv = FIELD_EX32(tb_flags, TBFLAG_A64, UNPRIV);
+    dc->ata = FIELD_EX32(tb_flags, TBFLAG_A64, ATA);
+    dc->mte_active[0] = FIELD_EX32(tb_flags, TBFLAG_A64, MTE_ACTIVE);
+    dc->mte_active[1] = FIELD_EX32(tb_flags, TBFLAG_A64, MTE0_ACTIVE);
     dc->vec_len = 0;
     dc->vec_stride = 0;
     dc->cp_regs = arm_cpu->cp_regs;
-- 
2.25.1

[PATCH v8 09/45] target/arm: Implement the IRG instruction

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Update to 00eac5.
    Merge choose_random_nonexcluded_tag into helper_irg since
    that pseudo function no longer exists separately.
v6: Remove obsolete logical/physical tag distinction;
    implement inline for !ATA.
---
 target/arm/helper-a64.h    |  2 ++
 target/arm/internals.h     |  5 +++
 target/arm/mte_helper.c    | 72 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c | 18 ++++++++++
 target/arm/Makefile.objs   |  1 +
 5 files changed, 98 insertions(+)
 create mode 100644 target/arm/mte_helper.c

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 3df7c185aa..587ccbe42f 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -103,3 +103,5 @@ DEF_HELPER_FLAGS_3(autda, TCG_CALL_NO_WG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_3(autdb, TCG_CALL_NO_WG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64)
 DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
+
+DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 53e249687b..ae611a6ff5 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1261,4 +1261,9 @@ void arm_log_exception(int idx);
  */
 #define GMID_EL1_BS  6
 
+static inline uint64_t address_with_allocation_tag(uint64_t ptr, int rtag)
+{
+    return deposit64(ptr, 56, 4, rtag);
+}
+
 #endif
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
new file mode 100644
index 0000000000..539a04de84
--- /dev/null
+++ b/target/arm/mte_helper.c
@@ -0,0 +1,72 @@
+/*
+ * ARM v8.5-MemTag Operations
+ *
+ * Copyright (c) 2020 Linaro, Ltd.
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "cpu.h"
+#include "internals.h"
+#include "exec/exec-all.h"
+#include "exec/cpu_ldst.h"
+#include "exec/helper-proto.h"
+
+
+static int choose_nonexcluded_tag(int tag, int offset, uint16_t exclude)
+{
+    if (exclude == 0xffff) {
+        return 0;
+    }
+    if (offset == 0) {
+        while (exclude & (1 << tag)) {
+            tag = (tag + 1) & 15;
+        }
+    } else {
+        do {
+            do {
+                tag = (tag + 1) & 15;
+            } while (exclude & (1 << tag));
+        } while (--offset > 0);
+    }
+    return tag;
+}
+
+uint64_t HELPER(irg)(CPUARMState *env, uint64_t rn, uint64_t rm)
+{
+    int rtag;
+
+    /*
+     * Our IMPDEF choice for GCR_EL1.RRND==1 is to behave as if
+     * GCR_EL1.RRND==0, always producing deterministic results.
+     */
+    uint16_t exclude = extract32(rm | env->cp15.gcr_el1, 0, 16);
+    int start = extract32(env->cp15.rgsr_el1, 0, 4);
+    int seed = extract32(env->cp15.rgsr_el1, 8, 16);
+    int offset, i;
+
+    /* RandomTag */
+    for (i = offset = 0; i < 4; ++i) {
+        /* NextRandomTagBit */
+        int top = (extract32(seed, 5, 1) ^ extract32(seed, 3, 1) ^
+                   extract32(seed, 2, 1) ^ extract32(seed, 0, 1));
+        seed = (top << 15) | (seed >> 1);
+        offset |= top << i;
+    }
+    rtag = choose_nonexcluded_tag(start, offset, exclude);
+    env->cp15.rgsr_el1 = rtag | (seed << 8);
+
+    return address_with_allocation_tag(rn, rtag);
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 717cb96a40..14d51c0dd1 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -226,6 +226,12 @@ static TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr)
     return clean;
 }
 
+/* Insert a zero tag into src, with the result at dst. */
+static void gen_address_with_allocation_tag0(TCGv_i64 dst, TCGv_i64 src)
+{
+    tcg_gen_andi_i64(dst, src, ~MAKE_64BIT_MASK(56, 4));
+}
+
 typedef struct DisasCompare64 {
     TCGCond cond;
     TCGv_i64 value;
@@ -5284,6 +5290,18 @@ static void disas_data_proc_2src(DisasContext *s, uint32_t insn)
     case 3: /* SDIV */
         handle_div(s, true, sf, rm, rn, rd);
         break;
+    case 4: /* IRG */
+        if (sf == 0 || !dc_isar_feature(aa64_mte_insn_reg, s)) {
+            goto do_unallocated;
+        }
+        if (s->ata) {
+            gen_helper_irg(cpu_reg_sp(s, rd), cpu_env,
+                           cpu_reg_sp(s, rn), cpu_reg(s, rm));
+        } else {
+            gen_address_with_allocation_tag0(cpu_reg_sp(s, rd),
+                                             cpu_reg_sp(s, rn));
+        }
+        break;
     case 8: /* LSLV */
         handle_shift_reg(s, A64_SHIFT_TYPE_LSL, sf, rm, rn, rd);
         break;
diff --git a/target/arm/Makefile.objs b/target/arm/Makefile.objs
index 83febd232c..fa39fd7c83 100644
--- a/target/arm/Makefile.objs
+++ b/target/arm/Makefile.objs
@@ -86,3 +86,4 @@ obj-$(CONFIG_SOFTMMU) += psci.o
 obj-$(TARGET_AARCH64) += translate-a64.o helper-a64.o
 obj-$(TARGET_AARCH64) += translate-sve.o sve_helper.o
 obj-$(TARGET_AARCH64) += pauth_helper.o
+obj-$(TARGET_AARCH64) += mte_helper.o
-- 
2.25.1

[PATCH v8 10/45] target/arm: Revise decoding for disas_add_sub_imm

[Richard Henderson]

The current Arm ARM has adjusted the official decode of
"Add/subtract (immediate)" so that the shift field is only bit 22,
and bit 23 is part of the op1 field of the parent category
"Data processing - immediate".

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate-a64.c | 23 ++++++++---------------
 1 file changed, 8 insertions(+), 15 deletions(-)

diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 14d51c0dd1..59f44fc412 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -3754,22 +3754,22 @@ static void disas_pc_rel_adr(DisasContext *s, uint32_t insn)
 /*
  * Add/subtract (immediate)
  *
- *  31 30 29 28       24 23 22 21         10 9   5 4   0
- * +--+--+--+-----------+-----+-------------+-----+-----+
- * |sf|op| S| 1 0 0 0 1 |shift|    imm12    |  Rn | Rd  |
- * +--+--+--+-----------+-----+-------------+-----+-----+
+ *  31 30 29 28         23 22 21         10 9   5 4   0
+ * +--+--+--+-------------+--+-------------+-----+-----+
+ * |sf|op| S| 1 0 0 0 1 0 |sh|    imm12    |  Rn | Rd  |
+ * +--+--+--+-------------+--+-------------+-----+-----+
  *
  *    sf: 0 -> 32bit, 1 -> 64bit
  *    op: 0 -> add  , 1 -> sub
  *     S: 1 -> set flags
- * shift: 00 -> LSL imm by 0, 01 -> LSL imm by 12
+ *    sh: 1 -> LSL imm by 12
  */
 static void disas_add_sub_imm(DisasContext *s, uint32_t insn)
 {
     int rd = extract32(insn, 0, 5);
     int rn = extract32(insn, 5, 5);
     uint64_t imm = extract32(insn, 10, 12);
-    int shift = extract32(insn, 22, 2);
+    bool shift = extract32(insn, 22, 1);
     bool setflags = extract32(insn, 29, 1);
     bool sub_op = extract32(insn, 30, 1);
     bool is_64bit = extract32(insn, 31, 1);
@@ -3778,15 +3778,8 @@ static void disas_add_sub_imm(DisasContext *s, uint32_t insn)
     TCGv_i64 tcg_rd = setflags ? cpu_reg(s, rd) : cpu_reg_sp(s, rd);
     TCGv_i64 tcg_result;
 
-    switch (shift) {
-    case 0x0:
-        break;
-    case 0x1:
+    if (shift) {
         imm <<= 12;
-        break;
-    default:
-        unallocated_encoding(s);
-        return;
     }
 
     tcg_result = tcg_temp_new_i64();
@@ -4174,7 +4167,7 @@ static void disas_data_proc_imm(DisasContext *s, uint32_t insn)
     case 0x20: case 0x21: /* PC-rel. addressing */
         disas_pc_rel_adr(s, insn);
         break;
-    case 0x22: case 0x23: /* Add/subtract (immediate) */
+    case 0x22: /* Add/subtract (immediate) */
         disas_add_sub_imm(s, insn);
         break;
     case 0x24: /* Logical (immediate) */
-- 
2.25.1

[PATCH v8 11/45] target/arm: Implement the ADDG, SUBG instructions

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Shift offset in translate; use extract32.
v6: Implement inline for !ATA.
v8: Use separate decode function.
---
 target/arm/helper-a64.h    |  1 +
 target/arm/internals.h     |  9 +++++++
 target/arm/mte_helper.c    | 10 ++++++++
 target/arm/translate-a64.c | 51 ++++++++++++++++++++++++++++++++++++++
 4 files changed, 71 insertions(+)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 587ccbe42f..6c116481e8 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -105,3 +105,4 @@ DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64)
 DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
 
 DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
+DEF_HELPER_FLAGS_4(addsubg, TCG_CALL_NO_RWG_SE, i64, env, i64, s32, i32)
diff --git a/target/arm/internals.h b/target/arm/internals.h
index ae611a6ff5..5c69d4e5a5 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1261,6 +1261,15 @@ void arm_log_exception(int idx);
  */
 #define GMID_EL1_BS  6
 
+/* We associate one allocation tag per 16 bytes, the minimum.  */
+#define LOG2_TAG_GRANULE 4
+#define TAG_GRANULE      (1 << LOG2_TAG_GRANULE)
+
+static inline int allocation_tag_from_addr(uint64_t ptr)
+{
+    return extract64(ptr, 56, 4);
+}
+
 static inline uint64_t address_with_allocation_tag(uint64_t ptr, int rtag)
 {
     return deposit64(ptr, 56, 4, rtag);
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 539a04de84..9ab9ed749d 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -70,3 +70,13 @@ uint64_t HELPER(irg)(CPUARMState *env, uint64_t rn, uint64_t rm)
 
     return address_with_allocation_tag(rn, rtag);
 }
+
+uint64_t HELPER(addsubg)(CPUARMState *env, uint64_t ptr,
+                         int32_t offset, uint32_t tag_offset)
+{
+    int start_tag = allocation_tag_from_addr(ptr);
+    uint16_t exclude = extract32(env->cp15.gcr_el1, 0, 16);
+    int rtag = choose_nonexcluded_tag(start_tag, tag_offset, exclude);
+
+    return address_with_allocation_tag(ptr + offset, rtag);
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 59f44fc412..e9bc7e90af 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -3808,6 +3808,54 @@ static void disas_add_sub_imm(DisasContext *s, uint32_t insn)
     tcg_temp_free_i64(tcg_result);
 }
 
+/*
+ * Add/subtract (immediate, with tags)
+ *
+ *  31 30 29 28         23 22 21     16 14      10 9   5 4   0
+ * +--+--+--+-------------+--+---------+--+-------+-----+-----+
+ * |sf|op| S| 1 0 0 0 1 0 |o2|  uimm6  |o3| uimm4 |  Rn | Rd  |
+ * +--+--+--+-------------+--+---------+--+-------+-----+-----+
+ *
+ *    op: 0 -> add, 1 -> sub
+ */
+static void disas_add_sub_imm_with_tags(DisasContext *s, uint32_t insn)
+{
+    int rd = extract32(insn, 0, 5);
+    int rn = extract32(insn, 5, 5);
+    int uimm4 = extract32(insn, 10, 4);
+    int uimm6 = extract32(insn, 16, 6);
+    bool sub_op = extract32(insn, 30, 1);
+    TCGv_i64 tcg_rn, tcg_rd;
+    int imm;
+
+    /* Test all of sf=1, S=0, o2=0, o3=0.  */
+    if ((insn & 0xc040e000u) != 0x80000000u ||
+        !dc_isar_feature(aa64_mte_insn_reg, s)) {
+        unallocated_encoding(s);
+        return;
+    }
+
+    imm = uimm6 << LOG2_TAG_GRANULE;
+    if (sub_op) {
+        imm = -imm;
+    }
+
+    tcg_rn = cpu_reg_sp(s, rn);
+    tcg_rd = cpu_reg_sp(s, rd);
+
+    if (s->ata) {
+        TCGv_i32 offset = tcg_const_i32(imm);
+        TCGv_i32 tag_offset = tcg_const_i32(uimm4);
+
+        gen_helper_addsubg(tcg_rd, cpu_env, tcg_rn, offset, tag_offset);
+        tcg_temp_free_i32(tag_offset);
+        tcg_temp_free_i32(offset);
+    } else {
+        tcg_gen_addi_i64(tcg_rd, tcg_rn, imm);
+        gen_address_with_allocation_tag0(tcg_rd, tcg_rd);
+    }
+}
+
 /* The input should be a value in the bottom e bits (with higher
  * bits zero); returns that value replicated into every element
  * of size e in a 64 bit integer.
@@ -4170,6 +4218,9 @@ static void disas_data_proc_imm(DisasContext *s, uint32_t insn)
     case 0x22: /* Add/subtract (immediate) */
         disas_add_sub_imm(s, insn);
         break;
+    case 0x23: /* Add/subtract (immediate, with tags) */
+        disas_add_sub_imm_with_tags(s, insn);
+        break;
     case 0x24: /* Logical (immediate) */
         disas_logic_imm(s, insn);
         break;
-- 
2.25.1

[PATCH v8 12/45] target/arm: Implement the GMI instruction

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v6: Inline the operation.
---
 target/arm/translate-a64.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index e9bc7e90af..255365e76c 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -5346,6 +5346,21 @@ static void disas_data_proc_2src(DisasContext *s, uint32_t insn)
                                              cpu_reg_sp(s, rn));
         }
         break;
+    case 5: /* GMI */
+        if (sf == 0 || !dc_isar_feature(aa64_mte_insn_reg, s)) {
+            goto do_unallocated;
+        } else {
+            TCGv_i64 t1 = tcg_const_i64(1);
+            TCGv_i64 t2 = tcg_temp_new_i64();
+
+            tcg_gen_extract_i64(t2, cpu_reg_sp(s, rn), 56, 4);
+            tcg_gen_shl_i64(t1, t1, t2);
+            tcg_gen_or_i64(cpu_reg(s, rd), cpu_reg(s, rm), t1);
+
+            tcg_temp_free_i64(t1);
+            tcg_temp_free_i64(t2);
+        }
+        break;
     case 8: /* LSLV */
         handle_shift_reg(s, A64_SHIFT_TYPE_LSL, sf, rm, rn, rd);
         break;
-- 
2.25.1

[PATCH v8 13/45] target/arm: Implement the SUBP instruction

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Fix extraction length.
---
 target/arm/translate-a64.c | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 255365e76c..5400516eef 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -5315,19 +5315,39 @@ static void handle_crc32(DisasContext *s,
  */
 static void disas_data_proc_2src(DisasContext *s, uint32_t insn)
 {
-    unsigned int sf, rm, opcode, rn, rd;
+    unsigned int sf, rm, opcode, rn, rd, setflag;
     sf = extract32(insn, 31, 1);
+    setflag = extract32(insn, 29, 1);
     rm = extract32(insn, 16, 5);
     opcode = extract32(insn, 10, 6);
     rn = extract32(insn, 5, 5);
     rd = extract32(insn, 0, 5);
 
-    if (extract32(insn, 29, 1)) {
+    if (setflag && opcode != 0) {
         unallocated_encoding(s);
         return;
     }
 
     switch (opcode) {
+    case 0: /* SUBP(S) */
+        if (sf == 0 || !dc_isar_feature(aa64_mte_insn_reg, s)) {
+            goto do_unallocated;
+        } else {
+            TCGv_i64 tcg_n, tcg_m, tcg_d;
+
+            tcg_n = read_cpu_reg_sp(s, rn, true);
+            tcg_m = read_cpu_reg_sp(s, rm, true);
+            tcg_gen_sextract_i64(tcg_n, tcg_n, 0, 56);
+            tcg_gen_sextract_i64(tcg_m, tcg_m, 0, 56);
+            tcg_d = cpu_reg(s, rd);
+
+            if (setflag) {
+                gen_sub_CC(true, tcg_d, tcg_n, tcg_m);
+            } else {
+                tcg_gen_sub_i64(tcg_d, tcg_n, tcg_m);
+            }
+        }
+        break;
     case 2: /* UDIV */
         handle_div(s, false, sf, rm, rn, rd);
         break;
-- 
2.25.1

[PATCH v8 14/45] target/arm: Define arm_cpu_do_unaligned_access for user-only

[Richard Henderson]

Use the same code as system mode, so that we generate the same
exception + syndrome for the unaligned access.

For the moment, if MTE is enabled so that this path is reachable,
this would generate a SIGSEGV in the user-only cpu_loop.  Decoding
the syndrome to produce the proper SIGBUS will be done later.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Raise the normal data exception + syndrome.
---
 target/arm/cpu.c        | 2 +-
 target/arm/tlb_helper.c | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 5b7a36b5d7..10677c0c23 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -2169,8 +2169,8 @@ static void arm_cpu_class_init(ObjectClass *oc, void *data)
     cc->tlb_fill = arm_cpu_tlb_fill;
     cc->debug_excp_handler = arm_debug_excp_handler;
     cc->debug_check_watchpoint = arm_debug_check_watchpoint;
-#if !defined(CONFIG_USER_ONLY)
     cc->do_unaligned_access = arm_cpu_do_unaligned_access;
+#if !defined(CONFIG_USER_ONLY)
     cc->do_transaction_failed = arm_cpu_do_transaction_failed;
     cc->adjust_watchpoint_address = arm_adjust_watchpoint_address;
 #endif /* CONFIG_TCG && !CONFIG_USER_ONLY */
diff --git a/target/arm/tlb_helper.c b/target/arm/tlb_helper.c
index 7388494a55..522a6442a4 100644
--- a/target/arm/tlb_helper.c
+++ b/target/arm/tlb_helper.c
@@ -10,8 +10,6 @@
 #include "internals.h"
 #include "exec/exec-all.h"
 
-#if !defined(CONFIG_USER_ONLY)
-
 static inline uint32_t merge_syn_data_abort(uint32_t template_syn,
                                             unsigned int target_el,
                                             bool same_el, bool ea,
@@ -122,6 +120,8 @@ void arm_cpu_do_unaligned_access(CPUState *cs, vaddr vaddr,
     arm_deliver_fault(cpu, vaddr, access_type, mmu_idx, &fi);
 }
 
+#if !defined(CONFIG_USER_ONLY)
+
 /*
  * arm_cpu_do_transaction_failed: handle a memory system error response
  * (eg "no device/memory present at address") by raising an external abort
-- 
2.25.1

[PATCH v8 15/45] target/arm: Implement LDG, STG, ST2G instructions

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Split out allocation_tag_mem.  Handle atomicity of stores.
v3: Add X[t] input to these insns; require pre-cleaned addresses.
v5: Fix !32-byte aligned operation of st2g.
v6: Fix op2 extract, stg pre/post-index, stores vs sp, commentary;
    use pre-computed ata.
v7: Fix STZG iteration (stephen long)
v8: Merge gen_probe_access patch; align address for ldg (pmm)
---
 target/arm/helper-a64.h    |   7 ++
 target/arm/helper.h        |   2 +
 target/arm/mte_helper.c    | 194 +++++++++++++++++++++++++++++++++++++
 target/arm/op_helper.c     |  16 +++
 target/arm/translate-a64.c | 172 +++++++++++++++++++++++++++++++-
 5 files changed, 386 insertions(+), 5 deletions(-)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 6c116481e8..2fa61b86fa 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -106,3 +106,10 @@ DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
 
 DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_4(addsubg, TCG_CALL_NO_RWG_SE, i64, env, i64, s32, i32)
+DEF_HELPER_FLAGS_3(ldg, TCG_CALL_NO_WG, i64, env, i64, i64)
+DEF_HELPER_FLAGS_3(stg, TCG_CALL_NO_WG, void, env, i64, i64)
+DEF_HELPER_FLAGS_3(stg_parallel, TCG_CALL_NO_WG, void, env, i64, i64)
+DEF_HELPER_FLAGS_2(stg_stub, TCG_CALL_NO_WG, void, env, i64)
+DEF_HELPER_FLAGS_3(st2g, TCG_CALL_NO_WG, void, env, i64, i64)
+DEF_HELPER_FLAGS_3(st2g_parallel, TCG_CALL_NO_WG, void, env, i64, i64)
+DEF_HELPER_FLAGS_2(st2g_stub, TCG_CALL_NO_WG, void, env, i64)
diff --git a/target/arm/helper.h b/target/arm/helper.h
index 2a20c8174c..759639a63a 100644
--- a/target/arm/helper.h
+++ b/target/arm/helper.h
@@ -96,6 +96,8 @@ DEF_HELPER_FLAGS_1(rebuild_hflags_a32_newel, TCG_CALL_NO_RWG, void, env)
 DEF_HELPER_FLAGS_2(rebuild_hflags_a32, TCG_CALL_NO_RWG, void, env, int)
 DEF_HELPER_FLAGS_2(rebuild_hflags_a64, TCG_CALL_NO_RWG, void, env, int)
 
+DEF_HELPER_FLAGS_5(probe_access, TCG_CALL_NO_WG, void, env, tl, i32, i32, i32)
+
 DEF_HELPER_1(vfp_get_fpscr, i32, env)
 DEF_HELPER_2(vfp_set_fpscr, void, env, i32)
 
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 9ab9ed749d..7ec7930dfc 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -44,6 +44,40 @@ static int choose_nonexcluded_tag(int tag, int offset, uint16_t exclude)
     return tag;
 }
 
+/**
+ * allocation_tag_mem:
+ * @env: the cpu environment
+ * @ptr_mmu_idx: the addressing regime to use for the virtual address
+ * @ptr: the virtual address for which to look up tag memory
+ * @ptr_access: the access to use for the virtual address
+ * @ptr_size: the number of bytes in the normal memory access
+ * @tag_access: the access to use for the tag memory
+ * @tag_size: the number of bytes in the tag memory access
+ * @ra: the return address for exception handling
+ *
+ * Our tag memory is formatted as a sequence of little-endian nibbles.
+ * That is, the byte at (addr >> (LOG2_TAG_GRANULE + 1)) contains two
+ * tags, with the tag at [3:0] for the lower addr and the tag at [7:4]
+ * for the higher addr.
+ *
+ * Here, resolve the physical address from the virtual address, and return
+ * a pointer to the corresponding tag byte.  Exit with exception if the
+ * virtual address is not accessible for @ptr_access.
+ *
+ * The @ptr_size and @tag_size values may not have an obvious relation
+ * due to the alignment of @ptr, and the number of tag checks required.
+ *
+ * If there is no tag storage corresponding to @ptr, return NULL.
+ */
+static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx,
+                                   uint64_t ptr, MMUAccessType ptr_access,
+                                   int ptr_size, MMUAccessType tag_access,
+                                   int tag_size, uintptr_t ra)
+{
+    /* Tag storage not implemented.  */
+    return NULL;
+}
+
 uint64_t HELPER(irg)(CPUARMState *env, uint64_t rn, uint64_t rm)
 {
     int rtag;
@@ -80,3 +114,163 @@ uint64_t HELPER(addsubg)(CPUARMState *env, uint64_t ptr,
 
     return address_with_allocation_tag(ptr + offset, rtag);
 }
+
+static int load_tag1(uint64_t ptr, uint8_t *mem)
+{
+    int ofs = extract32(ptr, LOG2_TAG_GRANULE, 1) * 4;
+    return extract32(*mem, ofs, 4);
+}
+
+uint64_t HELPER(ldg)(CPUARMState *env, uint64_t ptr, uint64_t xt)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uint8_t *mem;
+    int rtag = 0;
+
+    /* Trap if accessing an invalid page.  */
+    mem = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_LOAD, 1,
+                             MMU_DATA_LOAD, 1, GETPC());
+
+    /* Load if page supports tags. */
+    if (mem) {
+        rtag = load_tag1(ptr, mem);
+    }
+
+    return address_with_allocation_tag(xt, rtag);
+}
+
+static void check_tag_aligned(CPUARMState *env, uint64_t ptr, uintptr_t ra)
+{
+    if (unlikely(!QEMU_IS_ALIGNED(ptr, TAG_GRANULE))) {
+        arm_cpu_do_unaligned_access(env_cpu(env), ptr, MMU_DATA_STORE,
+                                    cpu_mmu_index(env, false), ra);
+        g_assert_not_reached();
+    }
+}
+
+/* For use in a non-parallel context, store to the given nibble.  */
+static void store_tag1(uint64_t ptr, uint8_t *mem, int tag)
+{
+    int ofs = extract32(ptr, LOG2_TAG_GRANULE, 1) * 4;
+    *mem = deposit32(*mem, ofs, 4, tag);
+}
+
+/* For use in a parallel context, atomically store to the given nibble.  */
+static void store_tag1_parallel(uint64_t ptr, uint8_t *mem, int tag)
+{
+    int ofs = extract32(ptr, LOG2_TAG_GRANULE, 1) * 4;
+    uint8_t old = atomic_read(mem);
+
+    while (1) {
+        uint8_t new = deposit32(old, ofs, 4, tag);
+        uint8_t cmp = atomic_cmpxchg(mem, old, new);
+        if (likely(cmp == old)) {
+            return;
+        }
+        old = cmp;
+    }
+}
+
+typedef void stg_store1(uint64_t, uint8_t *, int);
+
+static inline void do_stg(CPUARMState *env, uint64_t ptr, uint64_t xt,
+                          uintptr_t ra, stg_store1 store1)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uint8_t *mem;
+
+    check_tag_aligned(env, ptr, ra);
+
+    /* Trap if accessing an invalid page.  */
+    mem = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_STORE, TAG_GRANULE,
+                             MMU_DATA_STORE, 1, ra);
+
+    /* Store if page supports tags. */
+    if (mem) {
+        store1(ptr, mem, allocation_tag_from_addr(xt));
+    }
+}
+
+void HELPER(stg)(CPUARMState *env, uint64_t ptr, uint64_t xt)
+{
+    do_stg(env, ptr, xt, GETPC(), store_tag1);
+}
+
+void HELPER(stg_parallel)(CPUARMState *env, uint64_t ptr, uint64_t xt)
+{
+    do_stg(env, ptr, xt, GETPC(), store_tag1_parallel);
+}
+
+void HELPER(stg_stub)(CPUARMState *env, uint64_t ptr)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uintptr_t ra = GETPC();
+
+    check_tag_aligned(env, ptr, ra);
+    probe_write(env, ptr, TAG_GRANULE, mmu_idx, ra);
+}
+
+static inline void do_st2g(CPUARMState *env, uint64_t ptr, uint64_t xt,
+                           uintptr_t ra, stg_store1 store1)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    int tag = allocation_tag_from_addr(xt);
+    uint8_t *mem1, *mem2;
+
+    check_tag_aligned(env, ptr, ra);
+
+    /*
+     * Trap if accessing an invalid page(s).
+     * This takes priority over !allocation_tag_access_enabled.
+     */
+    if (ptr & TAG_GRANULE) {
+        /* Two stores unaligned mod TAG_GRANULE*2 -- modify two bytes. */
+        mem1 = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_STORE,
+                                  TAG_GRANULE, MMU_DATA_STORE, 1, ra);
+        mem2 = allocation_tag_mem(env, mmu_idx, ptr + TAG_GRANULE,
+                                  MMU_DATA_STORE, TAG_GRANULE,
+                                  MMU_DATA_STORE, 1, ra);
+
+        /* Store if page(s) support tags. */
+        if (mem1) {
+            store1(TAG_GRANULE, mem1, tag);
+        }
+        if (mem2) {
+            store1(0, mem2, tag);
+        }
+    } else {
+        /* Two stores aligned mod TAG_GRANULE*2 -- modify one byte. */
+        mem1 = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_STORE,
+                                  2 * TAG_GRANULE, MMU_DATA_STORE, 1, ra);
+        if (mem1) {
+            tag |= tag << 4;
+            atomic_set(mem1, tag);
+        }
+    }
+}
+
+void HELPER(st2g)(CPUARMState *env, uint64_t ptr, uint64_t xt)
+{
+    do_st2g(env, ptr, xt, GETPC(), store_tag1);
+}
+
+void HELPER(st2g_parallel)(CPUARMState *env, uint64_t ptr, uint64_t xt)
+{
+    do_st2g(env, ptr, xt, GETPC(), store_tag1_parallel);
+}
+
+void HELPER(st2g_stub)(CPUARMState *env, uint64_t ptr)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uintptr_t ra = GETPC();
+    int in_page = -(ptr | TARGET_PAGE_MASK);
+
+    check_tag_aligned(env, ptr, ra);
+
+    if (likely(in_page >= 2 * TAG_GRANULE)) {
+        probe_write(env, ptr, 2 * TAG_GRANULE, mmu_idx, ra);
+    } else {
+        probe_write(env, ptr, TAG_GRANULE, mmu_idx, ra);
+        probe_write(env, ptr + TAG_GRANULE, TAG_GRANULE, mmu_idx, ra);
+    }
+}
diff --git a/target/arm/op_helper.c b/target/arm/op_helper.c
index eb0de080f1..b1065216b2 100644
--- a/target/arm/op_helper.c
+++ b/target/arm/op_helper.c
@@ -935,3 +935,19 @@ uint32_t HELPER(ror_cc)(CPUARMState *env, uint32_t x, uint32_t i)
         return ((uint32_t)x >> shift) | (x << (32 - shift));
     }
 }
+
+void HELPER(probe_access)(CPUARMState *env, target_ulong ptr,
+                          uint32_t access_type, uint32_t mmu_idx,
+                          uint32_t size)
+{
+    uint32_t in_page = -((uint32_t)ptr | TARGET_PAGE_SIZE);
+    uintptr_t ra = GETPC();
+
+    if (likely(size <= in_page)) {
+        probe_access(env, ptr, size, access_type, mmu_idx, ra);
+    } else {
+        probe_access(env, ptr, in_page, access_type, mmu_idx, ra);
+        probe_access(env, ptr + in_page, size - in_page,
+                     access_type, mmu_idx, ra);
+    }
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 5400516eef..c6152b3cc0 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -232,6 +232,19 @@ static void gen_address_with_allocation_tag0(TCGv_i64 dst, TCGv_i64 src)
     tcg_gen_andi_i64(dst, src, ~MAKE_64BIT_MASK(56, 4));
 }
 
+static void gen_probe_access(DisasContext *s, TCGv_i64 ptr,
+                             MMUAccessType acc, int log2_size)
+{
+    TCGv_i32 t_acc = tcg_const_i32(acc);
+    TCGv_i32 t_idx = tcg_const_i32(get_mem_index(s));
+    TCGv_i32 t_size = tcg_const_i32(1 << log2_size);
+
+    gen_helper_probe_access(cpu_env, ptr, t_acc, t_idx, t_size);
+    tcg_temp_free_i32(t_acc);
+    tcg_temp_free_i32(t_idx);
+    tcg_temp_free_i32(t_size);
+}
+
 typedef struct DisasCompare64 {
     TCGCond cond;
     TCGv_i64 value;
@@ -3685,6 +3698,154 @@ static void disas_ldst_single_struct(DisasContext *s, uint32_t insn)
     }
 }
 
+/*
+ * Load/Store memory tags
+ *
+ *  31 30 29         24     22  21     12    10      5      0
+ * +-----+-------------+-----+---+------+-----+------+------+
+ * | 1 1 | 0 1 1 0 0 1 | op1 | 1 | imm9 | op2 |  Rn  |  Rt  |
+ * +-----+-------------+-----+---+------+-----+------+------+
+ */
+static void disas_ldst_tag(DisasContext *s, uint32_t insn)
+{
+    int rt = extract32(insn, 0, 5);
+    int rn = extract32(insn, 5, 5);
+    uint64_t offset = sextract64(insn, 12, 9) << LOG2_TAG_GRANULE;
+    int op2 = extract32(insn, 10, 2);
+    int op1 = extract32(insn, 22, 2);
+    bool is_load = false, is_pair = false, is_zero = false;
+    int index = 0;
+    TCGv_i64 addr, clean_addr, tcg_rt;
+
+    /* We checked insn bits [29:24,21] in the caller.  */
+    if (extract32(insn, 30, 2) != 3) {
+        goto do_unallocated;
+    }
+
+    /*
+     * @index is a tri-state variable which has 3 states:
+     * < 0 : post-index, writeback
+     * = 0 : signed offset
+     * > 0 : pre-index, writeback
+     */
+    switch (op1) {
+    case 0:
+        if (op2 != 0) {
+            /* STG */
+            index = op2 - 2;
+            break;
+        }
+        goto do_unallocated;
+    case 1:
+        if (op2 != 0) {
+            /* STZG */
+            is_zero = true;
+            index = op2 - 2;
+        } else {
+            /* LDG */
+            is_load = true;
+        }
+        break;
+    case 2:
+        if (op2 != 0) {
+            /* ST2G */
+            is_pair = true;
+            index = op2 - 2;
+            break;
+        }
+        goto do_unallocated;
+    case 3:
+        if (op2 != 0) {
+            /* STZ2G */
+            is_pair = is_zero = true;
+            index = op2 - 2;
+            break;
+        }
+        goto do_unallocated;
+
+    default:
+    do_unallocated:
+        unallocated_encoding(s);
+        return;
+    }
+
+    if (!dc_isar_feature(aa64_mte_insn_reg, s)) {
+        goto do_unallocated;
+    }
+
+    if (rn == 31) {
+        gen_check_sp_alignment(s);
+    }
+
+    addr = read_cpu_reg_sp(s, rn, true);
+    if (index >= 0) {
+        /* pre-index or signed offset */
+        tcg_gen_addi_i64(addr, addr, offset);
+    }
+
+    if (is_load) {
+        tcg_gen_andi_i64(addr, addr, -TAG_GRANULE);
+        tcg_rt = cpu_reg(s, rt);
+        if (s->ata) {
+            gen_helper_ldg(tcg_rt, cpu_env, addr, tcg_rt);
+        } else {
+            clean_addr = clean_data_tbi(s, addr);
+            gen_probe_access(s, clean_addr, MMU_DATA_LOAD, MO_8);
+            gen_address_with_allocation_tag0(tcg_rt, addr);
+        }
+    } else {
+        tcg_rt = cpu_reg_sp(s, rt);
+        if (!s->ata) {
+            /*
+             * For STG and ST2G, we need to check alignment and probe memory.
+             * TODO: For STZG and STZ2G, we could rely on the stores below,
+             * at least for system mode; user-only won't enforce alignment.
+             */
+            if (is_pair) {
+                gen_helper_st2g_stub(cpu_env, addr);
+            } else {
+                gen_helper_stg_stub(cpu_env, addr);
+            }
+        } else if (tb_cflags(s->base.tb) & CF_PARALLEL) {
+            if (is_pair) {
+                gen_helper_st2g_parallel(cpu_env, addr, tcg_rt);
+            } else {
+                gen_helper_stg_parallel(cpu_env, addr, tcg_rt);
+            }
+        } else {
+            if (is_pair) {
+                gen_helper_st2g(cpu_env, addr, tcg_rt);
+            } else {
+                gen_helper_stg(cpu_env, addr, tcg_rt);
+            }
+        }
+    }
+
+    if (is_zero) {
+        TCGv_i64 clean_addr = clean_data_tbi(s, addr);
+        TCGv_i64 tcg_zero = tcg_const_i64(0);
+        int mem_index = get_mem_index(s);
+        int i, n = (1 + is_pair) << LOG2_TAG_GRANULE;
+
+        tcg_gen_qemu_st_i64(tcg_zero, clean_addr, mem_index,
+                            MO_Q | MO_ALIGN_16);
+        for (i = 8; i < n; i += 8) {
+            tcg_gen_addi_i64(clean_addr, clean_addr, 8);
+            tcg_gen_qemu_st_i64(tcg_zero, clean_addr, mem_index, MO_Q);
+        }
+        tcg_temp_free_i64(tcg_zero);
+    }
+
+    if (index != 0) {
+        /* pre-index or post-index */
+        if (index < 0) {
+            /* post-index */
+            tcg_gen_addi_i64(addr, addr, offset);
+        }
+        tcg_gen_mov_i64(cpu_reg_sp(s, rn), addr);
+    }
+}
+
 /* Loads and stores */
 static void disas_ldst(DisasContext *s, uint32_t insn)
 {
@@ -3709,13 +3870,14 @@ static void disas_ldst(DisasContext *s, uint32_t insn)
     case 0x0d: /* AdvSIMD load/store single structure */
         disas_ldst_single_struct(s, insn);
         break;
-    case 0x19: /* LDAPR/STLR (unscaled immediate) */
-        if (extract32(insn, 10, 2) != 0 ||
-            extract32(insn, 21, 1) != 0) {
+    case 0x19:
+        if (extract32(insn, 21, 1) != 0) {
+            disas_ldst_tag(s, insn);
+        } else if (extract32(insn, 10, 2) == 0) {
+            disas_ldst_ldapr_stlr(s, insn);
+        } else {
             unallocated_encoding(s);
-            break;
         }
-        disas_ldst_ldapr_stlr(s, insn);
         break;
     default:
         unallocated_encoding(s);
-- 
2.25.1

[PATCH v8 16/45] target/arm: Implement the STGP instruction

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v3: Handle atomicity, require pre-cleaned address.
v6: Fix constant offset shift, non-checked address, use pre-computed ata.
---
 target/arm/translate-a64.c | 29 ++++++++++++++++++++++++++---
 1 file changed, 26 insertions(+), 3 deletions(-)

diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index c6152b3cc0..5d8c9483bc 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -2690,7 +2690,7 @@ static void disas_ld_lit(DisasContext *s, uint32_t insn)
  * +-----+-------+---+---+-------+---+-------+-------+------+------+
  *
  * opc: LDP/STP/LDNP/STNP        00 -> 32 bit, 10 -> 64 bit
- *      LDPSW                    01
+ *      LDPSW/STGP               01
  *      LDP/STP/LDNP/STNP (SIMD) 00 -> 32 bit, 01 -> 64 bit, 10 -> 128 bit
  *   V: 0 -> GPR, 1 -> Vector
  * idx: 00 -> signed offset with non-temporal hint, 01 -> post-index,
@@ -2715,6 +2715,7 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
     bool is_signed = false;
     bool postindex = false;
     bool wback = false;
+    bool set_tag = false;
 
     TCGv_i64 clean_addr, dirty_addr;
 
@@ -2727,6 +2728,14 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
 
     if (is_vector) {
         size = 2 + opc;
+    } else if (opc == 1 && !is_load) {
+        /* STGP */
+        if (!dc_isar_feature(aa64_mte_insn_reg, s) || index == 0) {
+            unallocated_encoding(s);
+            return;
+        }
+        size = 3;
+        set_tag = true;
     } else {
         size = 2 + extract32(opc, 1, 1);
         is_signed = extract32(opc, 0, 1);
@@ -2767,7 +2776,7 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
         return;
     }
 
-    offset <<= size;
+    offset <<= (set_tag ? LOG2_TAG_GRANULE : size);
 
     if (rn == 31) {
         gen_check_sp_alignment(s);
@@ -2777,8 +2786,22 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
     if (!postindex) {
         tcg_gen_addi_i64(dirty_addr, dirty_addr, offset);
     }
-    clean_addr = clean_data_tbi(s, dirty_addr);
 
+    if (set_tag) {
+        if (!s->ata) {
+            /*
+             * TODO: We could rely on the stores below, at least for
+             * system mode, if we arrange to add MO_ALIGN_16.
+             */
+            gen_helper_stg_stub(cpu_env, dirty_addr);
+        } else if (tb_cflags(s->base.tb) & CF_PARALLEL) {
+            gen_helper_stg_parallel(cpu_env, dirty_addr, dirty_addr);
+        } else {
+            gen_helper_stg(cpu_env, dirty_addr, dirty_addr);
+        }
+    }
+
+    clean_addr = clean_data_tbi(s, dirty_addr);
     if (is_vector) {
         if (is_load) {
             do_fp_ld(s, rt, clean_addr, size);
-- 
2.25.1

[PATCH v8 17/45] target/arm: Restrict the values of DCZID.BS under TCG

[Richard Henderson]

We can simplify our DC_ZVA if we recognize that the largest BS
that we actually use in system mode is 64.  Let us just assert
that it fits within TARGET_PAGE_SIZE.

For DC_GVA and STZGM, we want to be able to write whole bytes
of tag memory, so assert that BS is >= 2 * TAG_GRANULE, or 32.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/cpu.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 10677c0c23..f09efc4370 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -1758,6 +1758,30 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
     }
 #endif
 
+    if (tcg_enabled()) {
+        int dcz_blocklen = 4 << cpu->dcz_blocksize;
+
+        /*
+         * We only support DCZ blocklen that fits on one page.
+         *
+         * Architectually this is always true.  However TARGET_PAGE_SIZE
+         * is variable and, for compatibility with -machine virt-2.7,
+         * is only 1KiB, as an artifact of legacy ARMv5 subpage support.
+         * But even then, while the largest architectural DCZ blocklen
+         * is 2KiB, no cpu actually uses such a large blocklen.
+         */
+        assert(dcz_blocklen <= TARGET_PAGE_SIZE);
+
+        /*
+         * We only support DCZ blocksize >= 2*TAG_GRANULE, which is to say
+         * both nibbles of each byte storing tag data may be written at once.
+         * Since TAG_GRANULE is 16, this means that blocklen must be >= 32.
+         */
+        if (cpu_isar_feature(aa64_mte, cpu)) {
+            assert(dcz_blocklen >= 2 * TAG_GRANULE);
+        }
+    }
+
     qemu_init_vcpu(cs);
     cpu_reset(cs);
 
-- 
2.25.1

[PATCH v8 18/45] target/arm: Simplify DC_ZVA

[Richard Henderson]

Now that we know that the operation is on a single page,
we need not loop over pages while probing.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-a64.c | 94 +++++++++++------------------------------
 1 file changed, 25 insertions(+), 69 deletions(-)

diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c
index bc0649a44a..8682630ff6 100644
--- a/target/arm/helper-a64.c
+++ b/target/arm/helper-a64.c
@@ -1119,85 +1119,41 @@ void HELPER(dc_zva)(CPUARMState *env, uint64_t vaddr_in)
      * (which matches the usual QEMU behaviour of not implementing either
      * alignment faults or any memory attribute handling).
      */
-
-    ARMCPU *cpu = env_archcpu(env);
-    uint64_t blocklen = 4 << cpu->dcz_blocksize;
+    int blocklen = 4 << env_archcpu(env)->dcz_blocksize;
     uint64_t vaddr = vaddr_in & ~(blocklen - 1);
+    int mmu_idx = cpu_mmu_index(env, false);
+    void *mem;
+
+    /*
+     * Trapless lookup.  In addition to actual invalid page, may
+     * return NULL for I/O, watchpoints, clean pages, etc.
+     */
+    mem = tlb_vaddr_to_host(env, vaddr, MMU_DATA_STORE, mmu_idx);
 
 #ifndef CONFIG_USER_ONLY
-    {
+    if (unlikely(!mem)) {
+        uintptr_t ra = GETPC();
+
         /*
-         * Slightly awkwardly, QEMU's TARGET_PAGE_SIZE may be less than
-         * the block size so we might have to do more than one TLB lookup.
-         * We know that in fact for any v8 CPU the page size is at least 4K
-         * and the block size must be 2K or less, but TARGET_PAGE_SIZE is only
-         * 1K as an artefact of legacy v5 subpage support being present in the
-         * same QEMU executable. So in practice the hostaddr[] array has
-         * two entries, given the current setting of TARGET_PAGE_BITS_MIN.
+         * Trap if accessing an invalid page.  DC_ZVA requires that we supply
+         * the original pointer for an invalid page.  But watchpoints require
+         * that we probe the actual space.  So do both.
          */
-        int maxidx = DIV_ROUND_UP(blocklen, TARGET_PAGE_SIZE);
-        void *hostaddr[DIV_ROUND_UP(2 * KiB, 1 << TARGET_PAGE_BITS_MIN)];
-        int try, i;
-        unsigned mmu_idx = cpu_mmu_index(env, false);
-        TCGMemOpIdx oi = make_memop_idx(MO_UB, mmu_idx);
+        (void) probe_write(env, vaddr_in, 1, mmu_idx, ra);
+        mem = probe_write(env, vaddr, blocklen, mmu_idx, ra);
 
-        assert(maxidx <= ARRAY_SIZE(hostaddr));
-
-        for (try = 0; try < 2; try++) {
-
-            for (i = 0; i < maxidx; i++) {
-                hostaddr[i] = tlb_vaddr_to_host(env,
-                                                vaddr + TARGET_PAGE_SIZE * i,
-                                                1, mmu_idx);
-                if (!hostaddr[i]) {
-                    break;
-                }
-            }
-            if (i == maxidx) {
-                /*
-                 * If it's all in the TLB it's fair game for just writing to;
-                 * we know we don't need to update dirty status, etc.
-                 */
-                for (i = 0; i < maxidx - 1; i++) {
-                    memset(hostaddr[i], 0, TARGET_PAGE_SIZE);
-                }
-                memset(hostaddr[i], 0, blocklen - (i * TARGET_PAGE_SIZE));
-                return;
-            }
+        if (unlikely(!mem)) {
             /*
-             * OK, try a store and see if we can populate the tlb. This
-             * might cause an exception if the memory isn't writable,
-             * in which case we will longjmp out of here. We must for
-             * this purpose use the actual register value passed to us
-             * so that we get the fault address right.
+             * The only remaining reason for mem == NULL is I/O.
+             * Just do a series of byte writes as the architecture demands.
              */
-            helper_ret_stb_mmu(env, vaddr_in, 0, oi, GETPC());
-            /* Now we can populate the other TLB entries, if any */
-            for (i = 0; i < maxidx; i++) {
-                uint64_t va = vaddr + TARGET_PAGE_SIZE * i;
-                if (va != (vaddr_in & TARGET_PAGE_MASK)) {
-                    helper_ret_stb_mmu(env, va, 0, oi, GETPC());
-                }
+            for (int i = 0; i < blocklen; i++) {
+                cpu_stb_mmuidx_ra(env, vaddr + i, 0, mmu_idx, ra);
             }
-        }
-
-        /*
-         * Slow path (probably attempt to do this to an I/O device or
-         * similar, or clearing of a block of code we have translations
-         * cached for). Just do a series of byte writes as the architecture
-         * demands. It's not worth trying to use a cpu_physical_memory_map(),
-         * memset(), unmap() sequence here because:
-         *  + we'd need to account for the blocksize being larger than a page
-         *  + the direct-RAM access case is almost always going to be dealt
-         *    with in the fastpath code above, so there's no speed benefit
-         *  + we would have to deal with the map returning NULL because the
-         *    bounce buffer was in use
-         */
-        for (i = 0; i < blocklen; i++) {
-            helper_ret_stb_mmu(env, vaddr + i, 0, oi, GETPC());
+            return;
         }
     }
-#else
-    memset(g2h(vaddr), 0, blocklen);
 #endif
+
+    memset(mem, 0, blocklen);
 }
-- 
2.25.1

[PATCH v8 19/45] target/arm: Implement the LDGM, STGM, STZGM instructions

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v3: Require pre-cleaned addresses.
v6: Check full mte enabled.  Reorg the helpers.
---
 target/arm/helper-a64.h    |  3 ++
 target/arm/translate.h     |  2 +
 target/arm/mte_helper.c    | 84 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c | 72 ++++++++++++++++++++++++++++----
 4 files changed, 153 insertions(+), 8 deletions(-)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 2fa61b86fa..7b628d100e 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -113,3 +113,6 @@ DEF_HELPER_FLAGS_2(stg_stub, TCG_CALL_NO_WG, void, env, i64)
 DEF_HELPER_FLAGS_3(st2g, TCG_CALL_NO_WG, void, env, i64, i64)
 DEF_HELPER_FLAGS_3(st2g_parallel, TCG_CALL_NO_WG, void, env, i64, i64)
 DEF_HELPER_FLAGS_2(st2g_stub, TCG_CALL_NO_WG, void, env, i64)
+DEF_HELPER_FLAGS_2(ldgm, TCG_CALL_NO_WG, i64, env, i64)
+DEF_HELPER_FLAGS_3(stgm, TCG_CALL_NO_WG, void, env, i64, i64)
+DEF_HELPER_FLAGS_3(stzgm_tags, TCG_CALL_NO_WG, void, env, i64, i64)
diff --git a/target/arm/translate.h b/target/arm/translate.h
index dbbf6145cb..82c598beff 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -91,6 +91,8 @@ typedef struct DisasContext {
      *  < 0, set by the current instruction.
      */
     int8_t btype;
+    /* A copy of cpu->dcz_blocksize. */
+    uint8_t dcz_blocksize;
     /* True if this page is guarded.  */
     bool guarded_page;
     /* Bottom two bits of XScale c15_cpar coprocessor access control reg */
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 7ec7930dfc..27d4b4536c 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -274,3 +274,87 @@ void HELPER(st2g_stub)(CPUARMState *env, uint64_t ptr)
         probe_write(env, ptr + TAG_GRANULE, TAG_GRANULE, mmu_idx, ra);
     }
 }
+
+#define LDGM_STGM_SIZE  (4 << GMID_EL1_BS)
+
+uint64_t HELPER(ldgm)(CPUARMState *env, uint64_t ptr)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uintptr_t ra = GETPC();
+    void *tag_mem;
+
+    ptr = QEMU_ALIGN_DOWN(ptr, LDGM_STGM_SIZE);
+
+    /* Trap if accessing an invalid page.  */
+    tag_mem = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_LOAD,
+                                 LDGM_STGM_SIZE, MMU_DATA_LOAD,
+                                 LDGM_STGM_SIZE / (2 * TAG_GRANULE), ra);
+
+    /* The tag is squashed to zero if the page does not support tags.  */
+    if (!tag_mem) {
+        return 0;
+    }
+
+    QEMU_BUILD_BUG_ON(GMID_EL1_BS != 6);
+    /*
+     * We are loading 64-bits worth of tags.  The ordering of elements
+     * within the word corresponds to a 64-bit little-endian operation.
+     */
+    return ldq_le_p(tag_mem);
+}
+
+void HELPER(stgm)(CPUARMState *env, uint64_t ptr, uint64_t val)
+{
+    int mmu_idx = cpu_mmu_index(env, false);
+    uintptr_t ra = GETPC();
+    void *tag_mem;
+
+    ptr = QEMU_ALIGN_DOWN(ptr, LDGM_STGM_SIZE);
+
+    /* Trap if accessing an invalid page.  */
+    tag_mem = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_STORE,
+                                 LDGM_STGM_SIZE, MMU_DATA_LOAD,
+                                 LDGM_STGM_SIZE / (2 * TAG_GRANULE), ra);
+
+    /*
+     * Tag store only happens if the page support tags,
+     * and if the OS has enabled access to the tags.
+     */
+    if (!tag_mem) {
+        return;
+    }
+
+    QEMU_BUILD_BUG_ON(GMID_EL1_BS != 6);
+    /*
+     * We are storing 64-bits worth of tags.  The ordering of elements
+     * within the word corresponds to a 64-bit little-endian operation.
+     */
+    stq_le_p(tag_mem, val);
+}
+
+void HELPER(stzgm_tags)(CPUARMState *env, uint64_t ptr, uint64_t val)
+{
+    uintptr_t ra = GETPC();
+    int mmu_idx = cpu_mmu_index(env, false);
+    int log2_dcz_bytes, log2_tag_bytes;
+    intptr_t dcz_bytes, tag_bytes;
+    uint8_t *mem;
+
+    /*
+     * In arm_cpu_realizefn, we assert that dcz > LOG2_TAG_GRANULE+1,
+     * i.e. 32 bytes, which is an unreasonably small dcz anyway,
+     * to make sure that we can access one complete tag byte here.
+     */
+    log2_dcz_bytes = env_archcpu(env)->dcz_blocksize + 2;
+    log2_tag_bytes = log2_dcz_bytes - (LOG2_TAG_GRANULE + 1);
+    dcz_bytes = (intptr_t)1 << log2_dcz_bytes;
+    tag_bytes = (intptr_t)1 << log2_tag_bytes;
+    ptr &= -dcz_bytes;
+
+    mem = allocation_tag_mem(env, mmu_idx, ptr, MMU_DATA_STORE, dcz_bytes,
+                             MMU_DATA_STORE, tag_bytes, ra);
+    if (mem) {
+        int tag_pair = (val & 0xf) * 0x11;
+        memset(mem, tag_pair, tag_bytes);
+    }
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 5d8c9483bc..b7b2331ccc 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -3736,7 +3736,7 @@ static void disas_ldst_tag(DisasContext *s, uint32_t insn)
     uint64_t offset = sextract64(insn, 12, 9) << LOG2_TAG_GRANULE;
     int op2 = extract32(insn, 10, 2);
     int op1 = extract32(insn, 22, 2);
-    bool is_load = false, is_pair = false, is_zero = false;
+    bool is_load = false, is_pair = false, is_zero = false, is_mult = false;
     int index = 0;
     TCGv_i64 addr, clean_addr, tcg_rt;
 
@@ -3756,9 +3756,14 @@ static void disas_ldst_tag(DisasContext *s, uint32_t insn)
         if (op2 != 0) {
             /* STG */
             index = op2 - 2;
-            break;
+        } else {
+            /* STZGM */
+            if (s->current_el == 0 || offset != 0) {
+                goto do_unallocated;
+            }
+            is_mult = is_zero = true;
         }
-        goto do_unallocated;
+        break;
     case 1:
         if (op2 != 0) {
             /* STZG */
@@ -3774,17 +3779,27 @@ static void disas_ldst_tag(DisasContext *s, uint32_t insn)
             /* ST2G */
             is_pair = true;
             index = op2 - 2;
-            break;
+        } else {
+            /* STGM */
+            if (s->current_el == 0 || offset != 0) {
+                goto do_unallocated;
+            }
+            is_mult = true;
         }
-        goto do_unallocated;
+        break;
     case 3:
         if (op2 != 0) {
             /* STZ2G */
             is_pair = is_zero = true;
             index = op2 - 2;
-            break;
+        } else {
+            /* LDGM */
+            if (s->current_el == 0 || offset != 0) {
+                goto do_unallocated;
+            }
+            is_mult = is_load = true;
         }
-        goto do_unallocated;
+        break;
 
     default:
     do_unallocated:
@@ -3792,7 +3807,9 @@ static void disas_ldst_tag(DisasContext *s, uint32_t insn)
         return;
     }
 
-    if (!dc_isar_feature(aa64_mte_insn_reg, s)) {
+    if (is_mult
+        ? !dc_isar_feature(aa64_mte, s)
+        : !dc_isar_feature(aa64_mte_insn_reg, s)) {
         goto do_unallocated;
     }
 
@@ -3806,6 +3823,44 @@ static void disas_ldst_tag(DisasContext *s, uint32_t insn)
         tcg_gen_addi_i64(addr, addr, offset);
     }
 
+    if (is_mult) {
+        tcg_rt = cpu_reg(s, rt);
+
+        if (is_zero) {
+            int size = 4 << s->dcz_blocksize;
+
+            if (s->ata) {
+                gen_helper_stzgm_tags(cpu_env, addr, tcg_rt);
+            }
+            /*
+             * The non-tags portion of STZGM is mostly like DC_ZVA,
+             * except the alignment happens before the access.
+             */
+            clean_addr = clean_data_tbi(s, addr);
+            tcg_gen_andi_i64(clean_addr, clean_addr, -size);
+            gen_helper_dc_zva(cpu_env, clean_addr);
+        } else if (s->ata) {
+            if (is_load) {
+                gen_helper_ldgm(tcg_rt, cpu_env, addr);
+            } else {
+                gen_helper_stgm(cpu_env, addr, tcg_rt);
+            }
+        } else {
+            MMUAccessType acc = is_load ? MMU_DATA_LOAD : MMU_DATA_STORE;
+            int size = 4 << GMID_EL1_BS;
+
+            clean_addr = clean_data_tbi(s, addr);
+            tcg_gen_andi_i64(clean_addr, clean_addr, -size);
+            gen_probe_access(s, clean_addr, acc, size);
+
+            if (is_load) {
+                /* The result tags are zeros.  */
+                tcg_gen_movi_i64(tcg_rt, 0);
+            }
+        }
+        return;
+    }
+
     if (is_load) {
         tcg_gen_andi_i64(addr, addr, -TAG_GRANULE);
         tcg_rt = cpu_reg(s, rt);
@@ -14484,6 +14539,7 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->vec_stride = 0;
     dc->cp_regs = arm_cpu->cp_regs;
     dc->features = env->features;
+    dc->dcz_blocksize = arm_cpu->dcz_blocksize;
 
     /* Single step state. The code-generation logic here is:
      *  SS_ACTIVE == 0:
-- 
2.25.1

[PATCH v8 20/45] target/arm: Implement the access tag cache flushes

[Richard Henderson]

Like the regular data cache flushes, these are nops within qemu.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v6: Split out and handle el0 cache ops properly.
---
 target/arm/helper.c | 65 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 65 insertions(+)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 2c6ec244af..d8c31d03da 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6929,6 +6929,32 @@ static const ARMCPRegInfo mte_reginfo[] = {
       .opc0 = 3, .opc1 = 3, .crn = 4, .crm = 2, .opc2 = 7,
       .type = ARM_CP_NO_RAW,
       .access = PL0_RW, .readfn = tco_read, .writefn = tco_write },
+    { .name = "DC_IGVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 6, .opc2 = 3,
+      .type = ARM_CP_NOP, .access = PL1_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_IGSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 6, .opc2 = 4,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
+    { .name = "DC_IGDVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 6, .opc2 = 5,
+      .type = ARM_CP_NOP, .access = PL1_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_IGDSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 6, .opc2 = 6,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
+    { .name = "DC_CGSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 10, .opc2 = 4,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
+    { .name = "DC_CGDSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 10, .opc2 = 6,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
+    { .name = "DC_CIGSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 14, .opc2 = 4,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
+    { .name = "DC_CIGDSW", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 0, .crn = 7, .crm = 14, .opc2 = 6,
+      .type = ARM_CP_NOP, .access = PL1_W, .accessfn = access_tsw },
     REGINFO_SENTINEL
 };
 
@@ -6938,6 +6964,43 @@ static const ARMCPRegInfo mte_tco_ro_reginfo[] = {
       .type = ARM_CP_CONST, .access = PL0_RW, },
     REGINFO_SENTINEL
 };
+
+static const ARMCPRegInfo mte_el0_cacheop_reginfo[] = {
+    { .name = "DC_CGVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 10, .opc2 = 3,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CGDVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 10, .opc2 = 5,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CGVAP", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 12, .opc2 = 3,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CGDVAP", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 12, .opc2 = 5,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CGVADP", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 13, .opc2 = 3,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CGDVADP", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 13, .opc2 = 5,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CIGVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 14, .opc2 = 3,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_CIGDVAC", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 14, .opc2 = 5,
+      .type = ARM_CP_NOP, .access = PL0_W,
+      .accessfn = aa64_cacheop_poc_access },
+    REGINFO_SENTINEL
+};
+
 #endif
 
 static CPAccessResult access_predinv(CPUARMState *env, const ARMCPRegInfo *ri,
@@ -8071,8 +8134,10 @@ void register_cp_regs_for_features(ARMCPU *cpu)
      */
     if (cpu_isar_feature(aa64_mte, cpu)) {
         define_arm_cp_regs(cpu, mte_reginfo);
+        define_arm_cp_regs(cpu, mte_el0_cacheop_reginfo);
     } else if (cpu_isar_feature(aa64_mte_insn_reg, cpu)) {
         define_arm_cp_regs(cpu, mte_tco_ro_reginfo);
+        define_arm_cp_regs(cpu, mte_el0_cacheop_reginfo);
     }
 #endif
 
-- 
2.25.1

[PATCH v8 21/45] target/arm: Move regime_el to internals.h

[Richard Henderson]

We will shortly need this in mte_helper.c as well.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/internals.h | 36 ++++++++++++++++++++++++++++++++++++
 target/arm/helper.c    | 36 ------------------------------------
 2 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/target/arm/internals.h b/target/arm/internals.h
index 5c69d4e5a5..c36fcb151b 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -913,6 +913,42 @@ static inline bool regime_is_pan(CPUARMState *env, ARMMMUIdx mmu_idx)
     }
 }
 
+/* Return the exception level which controls this address translation regime */
+static inline uint32_t regime_el(CPUARMState *env, ARMMMUIdx mmu_idx)
+{
+    switch (mmu_idx) {
+    case ARMMMUIdx_E20_0:
+    case ARMMMUIdx_E20_2:
+    case ARMMMUIdx_E20_2_PAN:
+    case ARMMMUIdx_Stage2:
+    case ARMMMUIdx_E2:
+        return 2;
+    case ARMMMUIdx_SE3:
+        return 3;
+    case ARMMMUIdx_SE10_0:
+        return arm_el_is_aa64(env, 3) ? 1 : 3;
+    case ARMMMUIdx_SE10_1:
+    case ARMMMUIdx_SE10_1_PAN:
+    case ARMMMUIdx_Stage1_E0:
+    case ARMMMUIdx_Stage1_E1:
+    case ARMMMUIdx_Stage1_E1_PAN:
+    case ARMMMUIdx_E10_0:
+    case ARMMMUIdx_E10_1:
+    case ARMMMUIdx_E10_1_PAN:
+    case ARMMMUIdx_MPrivNegPri:
+    case ARMMMUIdx_MUserNegPri:
+    case ARMMMUIdx_MPriv:
+    case ARMMMUIdx_MUser:
+    case ARMMMUIdx_MSPrivNegPri:
+    case ARMMMUIdx_MSUserNegPri:
+    case ARMMMUIdx_MSPriv:
+    case ARMMMUIdx_MSUser:
+        return 1;
+    default:
+        g_assert_not_reached();
+    }
+}
+
 /* Return the FSR value for a debug exception (watchpoint, hardware
  * breakpoint or BKPT insn) targeting the specified exception level.
  */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index d8c31d03da..d14313de66 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -9793,42 +9793,6 @@ void arm_cpu_do_interrupt(CPUState *cs)
 }
 #endif /* !CONFIG_USER_ONLY */
 
-/* Return the exception level which controls this address translation regime */
-static uint32_t regime_el(CPUARMState *env, ARMMMUIdx mmu_idx)
-{
-    switch (mmu_idx) {
-    case ARMMMUIdx_E20_0:
-    case ARMMMUIdx_E20_2:
-    case ARMMMUIdx_E20_2_PAN:
-    case ARMMMUIdx_Stage2:
-    case ARMMMUIdx_E2:
-        return 2;
-    case ARMMMUIdx_SE3:
-        return 3;
-    case ARMMMUIdx_SE10_0:
-        return arm_el_is_aa64(env, 3) ? 1 : 3;
-    case ARMMMUIdx_SE10_1:
-    case ARMMMUIdx_SE10_1_PAN:
-    case ARMMMUIdx_Stage1_E0:
-    case ARMMMUIdx_Stage1_E1:
-    case ARMMMUIdx_Stage1_E1_PAN:
-    case ARMMMUIdx_E10_0:
-    case ARMMMUIdx_E10_1:
-    case ARMMMUIdx_E10_1_PAN:
-    case ARMMMUIdx_MPrivNegPri:
-    case ARMMMUIdx_MUserNegPri:
-    case ARMMMUIdx_MPriv:
-    case ARMMMUIdx_MUser:
-    case ARMMMUIdx_MSPrivNegPri:
-    case ARMMMUIdx_MSUserNegPri:
-    case ARMMMUIdx_MSPriv:
-    case ARMMMUIdx_MSUser:
-        return 1;
-    default:
-        g_assert_not_reached();
-    }
-}
-
 uint64_t arm_sctlr(CPUARMState *env, int el)
 {
     /* Only EL0 needs to be adjusted for EL1&0 or EL2&0. */
-- 
2.25.1

[PATCH v8 22/45] target/arm: Move regime_tcr to internals.h

[Richard Henderson]

We will shortly need this in mte_helper.c as well.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/internals.h | 9 +++++++++
 target/arm/helper.c    | 9 ---------
 2 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/target/arm/internals.h b/target/arm/internals.h
index c36fcb151b..7c9abbabc9 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -949,6 +949,15 @@ static inline uint32_t regime_el(CPUARMState *env, ARMMMUIdx mmu_idx)
     }
 }
 
+/* Return the TCR controlling this translation regime */
+static inline TCR *regime_tcr(CPUARMState *env, ARMMMUIdx mmu_idx)
+{
+    if (mmu_idx == ARMMMUIdx_Stage2) {
+        return &env->cp15.vtcr_el2;
+    }
+    return &env->cp15.tcr_el[regime_el(env, mmu_idx)];
+}
+
 /* Return the FSR value for a debug exception (watchpoint, hardware
  * breakpoint or BKPT insn) targeting the specified exception level.
  */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index d14313de66..33f902387b 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -9875,15 +9875,6 @@ static inline uint64_t regime_ttbr(CPUARMState *env, ARMMMUIdx mmu_idx,
 
 #endif /* !CONFIG_USER_ONLY */
 
-/* Return the TCR controlling this translation regime */
-static inline TCR *regime_tcr(CPUARMState *env, ARMMMUIdx mmu_idx)
-{
-    if (mmu_idx == ARMMMUIdx_Stage2) {
-        return &env->cp15.vtcr_el2;
-    }
-    return &env->cp15.tcr_el[regime_el(env, mmu_idx)];
-}
-
 /* Convert a possible stage1+2 MMU index into the appropriate
  * stage 1 MMU index
  */
-- 
2.25.1

[PATCH v8 23/45] target/arm: Add gen_mte_check1

[Richard Henderson]

Replace existing uses of check_data_tbi in translate-a64.c that
perform a single logical memory access.  Leave the helper blank
for now to reduce the patch size.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-a64.h    |   1 +
 target/arm/internals.h     |   8 +++
 target/arm/translate-a64.h |   2 +
 target/arm/mte_helper.c    |   8 +++
 target/arm/translate-a64.c | 100 ++++++++++++++++++++++++++++---------
 5 files changed, 95 insertions(+), 24 deletions(-)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 7b628d100e..2faa49d0a3 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -104,6 +104,7 @@ DEF_HELPER_FLAGS_3(autdb, TCG_CALL_NO_WG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64)
 DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
 
+DEF_HELPER_FLAGS_3(mte_check1, TCG_CALL_NO_WG, i64, env, i32, i64)
 DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_4(addsubg, TCG_CALL_NO_RWG_SE, i64, env, i64, s32, i32)
 DEF_HELPER_FLAGS_3(ldg, TCG_CALL_NO_WG, i64, env, i64, i64)
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 7c9abbabc9..fb92ef6b84 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1310,6 +1310,14 @@ void arm_log_exception(int idx);
 #define LOG2_TAG_GRANULE 4
 #define TAG_GRANULE      (1 << LOG2_TAG_GRANULE)
 
+/* Bits within a descriptor passed to the helper_mte_check* functions. */
+FIELD(MTEDESC, MIDX,  0, 4)
+FIELD(MTEDESC, TBI,   4, 2)
+FIELD(MTEDESC, TCMA,  6, 2)
+FIELD(MTEDESC, WRITE, 8, 1)
+FIELD(MTEDESC, ESIZE, 9, 5)
+FIELD(MTEDESC, TSIZE, 14, 10)  /* mte_checkN only */
+
 static inline int allocation_tag_from_addr(uint64_t ptr)
 {
     return extract64(ptr, 56, 4);
diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h
index da0f59a2ce..daab6a9666 100644
--- a/target/arm/translate-a64.h
+++ b/target/arm/translate-a64.h
@@ -40,6 +40,8 @@ TCGv_ptr get_fpstatus_ptr(bool);
 bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn,
                             unsigned int imms, unsigned int immr);
 bool sve_access_check(DisasContext *s);
+TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write,
+                        bool tag_checked, int log2_size);
 
 /* We should have at some point before trying to access an FP register
  * done the necessary access check, so assert that
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 27d4b4536c..ec12768dfc 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -358,3 +358,11 @@ void HELPER(stzgm_tags)(CPUARMState *env, uint64_t ptr, uint64_t val)
         memset(mem, tag_pair, tag_bytes);
     }
 }
+
+/*
+ * Perform an MTE checked access for a single logical or atomic access.
+ */
+uint64_t HELPER(mte_check1)(CPUARMState *env, uint32_t desc, uint64_t ptr)
+{
+    return ptr;
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index b7b2331ccc..0720c39e89 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -204,20 +204,20 @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src)
 }
 
 /*
- * Return a "clean" address for ADDR according to TBID.
- * This is always a fresh temporary, as we need to be able to
- * increment this independently of a dirty write-back address.
+ * Handle MTE and/or TBI.
+ *
+ * For TBI, ideally, we would do nothing.  Proper behaviour on fault is
+ * for the tag to be present in the FAR_ELx register.  But for user-only
+ * mode we do not have a TLB with which to implement this, so we must
+ * remove the top byte now.
+ *
+ * Always return a fresh temporary that we can increment independently
+ * of the write-back address.
  */
+
 static TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr)
 {
     TCGv_i64 clean = new_tmp_a64(s);
-    /*
-     * In order to get the correct value in the FAR_ELx register,
-     * we must present the memory subsystem with the "dirty" address
-     * including the TBI.  In system mode we can make this work via
-     * the TLB, dropping the TBI during translation.  But for user-only
-     * mode we don't have that option, and must remove the top byte now.
-     */
 #ifdef CONFIG_USER_ONLY
     gen_top_byte_ignore(s, clean, addr, s->tbid);
 #else
@@ -245,6 +245,45 @@ static void gen_probe_access(DisasContext *s, TCGv_i64 ptr,
     tcg_temp_free_i32(t_size);
 }
 
+/*
+ * For MTE, check a single logical or atomic access.  This probes a single
+ * address, the exact one specified.  The size and alignment of the access
+ * is not relevant to MTE, per se, but watchpoints do require the size,
+ * and we want to recognize those before making any other changes to state.
+ */
+static TCGv_i64 gen_mte_check1_mmuidx(DisasContext *s, TCGv_i64 addr,
+                                      bool is_write, bool tag_checked,
+                                      int log2_size, bool is_unpriv,
+                                      int core_idx)
+{
+    if (tag_checked && s->mte_active[is_unpriv]) {
+        TCGv_i32 tcg_desc;
+        TCGv_i64 ret;
+        int desc = 0;
+
+        desc = FIELD_DP32(desc, MTEDESC, MIDX, core_idx);
+        desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid);
+        desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma);
+        desc = FIELD_DP32(desc, MTEDESC, WRITE, is_write);
+        desc = FIELD_DP32(desc, MTEDESC, ESIZE, 1 << log2_size);
+        tcg_desc = tcg_const_i32(desc);
+
+        ret = new_tmp_a64(s);
+        gen_helper_mte_check1(ret, cpu_env, tcg_desc, addr);
+        tcg_temp_free_i32(tcg_desc);
+
+        return ret;
+    }
+    return clean_data_tbi(s, addr);
+}
+
+TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write,
+                        bool tag_checked, int log2_size)
+{
+    return gen_mte_check1_mmuidx(s, addr, is_write, tag_checked, log2_size,
+                                 false, get_mem_index(s));
+}
+
 typedef struct DisasCompare64 {
     TCGCond cond;
     TCGv_i64 value;
@@ -2367,7 +2406,7 @@ static void gen_compare_and_swap(DisasContext *s, int rs, int rt,
     if (rn == 31) {
         gen_check_sp_alignment(s);
     }
-    clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+    clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn), true, rn != 31, size);
     tcg_gen_atomic_cmpxchg_i64(tcg_rs, clean_addr, tcg_rs, tcg_rt, memidx,
                                size | MO_ALIGN | s->be_data);
 }
@@ -2385,7 +2424,9 @@ static void gen_compare_and_swap_pair(DisasContext *s, int rs, int rt,
     if (rn == 31) {
         gen_check_sp_alignment(s);
     }
-    clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+
+    /* This is a single atomic access, despite the "pair". */
+    clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn), true, rn != 31, size + 1);
 
     if (size == 2) {
         TCGv_i64 cmp = tcg_temp_new_i64();
@@ -2510,7 +2551,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
         if (is_lasr) {
             tcg_gen_mb(TCG_MO_ALL | TCG_BAR_STRL);
         }
-        clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+        clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                    true, rn != 31, size);
         gen_store_exclusive(s, rs, rt, rt2, clean_addr, size, false);
         return;
 
@@ -2519,7 +2561,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
         if (rn == 31) {
             gen_check_sp_alignment(s);
         }
-        clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+        clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                    false, rn != 31, size);
         s->is_ldex = true;
         gen_load_exclusive(s, rt, rt2, clean_addr, size, false);
         if (is_lasr) {
@@ -2539,7 +2582,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
             gen_check_sp_alignment(s);
         }
         tcg_gen_mb(TCG_MO_ALL | TCG_BAR_STRL);
-        clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+        clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                    true, rn != 31, size);
         do_gpr_st(s, cpu_reg(s, rt), clean_addr, size, true, rt,
                   disas_ldst_compute_iss_sf(size, false, 0), is_lasr);
         return;
@@ -2555,7 +2599,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
         if (rn == 31) {
             gen_check_sp_alignment(s);
         }
-        clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+        clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                    false, rn != 31, size);
         do_gpr_ld(s, cpu_reg(s, rt), clean_addr, size, false, false, true, rt,
                   disas_ldst_compute_iss_sf(size, false, 0), is_lasr);
         tcg_gen_mb(TCG_MO_ALL | TCG_BAR_LDAQ);
@@ -2569,7 +2614,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
             if (is_lasr) {
                 tcg_gen_mb(TCG_MO_ALL | TCG_BAR_STRL);
             }
-            clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+            clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                        true, rn != 31, size);
             gen_store_exclusive(s, rs, rt, rt2, clean_addr, size, true);
             return;
         }
@@ -2587,7 +2633,8 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn)
             if (rn == 31) {
                 gen_check_sp_alignment(s);
             }
-            clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+            clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn),
+                                        false, rn != 31, size);
             s->is_ldex = true;
             gen_load_exclusive(s, rt, rt2, clean_addr, size, true);
             if (is_lasr) {
@@ -2881,6 +2928,7 @@ static void disas_ldst_reg_imm9(DisasContext *s, uint32_t insn,
     bool iss_valid = !is_vector;
     bool post_index;
     bool writeback;
+    int memidx;
 
     TCGv_i64 clean_addr, dirty_addr;
 
@@ -2938,7 +2986,11 @@ static void disas_ldst_reg_imm9(DisasContext *s, uint32_t insn,
     if (!post_index) {
         tcg_gen_addi_i64(dirty_addr, dirty_addr, imm9);
     }
-    clean_addr = clean_data_tbi(s, dirty_addr);
+
+    memidx = is_unpriv ? get_a64_user_mem_index(s) : get_mem_index(s);
+    clean_addr = gen_mte_check1_mmuidx(s, dirty_addr, is_store,
+                                       writeback || rn != 31,
+                                       size, is_unpriv, memidx);
 
     if (is_vector) {
         if (is_store) {
@@ -2948,7 +3000,6 @@ static void disas_ldst_reg_imm9(DisasContext *s, uint32_t insn,
         }
     } else {
         TCGv_i64 tcg_rt = cpu_reg(s, rt);
-        int memidx = is_unpriv ? get_a64_user_mem_index(s) : get_mem_index(s);
         bool iss_sf = disas_ldst_compute_iss_sf(size, is_signed, opc);
 
         if (is_store) {
@@ -3045,7 +3096,7 @@ static void disas_ldst_reg_roffset(DisasContext *s, uint32_t insn,
     ext_and_shift_reg(tcg_rm, tcg_rm, opt, shift ? size : 0);
 
     tcg_gen_add_i64(dirty_addr, dirty_addr, tcg_rm);
-    clean_addr = clean_data_tbi(s, dirty_addr);
+    clean_addr = gen_mte_check1(s, dirty_addr, is_store, true, size);
 
     if (is_vector) {
         if (is_store) {
@@ -3130,7 +3181,7 @@ static void disas_ldst_reg_unsigned_imm(DisasContext *s, uint32_t insn,
     dirty_addr = read_cpu_reg_sp(s, rn, 1);
     offset = imm12 << size;
     tcg_gen_addi_i64(dirty_addr, dirty_addr, offset);
-    clean_addr = clean_data_tbi(s, dirty_addr);
+    clean_addr = gen_mte_check1(s, dirty_addr, is_store, rn != 31, size);
 
     if (is_vector) {
         if (is_store) {
@@ -3223,7 +3274,7 @@ static void disas_ldst_atomic(DisasContext *s, uint32_t insn,
     if (rn == 31) {
         gen_check_sp_alignment(s);
     }
-    clean_addr = clean_data_tbi(s, cpu_reg_sp(s, rn));
+    clean_addr = gen_mte_check1(s, cpu_reg_sp(s, rn), false, rn != 31, size);
 
     if (o3_opc == 014) {
         /*
@@ -3300,7 +3351,8 @@ static void disas_ldst_pac(DisasContext *s, uint32_t insn,
     tcg_gen_addi_i64(dirty_addr, dirty_addr, offset);
 
     /* Note that "clean" and "dirty" here refer to TBI not PAC.  */
-    clean_addr = clean_data_tbi(s, dirty_addr);
+    clean_addr = gen_mte_check1(s, dirty_addr, false,
+                                is_wback || rn != 31, size);
 
     tcg_rt = cpu_reg(s, rt);
     do_gpr_ld(s, tcg_rt, clean_addr, size, /* is_signed */ false,
-- 
2.25.1

[PATCH v8 24/45] target/arm: Add gen_mte_checkN

[Richard Henderson]

Replace existing uses of check_data_tbi in translate-a64.c that
perform multiple logical memory access.  Leave the helper blank
for now to reduce the patch size.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-a64.h    |  1 +
 target/arm/translate-a64.h |  2 ++
 target/arm/mte_helper.c    |  8 +++++
 target/arm/translate-a64.c | 71 +++++++++++++++++++++++++++++---------
 4 files changed, 66 insertions(+), 16 deletions(-)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 2faa49d0a3..005af678c7 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -105,6 +105,7 @@ DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64)
 DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
 
 DEF_HELPER_FLAGS_3(mte_check1, TCG_CALL_NO_WG, i64, env, i32, i64)
+DEF_HELPER_FLAGS_3(mte_checkN, TCG_CALL_NO_WG, i64, env, i32, i64)
 DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_4(addsubg, TCG_CALL_NO_RWG_SE, i64, env, i64, s32, i32)
 DEF_HELPER_FLAGS_3(ldg, TCG_CALL_NO_WG, i64, env, i64, i64)
diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h
index daab6a9666..781c441399 100644
--- a/target/arm/translate-a64.h
+++ b/target/arm/translate-a64.h
@@ -42,6 +42,8 @@ bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn,
 bool sve_access_check(DisasContext *s);
 TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write,
                         bool tag_checked, int log2_size);
+TCGv_i64 gen_mte_checkN(DisasContext *s, TCGv_i64 addr, bool is_write,
+                        bool tag_checked, int count, int log2_esize);
 
 /* We should have at some point before trying to access an FP register
  * done the necessary access check, so assert that
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index ec12768dfc..907a12b366 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -366,3 +366,11 @@ uint64_t HELPER(mte_check1)(CPUARMState *env, uint32_t desc, uint64_t ptr)
 {
     return ptr;
 }
+
+/*
+ * Perform an MTE checked access for multiple logical accesses.
+ */
+uint64_t HELPER(mte_checkN)(CPUARMState *env, uint32_t desc, uint64_t ptr)
+{
+    return ptr;
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 0720c39e89..5d62aff5d6 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -284,6 +284,34 @@ TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write,
                                  false, get_mem_index(s));
 }
 
+/*
+ * For MTE, check multiple logical sequential accesses.
+ */
+TCGv_i64 gen_mte_checkN(DisasContext *s, TCGv_i64 addr, bool is_write,
+                        bool tag_checked, int log2_esize, int total_size)
+{
+    if (tag_checked && s->mte_active[0] && total_size != (1 << log2_esize)) {
+        TCGv_i32 tcg_desc;
+        TCGv_i64 ret;
+        int desc = 0;
+
+        desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s));
+        desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid);
+        desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma);
+        desc = FIELD_DP32(desc, MTEDESC, WRITE, is_write);
+        desc = FIELD_DP32(desc, MTEDESC, ESIZE, 1 << log2_esize);
+        desc = FIELD_DP32(desc, MTEDESC, TSIZE, total_size);
+        tcg_desc = tcg_const_i32(desc);
+
+        ret = new_tmp_a64(s);
+        gen_helper_mte_checkN(ret, cpu_env, tcg_desc, addr);
+        tcg_temp_free_i32(tcg_desc);
+
+        return ret;
+    }
+    return gen_mte_check1(s, addr, is_write, tag_checked, log2_esize);
+}
+
 typedef struct DisasCompare64 {
     TCGCond cond;
     TCGv_i64 value;
@@ -2848,7 +2876,10 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
         }
     }
 
-    clean_addr = clean_data_tbi(s, dirty_addr);
+    clean_addr = gen_mte_checkN(s, dirty_addr, !is_load,
+                                (wback || rn != 31) && !set_tag,
+                                size, 2 << size);
+
     if (is_vector) {
         if (is_load) {
             do_fp_ld(s, rt, clean_addr, size);
@@ -3514,7 +3545,7 @@ static void disas_ldst_multiple_struct(DisasContext *s, uint32_t insn)
     TCGv_i64 clean_addr, tcg_rn, tcg_ebytes;
     MemOp endian = s->be_data;
 
-    int ebytes;   /* bytes per element */
+    int total;    /* total bytes */
     int elements; /* elements per vector */
     int rpt;    /* num iterations */
     int selem;  /* structure elements */
@@ -3584,19 +3615,26 @@ static void disas_ldst_multiple_struct(DisasContext *s, uint32_t insn)
         endian = MO_LE;
     }
 
-    /* Consecutive little-endian elements from a single register
+    total = rpt * selem * (is_q ? 16 : 8);
+    tcg_rn = cpu_reg_sp(s, rn);
+
+    /*
+     * Issue the MTE check vs the logical repeat count, before we
+     * promote consecutive little-endian elements below.
+     */
+    clean_addr = gen_mte_checkN(s, tcg_rn, is_store, is_postidx || rn != 31,
+                                size, total);
+
+    /*
+     * Consecutive little-endian elements from a single register
      * can be promoted to a larger little-endian operation.
      */
     if (selem == 1 && endian == MO_LE) {
         size = 3;
     }
-    ebytes = 1 << size;
-    elements = (is_q ? 16 : 8) / ebytes;
-
-    tcg_rn = cpu_reg_sp(s, rn);
-    clean_addr = clean_data_tbi(s, tcg_rn);
-    tcg_ebytes = tcg_const_i64(ebytes);
+    elements = (is_q ? 16 : 8) >> size;
 
+    tcg_ebytes = tcg_const_i64(1 << size);
     for (r = 0; r < rpt; r++) {
         int e;
         for (e = 0; e < elements; e++) {
@@ -3630,7 +3668,7 @@ static void disas_ldst_multiple_struct(DisasContext *s, uint32_t insn)
 
     if (is_postidx) {
         if (rm == 31) {
-            tcg_gen_addi_i64(tcg_rn, tcg_rn, rpt * elements * selem * ebytes);
+            tcg_gen_addi_i64(tcg_rn, tcg_rn, total);
         } else {
             tcg_gen_add_i64(tcg_rn, tcg_rn, cpu_reg(s, rm));
         }
@@ -3676,7 +3714,7 @@ static void disas_ldst_single_struct(DisasContext *s, uint32_t insn)
     int selem = (extract32(opc, 0, 1) << 1 | R) + 1;
     bool replicate = false;
     int index = is_q << 3 | S << 2 | size;
-    int ebytes, xs;
+    int xs, total;
     TCGv_i64 clean_addr, tcg_rn, tcg_ebytes;
 
     if (extract32(insn, 31, 1)) {
@@ -3730,16 +3768,17 @@ static void disas_ldst_single_struct(DisasContext *s, uint32_t insn)
         return;
     }
 
-    ebytes = 1 << scale;
-
     if (rn == 31) {
         gen_check_sp_alignment(s);
     }
 
+    total = selem << scale;
     tcg_rn = cpu_reg_sp(s, rn);
-    clean_addr = clean_data_tbi(s, tcg_rn);
-    tcg_ebytes = tcg_const_i64(ebytes);
 
+    clean_addr = gen_mte_checkN(s, tcg_rn, !is_load, is_postidx || rn != 31,
+                                scale, total);
+
+    tcg_ebytes = tcg_const_i64(1 << scale);
     for (xs = 0; xs < selem; xs++) {
         if (replicate) {
             /* Load and replicate to all elements */
@@ -3766,7 +3805,7 @@ static void disas_ldst_single_struct(DisasContext *s, uint32_t insn)
 
     if (is_postidx) {
         if (rm == 31) {
-            tcg_gen_addi_i64(tcg_rn, tcg_rn, selem * ebytes);
+            tcg_gen_addi_i64(tcg_rn, tcg_rn, total);
         } else {
             tcg_gen_add_i64(tcg_rn, tcg_rn, cpu_reg(s, rm));
         }
-- 
2.25.1

[PATCH v8 25/45] target/arm: Implement helper_mte_check1

[Richard Henderson]

Fill out the stub that was added earlier.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Remove ra argument to mte_probe1 (pmm).
---
 target/arm/internals.h  |  48 +++++++++++++++
 target/arm/mte_helper.c | 132 +++++++++++++++++++++++++++++++++++++++-
 2 files changed, 179 insertions(+), 1 deletion(-)

diff --git a/target/arm/internals.h b/target/arm/internals.h
index fb92ef6b84..807830cc40 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1318,6 +1318,10 @@ FIELD(MTEDESC, WRITE, 8, 1)
 FIELD(MTEDESC, ESIZE, 9, 5)
 FIELD(MTEDESC, TSIZE, 14, 10)  /* mte_checkN only */
 
+bool mte_probe1(CPUARMState *env, uint32_t desc, uint64_t ptr);
+uint64_t mte_check1(CPUARMState *env, uint32_t desc,
+                    uint64_t ptr, uintptr_t ra);
+
 static inline int allocation_tag_from_addr(uint64_t ptr)
 {
     return extract64(ptr, 56, 4);
@@ -1328,4 +1332,48 @@ static inline uint64_t address_with_allocation_tag(uint64_t ptr, int rtag)
     return deposit64(ptr, 56, 4, rtag);
 }
 
+/* Return true if tbi bits mean that the access is checked.  */
+static inline bool tbi_check(uint32_t desc, int bit55)
+{
+    return (desc >> (R_MTEDESC_TBI_SHIFT + bit55)) & 1;
+}
+
+/* Return true if tcma bits mean that the access is unchecked.  */
+static inline bool tcma_check(uint32_t desc, int bit55, int ptr_tag)
+{
+    /*
+     * We had extracted bit55 and ptr_tag for other reasons, so fold
+     * (ptr<59:55> == 00000 || ptr<59:55> == 11111) into a single test.
+     */
+    bool match = ((ptr_tag + bit55) & 0xf) == 0;
+    bool tcma = (desc >> (R_MTEDESC_TCMA_SHIFT + bit55)) & 1;
+    return tcma && match;
+}
+
+/*
+ * For TBI, ideally, we would do nothing.  Proper behaviour on fault is
+ * for the tag to be present in the FAR_ELx register.  But for user-only
+ * mode, we do not have a TLB with which to implement this, so we must
+ * remove the top byte.
+ */
+static inline uint64_t useronly_clean_ptr(uint64_t ptr)
+{
+    /* TBI is known to be enabled. */
+#ifdef CONFIG_USER_ONLY
+    ptr = sextract64(ptr, 0, 56);
+#endif
+    return ptr;
+}
+
+static inline uint64_t useronly_maybe_clean_ptr(uint32_t desc, uint64_t ptr)
+{
+#ifdef CONFIG_USER_ONLY
+    int64_t clean_ptr = sextract64(ptr, 0, 56);
+    if (tbi_check(desc, clean_ptr < 0)) {
+        ptr = clean_ptr;
+    }
+#endif
+    return ptr;
+}
+
 #endif
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 907a12b366..c8a5e7c0ed 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -359,12 +359,142 @@ void HELPER(stzgm_tags)(CPUARMState *env, uint64_t ptr, uint64_t val)
     }
 }
 
+/* Record a tag check failure.  */
+static void mte_check_fail(CPUARMState *env, int mmu_idx,
+                           uint64_t dirty_ptr, uintptr_t ra)
+{
+    ARMMMUIdx arm_mmu_idx = core_to_aa64_mmu_idx(mmu_idx);
+    int el, reg_el, tcf, select;
+    uint64_t sctlr;
+
+    reg_el = regime_el(env, arm_mmu_idx);
+    sctlr = env->cp15.sctlr_el[reg_el];
+
+    switch (arm_mmu_idx) {
+    case ARMMMUIdx_E10_0:
+    case ARMMMUIdx_E20_0:
+        el = 0;
+        tcf = extract64(sctlr, 38, 2);
+        break;
+    default:
+        el = reg_el;
+        tcf = extract64(sctlr, 40, 2);
+    }
+
+    switch (tcf) {
+    case 1:
+        /*
+         * Tag check fail causes a synchronous exception.
+         *
+         * In restore_state_to_opc, we set the exception syndrome
+         * for the load or store operation.  Unwind first so we
+         * may overwrite that with the syndrome for the tag check.
+         */
+        cpu_restore_state(env_cpu(env), ra, true);
+        env->exception.vaddress = dirty_ptr;
+        raise_exception(env, EXCP_DATA_ABORT,
+                        syn_data_abort_no_iss(el != 0, 0, 0, 0, 0, 0, 0x11),
+                        exception_target_el(env));
+        /* noreturn, but fall through to the assert anyway */
+
+    case 0:
+        /*
+         * Tag check fail does not affect the PE.
+         * We eliminate this case by not setting MTE_ACTIVE
+         * in tb_flags, so that we never make this runtime call.
+         */
+        g_assert_not_reached();
+
+    case 2:
+        /* Tag check fail causes asynchronous flag set.  */
+        mmu_idx = arm_mmu_idx_el(env, el);
+        if (regime_has_2_ranges(mmu_idx)) {
+            select = extract64(dirty_ptr, 55, 1);
+        } else {
+            select = 0;
+        }
+        env->cp15.tfsr_el[el] |= 1 << select;
+        break;
+
+    default:
+        /* Case 3: Reserved. */
+        qemu_log_mask(LOG_GUEST_ERROR,
+                      "Tag check failure with SCTLR_EL%d.TCF%s "
+                      "set to reserved value %d\n",
+                      reg_el, el ? "" : "0", tcf);
+        break;
+    }
+}
+
 /*
  * Perform an MTE checked access for a single logical or atomic access.
  */
+static bool mte_probe1_int(CPUARMState *env, uint32_t desc, uint64_t ptr,
+                           uintptr_t ra, int bit55)
+{
+    int mem_tag, mmu_idx, ptr_tag, size;
+    MMUAccessType type;
+    uint8_t *mem;
+
+    ptr_tag = allocation_tag_from_addr(ptr);
+
+    if (tcma_check(desc, bit55, ptr_tag)) {
+        return true;
+    }
+
+    mmu_idx = FIELD_EX32(desc, MTEDESC, MIDX);
+    type = FIELD_EX32(desc, MTEDESC, WRITE) ? MMU_DATA_STORE : MMU_DATA_LOAD;
+    size = FIELD_EX32(desc, MTEDESC, ESIZE);
+
+    mem = allocation_tag_mem(env, mmu_idx, ptr, type, size,
+                             MMU_DATA_LOAD, 1, ra);
+    if (!mem) {
+        return true;
+    }
+
+    mem_tag = load_tag1(ptr, mem);
+    return ptr_tag == mem_tag;
+}
+
+/*
+ * No-fault version of mte_check1, to be used by SVE for MemSingleNF.
+ * Returns false if the access is Checked and the check failed.  This
+ * is only intended to probe the tag -- the validity of the page must
+ * be checked beforehand.
+ */
+bool mte_probe1(CPUARMState *env, uint32_t desc, uint64_t ptr)
+{
+    int bit55 = extract64(ptr, 55, 1);
+
+    /* If TBI is disabled, the access is unchecked. */
+    if (unlikely(!tbi_check(desc, bit55))) {
+        return true;
+    }
+
+    return mte_probe1_int(env, desc, ptr, 0, bit55);
+}
+
+uint64_t mte_check1(CPUARMState *env, uint32_t desc,
+                    uint64_t ptr, uintptr_t ra)
+{
+    int bit55 = extract64(ptr, 55, 1);
+
+    /* If TBI is disabled, the access is unchecked, and ptr is not dirty. */
+    if (unlikely(!tbi_check(desc, bit55))) {
+        return ptr;
+    }
+
+    if (unlikely(!mte_probe1_int(env, desc, ptr, ra, bit55))) {
+        int mmu_idx = FIELD_EX32(desc, MTEDESC, MIDX);
+        mte_check_fail(env, mmu_idx, ptr, ra);
+    }
+
+    return useronly_clean_ptr(ptr);
+}
+
 uint64_t HELPER(mte_check1)(CPUARMState *env, uint32_t desc, uint64_t ptr)
 {
-    return ptr;
+    return mte_check1(env, desc, ptr, GETPC());
 }
 
 /*
-- 
2.25.1

[PATCH v8 26/45] target/arm: Implement helper_mte_checkN

[Richard Henderson]

Fill out the stub that was added earlier.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v7: Fix page crossing test (szabolcs nagy).
---
 target/arm/internals.h  |   2 +
 target/arm/mte_helper.c | 165 +++++++++++++++++++++++++++++++++++++++-
 2 files changed, 166 insertions(+), 1 deletion(-)

diff --git a/target/arm/internals.h b/target/arm/internals.h
index 807830cc40..c763a23dfb 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1321,6 +1321,8 @@ FIELD(MTEDESC, TSIZE, 14, 10)  /* mte_checkN only */
 bool mte_probe1(CPUARMState *env, uint32_t desc, uint64_t ptr);
 uint64_t mte_check1(CPUARMState *env, uint32_t desc,
                     uint64_t ptr, uintptr_t ra);
+uint64_t mte_checkN(CPUARMState *env, uint32_t desc,
+                    uint64_t ptr, uintptr_t ra);
 
 static inline int allocation_tag_from_addr(uint64_t ptr)
 {
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index c8a5e7c0ed..abe6af6b79 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -500,7 +500,170 @@ uint64_t HELPER(mte_check1)(CPUARMState *env, uint32_t desc, uint64_t ptr)
 /*
  * Perform an MTE checked access for multiple logical accesses.
  */
+
+/**
+ * checkN:
+ * @tag: tag memory to test
+ * @odd: true to begin testing at tags at odd nibble
+ * @cmp: the tag to compare against
+ * @count: number of tags to test
+ *
+ * Return the number of successful tests.
+ * Thus a return value < @count indicates a failure.
+ *
+ * A note about sizes: count is expected to be small.
+ *
+ * The most common use will be LDP/STP of two integer registers,
+ * which means 16 bytes of memory touching at most 2 tags, but
+ * often the access is aligned and thus just 1 tag.
+ *
+ * Using AdvSIMD LD/ST (multiple), one can access 64 bytes of memory,
+ * touching at most 5 tags.  SVE LDR/STR (vector) with the default
+ * vector length is also 64 bytes; the maximum architectural length
+ * is 256 bytes touching at most 9 tags.
+ *
+ * The loop below uses 7 logical operations and 1 memory operation
+ * per tag pair.  An implementation that loads an aligned word and
+ * uses masking to ignore adjacent tags requires 18 logical operations
+ * and thus does not begin to pay off until 6 tags.
+ * Which, according to the survey above, is unlikely to be common.
+ */
+static int checkN(uint8_t *mem, int odd, int cmp, int count)
+{
+    int n = 0, diff;
+
+    /* Replicate the test tag and compare.  */
+    cmp *= 0x11;
+    diff = *mem++ ^ cmp;
+
+    if (odd) {
+        goto start_odd;
+    }
+
+    while (1) {
+        /* Test even tag. */
+        if (unlikely((diff) & 0x0f)) {
+            break;
+        }
+        if (++n == count) {
+            break;
+        }
+
+    start_odd:
+        /* Test odd tag. */
+        if (unlikely((diff) & 0xf0)) {
+            break;
+        }
+        if (++n == count) {
+            break;
+        }
+
+        diff = *mem++ ^ cmp;
+    }
+    return n;
+}
+
+uint64_t mte_checkN(CPUARMState *env, uint32_t desc,
+                    uint64_t ptr, uintptr_t ra)
+{
+    int mmu_idx, ptr_tag, bit55;
+    uint64_t ptr_last, ptr_end, prev_page, next_page;
+    uint64_t tag_first, tag_end;
+    uint64_t tag_byte_first, tag_byte_end;
+    uint32_t esize, total, tag_count, tag_size, n, c;
+    uint8_t *mem1, *mem2;
+    MMUAccessType type;
+
+    bit55 = extract64(ptr, 55, 1);
+
+    /* If TBI is disabled, the access is unchecked, and ptr is not dirty. */
+    if (unlikely(!tbi_check(desc, bit55))) {
+        return ptr;
+    }
+
+    ptr_tag = allocation_tag_from_addr(ptr);
+
+    if (tcma_check(desc, bit55, ptr_tag)) {
+        goto done;
+    }
+
+    mmu_idx = FIELD_EX32(desc, MTEDESC, MIDX);
+    type = FIELD_EX32(desc, MTEDESC, WRITE) ? MMU_DATA_STORE : MMU_DATA_LOAD;
+    esize = FIELD_EX32(desc, MTEDESC, ESIZE);
+    total = FIELD_EX32(desc, MTEDESC, TSIZE);
+
+    /* Find the addr of the end of the access, and of the last element. */
+    ptr_end = ptr + total;
+    ptr_last = ptr_end - esize;
+
+    /* Round the bounds to the tag granule, and compute the number of tags. */
+    tag_first = QEMU_ALIGN_DOWN(ptr, TAG_GRANULE);
+    tag_end = QEMU_ALIGN_UP(ptr_last, TAG_GRANULE);
+    tag_count = (tag_end - tag_first) / TAG_GRANULE;
+
+    /* Round the bounds to twice the tag granule, and compute the bytes. */
+    tag_byte_first = QEMU_ALIGN_DOWN(ptr, 2 * TAG_GRANULE);
+    tag_byte_end = QEMU_ALIGN_UP(ptr_last, 2 * TAG_GRANULE);
+
+    /* Locate the page boundaries. */
+    prev_page = ptr & TARGET_PAGE_MASK;
+    next_page = prev_page + TARGET_PAGE_SIZE;
+
+    if (likely(tag_end - prev_page <= TARGET_PAGE_SIZE)) {
+        /* Memory access stays on one page. */
+        tag_size = (tag_byte_end - tag_byte_first) / (2 * TAG_GRANULE);
+        mem1 = allocation_tag_mem(env, mmu_idx, ptr, type, total,
+                                  MMU_DATA_LOAD, tag_size, ra);
+        if (!mem1) {
+            goto done;
+        }
+        /* Perform all of the comparisons. */
+        n = checkN(mem1, ptr & TAG_GRANULE, ptr_tag, tag_count);
+    } else {
+        /* Memory access crosses to next page. */
+        tag_size = (next_page - tag_byte_first) / (2 * TAG_GRANULE);
+        mem1 = allocation_tag_mem(env, mmu_idx, ptr, type, next_page - ptr,
+                                  MMU_DATA_LOAD, tag_size, ra);
+
+        tag_size = (tag_byte_end - next_page) / (2 * TAG_GRANULE);
+        mem2 = allocation_tag_mem(env, mmu_idx, next_page, type,
+                                  ptr_end - next_page,
+                                  MMU_DATA_LOAD, tag_size, ra);
+
+        /*
+         * Perform all of the comparisons.
+         * Note the possible but unlikely case of the operation spanning
+         * two pages that do not both have tagging enabled.
+         */
+        n = c = (next_page - tag_first) / TAG_GRANULE;
+        if (mem1) {
+            n = checkN(mem1, ptr & TAG_GRANULE, ptr_tag, c);
+        }
+        if (n == c) {
+            if (!mem2) {
+                goto done;
+            }
+            n += checkN(mem2, 0, ptr_tag, tag_count - c);
+        }
+    }
+
+    /*
+     * If we failed, we know which granule.  Compute the element that
+     * is first in that granule, and signal failure on that element.
+     */
+    if (unlikely(n < tag_count)) {
+        uint64_t fail_ofs;
+
+        fail_ofs = tag_first + n * TAG_GRANULE - ptr;
+        fail_ofs = ROUND_UP(fail_ofs, esize);
+        mte_check_fail(env, mmu_idx, ptr + fail_ofs, ra);
+    }
+
+ done:
+    return useronly_clean_ptr(ptr);
+}
+
 uint64_t HELPER(mte_checkN)(CPUARMState *env, uint32_t desc, uint64_t ptr)
 {
-    return ptr;
+    return mte_checkN(env, desc, ptr, GETPC());
 }
-- 
2.25.1

[PATCH v8 27/45] target/arm: Add helper_mte_check_zva

[Richard Henderson]

Use a special helper for DC_ZVA, rather than the more
general mte_checkN.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-a64.h    |   1 +
 target/arm/mte_helper.c    | 106 +++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c |  16 +++++-
 3 files changed, 122 insertions(+), 1 deletion(-)

diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
index 005af678c7..5b0b699a50 100644
--- a/target/arm/helper-a64.h
+++ b/target/arm/helper-a64.h
@@ -106,6 +106,7 @@ DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
 
 DEF_HELPER_FLAGS_3(mte_check1, TCG_CALL_NO_WG, i64, env, i32, i64)
 DEF_HELPER_FLAGS_3(mte_checkN, TCG_CALL_NO_WG, i64, env, i32, i64)
+DEF_HELPER_FLAGS_3(mte_check_zva, TCG_CALL_NO_WG, i64, env, i32, i64)
 DEF_HELPER_FLAGS_3(irg, TCG_CALL_NO_RWG, i64, env, i64, i64)
 DEF_HELPER_FLAGS_4(addsubg, TCG_CALL_NO_RWG_SE, i64, env, i64, s32, i32)
 DEF_HELPER_FLAGS_3(ldg, TCG_CALL_NO_WG, i64, env, i64, i64)
diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index abe6af6b79..4f9bd3add3 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -667,3 +667,109 @@ uint64_t HELPER(mte_checkN)(CPUARMState *env, uint32_t desc, uint64_t ptr)
 {
     return mte_checkN(env, desc, ptr, GETPC());
 }
+
+/*
+ * Perform an MTE checked access for DC_ZVA.
+ */
+uint64_t HELPER(mte_check_zva)(CPUARMState *env, uint32_t desc, uint64_t ptr)
+{
+    uintptr_t ra = GETPC();
+    int log2_dcz_bytes, log2_tag_bytes;
+    int mmu_idx, bit55;
+    intptr_t dcz_bytes, tag_bytes, i;
+    void *mem;
+    uint64_t ptr_tag, mem_tag, align_ptr;
+
+    bit55 = extract64(ptr, 55, 1);
+
+    /* If TBI is disabled, the access is unchecked, and ptr is not dirty. */
+    if (unlikely(!tbi_check(desc, bit55))) {
+        return ptr;
+    }
+
+    ptr_tag = allocation_tag_from_addr(ptr);
+
+    if (tcma_check(desc, bit55, ptr_tag)) {
+        goto done;
+    }
+
+    /*
+     * In arm_cpu_realizefn, we asserted that dcz > LOG2_TAG_GRANULE+1,
+     * i.e. 32 bytes, which is an unreasonably small dcz anyway, to make
+     * sure that we can access one complete tag byte here.
+     */
+    log2_dcz_bytes = env_archcpu(env)->dcz_blocksize + 2;
+    log2_tag_bytes = log2_dcz_bytes - (LOG2_TAG_GRANULE + 1);
+    dcz_bytes = (intptr_t)1 << log2_dcz_bytes;
+    tag_bytes = (intptr_t)1 << log2_tag_bytes;
+    align_ptr = ptr & -dcz_bytes;
+
+    /*
+     * Trap if accessing an invalid page.  DC_ZVA requires that we supply
+     * the original pointer for an invalid page.  But watchpoints require
+     * that we probe the actual space.  So do both.
+     */
+    mmu_idx = FIELD_EX32(desc, MTEDESC, MIDX);
+    (void) probe_write(env, ptr, 1, mmu_idx, ra);
+    mem = allocation_tag_mem(env, mmu_idx, align_ptr, MMU_DATA_STORE,
+                             dcz_bytes, MMU_DATA_LOAD, tag_bytes, ra);
+    if (!mem) {
+        goto done;
+    }
+
+    /*
+     * Unlike the reasoning for checkN, DC_ZVA is always aligned, and thus
+     * it is quite easy to perform all of the comparisons at once without
+     * any extra masking.
+     *
+     * The most common zva block size is 64; some of the thunderx cpus use
+     * a block size of 128.  For user-only, aarch64_max_initfn will set the
+     * block size to 512.  Fill out the other cases for future-proofing.
+     *
+     * In order to be able to find the first miscompare later, we want the
+     * tag bytes to be in little-endian order.
+     */
+    switch (log2_tag_bytes) {
+    case 0: /* zva_blocksize 32 */
+        mem_tag = *(uint8_t *)mem;
+        ptr_tag *= 0x11u;
+        break;
+    case 1: /* zva_blocksize 64 */
+        mem_tag = cpu_to_le16(*(uint16_t *)mem);
+        ptr_tag *= 0x1111u;
+        break;
+    case 2: /* zva_blocksize 128 */
+        mem_tag = cpu_to_le32(*(uint32_t *)mem);
+        ptr_tag *= 0x11111111u;
+        break;
+    case 3: /* zva_blocksize 256 */
+        mem_tag = cpu_to_le64(*(uint64_t *)mem);
+        ptr_tag *= 0x1111111111111111ull;
+        break;
+
+    default: /* zva_blocksize 512, 1024, 2048 */
+        ptr_tag *= 0x1111111111111111ull;
+        i = 0;
+        do {
+            mem_tag = cpu_to_le64(*(uint64_t *)(mem + i));
+            if (unlikely(mem_tag != ptr_tag)) {
+                goto fail;
+            }
+            i += 8;
+            align_ptr += 16 * TAG_GRANULE;
+        } while (i < tag_bytes);
+        goto done;
+    }
+
+    if (likely(mem_tag == ptr_tag)) {
+        goto done;
+    }
+
+ fail:
+    /* Locate the first nibble that differs. */
+    i = ctz64(mem_tag ^ ptr_tag) >> 4;
+    mte_check_fail(env, mmu_idx, align_ptr + i * TAG_GRANULE, ra);
+
+ done:
+    return useronly_clean_ptr(ptr);
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 5d62aff5d6..7e8263e86f 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1857,7 +1857,21 @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread,
         return;
     case ARM_CP_DC_ZVA:
         /* Writes clear the aligned block of memory which rt points into. */
-        tcg_rt = clean_data_tbi(s, cpu_reg(s, rt));
+        if (s->mte_active[0]) {
+            TCGv_i32 t_desc;
+            int desc = 0;
+
+            desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s));
+            desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid);
+            desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma);
+            t_desc = tcg_const_i32(desc);
+
+            tcg_rt = new_tmp_a64(s);
+            gen_helper_mte_check_zva(tcg_rt, cpu_env, t_desc, cpu_reg(s, rt));
+            tcg_temp_free_i32(t_desc);
+        } else {
+            tcg_rt = clean_data_tbi(s, cpu_reg(s, rt));
+        }
         gen_helper_dc_zva(cpu_env, tcg_rt);
         return;
     default:
-- 
2.25.1

[PATCH v8 28/45] target/arm: Use mte_checkN for sve unpredicated loads

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Drop the out-of-line helper (pmm).
---
 target/arm/translate-sve.c | 61 +++++++++++++++++++++-----------------
 1 file changed, 33 insertions(+), 28 deletions(-)

diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index ac7b3119e5..11e0dfc210 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4342,71 +4342,76 @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm)
     int len_remain = len % 8;
     int nparts = len / 8 + ctpop8(len_remain);
     int midx = get_mem_index(s);
-    TCGv_i64 addr, t0, t1;
+    TCGv_i64 dirty_addr, clean_addr, t0, t1;
 
-    addr = tcg_temp_new_i64();
-    t0 = tcg_temp_new_i64();
+    dirty_addr = tcg_temp_new_i64();
+    tcg_gen_addi_i64(dirty_addr, cpu_reg_sp(s, rn), imm);
+    clean_addr = gen_mte_checkN(s, dirty_addr, false, rn != 31, len, MO_8);
+    tcg_temp_free_i64(dirty_addr);
 
-    /* Note that unpredicated load/store of vector/predicate registers
+    /*
+     * Note that unpredicated load/store of vector/predicate registers
      * are defined as a stream of bytes, which equates to little-endian
-     * operations on larger quantities.  There is no nice way to force
-     * a little-endian load for aarch64_be-linux-user out of line.
-     *
+     * operations on larger quantities.
      * Attempt to keep code expansion to a minimum by limiting the
      * amount of unrolling done.
      */
     if (nparts <= 4) {
         int i;
 
+        t0 = tcg_temp_new_i64();
         for (i = 0; i < len_align; i += 8) {
-            tcg_gen_addi_i64(addr, cpu_reg_sp(s, rn), imm + i);
-            tcg_gen_qemu_ld_i64(t0, addr, midx, MO_LEQ);
+            tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEQ);
             tcg_gen_st_i64(t0, cpu_env, vofs + i);
+            tcg_gen_addi_i64(clean_addr, cpu_reg_sp(s, rn), 8);
         }
+        tcg_temp_free_i64(t0);
     } else {
         TCGLabel *loop = gen_new_label();
         TCGv_ptr tp, i = tcg_const_local_ptr(0);
 
+        /* Copy the clean address into a local temp, live across the loop. */
+        t0 = clean_addr;
+        clean_addr = tcg_temp_local_new_i64();
+        tcg_gen_mov_i64(clean_addr, t0);
+        tcg_temp_free_i64(t0);
+
         gen_set_label(loop);
 
-        /* Minimize the number of local temps that must be re-read from
-         * the stack each iteration.  Instead, re-compute values other
-         * than the loop counter.
-         */
+        t0 = tcg_temp_new_i64();
+        tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEQ);
+        tcg_gen_addi_i64(clean_addr, clean_addr, 8);
+
         tp = tcg_temp_new_ptr();
-        tcg_gen_addi_ptr(tp, i, imm);
-        tcg_gen_extu_ptr_i64(addr, tp);
-        tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, rn));
-
-        tcg_gen_qemu_ld_i64(t0, addr, midx, MO_LEQ);
-
         tcg_gen_add_ptr(tp, cpu_env, i);
         tcg_gen_addi_ptr(i, i, 8);
         tcg_gen_st_i64(t0, tp, vofs);
         tcg_temp_free_ptr(tp);
+        tcg_temp_free_i64(t0);
 
         tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop);
         tcg_temp_free_ptr(i);
     }
 
-    /* Predicate register loads can be any multiple of 2.
+    /*
+     * Predicate register loads can be any multiple of 2.
      * Note that we still store the entire 64-bit unit into cpu_env.
      */
     if (len_remain) {
-        tcg_gen_addi_i64(addr, cpu_reg_sp(s, rn), imm + len_align);
-
+        t0 = tcg_temp_new_i64();
         switch (len_remain) {
         case 2:
         case 4:
         case 8:
-            tcg_gen_qemu_ld_i64(t0, addr, midx, MO_LE | ctz32(len_remain));
+            tcg_gen_qemu_ld_i64(t0, clean_addr, midx,
+                                MO_LE | ctz32(len_remain));
             break;
 
         case 6:
             t1 = tcg_temp_new_i64();
-            tcg_gen_qemu_ld_i64(t0, addr, midx, MO_LEUL);
-            tcg_gen_addi_i64(addr, addr, 4);
-            tcg_gen_qemu_ld_i64(t1, addr, midx, MO_LEUW);
+            tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEUL);
+            tcg_gen_addi_i64(clean_addr, clean_addr, 4);
+            tcg_gen_qemu_ld_i64(t1, clean_addr, midx, MO_LEUW);
             tcg_gen_deposit_i64(t0, t0, t1, 32, 32);
             tcg_temp_free_i64(t1);
             break;
@@ -4415,9 +4420,9 @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm)
             g_assert_not_reached();
         }
         tcg_gen_st_i64(t0, cpu_env, vofs + len_align);
+        tcg_temp_free_i64(t0);
     }
-    tcg_temp_free_i64(addr);
-    tcg_temp_free_i64(t0);
+    tcg_temp_free_i64(clean_addr);
 }
 
 /* Similarly for stores.  */
-- 
2.25.1

[PATCH v8 29/45] target/arm: Use mte_checkN for sve unpredicated stores

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Drop the out-of-line helper (pmm).
---
 target/arm/translate-sve.c | 61 +++++++++++++++++++++-----------------
 1 file changed, 33 insertions(+), 28 deletions(-)

diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 11e0dfc210..4a613ca689 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4432,10 +4432,12 @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm)
     int len_remain = len % 8;
     int nparts = len / 8 + ctpop8(len_remain);
     int midx = get_mem_index(s);
-    TCGv_i64 addr, t0;
+    TCGv_i64 dirty_addr, clean_addr, t0;
 
-    addr = tcg_temp_new_i64();
-    t0 = tcg_temp_new_i64();
+    dirty_addr = tcg_temp_new_i64();
+    tcg_gen_addi_i64(dirty_addr, cpu_reg_sp(s, rn), imm);
+    clean_addr = gen_mte_checkN(s, dirty_addr, false, rn != 31, len, MO_8);
+    tcg_temp_free_i64(dirty_addr);
 
     /* Note that unpredicated load/store of vector/predicate registers
      * are defined as a stream of bytes, which equates to little-endian
@@ -4448,33 +4450,35 @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm)
     if (nparts <= 4) {
         int i;
 
+        t0 = tcg_temp_new_i64();
         for (i = 0; i < len_align; i += 8) {
             tcg_gen_ld_i64(t0, cpu_env, vofs + i);
-            tcg_gen_addi_i64(addr, cpu_reg_sp(s, rn), imm + i);
-            tcg_gen_qemu_st_i64(t0, addr, midx, MO_LEQ);
+            tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEQ);
+            tcg_gen_addi_i64(clean_addr, cpu_reg_sp(s, rn), 8);
         }
+        tcg_temp_free_i64(t0);
     } else {
         TCGLabel *loop = gen_new_label();
-        TCGv_ptr t2, i = tcg_const_local_ptr(0);
+        TCGv_ptr tp, i = tcg_const_local_ptr(0);
+
+        /* Copy the clean address into a local temp, live across the loop. */
+        t0 = clean_addr;
+        clean_addr = tcg_temp_local_new_i64();
+        tcg_gen_mov_i64(clean_addr, t0);
+        tcg_temp_free_i64(t0);
 
         gen_set_label(loop);
 
-        t2 = tcg_temp_new_ptr();
-        tcg_gen_add_ptr(t2, cpu_env, i);
-        tcg_gen_ld_i64(t0, t2, vofs);
-
-        /* Minimize the number of local temps that must be re-read from
-         * the stack each iteration.  Instead, re-compute values other
-         * than the loop counter.
-         */
-        tcg_gen_addi_ptr(t2, i, imm);
-        tcg_gen_extu_ptr_i64(addr, t2);
-        tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, rn));
-        tcg_temp_free_ptr(t2);
-
-        tcg_gen_qemu_st_i64(t0, addr, midx, MO_LEQ);
-
+        t0 = tcg_temp_new_i64();
+        tp = tcg_temp_new_ptr();
+        tcg_gen_add_ptr(tp, cpu_env, i);
+        tcg_gen_ld_i64(t0, tp, vofs);
         tcg_gen_addi_ptr(i, i, 8);
+        tcg_temp_free_ptr(tp);
+
+        tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEQ);
+        tcg_gen_addi_i64(clean_addr, clean_addr, 8);
+        tcg_temp_free_i64(t0);
 
         tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop);
         tcg_temp_free_ptr(i);
@@ -4482,29 +4486,30 @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm)
 
     /* Predicate register stores can be any multiple of 2.  */
     if (len_remain) {
+        t0 = tcg_temp_new_i64();
         tcg_gen_ld_i64(t0, cpu_env, vofs + len_align);
-        tcg_gen_addi_i64(addr, cpu_reg_sp(s, rn), imm + len_align);
 
         switch (len_remain) {
         case 2:
         case 4:
         case 8:
-            tcg_gen_qemu_st_i64(t0, addr, midx, MO_LE | ctz32(len_remain));
+            tcg_gen_qemu_st_i64(t0, clean_addr, midx,
+                                MO_LE | ctz32(len_remain));
             break;
 
         case 6:
-            tcg_gen_qemu_st_i64(t0, addr, midx, MO_LEUL);
-            tcg_gen_addi_i64(addr, addr, 4);
+            tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEUL);
+            tcg_gen_addi_i64(clean_addr, clean_addr, 4);
             tcg_gen_shri_i64(t0, t0, 32);
-            tcg_gen_qemu_st_i64(t0, addr, midx, MO_LEUW);
+            tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEUW);
             break;
 
         default:
             g_assert_not_reached();
         }
+        tcg_temp_free_i64(t0);
     }
-    tcg_temp_free_i64(addr);
-    tcg_temp_free_i64(t0);
+    tcg_temp_free_i64(clean_addr);
 }
 
 static bool trans_LDR_zri(DisasContext *s, arg_rri *a)
-- 
2.25.1

[PATCH v8 30/45] target/arm: Use mte_check1 for sve LD1R

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate-sve.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 4a613ca689..4fa521989d 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4892,7 +4892,7 @@ static bool trans_LD1R_zpri(DisasContext *s, arg_rpri_load *a)
     unsigned esz = dtype_esz[a->dtype];
     unsigned msz = dtype_msz(a->dtype);
     TCGLabel *over = gen_new_label();
-    TCGv_i64 temp;
+    TCGv_i64 temp, clean_addr;
 
     /* If the guarding predicate has no bits set, no load occurs.  */
     if (psz <= 8) {
@@ -4915,7 +4915,9 @@ static bool trans_LD1R_zpri(DisasContext *s, arg_rpri_load *a)
     /* Load the data.  */
     temp = tcg_temp_new_i64();
     tcg_gen_addi_i64(temp, cpu_reg_sp(s, a->rn), a->imm << msz);
-    tcg_gen_qemu_ld_i64(temp, temp, get_mem_index(s),
+    clean_addr = gen_mte_check1(s, temp, false, true, msz);
+
+    tcg_gen_qemu_ld_i64(temp, clean_addr, get_mem_index(s),
                         s->be_data | dtype_mop[a->dtype]);
 
     /* Broadcast to *all* elements.  */
-- 
2.25.1

[PATCH v8 31/45] target/arm: Tidy trans_LD1R_zpri

[Richard Henderson]

Move the variable declarations to the top of the function,
but do not create a new label before sve_access_check.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v8: Split out from previous patch (pmm).
---
 target/arm/translate-sve.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 4fa521989d..a3a0b98fbc 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4883,17 +4883,19 @@ static bool trans_LD1RQ_zpri(DisasContext *s, arg_rpri_load *a)
 /* Load and broadcast element.  */
 static bool trans_LD1R_zpri(DisasContext *s, arg_rpri_load *a)
 {
-    if (!sve_access_check(s)) {
-        return true;
-    }
-
     unsigned vsz = vec_full_reg_size(s);
     unsigned psz = pred_full_reg_size(s);
     unsigned esz = dtype_esz[a->dtype];
     unsigned msz = dtype_msz(a->dtype);
-    TCGLabel *over = gen_new_label();
+    TCGLabel *over;
     TCGv_i64 temp, clean_addr;
 
+    if (!sve_access_check(s)) {
+        return true;
+    }
+
+    over = gen_new_label();
+
     /* If the guarding predicate has no bits set, no load occurs.  */
     if (psz <= 8) {
         /* Reduce the pred_esz_masks value simply to reduce the
-- 
2.25.1

[PATCH v8 32/45] target/arm: Add arm_tlb_bti_gp

[Richard Henderson]

Introduce an lvalue macro to wrap target_tlb_bit0.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/cpu.h           | 13 +++++++++++++
 target/arm/helper.c        |  2 +-
 target/arm/translate-a64.c |  2 +-
 3 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index a5d3b6c9ee..3121836bdc 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3393,6 +3393,19 @@ static inline uint64_t *aa64_vfp_qreg(CPUARMState *env, unsigned regno)
 /* Shared between translate-sve.c and sve_helper.c.  */
 extern const uint64_t pred_esz_masks[4];
 
+/* Helper for the macros below, validating the argument type. */
+static inline MemTxAttrs *typecheck_memtxattrs(MemTxAttrs *x)
+{
+    return x;
+}
+
+/*
+ * Lvalue macros for ARM TLB bits that we must cache in the TCG TLB.
+ * Using these should be a bit more self-documenting than using the
+ * generic target bits directly.
+ */
+#define arm_tlb_bti_gp(x) (typecheck_memtxattrs(x)->target_tlb_bit0)
+
 /*
  * Naming convention for isar_feature functions:
  * Functions which test 32-bit ID registers should have _aa32_ in
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 33f902387b..44a3f9fb48 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -11079,7 +11079,7 @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address,
     }
     /* When in aarch64 mode, and BTI is enabled, remember GP in the IOTLB.  */
     if (aarch64 && guarded && cpu_isar_feature(aa64_bti, cpu)) {
-        txattrs->target_tlb_bit0 = true;
+        arm_tlb_bti_gp(txattrs) = true;
     }
 
     if (cacheattrs != NULL) {
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 7e8263e86f..ec2295393d 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14446,7 +14446,7 @@ static bool is_guarded_page(CPUARMState *env, DisasContext *s)
      * table entry even for that case.
      */
     return (tlb_hit(entry->addr_code, addr) &&
-            env_tlb(env)->d[mmu_idx].iotlb[index].attrs.target_tlb_bit0);
+            arm_tlb_bti_gp(&env_tlb(env)->d[mmu_idx].iotlb[index].attrs));
 #endif
 }
 
-- 
2.25.1

[PATCH v8 33/45] target/arm: Add mte helpers for sve scalar + int loads

[Richard Henderson]

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.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/cpu.h           |   1 +
 target/arm/helper-sve.h    |  58 ++++++++++
 target/arm/internals.h     |   6 +
 target/arm/sve_helper.c    | 218 ++++++++++++++++++++++++++++++-------
 target/arm/translate-sve.c | 186 ++++++++++++++++++++++---------
 5 files changed, 378 insertions(+), 91 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 3121836bdc..76f2287314 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3405,6 +3405,7 @@ static inline MemTxAttrs *typecheck_memtxattrs(MemTxAttrs *x)
  * generic target bits directly.
  */
 #define arm_tlb_bti_gp(x) (typecheck_memtxattrs(x)->target_tlb_bit0)
+#define arm_tlb_mte_tagged(x) (typecheck_memtxattrs(x)->target_tlb_bit1)
 
 /*
  * Naming convention for isar_feature functions:
diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index 7a200755ac..1bc1974fc2 100644
--- a/target/arm/helper-sve.h
+++ b/target/arm/helper-sve.h
@@ -1196,6 +1196,64 @@ DEF_HELPER_FLAGS_4(sve_ld1sds_le_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ld1sdu_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ld1sds_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_4(sve_ld1bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld2dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld3dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld4dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1bhu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1bsu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1bdu_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1bhs_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1bss_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1bds_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1hsu_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hdu_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hds_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1hsu_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hdu_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1hds_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1sdu_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1sds_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_ld1sdu_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_ld1sds_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_4(sve_ldff1bb_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldff1bhu_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_ldff1bsu_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
diff --git a/target/arm/internals.h b/target/arm/internals.h
index c763a23dfb..3306c4f829 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1310,6 +1310,12 @@ void arm_log_exception(int idx);
 #define LOG2_TAG_GRANULE 4
 #define TAG_GRANULE      (1 << LOG2_TAG_GRANULE)
 
+/*
+ * The SVE simd_data field, for memory ops, contains either
+ * rd (5 bits) or a shift count (2 bits).
+ */
+#define SVE_MTEDESC_SHIFT 5
+
 /* Bits within a descriptor passed to the helper_mte_check* functions. */
 FIELD(MTEDESC, MIDX,  0, 4)
 FIELD(MTEDESC, TBI,   4, 2)
diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index e590db6637..767ecb399f 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -4393,15 +4393,89 @@ static void sve_cont_ldst_watchpoints(SVEContLdSt *info, CPUARMState *env,
 #endif
 }
 
+typedef uint64_t mte_check_fn(CPUARMState *, uint32_t, uint64_t, uintptr_t);
+
+static inline QEMU_ALWAYS_INLINE
+void sve_cont_ldst_mte_check_int(SVEContLdSt *info, CPUARMState *env,
+                                 uint64_t *vg, target_ulong addr, int esize,
+                                 int msize, uint32_t mtedesc, uintptr_t ra,
+                                 mte_check_fn *check)
+{
+    intptr_t mem_off, reg_off, reg_last;
+
+    /* Process the page only if MemAttr == Tagged. */
+    if (arm_tlb_mte_tagged(&info->page[0].attrs)) {
+        mem_off = info->mem_off_first[0];
+        reg_off = info->reg_off_first[0];
+        reg_last = info->reg_off_split;
+        if (reg_last < 0) {
+            reg_last = info->reg_off_last[0];
+        }
+
+        do {
+            uint64_t pg = vg[reg_off >> 6];
+            do {
+                if ((pg >> (reg_off & 63)) & 1) {
+                    check(env, mtedesc, addr, ra);
+                }
+                reg_off += esize;
+                mem_off += msize;
+            } while (reg_off <= reg_last && (reg_off & 63));
+        } while (reg_off <= reg_last);
+    }
+
+    mem_off = info->mem_off_first[1];
+    if (mem_off >= 0 && arm_tlb_mte_tagged(&info->page[1].attrs)) {
+        reg_off = info->reg_off_first[1];
+        reg_last = info->reg_off_last[1];
+
+        do {
+            uint64_t pg = vg[reg_off >> 6];
+            do {
+                if ((pg >> (reg_off & 63)) & 1) {
+                    check(env, mtedesc, addr, ra);
+                }
+                reg_off += esize;
+                mem_off += msize;
+            } while (reg_off & 63);
+        } while (reg_off <= reg_last);
+    }
+}
+
+typedef void sve_cont_ldst_mte_check_fn(SVEContLdSt *info, CPUARMState *env,
+                                        uint64_t *vg, target_ulong addr,
+                                        int esize, int msize, uint32_t mtedesc,
+                                        uintptr_t ra);
+
+static void sve_cont_ldst_mte_check1(SVEContLdSt *info, CPUARMState *env,
+                                     uint64_t *vg, target_ulong addr,
+                                     int esize, int msize, uint32_t mtedesc,
+                                     uintptr_t ra)
+{
+    sve_cont_ldst_mte_check_int(info, env, vg, addr, esize, msize,
+                                mtedesc, ra, mte_check1);
+}
+
+static void sve_cont_ldst_mte_checkN(SVEContLdSt *info, CPUARMState *env,
+                                     uint64_t *vg, target_ulong addr,
+                                     int esize, int msize, uint32_t mtedesc,
+                                     uintptr_t ra)
+{
+    sve_cont_ldst_mte_check_int(info, env, vg, addr, esize, msize,
+                                mtedesc, ra, mte_checkN);
+}
+
+
 /*
  * Common helper for all contiguous 1,2,3,4-register predicated stores.
  */
 static inline QEMU_ALWAYS_INLINE
 void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
                uint32_t desc, const uintptr_t retaddr,
-               const int esz, const int msz, const int N,
+               const int esz, const int msz, const int N, uint32_t mtedesc,
                sve_ldst1_host_fn *host_fn,
-               sve_ldst1_tlb_fn *tlb_fn)
+               sve_ldst1_tlb_fn *tlb_fn,
+               sve_cont_ldst_mte_check_fn *mte_check_fn)
 {
     const unsigned rd = simd_data(desc);
     const intptr_t reg_max = simd_oprsz(desc);
@@ -4426,7 +4500,14 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
     sve_cont_ldst_watchpoints(&info, env, vg, addr, 1 << esz, N << msz,
                               BP_MEM_READ, retaddr);
 
-    /* TODO: MTE check. */
+    /*
+     * Handle mte checks for all active elements.
+     * Since TBI must be set for MTE, !mtedesc => !mte_active.
+     */
+    if (mte_check_fn && mtedesc) {
+        mte_check_fn(&info, env, vg, addr, 1 << esz, N << msz,
+                     mtedesc, retaddr);
+    }
 
     flags = info.page[0].flags | info.page[1].flags;
     if (unlikely(flags != 0)) {
@@ -4532,26 +4613,67 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
     }
 }
 
-#define DO_LD1_1(NAME, ESZ) \
-void HELPER(sve_##NAME##_r)(CPUARMState *env, void *vg,        \
-                            target_ulong addr, uint32_t desc)  \
-{                                                              \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, 1,      \
-              sve_##NAME##_host, sve_##NAME##_tlb);            \
+static inline QEMU_ALWAYS_INLINE
+void sve_ldN_r_mte(CPUARMState *env, uint64_t *vg, target_ulong addr,
+                   uint32_t desc, const uintptr_t ra,
+                   const int esz, const int msz, const int N,
+                   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_ldN_r(env, vg, addr, desc, ra, esz, msz, N, mtedesc, host_fn, tlb_fn,
+              N == 1 ? sve_cont_ldst_mte_check1 : sve_cont_ldst_mte_checkN);
 }
 
-#define DO_LD1_2(NAME, ESZ, MSZ) \
-void HELPER(sve_##NAME##_le_r)(CPUARMState *env, void *vg,        \
-                               target_ulong addr, uint32_t desc)  \
-{                                                                 \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1,          \
-              sve_##NAME##_le_host, sve_##NAME##_le_tlb);         \
-}                                                                 \
-void HELPER(sve_##NAME##_be_r)(CPUARMState *env, void *vg,        \
-                               target_ulong addr, uint32_t desc)  \
-{                                                                 \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1,          \
-              sve_##NAME##_be_host, sve_##NAME##_be_tlb);         \
+#define DO_LD1_1(NAME, ESZ)                                             \
+void HELPER(sve_##NAME##_r)(CPUARMState *env, void *vg,                 \
+                            target_ulong addr, uint32_t desc)           \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, 1, 0,            \
+              sve_##NAME##_host, sve_##NAME##_tlb, NULL);               \
+}                                                                       \
+void HELPER(sve_##NAME##_r_mte)(CPUARMState *env, void *vg,             \
+                                target_ulong addr, uint32_t desc)       \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, 1,           \
+                  sve_##NAME##_host, sve_##NAME##_tlb);                 \
+}
+
+#define DO_LD1_2(NAME, ESZ, MSZ)                                        \
+void HELPER(sve_##NAME##_le_r)(CPUARMState *env, void *vg,              \
+                               target_ulong addr, uint32_t desc)        \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, 0,             \
+              sve_##NAME##_le_host, sve_##NAME##_le_tlb, NULL);         \
+}                                                                       \
+void HELPER(sve_##NAME##_be_r)(CPUARMState *env, void *vg,              \
+                               target_ulong addr, uint32_t desc)        \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, 0,             \
+              sve_##NAME##_be_host, sve_##NAME##_be_tlb, NULL);         \
+}                                                                       \
+void HELPER(sve_##NAME##_le_r_mte)(CPUARMState *env, void *vg,          \
+                                 target_ulong addr, uint32_t desc)      \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1,            \
+                  sve_##NAME##_le_host, sve_##NAME##_le_tlb);           \
+}                                                                       \
+void HELPER(sve_##NAME##_be_r_mte)(CPUARMState *env, void *vg,          \
+                                 target_ulong addr, uint32_t desc)      \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1,            \
+                  sve_##NAME##_be_host, sve_##NAME##_be_tlb);           \
 }
 
 DO_LD1_1(ld1bb,  MO_8)
@@ -4577,26 +4699,44 @@ DO_LD1_2(ld1dd,  MO_64, MO_64)
 #undef DO_LD1_1
 #undef DO_LD1_2
 
-#define DO_LDN_1(N) \
-void HELPER(sve_ld##N##bb_r)(CPUARMState *env, void *vg,        \
-                             target_ulong addr, uint32_t desc)  \
-{                                                               \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), MO_8, MO_8, N,      \
-              sve_ld1bb_host, sve_ld1bb_tlb);                   \
+#define DO_LDN_1(N)                                                     \
+void HELPER(sve_ld##N##bb_r)(CPUARMState *env, void *vg,                \
+                             target_ulong addr, uint32_t desc)          \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), MO_8, MO_8, N, 0,           \
+              sve_ld1bb_host, sve_ld1bb_tlb, NULL);                     \
+}                                                                       \
+void HELPER(sve_ld##N##bb_r_mte)(CPUARMState *env, void *vg,            \
+                                 target_ulong addr, uint32_t desc)      \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), MO_8, MO_8, N,          \
+                  sve_ld1bb_host, sve_ld1bb_tlb);                       \
 }
 
-#define DO_LDN_2(N, SUFF, ESZ) \
-void HELPER(sve_ld##N##SUFF##_le_r)(CPUARMState *env, void *vg,       \
-                                    target_ulong addr, uint32_t desc) \
-{                                                                     \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N,              \
-              sve_ld1##SUFF##_le_host, sve_ld1##SUFF##_le_tlb);       \
-}                                                                     \
-void HELPER(sve_ld##N##SUFF##_be_r)(CPUARMState *env, void *vg,       \
-                                    target_ulong addr, uint32_t desc) \
-{                                                                     \
-    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N,              \
-              sve_ld1##SUFF##_be_host, sve_ld1##SUFF##_be_tlb);       \
+#define DO_LDN_2(N, SUFF, ESZ)                                          \
+void HELPER(sve_ld##N##SUFF##_le_r)(CPUARMState *env, void *vg,         \
+                                    target_ulong addr, uint32_t desc)   \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, 0,             \
+              sve_ld1##SUFF##_le_host, sve_ld1##SUFF##_le_tlb, NULL);   \
+}                                                                       \
+void HELPER(sve_ld##N##SUFF##_be_r)(CPUARMState *env, void *vg,         \
+                                    target_ulong addr, uint32_t desc)   \
+{                                                                       \
+    sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, 0,             \
+              sve_ld1##SUFF##_be_host, sve_ld1##SUFF##_be_tlb, NULL);   \
+}                                                                       \
+void HELPER(sve_ld##N##SUFF##_le_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, ESZ, N,            \
+                  sve_ld1##SUFF##_le_host, sve_ld1##SUFF##_le_tlb);     \
+}                                                                       \
+void HELPER(sve_ld##N##SUFF##_be_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, ESZ, N,            \
+                  sve_ld1##SUFF##_be_host, sve_ld1##SUFF##_be_tlb);     \
 }
 
 DO_LDN_1(2)
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index a3a0b98fbc..2620c965f0 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4575,18 +4575,32 @@ static const uint8_t dtype_esz[16] = {
 };
 
 static void do_mem_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
-                       int dtype, gen_helper_gvec_mem *fn)
+                       int dtype, uint32_t mte_n, bool is_write,
+                       gen_helper_gvec_mem *fn)
 {
     unsigned vsz = vec_full_reg_size(s);
     TCGv_ptr t_pg;
     TCGv_i32 t_desc;
-    int desc;
+    int desc = 0;
 
-    /* For e.g. LD4, there are not enough arguments to pass all 4
+    /*
+     * For e.g. LD4, there are not enough arguments to pass all 4
      * registers as pointers, so encode the regno into the data field.
      * For consistency, do this even for LD1.
+     * TODO: mte_n check here while callers are updated.
      */
-    desc = simd_desc(vsz, vsz, zt);
+    if (mte_n && s->mte_active[0]) {
+        int msz = dtype_msz(dtype);
+
+        desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s));
+        desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid);
+        desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma);
+        desc = FIELD_DP32(desc, MTEDESC, WRITE, is_write);
+        desc = FIELD_DP32(desc, MTEDESC, ESIZE, 1 << msz);
+        desc = FIELD_DP32(desc, MTEDESC, TSIZE, mte_n << msz);
+        desc <<= SVE_MTEDESC_SHIFT;
+    }
+    desc = simd_desc(vsz, vsz, zt | desc);
     t_desc = tcg_const_i32(desc);
     t_pg = tcg_temp_new_ptr();
 
@@ -4600,64 +4614,132 @@ static void do_mem_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
 static void do_ld_zpa(DisasContext *s, int zt, int pg,
                       TCGv_i64 addr, int dtype, int nreg)
 {
-    static gen_helper_gvec_mem * const fns[2][16][4] = {
-        /* Little-endian */
-        { { gen_helper_sve_ld1bb_r, gen_helper_sve_ld2bb_r,
+    static gen_helper_gvec_mem * const fns[2][2][16][4] = {
+        { /* mte inactive, little-endian */
+          { { gen_helper_sve_ld1bb_r, gen_helper_sve_ld2bb_r,
             gen_helper_sve_ld3bb_r, gen_helper_sve_ld4bb_r },
-          { gen_helper_sve_ld1bhu_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bsu_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bdu_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bhu_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bsu_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bdu_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1sds_le_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hh_le_r, gen_helper_sve_ld2hh_le_r,
-            gen_helper_sve_ld3hh_le_r, gen_helper_sve_ld4hh_le_r },
-          { gen_helper_sve_ld1hsu_le_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hdu_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1sds_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hh_le_r, gen_helper_sve_ld2hh_le_r,
+              gen_helper_sve_ld3hh_le_r, gen_helper_sve_ld4hh_le_r },
+            { gen_helper_sve_ld1hsu_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hdu_le_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1hds_le_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hss_le_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1ss_le_r, gen_helper_sve_ld2ss_le_r,
-            gen_helper_sve_ld3ss_le_r, gen_helper_sve_ld4ss_le_r },
-          { gen_helper_sve_ld1sdu_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hds_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hss_le_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1ss_le_r, gen_helper_sve_ld2ss_le_r,
+              gen_helper_sve_ld3ss_le_r, gen_helper_sve_ld4ss_le_r },
+            { gen_helper_sve_ld1sdu_le_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1bds_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bss_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bhs_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1dd_le_r, gen_helper_sve_ld2dd_le_r,
-            gen_helper_sve_ld3dd_le_r, gen_helper_sve_ld4dd_le_r } },
+            { gen_helper_sve_ld1bds_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bss_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bhs_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1dd_le_r, gen_helper_sve_ld2dd_le_r,
+              gen_helper_sve_ld3dd_le_r, gen_helper_sve_ld4dd_le_r } },
 
-        /* Big-endian */
-        { { gen_helper_sve_ld1bb_r, gen_helper_sve_ld2bb_r,
-            gen_helper_sve_ld3bb_r, gen_helper_sve_ld4bb_r },
-          { gen_helper_sve_ld1bhu_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bsu_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bdu_r, NULL, NULL, NULL },
+          /* mte inactive, big-endian */
+          { { gen_helper_sve_ld1bb_r, gen_helper_sve_ld2bb_r,
+              gen_helper_sve_ld3bb_r, gen_helper_sve_ld4bb_r },
+            { gen_helper_sve_ld1bhu_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bsu_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bdu_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1sds_be_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hh_be_r, gen_helper_sve_ld2hh_be_r,
-            gen_helper_sve_ld3hh_be_r, gen_helper_sve_ld4hh_be_r },
-          { gen_helper_sve_ld1hsu_be_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hdu_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1sds_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hh_be_r, gen_helper_sve_ld2hh_be_r,
+              gen_helper_sve_ld3hh_be_r, gen_helper_sve_ld4hh_be_r },
+            { gen_helper_sve_ld1hsu_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hdu_be_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1hds_be_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1hss_be_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1ss_be_r, gen_helper_sve_ld2ss_be_r,
-            gen_helper_sve_ld3ss_be_r, gen_helper_sve_ld4ss_be_r },
-          { gen_helper_sve_ld1sdu_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hds_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hss_be_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1ss_be_r, gen_helper_sve_ld2ss_be_r,
+              gen_helper_sve_ld3ss_be_r, gen_helper_sve_ld4ss_be_r },
+            { gen_helper_sve_ld1sdu_be_r, NULL, NULL, NULL },
 
-          { gen_helper_sve_ld1bds_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bss_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1bhs_r, NULL, NULL, NULL },
-          { gen_helper_sve_ld1dd_be_r, gen_helper_sve_ld2dd_be_r,
-            gen_helper_sve_ld3dd_be_r, gen_helper_sve_ld4dd_be_r } }
+            { gen_helper_sve_ld1bds_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bss_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bhs_r, NULL, NULL, NULL },
+            { gen_helper_sve_ld1dd_be_r, gen_helper_sve_ld2dd_be_r,
+              gen_helper_sve_ld3dd_be_r, gen_helper_sve_ld4dd_be_r } } },
+
+        { /* mte active, little-endian */
+          { { gen_helper_sve_ld1bb_r_mte,
+              gen_helper_sve_ld2bb_r_mte,
+              gen_helper_sve_ld3bb_r_mte,
+              gen_helper_sve_ld4bb_r_mte },
+            { gen_helper_sve_ld1bhu_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bsu_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bdu_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1sds_le_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hh_le_r_mte,
+              gen_helper_sve_ld2hh_le_r_mte,
+              gen_helper_sve_ld3hh_le_r_mte,
+              gen_helper_sve_ld4hh_le_r_mte },
+            { gen_helper_sve_ld1hsu_le_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hdu_le_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1hds_le_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hss_le_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1ss_le_r_mte,
+              gen_helper_sve_ld2ss_le_r_mte,
+              gen_helper_sve_ld3ss_le_r_mte,
+              gen_helper_sve_ld4ss_le_r_mte },
+            { gen_helper_sve_ld1sdu_le_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1bds_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bss_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bhs_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1dd_le_r_mte,
+              gen_helper_sve_ld2dd_le_r_mte,
+              gen_helper_sve_ld3dd_le_r_mte,
+              gen_helper_sve_ld4dd_le_r_mte } },
+
+          /* mte active, big-endian */
+          { { gen_helper_sve_ld1bb_r_mte,
+              gen_helper_sve_ld2bb_r_mte,
+              gen_helper_sve_ld3bb_r_mte,
+              gen_helper_sve_ld4bb_r_mte },
+            { gen_helper_sve_ld1bhu_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bsu_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bdu_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1sds_be_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hh_be_r_mte,
+              gen_helper_sve_ld2hh_be_r_mte,
+              gen_helper_sve_ld3hh_be_r_mte,
+              gen_helper_sve_ld4hh_be_r_mte },
+            { gen_helper_sve_ld1hsu_be_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hdu_be_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1hds_be_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1hss_be_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1ss_be_r_mte,
+              gen_helper_sve_ld2ss_be_r_mte,
+              gen_helper_sve_ld3ss_be_r_mte,
+              gen_helper_sve_ld4ss_be_r_mte },
+            { gen_helper_sve_ld1sdu_be_r_mte, NULL, NULL, NULL },
+
+            { gen_helper_sve_ld1bds_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bss_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1bhs_r_mte, NULL, NULL, NULL },
+            { gen_helper_sve_ld1dd_be_r_mte,
+              gen_helper_sve_ld2dd_be_r_mte,
+              gen_helper_sve_ld3dd_be_r_mte,
+              gen_helper_sve_ld4dd_be_r_mte } } },
     };
-    gen_helper_gvec_mem *fn = fns[s->be_data == MO_BE][dtype][nreg];
+    gen_helper_gvec_mem *fn
+        = fns[s->mte_active[0]][s->be_data == MO_BE][dtype][nreg];
 
-    /* While there are holes in the table, they are not
+    /*
+     * While there are holes in the table, they are not
      * accessible via the instruction encoding.
      */
     assert(fn != NULL);
-    do_mem_zpa(s, zt, pg, addr, dtype, fn);
+    do_mem_zpa(s, zt, pg, addr, dtype, nreg, false, fn);
 }
 
 static bool trans_LD_zprr(DisasContext *s, arg_rprr_load *a)
@@ -4739,7 +4821,7 @@ static bool trans_LDFF1_zprr(DisasContext *s, arg_rprr_load *a)
         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,
+        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 0, false,
                    fns[s->be_data == MO_BE][a->dtype]);
     }
     return true;
@@ -4798,7 +4880,7 @@ 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,
+        do_mem_zpa(s, a->rd, a->pg, addr, a->dtype, 0, false,
                    fns[s->be_data == MO_BE][a->dtype]);
     }
     return true;
@@ -5002,7 +5084,7 @@ static void do_st_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
         fn = fn_multiple[be][nreg - 1][msz];
     }
     assert(fn != NULL);
-    do_mem_zpa(s, zt, pg, addr, msz_dtype(s, msz), fn);
+    do_mem_zpa(s, zt, pg, addr, msz_dtype(s, msz), 0, true, fn);
 }
 
 static bool trans_ST_zprr(DisasContext *s, arg_rprr_store *a)
-- 
2.25.1

[PATCH v8 34/45] target/arm: Add mte helpers for sve scalar + int stores

[Richard Henderson]

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.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-sve.h    |  47 +++++++++++
 target/arm/sve_helper.c    |  95 ++++++++++++++++------
 target/arm/translate-sve.c | 162 ++++++++++++++++++++++++-------------
 3 files changed, 226 insertions(+), 78 deletions(-)

diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index 1bc1974fc2..1425f33c92 100644
--- a/target/arm/helper-sve.h
+++ b/target/arm/helper-sve.h
@@ -1363,6 +1363,53 @@ DEF_HELPER_FLAGS_4(sve_st1hd_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_st1sd_le_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 DEF_HELPER_FLAGS_4(sve_st1sd_be_r, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_4(sve_st1bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4bb_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4hh_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4hh_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4ss_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4ss_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4dd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st2dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st3dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st4dd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1bh_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1bs_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1bd_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1hs_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1hd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1hs_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1hd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_4(sve_st1sd_le_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+DEF_HELPER_FLAGS_4(sve_st1sd_be_r_mte, TCG_CALL_NO_WG, void, env, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_6(sve_ldbsu_zsu, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 DEF_HELPER_FLAGS_6(sve_ldhsu_le_zsu, TCG_CALL_NO_WG,
diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index 767ecb399f..ded9cedd18 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -5022,11 +5022,12 @@ DO_LDFF1_LDNF1_2(dd,  MO_64, MO_64)
  */
 
 static inline QEMU_ALWAYS_INLINE
-void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc,
-               const uintptr_t retaddr, const int esz,
-               const int msz, const int N,
+void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr,
+               uint32_t desc, const uintptr_t retaddr,
+               const int esz, const int msz, const int N, uint32_t mtedesc,
                sve_ldst1_host_fn *host_fn,
-               sve_ldst1_tlb_fn *tlb_fn)
+               sve_ldst1_tlb_fn *tlb_fn,
+               sve_cont_ldst_mte_check_fn *mte_check_fn)
 {
     const unsigned rd = simd_data(desc);
     const intptr_t reg_max = simd_oprsz(desc);
@@ -5048,7 +5049,14 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc,
     sve_cont_ldst_watchpoints(&info, env, vg, addr, 1 << esz, N << msz,
                               BP_MEM_WRITE, retaddr);
 
-    /* TODO: MTE check. */
+    /*
+     * Handle mte checks for all active elements.
+     * Since TBI must be set for MTE, !mtedesc => !mte_active.
+     */
+    if (mte_check_fn && mtedesc) {
+        mte_check_fn(&info, env, vg, addr, 1 << esz, N << msz,
+                     mtedesc, retaddr);
+    }
 
     flags = info.page[0].flags | info.page[1].flags;
     if (unlikely(flags != 0)) {
@@ -5142,26 +5150,67 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc,
     }
 }
 
-#define DO_STN_1(N, NAME, ESZ) \
-void HELPER(sve_st##N##NAME##_r)(CPUARMState *env, void *vg,        \
-                                 target_ulong addr, uint32_t desc)  \
-{                                                                   \
-    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, N,           \
-              sve_st1##NAME##_host, sve_st1##NAME##_tlb);           \
+static inline QEMU_ALWAYS_INLINE
+void sve_stN_r_mte(CPUARMState *env, uint64_t *vg, target_ulong addr,
+                   uint32_t desc, const uintptr_t ra,
+                   const int esz, const int msz, const int N,
+                   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_stN_r(env, vg, addr, desc, ra, esz, msz, N, mtedesc, host_fn, tlb_fn,
+              N == 1 ? sve_cont_ldst_mte_check1 : sve_cont_ldst_mte_checkN);
 }
 
-#define DO_STN_2(N, NAME, ESZ, MSZ) \
-void HELPER(sve_st##N##NAME##_le_r)(CPUARMState *env, void *vg,       \
-                                    target_ulong addr, uint32_t desc) \
-{                                                                     \
-    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N,              \
-              sve_st1##NAME##_le_host, sve_st1##NAME##_le_tlb);       \
-}                                                                     \
-void HELPER(sve_st##N##NAME##_be_r)(CPUARMState *env, void *vg,       \
-                                    target_ulong addr, uint32_t desc) \
-{                                                                     \
-    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N,              \
-              sve_st1##NAME##_be_host, sve_st1##NAME##_be_tlb);       \
+#define DO_STN_1(N, NAME, ESZ)                                          \
+void HELPER(sve_st##N##NAME##_r)(CPUARMState *env, void *vg,            \
+                                 target_ulong addr, uint32_t desc)      \
+{                                                                       \
+    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, N, 0,            \
+              sve_st1##NAME##_host, sve_st1##NAME##_tlb, NULL);         \
+}                                                                       \
+void HELPER(sve_st##N##NAME##_r_mte)(CPUARMState *env, void *vg,        \
+                                     target_ulong addr, uint32_t desc)  \
+{                                                                       \
+    sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, N,           \
+                  sve_st1##NAME##_host, sve_st1##NAME##_tlb);           \
+}
+
+#define DO_STN_2(N, NAME, ESZ, MSZ)                                     \
+void HELPER(sve_st##N##NAME##_le_r)(CPUARMState *env, void *vg,         \
+                                    target_ulong addr, uint32_t desc)   \
+{                                                                       \
+    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, 0,             \
+              sve_st1##NAME##_le_host, sve_st1##NAME##_le_tlb, NULL);   \
+}                                                                       \
+void HELPER(sve_st##N##NAME##_be_r)(CPUARMState *env, void *vg,         \
+                                    target_ulong addr, uint32_t desc)   \
+{                                                                       \
+    sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, 0,             \
+              sve_st1##NAME##_be_host, sve_st1##NAME##_be_tlb, NULL);   \
+}                                                                       \
+void HELPER(sve_st##N##NAME##_le_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, N,            \
+                  sve_st1##NAME##_le_host, sve_st1##NAME##_le_tlb);     \
+}                                                                       \
+void HELPER(sve_st##N##NAME##_be_r_mte)(CPUARMState *env, void *vg,     \
+                                        target_ulong addr, uint32_t desc) \
+{                                                                       \
+    sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, N,            \
+                  sve_st1##NAME##_be_host, sve_st1##NAME##_be_tlb);     \
 }
 
 DO_STN_1(1, bb, MO_8)
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 2620c965f0..daac8589f3 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -5018,73 +5018,125 @@ static bool trans_LD1R_zpri(DisasContext *s, arg_rpri_load *a)
 static void do_st_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
                       int msz, int esz, int nreg)
 {
-    static gen_helper_gvec_mem * const fn_single[2][4][4] = {
-        { { gen_helper_sve_st1bb_r,
-            gen_helper_sve_st1bh_r,
-            gen_helper_sve_st1bs_r,
-            gen_helper_sve_st1bd_r },
-          { NULL,
-            gen_helper_sve_st1hh_le_r,
-            gen_helper_sve_st1hs_le_r,
-            gen_helper_sve_st1hd_le_r },
-          { NULL, NULL,
-            gen_helper_sve_st1ss_le_r,
-            gen_helper_sve_st1sd_le_r },
-          { NULL, NULL, NULL,
-            gen_helper_sve_st1dd_le_r } },
-        { { gen_helper_sve_st1bb_r,
-            gen_helper_sve_st1bh_r,
-            gen_helper_sve_st1bs_r,
-            gen_helper_sve_st1bd_r },
-          { NULL,
-            gen_helper_sve_st1hh_be_r,
-            gen_helper_sve_st1hs_be_r,
-            gen_helper_sve_st1hd_be_r },
-          { NULL, NULL,
-            gen_helper_sve_st1ss_be_r,
-            gen_helper_sve_st1sd_be_r },
-          { NULL, NULL, NULL,
-            gen_helper_sve_st1dd_be_r } },
+    static gen_helper_gvec_mem * const fn_single[2][2][4][4] = {
+        { { { gen_helper_sve_st1bb_r,
+              gen_helper_sve_st1bh_r,
+              gen_helper_sve_st1bs_r,
+              gen_helper_sve_st1bd_r },
+            { NULL,
+              gen_helper_sve_st1hh_le_r,
+              gen_helper_sve_st1hs_le_r,
+              gen_helper_sve_st1hd_le_r },
+            { NULL, NULL,
+              gen_helper_sve_st1ss_le_r,
+              gen_helper_sve_st1sd_le_r },
+            { NULL, NULL, NULL,
+              gen_helper_sve_st1dd_le_r } },
+          { { gen_helper_sve_st1bb_r,
+              gen_helper_sve_st1bh_r,
+              gen_helper_sve_st1bs_r,
+              gen_helper_sve_st1bd_r },
+            { NULL,
+              gen_helper_sve_st1hh_be_r,
+              gen_helper_sve_st1hs_be_r,
+              gen_helper_sve_st1hd_be_r },
+            { NULL, NULL,
+              gen_helper_sve_st1ss_be_r,
+              gen_helper_sve_st1sd_be_r },
+            { NULL, NULL, NULL,
+              gen_helper_sve_st1dd_be_r } } },
+
+        { { { gen_helper_sve_st1bb_r_mte,
+              gen_helper_sve_st1bh_r_mte,
+              gen_helper_sve_st1bs_r_mte,
+              gen_helper_sve_st1bd_r_mte },
+            { NULL,
+              gen_helper_sve_st1hh_le_r_mte,
+              gen_helper_sve_st1hs_le_r_mte,
+              gen_helper_sve_st1hd_le_r_mte },
+            { NULL, NULL,
+              gen_helper_sve_st1ss_le_r_mte,
+              gen_helper_sve_st1sd_le_r_mte },
+            { NULL, NULL, NULL,
+              gen_helper_sve_st1dd_le_r_mte } },
+          { { gen_helper_sve_st1bb_r_mte,
+              gen_helper_sve_st1bh_r_mte,
+              gen_helper_sve_st1bs_r_mte,
+              gen_helper_sve_st1bd_r_mte },
+            { NULL,
+              gen_helper_sve_st1hh_be_r_mte,
+              gen_helper_sve_st1hs_be_r_mte,
+              gen_helper_sve_st1hd_be_r_mte },
+            { NULL, NULL,
+              gen_helper_sve_st1ss_be_r_mte,
+              gen_helper_sve_st1sd_be_r_mte },
+            { NULL, NULL, NULL,
+              gen_helper_sve_st1dd_be_r_mte } } },
     };
-    static gen_helper_gvec_mem * const fn_multiple[2][3][4] = {
-        { { gen_helper_sve_st2bb_r,
-            gen_helper_sve_st2hh_le_r,
-            gen_helper_sve_st2ss_le_r,
-            gen_helper_sve_st2dd_le_r },
-          { gen_helper_sve_st3bb_r,
-            gen_helper_sve_st3hh_le_r,
-            gen_helper_sve_st3ss_le_r,
-            gen_helper_sve_st3dd_le_r },
-          { gen_helper_sve_st4bb_r,
-            gen_helper_sve_st4hh_le_r,
-            gen_helper_sve_st4ss_le_r,
-            gen_helper_sve_st4dd_le_r } },
-        { { gen_helper_sve_st2bb_r,
-            gen_helper_sve_st2hh_be_r,
-            gen_helper_sve_st2ss_be_r,
-            gen_helper_sve_st2dd_be_r },
-          { gen_helper_sve_st3bb_r,
-            gen_helper_sve_st3hh_be_r,
-            gen_helper_sve_st3ss_be_r,
-            gen_helper_sve_st3dd_be_r },
-          { gen_helper_sve_st4bb_r,
-            gen_helper_sve_st4hh_be_r,
-            gen_helper_sve_st4ss_be_r,
-            gen_helper_sve_st4dd_be_r } },
+    static gen_helper_gvec_mem * const fn_multiple[2][2][3][4] = {
+        { { { gen_helper_sve_st2bb_r,
+              gen_helper_sve_st2hh_le_r,
+              gen_helper_sve_st2ss_le_r,
+              gen_helper_sve_st2dd_le_r },
+            { gen_helper_sve_st3bb_r,
+              gen_helper_sve_st3hh_le_r,
+              gen_helper_sve_st3ss_le_r,
+              gen_helper_sve_st3dd_le_r },
+            { gen_helper_sve_st4bb_r,
+              gen_helper_sve_st4hh_le_r,
+              gen_helper_sve_st4ss_le_r,
+              gen_helper_sve_st4dd_le_r } },
+          { { gen_helper_sve_st2bb_r,
+              gen_helper_sve_st2hh_be_r,
+              gen_helper_sve_st2ss_be_r,
+              gen_helper_sve_st2dd_be_r },
+            { gen_helper_sve_st3bb_r,
+              gen_helper_sve_st3hh_be_r,
+              gen_helper_sve_st3ss_be_r,
+              gen_helper_sve_st3dd_be_r },
+            { gen_helper_sve_st4bb_r,
+              gen_helper_sve_st4hh_be_r,
+              gen_helper_sve_st4ss_be_r,
+              gen_helper_sve_st4dd_be_r } } },
+        { { { gen_helper_sve_st2bb_r_mte,
+              gen_helper_sve_st2hh_le_r_mte,
+              gen_helper_sve_st2ss_le_r_mte,
+              gen_helper_sve_st2dd_le_r_mte },
+            { gen_helper_sve_st3bb_r_mte,
+              gen_helper_sve_st3hh_le_r_mte,
+              gen_helper_sve_st3ss_le_r_mte,
+              gen_helper_sve_st3dd_le_r_mte },
+            { gen_helper_sve_st4bb_r_mte,
+              gen_helper_sve_st4hh_le_r_mte,
+              gen_helper_sve_st4ss_le_r_mte,
+              gen_helper_sve_st4dd_le_r_mte } },
+          { { gen_helper_sve_st2bb_r_mte,
+              gen_helper_sve_st2hh_be_r_mte,
+              gen_helper_sve_st2ss_be_r_mte,
+              gen_helper_sve_st2dd_be_r_mte },
+            { gen_helper_sve_st3bb_r_mte,
+              gen_helper_sve_st3hh_be_r_mte,
+              gen_helper_sve_st3ss_be_r_mte,
+              gen_helper_sve_st3dd_be_r_mte },
+            { gen_helper_sve_st4bb_r_mte,
+              gen_helper_sve_st4hh_be_r_mte,
+              gen_helper_sve_st4ss_be_r_mte,
+              gen_helper_sve_st4dd_be_r_mte } } },
     };
     gen_helper_gvec_mem *fn;
     int be = s->be_data == MO_BE;
 
     if (nreg == 0) {
         /* ST1 */
-        fn = fn_single[be][msz][esz];
+        fn = fn_single[s->mte_active[0]][be][msz][esz];
+        nreg = 1;
     } else {
         /* ST2, ST3, ST4 -- msz == esz, enforced by encoding */
         assert(msz == esz);
-        fn = fn_multiple[be][nreg - 1][msz];
+        fn = fn_multiple[s->mte_active[0]][be][nreg - 1][msz];
     }
     assert(fn != NULL);
-    do_mem_zpa(s, zt, pg, addr, msz_dtype(s, msz), 0, true, fn);
+    do_mem_zpa(s, zt, pg, addr, msz_dtype(s, msz), nreg, true, fn);
 }
 
 static bool trans_ST_zprr(DisasContext *s, arg_rprr_store *a)
-- 
2.25.1

[PATCH v8 35/45] target/arm: Add mte helpers for sve scalar + int ff/nf loads

[Richard Henderson]

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.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-sve.h    |  98 ++++++++++++++++
 target/arm/sve_helper.c    |  99 ++++++++++++++--
 target/arm/translate-sve.c | 232 +++++++++++++++++++++++++------------
 3 files changed, 343 insertions(+), 86 deletions(-)

diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index 1425f33c92..f48752eb42 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 ded9cedd18..7aca4ad384 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 daac8589f3..e4fbe48493 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,
+    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_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_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 },
+            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,
+          /* 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_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_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 },
+            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,
+    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_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_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 },
+            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,
+          /* 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_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_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 },
+            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;
 }
-- 
2.25.1

[PATCH v8 36/45] target/arm: Handle TBI for sve scalar + int memory ops

[Richard Henderson]

We still need to handle tbi for user-only when mte is inactive.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate-a64.h | 1 +
 target/arm/translate-a64.c | 2 +-
 target/arm/translate-sve.c | 6 ++++--
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h
index 781c441399..49e4865918 100644
--- a/target/arm/translate-a64.h
+++ b/target/arm/translate-a64.h
@@ -40,6 +40,7 @@ TCGv_ptr get_fpstatus_ptr(bool);
 bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn,
                             unsigned int imms, unsigned int immr);
 bool sve_access_check(DisasContext *s);
+TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr);
 TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write,
                         bool tag_checked, int log2_size);
 TCGv_i64 gen_mte_checkN(DisasContext *s, TCGv_i64 addr, bool is_write,
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index ec2295393d..f35b122ded 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -215,7 +215,7 @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src)
  * of the write-back address.
  */
 
-static TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr)
+TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr)
 {
     TCGv_i64 clean = new_tmp_a64(s);
 #ifdef CONFIG_USER_ONLY
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index e4fbe48493..04eda9a126 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -4587,9 +4587,8 @@ static void do_mem_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
      * For e.g. LD4, there are not enough arguments to pass all 4
      * registers as pointers, so encode the regno into the data field.
      * For consistency, do this even for LD1.
-     * TODO: mte_n check here while callers are updated.
      */
-    if (mte_n && s->mte_active[0]) {
+    if (s->mte_active[0]) {
         int msz = dtype_msz(dtype);
 
         desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s));
@@ -4599,7 +4598,10 @@ static void do_mem_zpa(DisasContext *s, int zt, int pg, TCGv_i64 addr,
         desc = FIELD_DP32(desc, MTEDESC, ESIZE, 1 << msz);
         desc = FIELD_DP32(desc, MTEDESC, TSIZE, mte_n << msz);
         desc <<= SVE_MTEDESC_SHIFT;
+    } else {
+        addr = clean_data_tbi(s, addr);
     }
+
     desc = simd_desc(vsz, vsz, zt | desc);
     t_desc = tcg_const_i32(desc);
     t_pg = tcg_temp_new_ptr();
-- 
2.25.1

[PATCH v8 37/45] target/arm: Add mte helpers for sve scatter/gather memory ops

[Richard Henderson]

Because the elements are non-sequential, we cannot eliminate many
tests straight away like we can for sequential operations.  But
we often have the PTE details handy, so we can test for Tagged.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-sve.h    | 285 ++++++++++++++++
 target/arm/sve_helper.c    | 185 +++++++++--
 target/arm/translate-sve.c | 650 +++++++++++++++++++++++++------------
 3 files changed, 872 insertions(+), 248 deletions(-)

diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index f48752eb42..63c4a087ca 100644
--- a/target/arm/helper-sve.h
+++ b/target/arm/helper-sve.h
@@ -1617,6 +1617,115 @@ DEF_HELPER_FLAGS_6(sve_ldsds_le_zd, TCG_CALL_NO_WG,
 DEF_HELPER_FLAGS_6(sve_ldsds_be_zd, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_6(sve_ldbsu_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhsu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhsu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldss_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldss_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldbss_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhss_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhss_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldbsu_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhsu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhsu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldss_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldss_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldbss_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhss_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhss_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldbdu_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldbds_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldbdu_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldbds_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldbdu_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhdu_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsdu_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_lddd_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldbds_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldhds_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldsds_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_6(sve_ldffbsu_zsu, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 DEF_HELPER_FLAGS_6(sve_ldffhsu_le_zsu, TCG_CALL_NO_WG,
@@ -1726,6 +1835,115 @@ DEF_HELPER_FLAGS_6(sve_ldffsds_le_zd, TCG_CALL_NO_WG,
 DEF_HELPER_FLAGS_6(sve_ldffsds_be_zd, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_6(sve_ldffbsu_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhsu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhsu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffss_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffss_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffbss_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhss_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhss_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldffbsu_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhsu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhsu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffss_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffss_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffbss_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhss_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhss_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldffbdu_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffbds_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldffbdu_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffbds_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_ldffbdu_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhdu_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsdu_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffdd_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffbds_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffhds_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_ldffsds_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_6(sve_stbs_zsu, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 DEF_HELPER_FLAGS_6(sve_sths_le_zsu, TCG_CALL_NO_WG,
@@ -1793,4 +2011,71 @@ DEF_HELPER_FLAGS_6(sve_stdd_le_zd, TCG_CALL_NO_WG,
 DEF_HELPER_FLAGS_6(sve_stdd_be_zd, TCG_CALL_NO_WG,
                    void, env, ptr, ptr, ptr, tl, i32)
 
+DEF_HELPER_FLAGS_6(sve_stbs_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sths_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sths_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stss_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stss_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_stbs_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sths_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sths_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stss_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stss_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_stbd_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_le_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_be_zsu_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_stbd_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_le_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_be_zss_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
+DEF_HELPER_FLAGS_6(sve_stbd_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_sthd_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stsd_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_le_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+DEF_HELPER_FLAGS_6(sve_stdd_be_zd_mte, TCG_CALL_NO_WG,
+                   void, env, ptr, ptr, ptr, tl, i32)
+
 DEF_HELPER_FLAGS_4(sve2_pmull_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index 7aca4ad384..ad974c2cc5 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -5354,7 +5354,8 @@ static target_ulong off_zd_d(void *reg, intptr_t reg_ofs)
 static inline QEMU_ALWAYS_INLINE
 void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                target_ulong base, uint32_t desc, uintptr_t retaddr,
-               int esize, int msize, zreg_off_fn *off_fn,
+               uint32_t mtedesc, int esize, int msize,
+               zreg_off_fn *off_fn,
                sve_ldst1_host_fn *host_fn,
                sve_ldst1_tlb_fn *tlb_fn)
 {
@@ -5382,7 +5383,9 @@ void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                         cpu_check_watchpoint(env_cpu(env), addr, msize,
                                              info.attrs, BP_MEM_READ, retaddr);
                     }
-                    /* TODO: MTE check */
+                    if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) {
+                        mte_check1(env, mtedesc, addr, retaddr);
+                    }
                     host_fn(&scratch, reg_off, info.host);
                 } else {
                     /* Element crosses the page boundary. */
@@ -5393,7 +5396,9 @@ void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                                              msize, info.attrs,
                                              BP_MEM_READ, retaddr);
                     }
-                    /* TODO: MTE check */
+                    if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) {
+                        mte_check1(env, mtedesc, addr, retaddr);
+                    }
                     tlb_fn(env, &scratch, reg_off, addr, retaddr);
                 }
             }
@@ -5406,20 +5411,53 @@ void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
     memcpy(vd, &scratch, reg_max);
 }
 
+static inline QEMU_ALWAYS_INLINE
+void sve_ld1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
+                   target_ulong base, uint32_t desc, uintptr_t retaddr,
+                   int esize, int msize, zreg_off_fn *off_fn,
+                   sve_ldst1_host_fn *host_fn,
+                   sve_ldst1_tlb_fn *tlb_fn)
+{
+    uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+    /* Remove mtedesc from the normal sve descriptor. */
+    desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+
+    /*
+     * ??? TODO: For the 32-bit offset extractions, base + ofs cannot
+     * offset base entirely over the address space hole to change the
+     * pointer tag, or change the bit55 selector.  So we could here
+     * examine TBI + TCMA like we do for sve_ldN_r_mte().
+     */
+    sve_ld1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc,
+              esize, msize, off_fn, host_fn, tlb_fn);
+}
+
 #define DO_LD1_ZPZ_S(MEM, OFS, MSZ) \
 void HELPER(sve_ld##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
                                  void *vm, target_ulong base, uint32_t desc) \
 {                                                                            \
-    sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ,             \
+    sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 4, 1 << MSZ,          \
               off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);       \
+}                                                                            \
+void HELPER(sve_ld##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \
+     void *vm, target_ulong base, uint32_t desc)                             \
+{                                                                            \
+    sve_ld1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ,         \
+                  off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);   \
 }
 
 #define DO_LD1_ZPZ_D(MEM, OFS, MSZ) \
 void HELPER(sve_ld##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
                                  void *vm, target_ulong base, uint32_t desc) \
 {                                                                            \
-    sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ,             \
+    sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 8, 1 << MSZ,          \
               off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);       \
+}                                                                            \
+void HELPER(sve_ld##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \
+    void *vm, target_ulong base, uint32_t desc)                              \
+{                                                                            \
+    sve_ld1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ,         \
+                  off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);   \
 }
 
 DO_LD1_ZPZ_S(bsu, zsu, MO_8)
@@ -5498,7 +5536,8 @@ DO_LD1_ZPZ_D(dd_be, zd, MO_64)
 static inline QEMU_ALWAYS_INLINE
 void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                  target_ulong base, uint32_t desc, uintptr_t retaddr,
-                 const int esz, const int msz, zreg_off_fn *off_fn,
+                 uint32_t mtedesc, const int esz, const int msz,
+                 zreg_off_fn *off_fn,
                  sve_ldst1_host_fn *host_fn,
                  sve_ldst1_tlb_fn *tlb_fn)
 {
@@ -5523,6 +5562,9 @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
      * Probe the first element, allowing faults.
      */
     addr = base + (off_fn(vm, reg_off) << scale);
+    if (mtedesc) {
+        mte_check1(env, mtedesc, addr, retaddr);
+    }
     tlb_fn(env, vd, reg_off, addr, retaddr);
 
     /* After any fault, zero the other elements. */
@@ -5555,7 +5597,11 @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                      (env_cpu(env), addr, msize) & BP_MEM_READ)) {
                     goto fault;
                 }
-                /* TODO: MTE check. */
+                if (mtedesc &&
+                    arm_tlb_mte_tagged(&info.attrs) &&
+                    !mte_probe1(env, mtedesc, addr)) {
+                    goto fault;
+                }
 
                 host_fn(vd, reg_off, info.host);
             }
@@ -5568,20 +5614,58 @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
     record_fault(env, reg_off, reg_max);
 }
 
-#define DO_LDFF1_ZPZ_S(MEM, OFS, MSZ) \
-void HELPER(sve_ldff##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
-                                   void *vm, target_ulong base, uint32_t desc) \
-{                                                                              \
-    sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), MO_32, MSZ,              \
-                off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);       \
+static inline QEMU_ALWAYS_INLINE
+void sve_ldff1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
+                     target_ulong base, uint32_t desc, uintptr_t retaddr,
+                     const int esz, const int msz,
+                     zreg_off_fn *off_fn,
+                     sve_ldst1_host_fn *host_fn,
+                     sve_ldst1_tlb_fn *tlb_fn)
+{
+    uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+    /* Remove mtedesc from the normal sve descriptor. */
+    desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+
+    /*
+     * ??? TODO: For the 32-bit offset extractions, base + ofs cannot
+     * offset base entirely over the address space hole to change the
+     * pointer tag, or change the bit55 selector.  So we could here
+     * examine TBI + TCMA like we do for sve_ldN_r_mte().
+     */
+    sve_ldff1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc,
+                esz, msz, off_fn, host_fn, tlb_fn);
 }
 
-#define DO_LDFF1_ZPZ_D(MEM, OFS, MSZ) \
-void HELPER(sve_ldff##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
-                                   void *vm, target_ulong base, uint32_t desc) \
-{                                                                              \
-    sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), MO_64, MSZ,              \
-                off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb);       \
+#define DO_LDFF1_ZPZ_S(MEM, OFS, MSZ)                                   \
+void HELPER(sve_ldff##MEM##_##OFS)                                      \
+    (CPUARMState *env, void *vd, void *vg,                              \
+     void *vm, target_ulong base, uint32_t desc)                        \
+{                                                                       \
+    sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), 0, MO_32, MSZ,    \
+                off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \
+}                                                                       \
+void HELPER(sve_ldff##MEM##_##OFS##_mte)                                \
+    (CPUARMState *env, void *vd, void *vg,                              \
+     void *vm, target_ulong base, uint32_t desc)                        \
+{                                                                       \
+    sve_ldff1_z_mte(env, vd, vg, vm, base, desc, GETPC(), MO_32, MSZ,   \
+                    off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \
+}
+
+#define DO_LDFF1_ZPZ_D(MEM, OFS, MSZ)                                   \
+void HELPER(sve_ldff##MEM##_##OFS)                                      \
+    (CPUARMState *env, void *vd, void *vg,                              \
+     void *vm, target_ulong base, uint32_t desc)                        \
+{                                                                       \
+    sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), 0, MO_64, MSZ,    \
+                off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \
+}                                                                       \
+void HELPER(sve_ldff##MEM##_##OFS##_mte)                                \
+    (CPUARMState *env, void *vd, void *vg,                              \
+     void *vm, target_ulong base, uint32_t desc)                        \
+{                                                                       \
+    sve_ldff1_z_mte(env, vd, vg, vm, base, desc, GETPC(), MO_64, MSZ,   \
+                    off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \
 }
 
 DO_LDFF1_ZPZ_S(bsu, zsu, MO_8)
@@ -5653,7 +5737,8 @@ DO_LDFF1_ZPZ_D(dd_be, zd, MO_64)
 static inline QEMU_ALWAYS_INLINE
 void sve_st1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                target_ulong base, uint32_t desc, uintptr_t retaddr,
-               int esize, int msize, zreg_off_fn *off_fn,
+               uint32_t mtedesc, int esize, int msize,
+               zreg_off_fn *off_fn,
                sve_ldst1_host_fn *host_fn,
                sve_ldst1_tlb_fn *tlb_fn)
 {
@@ -5697,7 +5782,10 @@ void sve_st1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                     cpu_check_watchpoint(env_cpu(env), addr, msize,
                                          info.attrs, BP_MEM_WRITE, retaddr);
                 }
-                /* TODO: MTE check. */
+
+                if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) {
+                    mte_check1(env, mtedesc, addr, retaddr);
+                }
             }
             i += 1;
             reg_off += esize;
@@ -5727,20 +5815,53 @@ void sve_st1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
     } while (reg_off < reg_max);
 }
 
-#define DO_ST1_ZPZ_S(MEM, OFS, MSZ) \
-void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
-                                 void *vm, target_ulong base, uint32_t desc) \
-{                                                                            \
-    sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ,             \
-              off_##OFS##_s, sve_st1##MEM##_host, sve_st1##MEM##_tlb);       \
+static inline QEMU_ALWAYS_INLINE
+void sve_st1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
+                   target_ulong base, uint32_t desc, uintptr_t retaddr,
+                   int esize, int msize, zreg_off_fn *off_fn,
+                   sve_ldst1_host_fn *host_fn,
+                   sve_ldst1_tlb_fn *tlb_fn)
+{
+    uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+    /* Remove mtedesc from the normal sve descriptor. */
+    desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT);
+
+    /*
+     * ??? TODO: For the 32-bit offset extractions, base + ofs cannot
+     * offset base entirely over the address space hole to change the
+     * pointer tag, or change the bit55 selector.  So we could here
+     * examine TBI + TCMA like we do for sve_ldN_r_mte().
+     */
+    sve_st1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc,
+              esize, msize, off_fn, host_fn, tlb_fn);
 }
 
-#define DO_ST1_ZPZ_D(MEM, OFS, MSZ) \
-void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,       \
+#define DO_ST1_ZPZ_S(MEM, OFS, MSZ)                                     \
+void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,  \
                                  void *vm, target_ulong base, uint32_t desc) \
-{                                                                            \
-    sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ,             \
-              off_##OFS##_d, sve_st1##MEM##_host, sve_st1##MEM##_tlb);       \
+{                                                                       \
+    sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 4, 1 << MSZ,     \
+              off_##OFS##_s, sve_st1##MEM##_host, sve_st1##MEM##_tlb);  \
+}                                                                       \
+void HELPER(sve_st##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \
+    void *vm, target_ulong base, uint32_t desc)                         \
+{                                                                       \
+    sve_st1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ,    \
+                  off_##OFS##_s, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \
+}
+
+#define DO_ST1_ZPZ_D(MEM, OFS, MSZ)                                     \
+void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg,  \
+                                 void *vm, target_ulong base, uint32_t desc) \
+{                                                                       \
+    sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 8, 1 << MSZ,     \
+              off_##OFS##_d, sve_st1##MEM##_host, sve_st1##MEM##_tlb);  \
+}                                                                       \
+void HELPER(sve_st##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \
+    void *vm, target_ulong base, uint32_t desc)                         \
+{                                                                       \
+    sve_st1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ,    \
+                  off_##OFS##_d, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \
 }
 
 DO_ST1_ZPZ_S(bs, zsu, MO_8)
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 04eda9a126..f318ca265f 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -5261,7 +5261,7 @@ static bool trans_ST_zpri(DisasContext *s, arg_rpri_store *a)
  */
 
 static void do_mem_zpz(DisasContext *s, int zt, int pg, int zm,
-                       int scale, TCGv_i64 scalar, int msz,
+                       int scale, TCGv_i64 scalar, int msz, bool is_write,
                        gen_helper_gvec_mem_scatter *fn)
 {
     unsigned vsz = vec_full_reg_size(s);
@@ -5269,8 +5269,16 @@ static void do_mem_zpz(DisasContext *s, int zt, int pg, int zm,
     TCGv_ptr t_pg = tcg_temp_new_ptr();
     TCGv_ptr t_zt = tcg_temp_new_ptr();
     TCGv_i32 t_desc;
-    int desc;
+    int desc = 0;
 
+    if (s->mte_active[0]) {
+        desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s));
+        desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid);
+        desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma);
+        desc = FIELD_DP32(desc, MTEDESC, WRITE, is_write);
+        desc = FIELD_DP32(desc, MTEDESC, ESIZE, 1 << msz);
+        desc <<= SVE_MTEDESC_SHIFT;
+    }
     desc = simd_desc(vsz, vsz, scale);
     t_desc = tcg_const_i32(desc);
 
@@ -5285,176 +5293,339 @@ static void do_mem_zpz(DisasContext *s, int zt, int pg, int zm,
     tcg_temp_free_i32(t_desc);
 }
 
-/* Indexed by [be][ff][xs][u][msz].  */
-static gen_helper_gvec_mem_scatter * const gather_load_fn32[2][2][2][2][3] = {
-    /* Little-endian */
-    { { { { gen_helper_sve_ldbss_zsu,
-            gen_helper_sve_ldhss_le_zsu,
-            NULL, },
-          { gen_helper_sve_ldbsu_zsu,
-            gen_helper_sve_ldhsu_le_zsu,
-            gen_helper_sve_ldss_le_zsu, } },
-        { { gen_helper_sve_ldbss_zss,
-            gen_helper_sve_ldhss_le_zss,
-            NULL, },
-          { gen_helper_sve_ldbsu_zss,
-            gen_helper_sve_ldhsu_le_zss,
-            gen_helper_sve_ldss_le_zss, } } },
+/* Indexed by [mte][be][ff][xs][u][msz].  */
+static gen_helper_gvec_mem_scatter * const
+gather_load_fn32[2][2][2][2][2][3] = {
+    { /* MTE Inactive */
+        { /* Little-endian */
+            { { { gen_helper_sve_ldbss_zsu,
+                  gen_helper_sve_ldhss_le_zsu,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zsu,
+                  gen_helper_sve_ldhsu_le_zsu,
+                  gen_helper_sve_ldss_le_zsu, } },
+              { { gen_helper_sve_ldbss_zss,
+                  gen_helper_sve_ldhss_le_zss,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zss,
+                  gen_helper_sve_ldhsu_le_zss,
+                  gen_helper_sve_ldss_le_zss, } } },
 
-      /* First-fault */
-      { { { gen_helper_sve_ldffbss_zsu,
-            gen_helper_sve_ldffhss_le_zsu,
-            NULL, },
-          { gen_helper_sve_ldffbsu_zsu,
-            gen_helper_sve_ldffhsu_le_zsu,
-            gen_helper_sve_ldffss_le_zsu, } },
-        { { gen_helper_sve_ldffbss_zss,
-            gen_helper_sve_ldffhss_le_zss,
-            NULL, },
-          { gen_helper_sve_ldffbsu_zss,
-            gen_helper_sve_ldffhsu_le_zss,
-            gen_helper_sve_ldffss_le_zss, } } } },
+            /* First-fault */
+            { { { gen_helper_sve_ldffbss_zsu,
+                  gen_helper_sve_ldffhss_le_zsu,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zsu,
+                  gen_helper_sve_ldffhsu_le_zsu,
+                  gen_helper_sve_ldffss_le_zsu, } },
+              { { gen_helper_sve_ldffbss_zss,
+                  gen_helper_sve_ldffhss_le_zss,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zss,
+                  gen_helper_sve_ldffhsu_le_zss,
+                  gen_helper_sve_ldffss_le_zss, } } } },
 
-    /* Big-endian */
-    { { { { gen_helper_sve_ldbss_zsu,
-            gen_helper_sve_ldhss_be_zsu,
-            NULL, },
-          { gen_helper_sve_ldbsu_zsu,
-            gen_helper_sve_ldhsu_be_zsu,
-            gen_helper_sve_ldss_be_zsu, } },
-        { { gen_helper_sve_ldbss_zss,
-            gen_helper_sve_ldhss_be_zss,
-            NULL, },
-          { gen_helper_sve_ldbsu_zss,
-            gen_helper_sve_ldhsu_be_zss,
-            gen_helper_sve_ldss_be_zss, } } },
+        { /* Big-endian */
+            { { { gen_helper_sve_ldbss_zsu,
+                  gen_helper_sve_ldhss_be_zsu,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zsu,
+                  gen_helper_sve_ldhsu_be_zsu,
+                  gen_helper_sve_ldss_be_zsu, } },
+              { { gen_helper_sve_ldbss_zss,
+                  gen_helper_sve_ldhss_be_zss,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zss,
+                  gen_helper_sve_ldhsu_be_zss,
+                  gen_helper_sve_ldss_be_zss, } } },
 
-      /* First-fault */
-      { { { gen_helper_sve_ldffbss_zsu,
-            gen_helper_sve_ldffhss_be_zsu,
-            NULL, },
-          { gen_helper_sve_ldffbsu_zsu,
-            gen_helper_sve_ldffhsu_be_zsu,
-            gen_helper_sve_ldffss_be_zsu, } },
-        { { gen_helper_sve_ldffbss_zss,
-            gen_helper_sve_ldffhss_be_zss,
-            NULL, },
-          { gen_helper_sve_ldffbsu_zss,
-            gen_helper_sve_ldffhsu_be_zss,
-            gen_helper_sve_ldffss_be_zss, } } } },
+            /* First-fault */
+            { { { gen_helper_sve_ldffbss_zsu,
+                  gen_helper_sve_ldffhss_be_zsu,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zsu,
+                  gen_helper_sve_ldffhsu_be_zsu,
+                  gen_helper_sve_ldffss_be_zsu, } },
+              { { gen_helper_sve_ldffbss_zss,
+                  gen_helper_sve_ldffhss_be_zss,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zss,
+                  gen_helper_sve_ldffhsu_be_zss,
+                  gen_helper_sve_ldffss_be_zss, } } } } },
+    { /* MTE Active */
+        { /* Little-endian */
+            { { { gen_helper_sve_ldbss_zsu_mte,
+                  gen_helper_sve_ldhss_le_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zsu_mte,
+                  gen_helper_sve_ldhsu_le_zsu_mte,
+                  gen_helper_sve_ldss_le_zsu_mte, } },
+              { { gen_helper_sve_ldbss_zss_mte,
+                  gen_helper_sve_ldhss_le_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zss_mte,
+                  gen_helper_sve_ldhsu_le_zss_mte,
+                  gen_helper_sve_ldss_le_zss_mte, } } },
+
+            /* First-fault */
+            { { { gen_helper_sve_ldffbss_zsu_mte,
+                  gen_helper_sve_ldffhss_le_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zsu_mte,
+                  gen_helper_sve_ldffhsu_le_zsu_mte,
+                  gen_helper_sve_ldffss_le_zsu_mte, } },
+              { { gen_helper_sve_ldffbss_zss_mte,
+                  gen_helper_sve_ldffhss_le_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zss_mte,
+                  gen_helper_sve_ldffhsu_le_zss_mte,
+                  gen_helper_sve_ldffss_le_zss_mte, } } } },
+
+        { /* Big-endian */
+            { { { gen_helper_sve_ldbss_zsu_mte,
+                  gen_helper_sve_ldhss_be_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zsu_mte,
+                  gen_helper_sve_ldhsu_be_zsu_mte,
+                  gen_helper_sve_ldss_be_zsu_mte, } },
+              { { gen_helper_sve_ldbss_zss_mte,
+                  gen_helper_sve_ldhss_be_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldbsu_zss_mte,
+                  gen_helper_sve_ldhsu_be_zss_mte,
+                  gen_helper_sve_ldss_be_zss_mte, } } },
+
+            /* First-fault */
+            { { { gen_helper_sve_ldffbss_zsu_mte,
+                  gen_helper_sve_ldffhss_be_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zsu_mte,
+                  gen_helper_sve_ldffhsu_be_zsu_mte,
+                  gen_helper_sve_ldffss_be_zsu_mte, } },
+              { { gen_helper_sve_ldffbss_zss_mte,
+                  gen_helper_sve_ldffhss_be_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbsu_zss_mte,
+                  gen_helper_sve_ldffhsu_be_zss_mte,
+                  gen_helper_sve_ldffss_be_zss_mte, } } } } },
 };
 
 /* Note that we overload xs=2 to indicate 64-bit offset.  */
-static gen_helper_gvec_mem_scatter * const gather_load_fn64[2][2][3][2][4] = {
-    /* Little-endian */
-    { { { { gen_helper_sve_ldbds_zsu,
-            gen_helper_sve_ldhds_le_zsu,
-            gen_helper_sve_ldsds_le_zsu,
-            NULL, },
-          { gen_helper_sve_ldbdu_zsu,
-            gen_helper_sve_ldhdu_le_zsu,
-            gen_helper_sve_ldsdu_le_zsu,
-            gen_helper_sve_lddd_le_zsu, } },
-        { { gen_helper_sve_ldbds_zss,
-            gen_helper_sve_ldhds_le_zss,
-            gen_helper_sve_ldsds_le_zss,
-            NULL, },
-          { gen_helper_sve_ldbdu_zss,
-            gen_helper_sve_ldhdu_le_zss,
-            gen_helper_sve_ldsdu_le_zss,
-            gen_helper_sve_lddd_le_zss, } },
-        { { gen_helper_sve_ldbds_zd,
-            gen_helper_sve_ldhds_le_zd,
-            gen_helper_sve_ldsds_le_zd,
-            NULL, },
-          { gen_helper_sve_ldbdu_zd,
-            gen_helper_sve_ldhdu_le_zd,
-            gen_helper_sve_ldsdu_le_zd,
-            gen_helper_sve_lddd_le_zd, } } },
+static gen_helper_gvec_mem_scatter * const
+gather_load_fn64[2][2][2][3][2][4] = {
+    { /* MTE Inactive */
+        { /* Little-endian */
+            { { { gen_helper_sve_ldbds_zsu,
+                  gen_helper_sve_ldhds_le_zsu,
+                  gen_helper_sve_ldsds_le_zsu,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zsu,
+                  gen_helper_sve_ldhdu_le_zsu,
+                  gen_helper_sve_ldsdu_le_zsu,
+                  gen_helper_sve_lddd_le_zsu, } },
+              { { gen_helper_sve_ldbds_zss,
+                  gen_helper_sve_ldhds_le_zss,
+                  gen_helper_sve_ldsds_le_zss,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zss,
+                  gen_helper_sve_ldhdu_le_zss,
+                  gen_helper_sve_ldsdu_le_zss,
+                  gen_helper_sve_lddd_le_zss, } },
+              { { gen_helper_sve_ldbds_zd,
+                  gen_helper_sve_ldhds_le_zd,
+                  gen_helper_sve_ldsds_le_zd,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zd,
+                  gen_helper_sve_ldhdu_le_zd,
+                  gen_helper_sve_ldsdu_le_zd,
+                  gen_helper_sve_lddd_le_zd, } } },
 
-      /* First-fault */
-      { { { gen_helper_sve_ldffbds_zsu,
-            gen_helper_sve_ldffhds_le_zsu,
-            gen_helper_sve_ldffsds_le_zsu,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zsu,
-            gen_helper_sve_ldffhdu_le_zsu,
-            gen_helper_sve_ldffsdu_le_zsu,
-            gen_helper_sve_ldffdd_le_zsu, } },
-        { { gen_helper_sve_ldffbds_zss,
-            gen_helper_sve_ldffhds_le_zss,
-            gen_helper_sve_ldffsds_le_zss,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zss,
-            gen_helper_sve_ldffhdu_le_zss,
-            gen_helper_sve_ldffsdu_le_zss,
-            gen_helper_sve_ldffdd_le_zss, } },
-        { { gen_helper_sve_ldffbds_zd,
-            gen_helper_sve_ldffhds_le_zd,
-            gen_helper_sve_ldffsds_le_zd,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zd,
-            gen_helper_sve_ldffhdu_le_zd,
-            gen_helper_sve_ldffsdu_le_zd,
-            gen_helper_sve_ldffdd_le_zd, } } } },
+            /* First-fault */
+            { { { gen_helper_sve_ldffbds_zsu,
+                  gen_helper_sve_ldffhds_le_zsu,
+                  gen_helper_sve_ldffsds_le_zsu,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zsu,
+                  gen_helper_sve_ldffhdu_le_zsu,
+                  gen_helper_sve_ldffsdu_le_zsu,
+                  gen_helper_sve_ldffdd_le_zsu, } },
+              { { gen_helper_sve_ldffbds_zss,
+                  gen_helper_sve_ldffhds_le_zss,
+                  gen_helper_sve_ldffsds_le_zss,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zss,
+                  gen_helper_sve_ldffhdu_le_zss,
+                  gen_helper_sve_ldffsdu_le_zss,
+                  gen_helper_sve_ldffdd_le_zss, } },
+              { { gen_helper_sve_ldffbds_zd,
+                  gen_helper_sve_ldffhds_le_zd,
+                  gen_helper_sve_ldffsds_le_zd,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zd,
+                  gen_helper_sve_ldffhdu_le_zd,
+                  gen_helper_sve_ldffsdu_le_zd,
+                  gen_helper_sve_ldffdd_le_zd, } } } },
+        { /* Big-endian */
+            { { { gen_helper_sve_ldbds_zsu,
+                  gen_helper_sve_ldhds_be_zsu,
+                  gen_helper_sve_ldsds_be_zsu,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zsu,
+                  gen_helper_sve_ldhdu_be_zsu,
+                  gen_helper_sve_ldsdu_be_zsu,
+                  gen_helper_sve_lddd_be_zsu, } },
+              { { gen_helper_sve_ldbds_zss,
+                  gen_helper_sve_ldhds_be_zss,
+                  gen_helper_sve_ldsds_be_zss,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zss,
+                  gen_helper_sve_ldhdu_be_zss,
+                  gen_helper_sve_ldsdu_be_zss,
+                  gen_helper_sve_lddd_be_zss, } },
+              { { gen_helper_sve_ldbds_zd,
+                  gen_helper_sve_ldhds_be_zd,
+                  gen_helper_sve_ldsds_be_zd,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zd,
+                  gen_helper_sve_ldhdu_be_zd,
+                  gen_helper_sve_ldsdu_be_zd,
+                  gen_helper_sve_lddd_be_zd, } } },
 
-    /* Big-endian */
-    { { { { gen_helper_sve_ldbds_zsu,
-            gen_helper_sve_ldhds_be_zsu,
-            gen_helper_sve_ldsds_be_zsu,
-            NULL, },
-          { gen_helper_sve_ldbdu_zsu,
-            gen_helper_sve_ldhdu_be_zsu,
-            gen_helper_sve_ldsdu_be_zsu,
-            gen_helper_sve_lddd_be_zsu, } },
-        { { gen_helper_sve_ldbds_zss,
-            gen_helper_sve_ldhds_be_zss,
-            gen_helper_sve_ldsds_be_zss,
-            NULL, },
-          { gen_helper_sve_ldbdu_zss,
-            gen_helper_sve_ldhdu_be_zss,
-            gen_helper_sve_ldsdu_be_zss,
-            gen_helper_sve_lddd_be_zss, } },
-        { { gen_helper_sve_ldbds_zd,
-            gen_helper_sve_ldhds_be_zd,
-            gen_helper_sve_ldsds_be_zd,
-            NULL, },
-          { gen_helper_sve_ldbdu_zd,
-            gen_helper_sve_ldhdu_be_zd,
-            gen_helper_sve_ldsdu_be_zd,
-            gen_helper_sve_lddd_be_zd, } } },
+            /* First-fault */
+            { { { gen_helper_sve_ldffbds_zsu,
+                  gen_helper_sve_ldffhds_be_zsu,
+                  gen_helper_sve_ldffsds_be_zsu,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zsu,
+                  gen_helper_sve_ldffhdu_be_zsu,
+                  gen_helper_sve_ldffsdu_be_zsu,
+                  gen_helper_sve_ldffdd_be_zsu, } },
+              { { gen_helper_sve_ldffbds_zss,
+                  gen_helper_sve_ldffhds_be_zss,
+                  gen_helper_sve_ldffsds_be_zss,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zss,
+                  gen_helper_sve_ldffhdu_be_zss,
+                  gen_helper_sve_ldffsdu_be_zss,
+                  gen_helper_sve_ldffdd_be_zss, } },
+              { { gen_helper_sve_ldffbds_zd,
+                  gen_helper_sve_ldffhds_be_zd,
+                  gen_helper_sve_ldffsds_be_zd,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zd,
+                  gen_helper_sve_ldffhdu_be_zd,
+                  gen_helper_sve_ldffsdu_be_zd,
+                  gen_helper_sve_ldffdd_be_zd, } } } } },
+    { /* MTE Active */
+        { /* Little-endian */
+            { { { gen_helper_sve_ldbds_zsu_mte,
+                  gen_helper_sve_ldhds_le_zsu_mte,
+                  gen_helper_sve_ldsds_le_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zsu_mte,
+                  gen_helper_sve_ldhdu_le_zsu_mte,
+                  gen_helper_sve_ldsdu_le_zsu_mte,
+                  gen_helper_sve_lddd_le_zsu_mte, } },
+              { { gen_helper_sve_ldbds_zss_mte,
+                  gen_helper_sve_ldhds_le_zss_mte,
+                  gen_helper_sve_ldsds_le_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zss_mte,
+                  gen_helper_sve_ldhdu_le_zss_mte,
+                  gen_helper_sve_ldsdu_le_zss_mte,
+                  gen_helper_sve_lddd_le_zss_mte, } },
+              { { gen_helper_sve_ldbds_zd_mte,
+                  gen_helper_sve_ldhds_le_zd_mte,
+                  gen_helper_sve_ldsds_le_zd_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zd_mte,
+                  gen_helper_sve_ldhdu_le_zd_mte,
+                  gen_helper_sve_ldsdu_le_zd_mte,
+                  gen_helper_sve_lddd_le_zd_mte, } } },
 
-      /* First-fault */
-      { { { gen_helper_sve_ldffbds_zsu,
-            gen_helper_sve_ldffhds_be_zsu,
-            gen_helper_sve_ldffsds_be_zsu,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zsu,
-            gen_helper_sve_ldffhdu_be_zsu,
-            gen_helper_sve_ldffsdu_be_zsu,
-            gen_helper_sve_ldffdd_be_zsu, } },
-        { { gen_helper_sve_ldffbds_zss,
-            gen_helper_sve_ldffhds_be_zss,
-            gen_helper_sve_ldffsds_be_zss,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zss,
-            gen_helper_sve_ldffhdu_be_zss,
-            gen_helper_sve_ldffsdu_be_zss,
-            gen_helper_sve_ldffdd_be_zss, } },
-        { { gen_helper_sve_ldffbds_zd,
-            gen_helper_sve_ldffhds_be_zd,
-            gen_helper_sve_ldffsds_be_zd,
-            NULL, },
-          { gen_helper_sve_ldffbdu_zd,
-            gen_helper_sve_ldffhdu_be_zd,
-            gen_helper_sve_ldffsdu_be_zd,
-            gen_helper_sve_ldffdd_be_zd, } } } },
+            /* First-fault */
+            { { { gen_helper_sve_ldffbds_zsu_mte,
+                  gen_helper_sve_ldffhds_le_zsu_mte,
+                  gen_helper_sve_ldffsds_le_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zsu_mte,
+                  gen_helper_sve_ldffhdu_le_zsu_mte,
+                  gen_helper_sve_ldffsdu_le_zsu_mte,
+                  gen_helper_sve_ldffdd_le_zsu_mte, } },
+              { { gen_helper_sve_ldffbds_zss_mte,
+                  gen_helper_sve_ldffhds_le_zss_mte,
+                  gen_helper_sve_ldffsds_le_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zss_mte,
+                  gen_helper_sve_ldffhdu_le_zss_mte,
+                  gen_helper_sve_ldffsdu_le_zss_mte,
+                  gen_helper_sve_ldffdd_le_zss_mte, } },
+              { { gen_helper_sve_ldffbds_zd_mte,
+                  gen_helper_sve_ldffhds_le_zd_mte,
+                  gen_helper_sve_ldffsds_le_zd_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zd_mte,
+                  gen_helper_sve_ldffhdu_le_zd_mte,
+                  gen_helper_sve_ldffsdu_le_zd_mte,
+                  gen_helper_sve_ldffdd_le_zd_mte, } } } },
+        { /* Big-endian */
+            { { { gen_helper_sve_ldbds_zsu_mte,
+                  gen_helper_sve_ldhds_be_zsu_mte,
+                  gen_helper_sve_ldsds_be_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zsu_mte,
+                  gen_helper_sve_ldhdu_be_zsu_mte,
+                  gen_helper_sve_ldsdu_be_zsu_mte,
+                  gen_helper_sve_lddd_be_zsu_mte, } },
+              { { gen_helper_sve_ldbds_zss_mte,
+                  gen_helper_sve_ldhds_be_zss_mte,
+                  gen_helper_sve_ldsds_be_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zss_mte,
+                  gen_helper_sve_ldhdu_be_zss_mte,
+                  gen_helper_sve_ldsdu_be_zss_mte,
+                  gen_helper_sve_lddd_be_zss_mte, } },
+              { { gen_helper_sve_ldbds_zd_mte,
+                  gen_helper_sve_ldhds_be_zd_mte,
+                  gen_helper_sve_ldsds_be_zd_mte,
+                  NULL, },
+                { gen_helper_sve_ldbdu_zd_mte,
+                  gen_helper_sve_ldhdu_be_zd_mte,
+                  gen_helper_sve_ldsdu_be_zd_mte,
+                  gen_helper_sve_lddd_be_zd_mte, } } },
+
+            /* First-fault */
+            { { { gen_helper_sve_ldffbds_zsu_mte,
+                  gen_helper_sve_ldffhds_be_zsu_mte,
+                  gen_helper_sve_ldffsds_be_zsu_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zsu_mte,
+                  gen_helper_sve_ldffhdu_be_zsu_mte,
+                  gen_helper_sve_ldffsdu_be_zsu_mte,
+                  gen_helper_sve_ldffdd_be_zsu_mte, } },
+              { { gen_helper_sve_ldffbds_zss_mte,
+                  gen_helper_sve_ldffhds_be_zss_mte,
+                  gen_helper_sve_ldffsds_be_zss_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zss_mte,
+                  gen_helper_sve_ldffhdu_be_zss_mte,
+                  gen_helper_sve_ldffsdu_be_zss_mte,
+                  gen_helper_sve_ldffdd_be_zss_mte, } },
+              { { gen_helper_sve_ldffbds_zd_mte,
+                  gen_helper_sve_ldffhds_be_zd_mte,
+                  gen_helper_sve_ldffsds_be_zd_mte,
+                  NULL, },
+                { gen_helper_sve_ldffbdu_zd_mte,
+                  gen_helper_sve_ldffhdu_be_zd_mte,
+                  gen_helper_sve_ldffsdu_be_zd_mte,
+                  gen_helper_sve_ldffdd_be_zd_mte, } } } } },
 };
 
 static bool trans_LD1_zprz(DisasContext *s, arg_LD1_zprz *a)
 {
     gen_helper_gvec_mem_scatter *fn = NULL;
-    int be = s->be_data == MO_BE;
+    bool be = s->be_data == MO_BE;
+    bool mte = s->mte_active[0];
 
     if (!sve_access_check(s)) {
         return true;
@@ -5462,23 +5633,24 @@ static bool trans_LD1_zprz(DisasContext *s, arg_LD1_zprz *a)
 
     switch (a->esz) {
     case MO_32:
-        fn = gather_load_fn32[be][a->ff][a->xs][a->u][a->msz];
+        fn = gather_load_fn32[mte][be][a->ff][a->xs][a->u][a->msz];
         break;
     case MO_64:
-        fn = gather_load_fn64[be][a->ff][a->xs][a->u][a->msz];
+        fn = gather_load_fn64[mte][be][a->ff][a->xs][a->u][a->msz];
         break;
     }
     assert(fn != NULL);
 
     do_mem_zpz(s, a->rd, a->pg, a->rm, a->scale * a->msz,
-               cpu_reg_sp(s, a->rn), a->msz, fn);
+               cpu_reg_sp(s, a->rn), a->msz, false, fn);
     return true;
 }
 
 static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a)
 {
     gen_helper_gvec_mem_scatter *fn = NULL;
-    int be = s->be_data == MO_BE;
+    bool be = s->be_data == MO_BE;
+    bool mte = s->mte_active[0];
     TCGv_i64 imm;
 
     if (a->esz < a->msz || (a->esz == a->msz && !a->u)) {
@@ -5490,10 +5662,10 @@ static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a)
 
     switch (a->esz) {
     case MO_32:
-        fn = gather_load_fn32[be][a->ff][0][a->u][a->msz];
+        fn = gather_load_fn32[mte][be][a->ff][0][a->u][a->msz];
         break;
     case MO_64:
-        fn = gather_load_fn64[be][a->ff][2][a->u][a->msz];
+        fn = gather_load_fn64[mte][be][a->ff][2][a->u][a->msz];
         break;
     }
     assert(fn != NULL);
@@ -5502,63 +5674,108 @@ static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a)
      * by loading the immediate into the scalar parameter.
      */
     imm = tcg_const_i64(a->imm << a->msz);
-    do_mem_zpz(s, a->rd, a->pg, a->rn, 0, imm, a->msz, fn);
+    do_mem_zpz(s, a->rd, a->pg, a->rn, 0, imm, a->msz, false, fn);
     tcg_temp_free_i64(imm);
     return true;
 }
 
-/* Indexed by [be][xs][msz].  */
-static gen_helper_gvec_mem_scatter * const scatter_store_fn32[2][2][3] = {
-    /* Little-endian */
-    { { gen_helper_sve_stbs_zsu,
-        gen_helper_sve_sths_le_zsu,
-        gen_helper_sve_stss_le_zsu, },
-      { gen_helper_sve_stbs_zss,
-        gen_helper_sve_sths_le_zss,
-        gen_helper_sve_stss_le_zss, } },
-    /* Big-endian */
-    { { gen_helper_sve_stbs_zsu,
-        gen_helper_sve_sths_be_zsu,
-        gen_helper_sve_stss_be_zsu, },
-      { gen_helper_sve_stbs_zss,
-        gen_helper_sve_sths_be_zss,
-        gen_helper_sve_stss_be_zss, } },
+/* Indexed by [mte][be][xs][msz].  */
+static gen_helper_gvec_mem_scatter * const scatter_store_fn32[2][2][2][3] = {
+    { /* MTE Inactive */
+        { /* Little-endian */
+            { gen_helper_sve_stbs_zsu,
+              gen_helper_sve_sths_le_zsu,
+              gen_helper_sve_stss_le_zsu, },
+            { gen_helper_sve_stbs_zss,
+              gen_helper_sve_sths_le_zss,
+              gen_helper_sve_stss_le_zss, } },
+        { /* Big-endian */
+            { gen_helper_sve_stbs_zsu,
+              gen_helper_sve_sths_be_zsu,
+              gen_helper_sve_stss_be_zsu, },
+            { gen_helper_sve_stbs_zss,
+              gen_helper_sve_sths_be_zss,
+              gen_helper_sve_stss_be_zss, } } },
+    { /* MTE Active */
+        { /* Little-endian */
+            { gen_helper_sve_stbs_zsu_mte,
+              gen_helper_sve_sths_le_zsu_mte,
+              gen_helper_sve_stss_le_zsu_mte, },
+            { gen_helper_sve_stbs_zss_mte,
+              gen_helper_sve_sths_le_zss_mte,
+              gen_helper_sve_stss_le_zss_mte, } },
+        { /* Big-endian */
+            { gen_helper_sve_stbs_zsu_mte,
+              gen_helper_sve_sths_be_zsu_mte,
+              gen_helper_sve_stss_be_zsu_mte, },
+            { gen_helper_sve_stbs_zss_mte,
+              gen_helper_sve_sths_be_zss_mte,
+              gen_helper_sve_stss_be_zss_mte, } } },
 };
 
 /* Note that we overload xs=2 to indicate 64-bit offset.  */
-static gen_helper_gvec_mem_scatter * const scatter_store_fn64[2][3][4] = {
-    /* Little-endian */
-    { { gen_helper_sve_stbd_zsu,
-        gen_helper_sve_sthd_le_zsu,
-        gen_helper_sve_stsd_le_zsu,
-        gen_helper_sve_stdd_le_zsu, },
-      { gen_helper_sve_stbd_zss,
-        gen_helper_sve_sthd_le_zss,
-        gen_helper_sve_stsd_le_zss,
-        gen_helper_sve_stdd_le_zss, },
-      { gen_helper_sve_stbd_zd,
-        gen_helper_sve_sthd_le_zd,
-        gen_helper_sve_stsd_le_zd,
-        gen_helper_sve_stdd_le_zd, } },
-    /* Big-endian */
-    { { gen_helper_sve_stbd_zsu,
-        gen_helper_sve_sthd_be_zsu,
-        gen_helper_sve_stsd_be_zsu,
-        gen_helper_sve_stdd_be_zsu, },
-      { gen_helper_sve_stbd_zss,
-        gen_helper_sve_sthd_be_zss,
-        gen_helper_sve_stsd_be_zss,
-        gen_helper_sve_stdd_be_zss, },
-      { gen_helper_sve_stbd_zd,
-        gen_helper_sve_sthd_be_zd,
-        gen_helper_sve_stsd_be_zd,
-        gen_helper_sve_stdd_be_zd, } },
+static gen_helper_gvec_mem_scatter * const scatter_store_fn64[2][2][3][4] = {
+    { /* MTE Inactive */
+         { /* Little-endian */
+             { gen_helper_sve_stbd_zsu,
+               gen_helper_sve_sthd_le_zsu,
+               gen_helper_sve_stsd_le_zsu,
+               gen_helper_sve_stdd_le_zsu, },
+             { gen_helper_sve_stbd_zss,
+               gen_helper_sve_sthd_le_zss,
+               gen_helper_sve_stsd_le_zss,
+               gen_helper_sve_stdd_le_zss, },
+             { gen_helper_sve_stbd_zd,
+               gen_helper_sve_sthd_le_zd,
+               gen_helper_sve_stsd_le_zd,
+               gen_helper_sve_stdd_le_zd, } },
+         { /* Big-endian */
+             { gen_helper_sve_stbd_zsu,
+               gen_helper_sve_sthd_be_zsu,
+               gen_helper_sve_stsd_be_zsu,
+               gen_helper_sve_stdd_be_zsu, },
+             { gen_helper_sve_stbd_zss,
+               gen_helper_sve_sthd_be_zss,
+               gen_helper_sve_stsd_be_zss,
+               gen_helper_sve_stdd_be_zss, },
+             { gen_helper_sve_stbd_zd,
+               gen_helper_sve_sthd_be_zd,
+               gen_helper_sve_stsd_be_zd,
+               gen_helper_sve_stdd_be_zd, } } },
+    { /* MTE Inactive */
+         { /* Little-endian */
+             { gen_helper_sve_stbd_zsu_mte,
+               gen_helper_sve_sthd_le_zsu_mte,
+               gen_helper_sve_stsd_le_zsu_mte,
+               gen_helper_sve_stdd_le_zsu_mte, },
+             { gen_helper_sve_stbd_zss_mte,
+               gen_helper_sve_sthd_le_zss_mte,
+               gen_helper_sve_stsd_le_zss_mte,
+               gen_helper_sve_stdd_le_zss_mte, },
+             { gen_helper_sve_stbd_zd_mte,
+               gen_helper_sve_sthd_le_zd_mte,
+               gen_helper_sve_stsd_le_zd_mte,
+               gen_helper_sve_stdd_le_zd_mte, } },
+         { /* Big-endian */
+             { gen_helper_sve_stbd_zsu_mte,
+               gen_helper_sve_sthd_be_zsu_mte,
+               gen_helper_sve_stsd_be_zsu_mte,
+               gen_helper_sve_stdd_be_zsu_mte, },
+             { gen_helper_sve_stbd_zss_mte,
+               gen_helper_sve_sthd_be_zss_mte,
+               gen_helper_sve_stsd_be_zss_mte,
+               gen_helper_sve_stdd_be_zss_mte, },
+             { gen_helper_sve_stbd_zd_mte,
+               gen_helper_sve_sthd_be_zd_mte,
+               gen_helper_sve_stsd_be_zd_mte,
+               gen_helper_sve_stdd_be_zd_mte, } } },
 };
 
 static bool trans_ST1_zprz(DisasContext *s, arg_ST1_zprz *a)
 {
     gen_helper_gvec_mem_scatter *fn;
-    int be = s->be_data == MO_BE;
+    bool be = s->be_data == MO_BE;
+    bool mte = s->mte_active[0];
 
     if (a->esz < a->msz || (a->msz == 0 && a->scale)) {
         return false;
@@ -5568,23 +5785,24 @@ static bool trans_ST1_zprz(DisasContext *s, arg_ST1_zprz *a)
     }
     switch (a->esz) {
     case MO_32:
-        fn = scatter_store_fn32[be][a->xs][a->msz];
+        fn = scatter_store_fn32[mte][be][a->xs][a->msz];
         break;
     case MO_64:
-        fn = scatter_store_fn64[be][a->xs][a->msz];
+        fn = scatter_store_fn64[mte][be][a->xs][a->msz];
         break;
     default:
         g_assert_not_reached();
     }
     do_mem_zpz(s, a->rd, a->pg, a->rm, a->scale * a->msz,
-               cpu_reg_sp(s, a->rn), a->msz, fn);
+               cpu_reg_sp(s, a->rn), a->msz, true, fn);
     return true;
 }
 
 static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a)
 {
     gen_helper_gvec_mem_scatter *fn = NULL;
-    int be = s->be_data == MO_BE;
+    bool be = s->be_data == MO_BE;
+    bool mte = s->mte_active[0];
     TCGv_i64 imm;
 
     if (a->esz < a->msz) {
@@ -5596,10 +5814,10 @@ static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a)
 
     switch (a->esz) {
     case MO_32:
-        fn = scatter_store_fn32[be][0][a->msz];
+        fn = scatter_store_fn32[mte][be][0][a->msz];
         break;
     case MO_64:
-        fn = scatter_store_fn64[be][2][a->msz];
+        fn = scatter_store_fn64[mte][be][2][a->msz];
         break;
     }
     assert(fn != NULL);
@@ -5608,7 +5826,7 @@ static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a)
      * by loading the immediate into the scalar parameter.
      */
     imm = tcg_const_i64(a->imm << a->msz);
-    do_mem_zpz(s, a->rd, a->pg, a->rn, 0, imm, a->msz, fn);
+    do_mem_zpz(s, a->rd, a->pg, a->rn, 0, imm, a->msz, true, fn);
     tcg_temp_free_i64(imm);
     return true;
 }
-- 
2.25.1

[PATCH v8 38/45] target/arm: Complete TBI clearing for user-only for SVE

[Richard Henderson]

There are a number of paths by which the TBI is still intact
for user-only in the SVE helpers.

Because we currently always set TBI for user-only, we do not
need to pass down the actual TBI setting from above, and we
can remove the top byte in the inner-most primitives, so that
none are forgotten.  Moreover, this keeps the "dirty" pointer
around at the higher levels, where we need it for any MTE checking.

Since the normal case, especially for user-only, goes through
RAM, this clearing merely adds two insns per page lookup, which
will be completely in the noise.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/sve_helper.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index ad974c2cc5..382fa82bc8 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -3966,14 +3966,16 @@ static void sve_##NAME##_host(void *vd, intptr_t reg_off, void *host)  \
 static void sve_##NAME##_tlb(CPUARMState *env, void *vd, intptr_t reg_off,  \
                              target_ulong addr, uintptr_t ra)               \
 {                                                                           \
-    *(TYPEE *)(vd + H(reg_off)) = (TYPEM)TLB(env, addr, ra);                \
+    *(TYPEE *)(vd + H(reg_off)) =                                           \
+        (TYPEM)TLB(env, useronly_clean_ptr(addr), ra);                      \
 }
 
 #define DO_ST_TLB(NAME, H, TYPEE, TYPEM, TLB) \
 static void sve_##NAME##_tlb(CPUARMState *env, void *vd, intptr_t reg_off,  \
                              target_ulong addr, uintptr_t ra)               \
 {                                                                           \
-    TLB(env, addr, (TYPEM)*(TYPEE *)(vd + H(reg_off)), ra);                 \
+    TLB(env, useronly_clean_ptr(addr),                                      \
+        (TYPEM)*(TYPEE *)(vd + H(reg_off)), ra);                            \
 }
 
 #define DO_LD_PRIM_1(NAME, H, TE, TM)                   \
@@ -4091,6 +4093,19 @@ static bool sve_probe_page(SVEHostPage *info, bool nofault,
     int flags;
 
     addr += mem_off;
+
+    /*
+     * User-only currently always issues with TBI.  See the comment
+     * above useronly_clean_ptr.  Usually we clean this top byte away
+     * during translation, but we can't do that for e.g. vector + imm
+     * addressing modes.
+     *
+     * We currently always enable TBI for user-only, and do not provide
+     * a way to turn it off.  So clean the pointer unconditionally here,
+     * rather than look it up here, or pass it down from above.
+     */
+    addr = useronly_clean_ptr(addr);
+
     flags = probe_access_flags(env, addr, access_type, mmu_idx, nofault,
                                &info->host, retaddr);
     info->flags = flags;
-- 
2.25.1

[PATCH v8 39/45] target/arm: Implement data cache set allocation tags

[Richard Henderson]

This is DC GVA and DC GZVA, and the tag check for DC ZVA.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Use allocation_tag_mem + memset.
v3: Require pre-cleaned addresses.
v6: Move DCZ block size assert to cpu realize.
    Perform a tag check for DC ZVA.
---
 target/arm/cpu.h           |  4 +++-
 target/arm/helper.c        | 16 ++++++++++++++++
 target/arm/translate-a64.c | 39 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 58 insertions(+), 1 deletion(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 76f2287314..10c4d031b1 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -2360,7 +2360,9 @@ static inline uint64_t cpreg_to_kvm_id(uint32_t cpregid)
 #define ARM_CP_NZCV              (ARM_CP_SPECIAL | 0x0300)
 #define ARM_CP_CURRENTEL         (ARM_CP_SPECIAL | 0x0400)
 #define ARM_CP_DC_ZVA            (ARM_CP_SPECIAL | 0x0500)
-#define ARM_LAST_SPECIAL         ARM_CP_DC_ZVA
+#define ARM_CP_DC_GVA            (ARM_CP_SPECIAL | 0x0600)
+#define ARM_CP_DC_GZVA           (ARM_CP_SPECIAL | 0x0700)
+#define ARM_LAST_SPECIAL         ARM_CP_DC_GZVA
 #define ARM_CP_FPU               0x1000
 #define ARM_CP_SVE               0x2000
 #define ARM_CP_NO_GDB            0x4000
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 44a3f9fb48..23cf44fcf4 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6998,6 +6998,22 @@ static const ARMCPRegInfo mte_el0_cacheop_reginfo[] = {
       .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 14, .opc2 = 5,
       .type = ARM_CP_NOP, .access = PL0_W,
       .accessfn = aa64_cacheop_poc_access },
+    { .name = "DC_GVA", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 4, .opc2 = 3,
+      .access = PL0_W, .type = ARM_CP_DC_GVA,
+#ifndef CONFIG_USER_ONLY
+      /* Avoid overhead of an access check that always passes in user-mode */
+      .accessfn = aa64_zva_access,
+#endif
+    },
+    { .name = "DC_GZVA", .state = ARM_CP_STATE_AA64,
+      .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 4, .opc2 = 4,
+      .access = PL0_W, .type = ARM_CP_DC_GZVA,
+#ifndef CONFIG_USER_ONLY
+      /* Avoid overhead of an access check that always passes in user-mode */
+      .accessfn = aa64_zva_access,
+#endif
+    },
     REGINFO_SENTINEL
 };
 
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index f35b122ded..1041ec29d4 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1874,6 +1874,45 @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread,
         }
         gen_helper_dc_zva(cpu_env, tcg_rt);
         return;
+    case ARM_CP_DC_GVA:
+        {
+            TCGv_i64 clean_addr, tag;
+
+            /*
+             * DC_GVA, like DC_ZVA, requires that we supply the original
+             * pointer for an invalid page.  Probe that address first.
+             */
+            tcg_rt = cpu_reg(s, rt);
+            clean_addr = clean_data_tbi(s, tcg_rt);
+            gen_probe_access(s, clean_addr, MMU_DATA_STORE, MO_8);
+
+            if (s->ata) {
+                /* Extract the tag from the register to match STZGM.  */
+                tag = tcg_temp_new_i64();
+                tcg_gen_shri_i64(tag, tcg_rt, 56);
+                gen_helper_stzgm_tags(cpu_env, clean_addr, tag);
+                tcg_temp_free_i64(tag);
+            }
+        }
+        return;
+    case ARM_CP_DC_GZVA:
+        {
+            TCGv_i64 clean_addr, tag;
+
+            /* For DC_GZVA, we can rely on DC_ZVA for the proper fault. */
+            tcg_rt = cpu_reg(s, rt);
+            clean_addr = clean_data_tbi(s, tcg_rt);
+            gen_helper_dc_zva(cpu_env, clean_addr);
+
+            if (s->ata) {
+                /* Extract the tag from the register to match STZGM.  */
+                tag = tcg_temp_new_i64();
+                tcg_gen_shri_i64(tag, tcg_rt, 56);
+                gen_helper_stzgm_tags(cpu_env, clean_addr, tag);
+                tcg_temp_free_i64(tag);
+            }
+        }
+        return;
     default:
         break;
     }
-- 
2.25.1

[PATCH v8 40/45] target/arm: Set PSTATE.TCO on exception entry

[Richard Henderson]

D1.10 specifies that exception handlers begin with tag checks overridden.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v2: Only set if MTE feature present.
---
 target/arm/helper.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 23cf44fcf4..d220612a20 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -9704,6 +9704,9 @@ static void arm_cpu_do_interrupt_aarch64(CPUState *cs)
             break;
         }
     }
+    if (cpu_isar_feature(aa64_mte, cpu)) {
+        new_mode |= PSTATE_TCO;
+    }
 
     pstate_write(env, PSTATE_DAIF | new_mode);
     env->aarch64 = 1;
-- 
2.25.1

[PATCH v8 41/45] target/arm: Always pass cacheattr to get_phys_addr

[Richard Henderson]

We need to check the memattr of a page in order to determine
whether it is Tagged for MTE.  Between Stage1 and Stage2,
this becomes simpler if we always collect this data, instead
of occasionally being presented with NULL.

Use the nonnull attribute to allow the compiler to check that
all pointer arguments are non-null.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/internals.h  |  3 ++-
 target/arm/helper.c     | 60 ++++++++++++++++++++---------------------
 target/arm/m_helper.c   | 11 +++++---
 target/arm/tlb_helper.c |  4 ++-
 4 files changed, 42 insertions(+), 36 deletions(-)

diff --git a/target/arm/internals.h b/target/arm/internals.h
index 3306c4f829..ae99725d2b 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1294,7 +1294,8 @@ bool get_phys_addr(CPUARMState *env, target_ulong address,
                    MMUAccessType access_type, ARMMMUIdx mmu_idx,
                    hwaddr *phys_ptr, MemTxAttrs *attrs, int *prot,
                    target_ulong *page_size,
-                   ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs);
+                   ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs)
+    __attribute__((nonnull));
 
 void arm_log_exception(int idx);
 
diff --git a/target/arm/helper.c b/target/arm/helper.c
index d220612a20..2072db2f92 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -44,7 +44,8 @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address,
                                bool s1_is_el0,
                                hwaddr *phys_ptr, MemTxAttrs *txattrs, int *prot,
                                target_ulong *page_size_ptr,
-                               ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs);
+                               ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs)
+    __attribute__((nonnull));
 #endif
 
 static void switch_mode(CPUARMState *env, int mode);
@@ -11101,19 +11102,16 @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address,
         arm_tlb_bti_gp(txattrs) = true;
     }
 
-    if (cacheattrs != NULL) {
-        if (mmu_idx == ARMMMUIdx_Stage2) {
-            cacheattrs->attrs = convert_stage2_attrs(env,
-                                                     extract32(attrs, 0, 4));
-        } else {
-            /* Index into MAIR registers for cache attributes */
-            uint8_t attrindx = extract32(attrs, 0, 3);
-            uint64_t mair = env->cp15.mair_el[regime_el(env, mmu_idx)];
-            assert(attrindx <= 7);
-            cacheattrs->attrs = extract64(mair, attrindx * 8, 8);
-        }
-        cacheattrs->shareability = extract32(attrs, 6, 2);
+    if (mmu_idx == ARMMMUIdx_Stage2) {
+        cacheattrs->attrs = convert_stage2_attrs(env, extract32(attrs, 0, 4));
+    } else {
+        /* Index into MAIR registers for cache attributes */
+        uint8_t attrindx = extract32(attrs, 0, 3);
+        uint64_t mair = env->cp15.mair_el[regime_el(env, mmu_idx)];
+        assert(attrindx <= 7);
+        cacheattrs->attrs = extract64(mair, attrindx * 8, 8);
     }
+    cacheattrs->shareability = extract32(attrs, 6, 2);
 
     *phys_ptr = descaddr;
     *page_size_ptr = page_size;
@@ -11948,28 +11946,29 @@ bool get_phys_addr(CPUARMState *env, target_ulong address,
             ret = get_phys_addr_lpae(env, ipa, access_type, ARMMMUIdx_Stage2,
                                      mmu_idx == ARMMMUIdx_E10_0,
                                      phys_ptr, attrs, &s2_prot,
-                                     page_size, fi,
-                                     cacheattrs != NULL ? &cacheattrs2 : NULL);
+                                     page_size, fi, &cacheattrs2);
             fi->s2addr = ipa;
             /* Combine the S1 and S2 perms.  */
             *prot &= s2_prot;
 
-            /* Combine the S1 and S2 cache attributes, if needed */
-            if (!ret && cacheattrs != NULL) {
-                if (env->cp15.hcr_el2 & HCR_DC) {
-                    /*
-                     * HCR.DC forces the first stage attributes to
-                     *  Normal Non-Shareable,
-                     *  Inner Write-Back Read-Allocate Write-Allocate,
-                     *  Outer Write-Back Read-Allocate Write-Allocate.
-                     */
-                    cacheattrs->attrs = 0xff;
-                    cacheattrs->shareability = 0;
-                }
-                *cacheattrs = combine_cacheattrs(*cacheattrs, cacheattrs2);
+            /* If S2 fails, return early.  */
+            if (ret) {
+                return ret;
             }
 
-            return ret;
+            /* Combine the S1 and S2 cache attributes. */
+            if (env->cp15.hcr_el2 & HCR_DC) {
+                /*
+                 * HCR.DC forces the first stage attributes to
+                 *  Normal Non-Shareable,
+                 *  Inner Write-Back Read-Allocate Write-Allocate,
+                 *  Outer Write-Back Read-Allocate Write-Allocate.
+                 */
+                cacheattrs->attrs = 0xff;
+                cacheattrs->shareability = 0;
+            }
+            *cacheattrs = combine_cacheattrs(*cacheattrs, cacheattrs2);
+            return 0;
         } else {
             /*
              * For non-EL2 CPUs a stage1+stage2 translation is just stage 1.
@@ -12094,11 +12093,12 @@ hwaddr arm_cpu_get_phys_page_attrs_debug(CPUState *cs, vaddr addr,
     bool ret;
     ARMMMUFaultInfo fi = {};
     ARMMMUIdx mmu_idx = arm_mmu_idx(env);
+    ARMCacheAttrs cacheattrs = {};
 
     *attrs = (MemTxAttrs) {};
 
     ret = get_phys_addr(env, addr, 0, mmu_idx, &phys_addr,
-                        attrs, &prot, &page_size, &fi, NULL);
+                        attrs, &prot, &page_size, &fi, &cacheattrs);
 
     if (ret) {
         return -1;
diff --git a/target/arm/m_helper.c b/target/arm/m_helper.c
index 5e8a795d20..036454234c 100644
--- a/target/arm/m_helper.c
+++ b/target/arm/m_helper.c
@@ -187,12 +187,13 @@ static bool v7m_stack_write(ARMCPU *cpu, uint32_t addr, uint32_t value,
     hwaddr physaddr;
     int prot;
     ARMMMUFaultInfo fi = {};
+    ARMCacheAttrs cacheattrs = {};
     bool secure = mmu_idx & ARM_MMU_IDX_M_S;
     int exc;
     bool exc_secure;
 
     if (get_phys_addr(env, addr, MMU_DATA_STORE, mmu_idx, &physaddr,
-                      &attrs, &prot, &page_size, &fi, NULL)) {
+                      &attrs, &prot, &page_size, &fi, &cacheattrs)) {
         /* MPU/SAU lookup failed */
         if (fi.type == ARMFault_QEMU_SFault) {
             if (mode == STACK_LAZYFP) {
@@ -279,13 +280,14 @@ static bool v7m_stack_read(ARMCPU *cpu, uint32_t *dest, uint32_t addr,
     hwaddr physaddr;
     int prot;
     ARMMMUFaultInfo fi = {};
+    ARMCacheAttrs cacheattrs = {};
     bool secure = mmu_idx & ARM_MMU_IDX_M_S;
     int exc;
     bool exc_secure;
     uint32_t value;
 
     if (get_phys_addr(env, addr, MMU_DATA_LOAD, mmu_idx, &physaddr,
-                      &attrs, &prot, &page_size, &fi, NULL)) {
+                      &attrs, &prot, &page_size, &fi, &cacheattrs)) {
         /* MPU/SAU lookup failed */
         if (fi.type == ARMFault_QEMU_SFault) {
             qemu_log_mask(CPU_LOG_INT,
@@ -1928,6 +1930,7 @@ static bool v7m_read_half_insn(ARMCPU *cpu, ARMMMUIdx mmu_idx,
     V8M_SAttributes sattrs = {};
     MemTxAttrs attrs = {};
     ARMMMUFaultInfo fi = {};
+    ARMCacheAttrs cacheattrs = {};
     MemTxResult txres;
     target_ulong page_size;
     hwaddr physaddr;
@@ -1945,8 +1948,8 @@ static bool v7m_read_half_insn(ARMCPU *cpu, ARMMMUIdx mmu_idx,
                       "...really SecureFault with SFSR.INVEP\n");
         return false;
     }
-    if (get_phys_addr(env, addr, MMU_INST_FETCH, mmu_idx,
-                      &physaddr, &attrs, &prot, &page_size, &fi, NULL)) {
+    if (get_phys_addr(env, addr, MMU_INST_FETCH, mmu_idx, &physaddr,
+                      &attrs, &prot, &page_size, &fi, &cacheattrs)) {
         /* the MPU lookup failed */
         env->v7m.cfsr[env->v7m.secure] |= R_V7M_CFSR_IACCVIOL_MASK;
         armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_MEM, env->v7m.secure);
diff --git a/target/arm/tlb_helper.c b/target/arm/tlb_helper.c
index 522a6442a4..89d90465a3 100644
--- a/target/arm/tlb_helper.c
+++ b/target/arm/tlb_helper.c
@@ -166,6 +166,7 @@ bool arm_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
     int prot, ret;
     MemTxAttrs attrs = {};
     ARMMMUFaultInfo fi = {};
+    ARMCacheAttrs cacheattrs = {};
 
     /*
      * Walk the page table and (if the mapping exists) add the page
@@ -175,7 +176,8 @@ bool arm_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
      */
     ret = get_phys_addr(&cpu->env, address, access_type,
                         core_to_arm_mmu_idx(&cpu->env, mmu_idx),
-                        &phys_addr, &attrs, &prot, &page_size, &fi, NULL);
+                        &phys_addr, &attrs, &prot, &page_size,
+                        &fi, &cacheattrs);
     if (likely(!ret)) {
         /*
          * Map a single [sub]page. Regions smaller than our declared
-- 
2.25.1

[PATCH v8 42/45] target/arm: Cache the Tagged bit for a page in MemTxAttrs

[Richard Henderson]

This "bit" is a particular value of the page's MemAttr.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v6: Test HCR_EL2.{DC,DCT}; test Stage2 attributes.
v8: Fill in cacheattrs for S1 disabled; retain tagging when
    combining attributes; set mte_tagging in arm_cpu_tlb_fill.
---
 target/arm/helper.c     | 48 ++++++++++++++++++++++++++++++++++++++---
 target/arm/tlb_helper.c |  5 +++++
 2 files changed, 50 insertions(+), 3 deletions(-)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 2072db2f92..e614d7c4bd 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -11834,9 +11834,19 @@ static uint8_t combine_cacheattr_nibble(uint8_t s1, uint8_t s2)
  */
 static ARMCacheAttrs combine_cacheattrs(ARMCacheAttrs s1, ARMCacheAttrs s2)
 {
-    uint8_t s1lo = extract32(s1.attrs, 0, 4), s2lo = extract32(s2.attrs, 0, 4);
-    uint8_t s1hi = extract32(s1.attrs, 4, 4), s2hi = extract32(s2.attrs, 4, 4);
+    uint8_t s1lo, s2lo, s1hi, s2hi;
     ARMCacheAttrs ret;
+    bool tagged = false;
+
+    if (s1.attrs == 0xf0) {
+        tagged = true;
+        s1.attrs = 0xff;
+    }
+
+    s1lo = extract32(s1.attrs, 0, 4);
+    s2lo = extract32(s2.attrs, 0, 4);
+    s1hi = extract32(s1.attrs, 4, 4);
+    s2hi = extract32(s2.attrs, 4, 4);
 
     /* Combine shareability attributes (table D4-43) */
     if (s1.shareability == 2 || s2.shareability == 2) {
@@ -11884,6 +11894,11 @@ static ARMCacheAttrs combine_cacheattrs(ARMCacheAttrs s1, ARMCacheAttrs s2)
         }
     }
 
+    /* TODO: CombineS1S2Desc does not consider transient, only WB, RWA. */
+    if (tagged && ret.attrs == 0xff) {
+        ret.attrs = 0xf0;
+    }
+
     return ret;
 }
 
@@ -11963,8 +11978,11 @@ bool get_phys_addr(CPUARMState *env, target_ulong address,
                  *  Normal Non-Shareable,
                  *  Inner Write-Back Read-Allocate Write-Allocate,
                  *  Outer Write-Back Read-Allocate Write-Allocate.
+                 * Do not overwrite Tagged within attrs.
                  */
-                cacheattrs->attrs = 0xff;
+                if (cacheattrs->attrs != 0xf0) {
+                    cacheattrs->attrs = 0xff;
+                }
                 cacheattrs->shareability = 0;
             }
             *cacheattrs = combine_cacheattrs(*cacheattrs, cacheattrs2);
@@ -12029,6 +12047,9 @@ bool get_phys_addr(CPUARMState *env, target_ulong address,
     /* Definitely a real MMU, not an MPU */
 
     if (regime_translation_disabled(env, mmu_idx)) {
+        uint64_t hcr;
+        uint8_t memattr;
+
         /*
          * MMU disabled.  S1 addresses within aa64 translation regimes are
          * still checked for bounds -- see AArch64.TranslateAddressS1Off.
@@ -12066,6 +12087,27 @@ bool get_phys_addr(CPUARMState *env, target_ulong address,
         *phys_ptr = address;
         *prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;
         *page_size = TARGET_PAGE_SIZE;
+
+        /* Fill in cacheattr a-la AArch64.TranslateAddressS1Off. */
+        hcr = arm_hcr_el2_eff(env);
+        cacheattrs->shareability = 0;
+        if (hcr & HCR_DC) {
+            if (hcr & HCR_DCT) {
+                memattr = 0xf0;  /* Tagged, Normal, WB, RWA */
+            } else {
+                memattr = 0xff;  /* Normal, WB, RWA */
+            }
+        } else if (access_type == MMU_INST_FETCH) {
+            if (regime_sctlr(env, mmu_idx) & SCTLR_I) {
+                memattr = 0xee;  /* Normal, WT, RA, NT */
+            } else {
+                memattr = 0x44;  /* Normal, NC, No */
+            }
+            cacheattrs->shareability = 2; /* outersharable */
+        } else {
+            memattr = 0x00;      /* Device, nGnRnE */
+        }
+        cacheattrs->attrs = memattr;
         return 0;
     }
 
diff --git a/target/arm/tlb_helper.c b/target/arm/tlb_helper.c
index 89d90465a3..b35dc8a011 100644
--- a/target/arm/tlb_helper.c
+++ b/target/arm/tlb_helper.c
@@ -188,6 +188,11 @@ bool arm_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
             phys_addr &= TARGET_PAGE_MASK;
             address &= TARGET_PAGE_MASK;
         }
+        /* Notice and record tagged memory. */
+        if (cpu_isar_feature(aa64_mte, cpu) && cacheattrs.attrs == 0xf0) {
+            arm_tlb_mte_tagged(&attrs) = true;
+        }
+
         tlb_set_page_with_attrs(cs, address, phys_addr, attrs,
                                 prot, mmu_idx, page_size);
         return true;
-- 
2.25.1

[PATCH v8 43/45] target/arm: Create tagged ram when MTE is enabled

[Richard Henderson]

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v5: Assign cs->num_ases to the final value first.
    Downgrade to ID_AA64PFR1.MTE=1 if tag memory is not available.
v6: Add secure tag memory for EL3.
v8: Add arm,armv8.5-memtag.
---
 include/hw/arm/boot.h |  3 +++
 target/arm/cpu.h      |  6 +++++
 hw/arm/boot.c         | 12 ++++++---
 hw/arm/virt.c         | 57 +++++++++++++++++++++++++++++++++++++++++--
 target/arm/cpu.c      | 51 +++++++++++++++++++++++++++++++++++---
 5 files changed, 121 insertions(+), 8 deletions(-)

diff --git a/include/hw/arm/boot.h b/include/hw/arm/boot.h
index ce2b48b88b..605446afe7 100644
--- a/include/hw/arm/boot.h
+++ b/include/hw/arm/boot.h
@@ -116,6 +116,9 @@ struct arm_boot_info {
      */
     bool secure_board_setup;
 
+    /* If set, all ram objects have tag memory objects. */
+    bool tag_memory;
+
     arm_endianness endianness;
 };
 
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 10c4d031b1..206e617c25 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -792,6 +792,10 @@ struct ARMCPU {
     /* MemoryRegion to use for secure physical accesses */
     MemoryRegion *secure_memory;
 
+    /* MemoryRegion to use for allocation tag accesses */
+    MemoryRegion *tag_memory;
+    MemoryRegion *secure_tag_memory;
+
     /* For v8M, pointer to the IDAU interface provided by board/SoC */
     Object *idau;
 
@@ -2985,6 +2989,8 @@ typedef enum ARMMMUIdxBit {
 typedef enum ARMASIdx {
     ARMASIdx_NS = 0,
     ARMASIdx_S = 1,
+    ARMASIdx_TagNS = 2,
+    ARMASIdx_TagS = 3,
 } ARMASIdx;
 
 /* Return the Exception Level targeted by debug exceptions. */
diff --git a/hw/arm/boot.c b/hw/arm/boot.c
index fef4072db1..4f96ce42fe 100644
--- a/hw/arm/boot.c
+++ b/hw/arm/boot.c
@@ -427,7 +427,7 @@ static void set_kernel_args_old(const struct arm_boot_info *info,
 
 static int fdt_add_memory_node(void *fdt, uint32_t acells, hwaddr mem_base,
                                uint32_t scells, hwaddr mem_len,
-                               int numa_node_id)
+                               int numa_node_id, bool tag_memory)
 {
     char *nodename;
     int ret;
@@ -446,6 +446,10 @@ static int fdt_add_memory_node(void *fdt, uint32_t acells, hwaddr mem_base,
         ret = qemu_fdt_setprop_cell(fdt, nodename,
                                     "numa-node-id", numa_node_id);
     }
+    if (tag_memory) {
+        qemu_fdt_setprop(fdt, nodename, "arm,armv8.5-memtag", "", 0);
+    }
+
 out:
     g_free(nodename);
     return ret;
@@ -534,6 +538,7 @@ int arm_load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
     hwaddr mem_base, mem_len;
     char **node_path;
     Error *err = NULL;
+    bool tag_memory;
 
     if (binfo->dtb_filename) {
         char *filename;
@@ -599,12 +604,13 @@ int arm_load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
     }
     g_strfreev(node_path);
 
+    tag_memory = binfo->tag_memory;
     if (ms->numa_state != NULL && ms->numa_state->num_nodes > 0) {
         mem_base = binfo->loader_start;
         for (i = 0; i < ms->numa_state->num_nodes; i++) {
             mem_len = ms->numa_state->nodes[i].node_mem;
             rc = fdt_add_memory_node(fdt, acells, mem_base,
-                                     scells, mem_len, i);
+                                     scells, mem_len, i, tag_memory);
             if (rc < 0) {
                 fprintf(stderr, "couldn't add /memory@%"PRIx64" node\n",
                         mem_base);
@@ -615,7 +621,7 @@ int arm_load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
         }
     } else {
         rc = fdt_add_memory_node(fdt, acells, binfo->loader_start,
-                                 scells, binfo->ram_size, -1);
+                                 scells, binfo->ram_size, -1, tag_memory);
         if (rc < 0) {
             fprintf(stderr, "couldn't add /memory@%"PRIx64" node\n",
                     binfo->loader_start);
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
index caceb1e4a0..0d81e0d3ac 100644
--- a/hw/arm/virt.c
+++ b/hw/arm/virt.c
@@ -1390,8 +1390,19 @@ static void create_platform_bus(VirtMachineState *vms)
                                 sysbus_mmio_get_region(s, 0));
 }
 
+static void create_tag_ram(MemoryRegion *tag_sysmem,
+                           hwaddr base, hwaddr size,
+                           const char *name)
+{
+    MemoryRegion *tagram = g_new(MemoryRegion, 1);
+
+    memory_region_init_ram(tagram, NULL, name, size / 32, &error_fatal);
+    memory_region_add_subregion(tag_sysmem, base / 32, tagram);
+}
+
 static void create_secure_ram(VirtMachineState *vms,
-                              MemoryRegion *secure_sysmem)
+                              MemoryRegion *secure_sysmem,
+                              MemoryRegion *secure_tag_sysmem)
 {
     MemoryRegion *secram = g_new(MemoryRegion, 1);
     char *nodename;
@@ -1409,6 +1420,11 @@ static void create_secure_ram(VirtMachineState *vms,
     qemu_fdt_setprop_string(vms->fdt, nodename, "status", "disabled");
     qemu_fdt_setprop_string(vms->fdt, nodename, "secure-status", "okay");
 
+    if (secure_tag_sysmem) {
+        create_tag_ram(secure_tag_sysmem, base, size, "mach-virt.secure-tag");
+        qemu_fdt_setprop(vms->fdt, nodename, "arm,armv8.5-memtag", "", 0);
+    }
+
     g_free(nodename);
 }
 
@@ -1665,6 +1681,8 @@ static void machvirt_init(MachineState *machine)
     const CPUArchIdList *possible_cpus;
     MemoryRegion *sysmem = get_system_memory();
     MemoryRegion *secure_sysmem = NULL;
+    MemoryRegion *tag_sysmem = NULL;
+    MemoryRegion *secure_tag_sysmem = NULL;
     int n, virt_max_cpus;
     bool firmware_loaded;
     bool aarch64 = true;
@@ -1819,6 +1837,36 @@ static void machvirt_init(MachineState *machine)
                                      "secure-memory", &error_abort);
         }
 
+        /*
+         * The cpu adds the property if and only if MemTag is supported.
+         * If it is, we must allocate the ram to back that up.
+         */
+        if (object_property_find(cpuobj, "tag-memory", NULL)) {
+            if (!tag_sysmem) {
+                vms->bootinfo.tag_memory = true;
+                tag_sysmem = g_new(MemoryRegion, 1);
+                memory_region_init(tag_sysmem, OBJECT(machine),
+                                   "tag-memory", UINT64_MAX / 32);
+
+                if (vms->secure) {
+                    secure_tag_sysmem = g_new(MemoryRegion, 1);
+                    memory_region_init(secure_tag_sysmem, OBJECT(machine),
+                                       "secure-tag-memory", UINT64_MAX / 32);
+
+                    /* As with ram, secure-tag takes precedence over tag.  */
+                    memory_region_add_subregion_overlap(secure_tag_sysmem, 0,
+                                                        tag_sysmem, -1);
+                }
+            }
+
+            object_property_set_link(cpuobj, OBJECT(tag_sysmem),
+                                     "tag-memory", &error_abort);
+            if (vms->secure) {
+                object_property_set_link(cpuobj, OBJECT(secure_tag_sysmem),
+                                         "secure-tag-memory", &error_abort);
+            }
+        }
+
         qdev_realize(DEVICE(cpuobj), NULL, &error_fatal);
         object_unref(cpuobj);
     }
@@ -1857,10 +1905,15 @@ static void machvirt_init(MachineState *machine)
     create_uart(vms, VIRT_UART, sysmem, serial_hd(0));
 
     if (vms->secure) {
-        create_secure_ram(vms, secure_sysmem);
+        create_secure_ram(vms, secure_sysmem, secure_tag_sysmem);
         create_uart(vms, VIRT_SECURE_UART, secure_sysmem, serial_hd(1));
     }
 
+    if (tag_sysmem) {
+        create_tag_ram(tag_sysmem, vms->memmap[VIRT_MEM].base,
+                       machine->ram_size, "mach-virt.tag");
+    }
+
     vms->highmem_ecam &= vms->highmem && (!firmware_loaded || aarch64);
 
     create_rtc(vms);
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index f09efc4370..7159188247 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -1249,6 +1249,25 @@ void arm_cpu_post_init(Object *obj)
     if (kvm_enabled()) {
         kvm_arm_add_vcpu_properties(obj);
     }
+
+#ifndef CONFIG_USER_ONLY
+    if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64) &&
+        cpu_isar_feature(aa64_mte, cpu)) {
+        object_property_add_link(obj, "tag-memory",
+                                 TYPE_MEMORY_REGION,
+                                 (Object **)&cpu->tag_memory,
+                                 qdev_prop_allow_set_link_before_realize,
+                                 OBJ_PROP_LINK_STRONG);
+
+        if (arm_feature(&cpu->env, ARM_FEATURE_EL3)) {
+            object_property_add_link(obj, "secure-tag-memory",
+                                     TYPE_MEMORY_REGION,
+                                     (Object **)&cpu->secure_tag_memory,
+                                     qdev_prop_allow_set_link_before_realize,
+                                     OBJ_PROP_LINK_STRONG);
+        }
+    }
+#endif
 }
 
 static void arm_cpu_finalizefn(Object *obj)
@@ -1739,17 +1758,43 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
     MachineState *ms = MACHINE(qdev_get_machine());
     unsigned int smp_cpus = ms->smp.cpus;
 
-    if (cpu->has_el3 || arm_feature(env, ARM_FEATURE_M_SECURITY)) {
+    /*
+     * We must set cs->num_ases to the final value before
+     * the first call to cpu_address_space_init.
+     */
+    if (cpu->tag_memory != NULL) {
+        cs->num_ases = 4;
+    } else if (cpu->has_el3 || arm_feature(env, ARM_FEATURE_M_SECURITY)) {
         cs->num_ases = 2;
+    } else {
+        cs->num_ases = 1;
+    }
 
+    if (cpu->has_el3 || arm_feature(env, ARM_FEATURE_M_SECURITY)) {
         if (!cpu->secure_memory) {
             cpu->secure_memory = cs->memory;
         }
         cpu_address_space_init(cs, ARMASIdx_S, "cpu-secure-memory",
                                cpu->secure_memory);
-    } else {
-        cs->num_ases = 1;
     }
+
+    if (cpu->tag_memory != NULL) {
+        cpu_address_space_init(cs, ARMASIdx_TagNS, "cpu-tag-memory",
+                               cpu->tag_memory);
+        if (cpu->has_el3) {
+            cpu_address_space_init(cs, ARMASIdx_TagS, "cpu-tag-memory",
+                                   cpu->secure_tag_memory);
+        }
+    } else if (cpu_isar_feature(aa64_mte, cpu)) {
+        /*
+         * Since there is no tag memory, we can't meaningfully support MTE
+         * to its fullest.  To avoid problems later, when we would come to
+         * use the tag memory, downgrade support to insns only.
+         */
+        cpu->isar.id_aa64pfr1 =
+            FIELD_DP64(cpu->isar.id_aa64pfr1, ID_AA64PFR1, MTE, 1);
+    }
+
     cpu_address_space_init(cs, ARMASIdx_NS, "cpu-memory", cs->memory);
 
     /* No core_count specified, default to smp_cpus. */
-- 
2.25.1

[PATCH v8 44/45] target/arm: Add allocation tag storage for system mode

[Richard Henderson]

Look up the physical address for the given virtual address,
convert that to a tag physical address, and finally return
the host address that backs it.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/mte_helper.c | 126 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 126 insertions(+)

diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c
index 4f9bd3add3..4911cbca50 100644
--- a/target/arm/mte_helper.c
+++ b/target/arm/mte_helper.c
@@ -21,6 +21,7 @@
 #include "cpu.h"
 #include "internals.h"
 #include "exec/exec-all.h"
+#include "exec/ram_addr.h"
 #include "exec/cpu_ldst.h"
 #include "exec/helper-proto.h"
 
@@ -74,8 +75,133 @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx,
                                    int ptr_size, MMUAccessType tag_access,
                                    int tag_size, uintptr_t ra)
 {
+#ifdef CONFIG_USER_ONLY
     /* Tag storage not implemented.  */
     return NULL;
+#else
+    uintptr_t index;
+    CPUIOTLBEntry *iotlbentry;
+    int in_page, flags;
+    ram_addr_t ptr_ra;
+    hwaddr ptr_paddr, tag_paddr, xlat;
+    MemoryRegion *mr;
+    ARMASIdx tag_asi;
+    AddressSpace *tag_as;
+    void *host;
+
+    /*
+     * Probe the first byte of the virtual address.  This raises an
+     * exception for inaccessible pages, and resolves the virtual address
+     * into the softmmu tlb.
+     *
+     * When RA == 0, this is for mte_probe1.  The page is expected to be
+     * valid.  Indicate to probe_access_flags no-fault, then assert that
+     * we received a valid page.
+     */
+    flags = probe_access_flags(env, ptr, ptr_access, ptr_mmu_idx,
+                               ra == 0, &host, ra);
+    assert(!(flags & TLB_INVALID_MASK));
+
+    /*
+     * Find the iotlbentry for ptr.  This *must* be present in the TLB
+     * because we just found the mapping.
+     * TODO: Perhaps there should be a cputlb helper that returns a
+     * matching tlb entry + iotlb entry.
+     */
+    index = tlb_index(env, ptr_mmu_idx, ptr);
+# ifdef CONFIG_DEBUG_TCG
+    {
+        CPUTLBEntry *entry = tlb_entry(env, ptr_mmu_idx, ptr);
+        target_ulong comparator = (ptr_access == MMU_DATA_LOAD
+                                   ? entry->addr_read
+                                   : tlb_addr_write(entry));
+        g_assert(tlb_hit(comparator, ptr));
+    }
+# endif
+    iotlbentry = &env_tlb(env)->d[ptr_mmu_idx].iotlb[index];
+
+    /* If the virtual page MemAttr != Tagged, access unchecked. */
+    if (!arm_tlb_mte_tagged(&iotlbentry->attrs)) {
+        return NULL;
+    }
+
+    /* If not normal memory, there is no tag storage: access unchecked. */
+    if (unlikely(flags & TLB_MMIO)) {
+        qemu_log_mask(LOG_GUEST_ERROR,
+                      "Page @ 0x%" PRIx64 " indicates Tagged Normal memory "
+                      "but is Device memory\n", ptr);
+        return NULL;
+    }
+
+    /*
+     * The Normal memory access can extend to the next page.  E.g. a single
+     * 8-byte access to the last byte of a page will check only the last
+     * tag on the first page.
+     * Any page access exception has priority over tag check exception.
+     */
+    in_page = -(ptr | TARGET_PAGE_MASK);
+    if (unlikely(ptr_size > in_page)) {
+        void *ignore;
+        flags |= probe_access_flags(env, ptr + in_page, ptr_access,
+                                    ptr_mmu_idx, false, &ignore, ra);
+    }
+
+    /* Any debug exception has priority over a tag check exception. */
+    if (unlikely(flags & TLB_WATCHPOINT)) {
+        int wp = ptr_access == MMU_DATA_LOAD ? BP_MEM_READ : BP_MEM_WRITE;
+        cpu_check_watchpoint(env_cpu(env), ptr, ptr_size,
+                             iotlbentry->attrs, wp, ra);
+    }
+
+    /*
+     * Find the physical address within the normal mem space.
+     * The memory region lookup must succeed because TLB_MMIO was
+     * not set in the cputlb lookup above.
+     */
+    mr = memory_region_from_host(host, &ptr_ra);
+    tcg_debug_assert(mr != NULL);
+    tcg_debug_assert(memory_region_is_ram(mr));
+    ptr_paddr = ptr_ra;
+    do {
+        ptr_paddr += mr->addr;
+        mr = mr->container;
+    } while (mr);
+
+    /* Convert to the physical address in tag space.  */
+    tag_paddr = ptr_paddr >> (LOG2_TAG_GRANULE + 1);
+
+    /* Look up the address in tag space. */
+    tag_asi = iotlbentry->attrs.secure ? ARMASIdx_TagS : ARMASIdx_TagNS;
+    tag_as = cpu_get_address_space(env_cpu(env), tag_asi);
+    mr = address_space_translate(tag_as, tag_paddr, &xlat, NULL,
+                                 tag_access == MMU_DATA_STORE,
+                                 iotlbentry->attrs);
+
+    /*
+     * Note that @mr will never be NULL.  If there is nothing in the address
+     * space at @tag_paddr, the translation will return the unallocated memory
+     * region.  For our purposes, the result must be ram.
+     */
+    if (unlikely(!memory_region_is_ram(mr))) {
+        /* ??? Failure is a board configuration error. */
+        qemu_log_mask(LOG_UNIMP,
+                      "Tag Memory @ 0x%" HWADDR_PRIx " not found for "
+                      "Normal Memory @ 0x%" HWADDR_PRIx "\n",
+                      tag_paddr, ptr_paddr);
+        return NULL;
+    }
+
+    /*
+     * Ensure the tag memory is dirty on write, for migration.
+     * Tag memory can never contain code or display memory (vga).
+     */
+    if (tag_access == MMU_DATA_STORE) {
+        ram_addr_t tag_ra = memory_region_get_ram_addr(mr) + xlat;
+        cpu_physical_memory_set_dirty_flag(tag_ra, DIRTY_MEMORY_MIGRATION);
+    }
+
+    return memory_region_get_ram_ptr(mr) + xlat;
+#endif
 }
 
 uint64_t HELPER(irg)(CPUARMState *env, uint64_t rn, uint64_t rm)
-- 
2.25.1

[PATCH v8 45/45] target/arm: Enable MTE

[Richard Henderson]

We now implement all of the components of MTE, without actually
supporting any tagged memory.  All MTE instructions will work,
trivially, so we can enable support.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
v6: Delay user-only cpu reset bits to the user-only patch set.
---
 target/arm/cpu64.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 778cecc2e6..45b0ca7188 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -654,6 +654,7 @@ static void aarch64_max_initfn(Object *obj)
 
         t = cpu->isar.id_aa64pfr1;
         t = FIELD_DP64(t, ID_AA64PFR1, BT, 1);
+        t = FIELD_DP64(t, ID_AA64PFR1, MTE, 2);
         cpu->isar.id_aa64pfr1 = t;
 
         t = cpu->isar.id_aa64mmfr1;
-- 
2.25.1

Re: [PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[Derrick McKee]

Hi Richard,

I am currently doing research with MTE, and Arm's FVP leaves a lot to
be desired.  Is there some way I can get access to your QEMU changes
that enable MTE for system emulation?  Thanks.

On Tue, Jun 23, 2020 at 3:37 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Version 8 incorporates quite a bit of review (thanks, PMM):
>
>  * Include HCR_DCT in the HCR_EL2 update.
>  * Use a separate decode function for ADDG/SUBG.
>  * Use the system arm_cpu_do_unaligned_access for user-only as well.
>  * Fix alignment for LDG.
>  * Fix no-fault for mte_probe1.
>  * Rearrage how tagged pages are recorded.
>
> and perhaps most visibly, for those tracking the kernel patches:
>
>  * Add arm,armv8.5-memtag to the dtb.
>
>
> r~
>
>
> In need of review:
> 0010-target-arm-Revise-decoding-for-disas_add_sub_imm.patch
> 0011-target-arm-Implement-the-ADDG-SUBG-instructions.patch
> 0014-target-arm-Define-arm_cpu_do_unaligned_access-for.patch
> 0015-target-arm-Implement-LDG-STG-ST2G-instructions.patch
> 0025-target-arm-Implement-helper_mte_check1.patch
> 0028-target-arm-Use-mte_checkN-for-sve-unpredicated-lo.patch
> 0029-target-arm-Use-mte_checkN-for-sve-unpredicated-st.patch
> 0030-target-arm-Use-mte_check1-for-sve-LD1R.patch
> 0031-target-arm-Tidy-trans_LD1R_zpri.patch
> 0032-target-arm-Add-arm_tlb_bti_gp.patch
> 0033-target-arm-Add-mte-helpers-for-sve-scalar-int-loa.patch
> 0034-target-arm-Add-mte-helpers-for-sve-scalar-int-sto.patch
> 0035-target-arm-Add-mte-helpers-for-sve-scalar-int-ff-.patch
> 0036-target-arm-Handle-TBI-for-sve-scalar-int-memory-o.patch
> 0037-target-arm-Add-mte-helpers-for-sve-scatter-gather.patch
> 0038-target-arm-Complete-TBI-clearing-for-user-only-fo.patch
> 0041-target-arm-Always-pass-cacheattr-to-get_phys_addr.patch
> 0042-target-arm-Cache-the-Tagged-bit-for-a-page-in-Mem.patch
> 0043-target-arm-Create-tagged-ram-when-MTE-is-enabled.patch
> 0044-target-arm-Add-allocation-tag-storage-for-system-.patch
> 0045-target-arm-Enable-MTE.patch
>
>
> Richard Henderson (45):
>   target/arm: Add isar tests for mte
>   target/arm: Improve masking of SCR RES0 bits
>   target/arm: Add support for MTE to SCTLR_ELx
>   target/arm: Add support for MTE to HCR_EL2 and SCR_EL3
>   target/arm: Rename DISAS_UPDATE to DISAS_UPDATE_EXIT
>   target/arm: Add DISAS_UPDATE_NOCHAIN
>   target/arm: Add MTE system registers
>   target/arm: Add MTE bits to tb_flags
>   target/arm: Implement the IRG instruction
>   target/arm: Revise decoding for disas_add_sub_imm
>   target/arm: Implement the ADDG, SUBG instructions
>   target/arm: Implement the GMI instruction
>   target/arm: Implement the SUBP instruction
>   target/arm: Define arm_cpu_do_unaligned_access for user-only
>   target/arm: Implement LDG, STG, ST2G instructions
>   target/arm: Implement the STGP instruction
>   target/arm: Restrict the values of DCZID.BS under TCG
>   target/arm: Simplify DC_ZVA
>   target/arm: Implement the LDGM, STGM, STZGM instructions
>   target/arm: Implement the access tag cache flushes
>   target/arm: Move regime_el to internals.h
>   target/arm: Move regime_tcr to internals.h
>   target/arm: Add gen_mte_check1
>   target/arm: Add gen_mte_checkN
>   target/arm: Implement helper_mte_check1
>   target/arm: Implement helper_mte_checkN
>   target/arm: Add helper_mte_check_zva
>   target/arm: Use mte_checkN for sve unpredicated loads
>   target/arm: Use mte_checkN for sve unpredicated stores
>   target/arm: Use mte_check1 for sve LD1R
>   target/arm: Tidy trans_LD1R_zpri
>   target/arm: Add arm_tlb_bti_gp
>   target/arm: Add mte helpers for sve scalar + int loads
>   target/arm: Add mte helpers for sve scalar + int stores
>   target/arm: Add mte helpers for sve scalar + int ff/nf loads
>   target/arm: Handle TBI for sve scalar + int memory ops
>   target/arm: Add mte helpers for sve scatter/gather memory ops
>   target/arm: Complete TBI clearing for user-only for SVE
>   target/arm: Implement data cache set allocation tags
>   target/arm: Set PSTATE.TCO on exception entry
>   target/arm: Always pass cacheattr to get_phys_addr
>   target/arm: Cache the Tagged bit for a page in MemTxAttrs
>   target/arm: Create tagged ram when MTE is enabled
>   target/arm: Add allocation tag storage for system mode
>   target/arm: Enable MTE
>
>  include/hw/arm/boot.h          |    3 +
>  target/arm/cpu.h               |   50 +-
>  target/arm/helper-a64.h        |   16 +
>  target/arm/helper-sve.h        |  488 ++++++++++++
>  target/arm/helper.h            |    2 +
>  target/arm/internals.h         |  153 +++-
>  target/arm/translate-a64.h     |    5 +
>  target/arm/translate.h         |   23 +-
>  hw/arm/boot.c                  |   12 +-
>  hw/arm/virt.c                  |   57 +-
>  target/arm/cpu.c               |   77 +-
>  target/arm/cpu64.c             |    1 +
>  target/arm/helper-a64.c        |   94 +--
>  target/arm/helper.c            |  423 +++++++---
>  target/arm/m_helper.c          |   11 +-
>  target/arm/mte_helper.c        |  901 +++++++++++++++++++++
>  target/arm/op_helper.c         |   16 +
>  target/arm/sve_helper.c        |  616 +++++++++++---
>  target/arm/tlb_helper.c        |   13 +-
>  target/arm/translate-a64.c     |  652 +++++++++++++--
>  target/arm/translate-sve.c     | 1366 +++++++++++++++++++++-----------
>  target/arm/translate-vfp.inc.c |    2 +-
>  target/arm/translate.c         |   16 +-
>  target/arm/Makefile.objs       |    1 +
>  24 files changed, 4164 insertions(+), 834 deletions(-)
>  create mode 100644 target/arm/mte_helper.c
>
> --
> 2.25.1
>
>


--
Derrick McKee
Phone: (703) 957-9362
Email: derrick.mckee@gmail.com

Re: [PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[Richard Henderson]

On 6/23/20 12:55 PM, Derrick McKee wrote:
> Hi Richard,
> 
> I am currently doing research with MTE, and Arm's FVP leaves a lot to
> be desired.  Is there some way I can get access to your QEMU changes
> that enable MTE for system emulation?  Thanks.

You can apply this patch set to qemu master.  ;-)

But more seriously, this patch set is in the tree at
https://github.com/rth7680/qemu/tree/tgt-arm-mte


r~

> 
> On Tue, Jun 23, 2020 at 3:37 PM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> Version 8 incorporates quite a bit of review (thanks, PMM):
>>
>>  * Include HCR_DCT in the HCR_EL2 update.
>>  * Use a separate decode function for ADDG/SUBG.
>>  * Use the system arm_cpu_do_unaligned_access for user-only as well.
>>  * Fix alignment for LDG.
>>  * Fix no-fault for mte_probe1.
>>  * Rearrage how tagged pages are recorded.
>>
>> and perhaps most visibly, for those tracking the kernel patches:
>>
>>  * Add arm,armv8.5-memtag to the dtb.

Re: [PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[no-reply]

Patchew URL: https://patchew.org/QEMU/20200623193658.623279-1-richard.henderson@linaro.org/



Hi,

This series seems to have some coding style problems. See output below for
more information:

Subject: [PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode
Type: series
Message-id: 20200623193658.623279-1-richard.henderson@linaro.org

=== TEST SCRIPT BEGIN ===
#!/bin/bash
git rev-parse base > /dev/null || exit 0
git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram
./scripts/checkpatch.pl --mailback base..
=== TEST SCRIPT END ===

From https://github.com/patchew-project/qemu
 * [new tag]         patchew/20200623193658.623279-1-richard.henderson@linaro.org -> patchew/20200623193658.623279-1-richard.henderson@linaro.org
Switched to a new branch 'test'
1231ec5 target/arm: Enable MTE
caf48f0 target/arm: Add allocation tag storage for system mode
86926ba target/arm: Create tagged ram when MTE is enabled
dc24d39 target/arm: Cache the Tagged bit for a page in MemTxAttrs
b3fdcfe target/arm: Always pass cacheattr to get_phys_addr
52dae54 target/arm: Set PSTATE.TCO on exception entry
9d92dcf target/arm: Implement data cache set allocation tags
985b40f target/arm: Complete TBI clearing for user-only for SVE
d22a925 target/arm: Add mte helpers for sve scatter/gather memory ops
26d1ba5 target/arm: Handle TBI for sve scalar + int memory ops
5dc33b2 target/arm: Add mte helpers for sve scalar + int ff/nf loads
457edf5 target/arm: Add mte helpers for sve scalar + int stores
7e68033 target/arm: Add mte helpers for sve scalar + int loads
df0ee49 target/arm: Add arm_tlb_bti_gp
46870d1 target/arm: Tidy trans_LD1R_zpri
661859f target/arm: Use mte_check1 for sve LD1R
c067218 target/arm: Use mte_checkN for sve unpredicated stores
abbc5d1 target/arm: Use mte_checkN for sve unpredicated loads
da2c53d target/arm: Add helper_mte_check_zva
a947bd3 target/arm: Implement helper_mte_checkN
017e09e target/arm: Implement helper_mte_check1
060a546 target/arm: Add gen_mte_checkN
349cca3 target/arm: Add gen_mte_check1
bcb99d2 target/arm: Move regime_tcr to internals.h
3e1b548 target/arm: Move regime_el to internals.h
e03f1b4 target/arm: Implement the access tag cache flushes
d6ad4b5 target/arm: Implement the LDGM, STGM, STZGM instructions
e16594d target/arm: Simplify DC_ZVA
dc1d79a target/arm: Restrict the values of DCZID.BS under TCG
1291525 target/arm: Implement the STGP instruction
1b4d84d target/arm: Implement LDG, STG, ST2G instructions
54c306b target/arm: Define arm_cpu_do_unaligned_access for user-only
56adc0c target/arm: Implement the SUBP instruction
e1d7cf7 target/arm: Implement the GMI instruction
2890f30 target/arm: Implement the ADDG, SUBG instructions
912b9bf target/arm: Revise decoding for disas_add_sub_imm
63945d2 target/arm: Implement the IRG instruction
22de2b3 target/arm: Add MTE bits to tb_flags
45f22ac target/arm: Add MTE system registers
2fbb1da target/arm: Add DISAS_UPDATE_NOCHAIN
e4d0611 target/arm: Rename DISAS_UPDATE to DISAS_UPDATE_EXIT
d0f228b target/arm: Add support for MTE to HCR_EL2 and SCR_EL3
a5b3055 target/arm: Add support for MTE to SCTLR_ELx
abdfb70 target/arm: Improve masking of SCR RES0 bits
40707f0 target/arm: Add isar tests for mte

=== OUTPUT BEGIN ===
1/45 Checking commit 40707f09f121 (target/arm: Add isar tests for mte)
2/45 Checking commit abdfb70dea3d (target/arm: Improve masking of SCR RES0 bits)
3/45 Checking commit a5b305518974 (target/arm: Add support for MTE to SCTLR_ELx)
4/45 Checking commit d0f228b9f5f4 (target/arm: Add support for MTE to HCR_EL2 and SCR_EL3)
5/45 Checking commit e4d06118aebc (target/arm: Rename DISAS_UPDATE to DISAS_UPDATE_EXIT)
6/45 Checking commit 2fbb1dafd99c (target/arm: Add DISAS_UPDATE_NOCHAIN)
7/45 Checking commit 45f22acdd599 (target/arm: Add MTE system registers)
8/45 Checking commit 22de2b35721f (target/arm: Add MTE bits to tb_flags)
9/45 Checking commit 63945d282302 (target/arm: Implement the IRG instruction)
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#45: 
new file mode 100644

total: 0 errors, 1 warnings, 120 lines checked

Patch 9/45 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
10/45 Checking commit 912b9bf152c9 (target/arm: Revise decoding for disas_add_sub_imm)
11/45 Checking commit 2890f308b67c (target/arm: Implement the ADDG, SUBG instructions)
12/45 Checking commit e1d7cf7b3195 (target/arm: Implement the GMI instruction)
13/45 Checking commit 56adc0c58443 (target/arm: Implement the SUBP instruction)
14/45 Checking commit 54c306b84778 (target/arm: Define arm_cpu_do_unaligned_access for user-only)
15/45 Checking commit 1b4d84d4bf40 (target/arm: Implement LDG, STG, ST2G instructions)
16/45 Checking commit 129152540330 (target/arm: Implement the STGP instruction)
17/45 Checking commit dc1d79a46ea9 (target/arm: Restrict the values of DCZID.BS under TCG)
18/45 Checking commit e16594db4ee6 (target/arm: Simplify DC_ZVA)
19/45 Checking commit d6ad4b51f761 (target/arm: Implement the LDGM, STGM, STZGM instructions)
20/45 Checking commit e03f1b492eb8 (target/arm: Implement the access tag cache flushes)
21/45 Checking commit 3e1b548f5665 (target/arm: Move regime_el to internals.h)
22/45 Checking commit bcb99d2cb0d5 (target/arm: Move regime_tcr to internals.h)
23/45 Checking commit 349cca39015e (target/arm: Add gen_mte_check1)
24/45 Checking commit 060a546b2de1 (target/arm: Add gen_mte_checkN)
25/45 Checking commit 017e09edd952 (target/arm: Implement helper_mte_check1)
26/45 Checking commit a947bd38a095 (target/arm: Implement helper_mte_checkN)
27/45 Checking commit da2c53dcc0e2 (target/arm: Add helper_mte_check_zva)
28/45 Checking commit abbc5d1dd1f3 (target/arm: Use mte_checkN for sve unpredicated loads)
29/45 Checking commit c067218f4474 (target/arm: Use mte_checkN for sve unpredicated stores)
30/45 Checking commit 661859ff9984 (target/arm: Use mte_check1 for sve LD1R)
31/45 Checking commit 46870d1ed93b (target/arm: Tidy trans_LD1R_zpri)
32/45 Checking commit df0ee49681ee (target/arm: Add arm_tlb_bti_gp)
33/45 Checking commit 7e68033a5c5e (target/arm: Add mte helpers for sve scalar + int loads)
34/45 Checking commit 457edf568b72 (target/arm: Add mte helpers for sve scalar + int stores)
35/45 Checking commit 5dc33b2a8bb2 (target/arm: Add mte helpers for sve scalar + int ff/nf loads)
36/45 Checking commit 26d1ba548fc0 (target/arm: Handle TBI for sve scalar + int memory ops)
37/45 Checking commit d22a9255942c (target/arm: Add mte helpers for sve scatter/gather memory ops)
38/45 Checking commit 985b40fb259c (target/arm: Complete TBI clearing for user-only for SVE)
ERROR: spaces required around that '*' (ctx:VxV)
#42: FILE: target/arm/sve_helper.c:3978:
+        (TYPEM)*(TYPEE *)(vd + H(reg_off)), ra);                            \
                ^

total: 1 errors, 0 warnings, 37 lines checked

Patch 38/45 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

39/45 Checking commit 9d92dcf8187b (target/arm: Implement data cache set allocation tags)
40/45 Checking commit 52dae54ca0f4 (target/arm: Set PSTATE.TCO on exception entry)
41/45 Checking commit b3fdcfe39e1d (target/arm: Always pass cacheattr to get_phys_addr)
42/45 Checking commit dc24d3963608 (target/arm: Cache the Tagged bit for a page in MemTxAttrs)
43/45 Checking commit 86926ba74d36 (target/arm: Create tagged ram when MTE is enabled)
44/45 Checking commit caf48f000954 (target/arm: Add allocation tag storage for system mode)
45/45 Checking commit 1231ec5cb04a (target/arm: Enable MTE)
=== OUTPUT END ===

Test command exited with code: 1


The full log is available at
http://patchew.org/logs/20200623193658.623279-1-richard.henderson@linaro.org/testing.checkpatch/?type=message.
---
Email generated automatically by Patchew [https://patchew.org/].
Please send your feedback to patchew-devel@redhat.com

Re: [PATCH v8 10/45] target/arm: Revise decoding for disas_add_sub_imm

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> The current Arm ARM has adjusted the official decode of
> "Add/subtract (immediate)" so that the shift field is only bit 22,
> and bit 23 is part of the op1 field of the parent category
> "Data processing - immediate".
>
> Suggested-by: Peter Maydell <peter.maydell@linaro.org>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/translate-a64.c | 23 ++++++++---------------
>  1 file changed, 8 insertions(+), 15 deletions(-)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 11/45] target/arm: Implement the ADDG, SUBG instructions

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> +/*
> + * Add/subtract (immediate, with tags)
> + *
> + *  31 30 29 28         23 22 21     16 14      10 9   5 4   0
> + * +--+--+--+-------------+--+---------+--+-------+-----+-----+
> + * |sf|op| S| 1 0 0 0 1 0 |o2|  uimm6  |o3| uimm4 |  Rn | Rd  |
> + * +--+--+--+-------------+--+---------+--+-------+-----+-----+

Bit 23 should be '1'.

> + *
> + *    op: 0 -> add, 1 -> sub
> + */
> +static void disas_add_sub_imm_with_tags(DisasContext *s, uint32_t insn)
> +{
> +    int rd = extract32(insn, 0, 5);
> +    int rn = extract32(insn, 5, 5);
> +    int uimm4 = extract32(insn, 10, 4);
> +    int uimm6 = extract32(insn, 16, 6);
> +    bool sub_op = extract32(insn, 30, 1);
> +    TCGv_i64 tcg_rn, tcg_rd;
> +    int imm;
> +
> +    /* Test all of sf=1, S=0, o2=0, o3=0.  */
> +    if ((insn & 0xc040e000u) != 0x80000000u ||

This bit pattern doesn't seem to match the comment:
0xc is 0b1100 so that's sf and op, not sf and S;
0xe is 0b1110 so that's testing the top bit of uimm4
as well as op3. I think it should be 0xa040c000.
Though the existence of this bug suggests that it would
be clearer to test the individual fields :-)

Otherwise
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>


Interestingly clang (6.0.0) gets pretty close to optimizing
the field test version:

bool foo(uint32_t insn) {
   return (insn & 0xa040c000) != 0x80000000u;
}

bool bar(uint32_t insn) {
   bool sf = extract32(insn, 31, 1);
   bool s = extract32(insn, 29, 1);
   bool o2 = extract32(insn, 22, 1);
   int op2 = extract32(insn, 14, 2);

   return sf != 1 || s != 0 || o2 != 0 || op2 != 0;
}

gives
0000000000000000 <foo>:
   0:   81 e7 00 c0 40 a0       and    $0xa040c000,%edi
   6:   81 ff 00 00 00 80       cmp    $0x80000000,%edi
   c:   0f 95 c0                setne  %al
   f:   c3                      retq

0000000000000010 <bar>:
  10:   89 f8                   mov    %edi,%eax
  12:   25 00 00 00 a0          and    $0xa0000000,%eax
  17:   3d 00 00 00 80          cmp    $0x80000000,%eax
  1c:   75 0a                   jne    28 <bar+0x18>
  1e:   f7 c7 00 c0 40 00       test   $0x40c000,%edi
  24:   0f 95 c0                setne  %al
  27:   c3                      retq
  28:   b0 01                   mov    $0x1,%al
  2a:   c3                      retq

(I don't know why it's split it into two tests: it's
not that it's testing "must be 1" in one part and "must
be 0" in the other because it has checked both sf and s
in the first comparison.)

gcc (7.4.0) makes more of a hash of it though:

  10:   89 f8                   mov    %edi,%eax
  12:   89 fa                   mov    %edi,%edx
  14:   c1 e8 1f                shr    $0x1f,%eax
  17:   c1 ea 1d                shr    $0x1d,%edx
  1a:   83 f0 01                xor    $0x1,%eax
  1d:   09 d0                   or     %edx,%eax
  1f:   83 e0 01                and    $0x1,%eax
  22:   75 13                   jne    37 <bar+0x27>
  24:   89 f8                   mov    %edi,%eax
  26:   c1 ef 16                shr    $0x16,%edi
  29:   c1 e8 0e                shr    $0xe,%eax
  2c:   83 e7 01                and    $0x1,%edi
  2f:   83 e0 03                and    $0x3,%eax
  32:   09 f8                   or     %edi,%eax
  34:   0f 95 c0                setne  %al
  37:   f3 c3                   repz retq

thanks
-- PMM

Re: [PATCH v8 14/45] target/arm: Define arm_cpu_do_unaligned_access for user-only

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Use the same code as system mode, so that we generate the same
> exception + syndrome for the unaligned access.
>
> For the moment, if MTE is enabled so that this path is reachable,
> this would generate a SIGSEGV in the user-only cpu_loop.  Decoding
> the syndrome to produce the proper SIGBUS will be done later.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v8: Raise the normal data exception + syndrome.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 15/45] target/arm: Implement LDG, STG, ST2G instructions

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v2: Split out allocation_tag_mem.  Handle atomicity of stores.
> v3: Add X[t] input to these insns; require pre-cleaned addresses.
> v5: Fix !32-byte aligned operation of st2g.
> v6: Fix op2 extract, stg pre/post-index, stores vs sp, commentary;
>     use pre-computed ata.
> v7: Fix STZG iteration (stephen long)
> v8: Merge gen_probe_access patch; align address for ldg (pmm)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 25/45] target/arm: Implement helper_mte_check1

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Fill out the stub that was added earlier.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v8: Remove ra argument to mte_probe1 (pmm).
> ---
>  target/arm/internals.h  |  48 +++++++++++++++
>  target/arm/mte_helper.c | 132 +++++++++++++++++++++++++++++++++++++++-
>  2 files changed, 179 insertions(+), 1 deletion(-)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 28/45] target/arm: Use mte_checkN for sve unpredicated loads

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v8: Drop the out-of-line helper (pmm).
> ---
>  target/arm/translate-sve.c | 61 +++++++++++++++++++++-----------------
>  1 file changed, 33 insertions(+), 28 deletions(-)
>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 29/45] target/arm: Use mte_checkN for sve unpredicated stores

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v8: Drop the out-of-line helper (pmm).
> ---
>  target/arm/translate-sve.c | 61 +++++++++++++++++++++-----------------
>  1 file changed, 33 insertions(+), 28 deletions(-)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 30/45] target/arm: Use mte_check1 for sve LD1R

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/translate-sve.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 31/45] target/arm: Tidy trans_LD1R_zpri

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Move the variable declarations to the top of the function,
> but do not create a new label before sve_access_check.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v8: Split out from previous patch (pmm).
> ---

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 32/45] target/arm: Add arm_tlb_bti_gp

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Introduce an lvalue macro to wrap target_tlb_bit0.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> +/* Helper for the macros below, validating the argument type. */
> +static inline MemTxAttrs *typecheck_memtxattrs(MemTxAttrs *x)
> +{
> +    return x;
> +}
> +
> +/*
> + * Lvalue macros for ARM TLB bits that we must cache in the TCG TLB.
> + * Using these should be a bit more self-documenting than using the
> + * generic target bits directly.
> + */
> +#define arm_tlb_bti_gp(x) (typecheck_memtxattrs(x)->target_tlb_bit0)

It seems unlikely that we wouldn't get a compile failure if
we passed something other than a MemTxAttrs* to this macro,
given the specificity of the field name, but I guess it
gives us a less confusing compiler error ?

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 33/45] target/arm: Add mte helpers for sve scalar + int loads

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> 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.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/cpu.h           |   1 +
>  target/arm/helper-sve.h    |  58 ++++++++++
>  target/arm/internals.h     |   6 +
>  target/arm/sve_helper.c    | 218 ++++++++++++++++++++++++++++++-------
>  target/arm/translate-sve.c | 186 ++++++++++++++++++++++---------
>  5 files changed, 378 insertions(+), 91 deletions(-)
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 34/45] target/arm: Add mte helpers for sve scalar + int stores

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> 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.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 35/45] target/arm: Add mte helpers for sve scalar + int ff/nf loads

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> 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.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/helper-sve.h    |  98 ++++++++++++++++
>  target/arm/sve_helper.c    |  99 ++++++++++++++--
>  target/arm/translate-sve.c | 232 +++++++++++++++++++++++++------------
>  3 files changed, 343 insertions(+), 86 deletions(-)

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 36/45] target/arm: Handle TBI for sve scalar + int memory ops

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> We still need to handle tbi for user-only when mte is inactive.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 37/45] target/arm: Add mte helpers for sve scatter/gather memory ops

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Because the elements are non-sequential, we cannot eliminate many
> tests straight away like we can for sequential operations.  But
> we often have the PTE details handy, so we can test for Tagged.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 38/45] target/arm: Complete TBI clearing for user-only for SVE

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> There are a number of paths by which the TBI is still intact
> for user-only in the SVE helpers.
>
> Because we currently always set TBI for user-only, we do not
> need to pass down the actual TBI setting from above, and we
> can remove the top byte in the inner-most primitives, so that
> none are forgotten.  Moreover, this keeps the "dirty" pointer
> around at the higher levels, where we need it for any MTE checking.
>
> Since the normal case, especially for user-only, goes through
> RAM, this clearing merely adds two insns per page lookup, which
> will be completely in the noise.

Can we have an assert() somewhere suitable that TBI is set?
That way if we ever do have an SVE-capable linux-user which
doesn't set TBI for some reason we'll get a useful reminder
that we need to fix something.

Otherwise
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 41/45] target/arm: Always pass cacheattr to get_phys_addr

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> We need to check the memattr of a page in order to determine
> whether it is Tagged for MTE.  Between Stage1 and Stage2,
> this becomes simpler if we always collect this data, instead
> of occasionally being presented with NULL.
>
> Use the nonnull attribute to allow the compiler to check that
> all pointer arguments are non-null.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 42/45] target/arm: Cache the Tagged bit for a page in MemTxAttrs

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> This "bit" is a particular value of the page's MemAttr.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v6: Test HCR_EL2.{DC,DCT}; test Stage2 attributes.
> v8: Fill in cacheattrs for S1 disabled; retain tagging when
>     combining attributes; set mte_tagging in arm_cpu_tlb_fill.
> ---
>  target/arm/helper.c     | 48 ++++++++++++++++++++++++++++++++++++++---
>  target/arm/tlb_helper.c |  5 +++++
>  2 files changed, 50 insertions(+), 3 deletions(-)

> +        } else if (access_type == MMU_INST_FETCH) {
> +            if (regime_sctlr(env, mmu_idx) & SCTLR_I) {
> +                memattr = 0xee;  /* Normal, WT, RA, NT */
> +            } else {
> +                memattr = 0x44;  /* Normal, NC, No */
> +            }
> +            cacheattrs->shareability = 2; /* outersharable */

"outer shareable"

Otherwise
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 44/45] target/arm: Add allocation tag storage for system mode

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:38, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Look up the physical address for the given virtual address,
> convert that to a tag physical address, and finally return
> the host address that backs it.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/mte_helper.c | 126 ++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 126 insertions(+)
>

> +    /* If not normal memory, there is no tag storage: access unchecked. */
> +    if (unlikely(flags & TLB_MMIO)) {

Comment says we're checking a memory attribute, but the code
is checking for TLB_MMIO, which isn't the same thing.

thanks
-- PMM

Re: [PATCH v8 45/45] target/arm: Enable MTE

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:38, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> We now implement all of the components of MTE, without actually
> supporting any tagged memory.  All MTE instructions will work,
> trivially, so we can enable support.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v6: Delay user-only cpu reset bits to the user-only patch set.
> ---
>  target/arm/cpu64.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
> index 778cecc2e6..45b0ca7188 100644
> --- a/target/arm/cpu64.c
> +++ b/target/arm/cpu64.c
> @@ -654,6 +654,7 @@ static void aarch64_max_initfn(Object *obj)
>
>          t = cpu->isar.id_aa64pfr1;
>          t = FIELD_DP64(t, ID_AA64PFR1, BT, 1);
> +        t = FIELD_DP64(t, ID_AA64PFR1, MTE, 2);
>          cpu->isar.id_aa64pfr1 = t;

Worth a brief comment
   /* Will be downgraded to 1 if board provides no tag memory */

?

Otherwise
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [PATCH v8 43/45] target/arm: Create tagged ram when MTE is enabled

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> v5: Assign cs->num_ases to the final value first.
>     Downgrade to ID_AA64PFR1.MTE=1 if tag memory is not available.
> v6: Add secure tag memory for EL3.
> v8: Add arm,armv8.5-memtag.

> +    if (tag_memory) {
> +        qemu_fdt_setprop(fdt, nodename, "arm,armv8.5-memtag", "", 0);
> +    }

Catalin mentioned in a recent email that the plan is not to
require the dt to specify whether memory nodes support tags.
Could we keep the dtb-modification parts in a separate patch
so it's easier to drop them (or to commit the actual
emulation support first and the dtb changes later), please?

thanks
-- PMM

Re: [PATCH v8 00/45] target/arm: Implement ARMv8.5-MemTag, system mode

[Peter Maydell]

On Tue, 23 Jun 2020 at 20:37, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Version 8 incorporates quite a bit of review (thanks, PMM):
>
>  * Include HCR_DCT in the HCR_EL2 update.
>  * Use a separate decode function for ADDG/SUBG.
>  * Use the system arm_cpu_do_unaligned_access for user-only as well.
>  * Fix alignment for LDG.
>  * Fix no-fault for mte_probe1.
>  * Rearrage how tagged pages are recorded.

I think I've reviewed everything that didn't already have review now.

thanks
-- PMM

Re: [PATCH v8 38/45] target/arm: Complete TBI clearing for user-only for SVE

[Richard Henderson]

On 6/25/20 5:52 AM, Peter Maydell wrote:
> On Tue, 23 Jun 2020 at 20:37, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> There are a number of paths by which the TBI is still intact
>> for user-only in the SVE helpers.
>>
>> Because we currently always set TBI for user-only, we do not
>> need to pass down the actual TBI setting from above, and we
>> can remove the top byte in the inner-most primitives, so that
>> none are forgotten.  Moreover, this keeps the "dirty" pointer
>> around at the higher levels, where we need it for any MTE checking.
>>
>> Since the normal case, especially for user-only, goes through
>> RAM, this clearing merely adds two insns per page lookup, which
>> will be completely in the noise.
> 
> Can we have an assert() somewhere suitable that TBI is set?
> That way if we ever do have an SVE-capable linux-user which
> doesn't set TBI for some reason we'll get a useful reminder
> that we need to fix something.

At what level would you like such an assert?
At present we have, in arm_cpu_reset,

      /*
       * Enable TBI0 and TBI1.  While the real kernel only enables TBI0,
       * turning on both here will produce smaller code and otherwise
       * make no difference to the user-level emulation.
       */
      env->cp15.tcr_el[1].raw_tcr = (3ULL << 37);


r~

Re: [PATCH v8 44/45] target/arm: Add allocation tag storage for system mode

[Richard Henderson]

On 6/25/20 6:03 AM, Peter Maydell wrote:
> On Tue, 23 Jun 2020 at 20:38, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> Look up the physical address for the given virtual address,
>> convert that to a tag physical address, and finally return
>> the host address that backs it.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>  target/arm/mte_helper.c | 126 ++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 126 insertions(+)
>>
> 
>> +    /* If not normal memory, there is no tag storage: access unchecked. */
>> +    if (unlikely(flags & TLB_MMIO)) {
> 
> Comment says we're checking a memory attribute, but the code
> is checking for TLB_MMIO, which isn't the same thing.

Comment is not trying to allude to Normal vs Device, but "ram" vs "mmio" in the
qemu sense.


r~

Re: [PATCH v8 38/45] target/arm: Complete TBI clearing for user-only for SVE

[Peter Maydell]

On Thu, 25 Jun 2020 at 17:54, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 6/25/20 5:52 AM, Peter Maydell wrote:
> > On Tue, 23 Jun 2020 at 20:37, Richard Henderson
> > <richard.henderson@linaro.org> wrote:
> >>
> >> There are a number of paths by which the TBI is still intact
> >> for user-only in the SVE helpers.
> >>
> >> Because we currently always set TBI for user-only, we do not
> >> need to pass down the actual TBI setting from above, and we
> >> can remove the top byte in the inner-most primitives, so that
> >> none are forgotten.  Moreover, this keeps the "dirty" pointer
> >> around at the higher levels, where we need it for any MTE checking.
> >>
> >> Since the normal case, especially for user-only, goes through
> >> RAM, this clearing merely adds two insns per page lookup, which
> >> will be completely in the noise.
> >
> > Can we have an assert() somewhere suitable that TBI is set?
> > That way if we ever do have an SVE-capable linux-user which
> > doesn't set TBI for some reason we'll get a useful reminder
> > that we need to fix something.
>
> At what level would you like such an assert?

I don't particularly care, as long as we get notified if our
assumption that TBI is always set gets broken (ie we change
that bit of CPU reset code for some reason).

thanks
-- PMM

Re: [PATCH v8 44/45] target/arm: Add allocation tag storage for system mode

[Peter Maydell]

On Thu, 25 Jun 2020 at 18:02, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 6/25/20 6:03 AM, Peter Maydell wrote:
> > On Tue, 23 Jun 2020 at 20:38, Richard Henderson
> > <richard.henderson@linaro.org> wrote:
> >>
> >> Look up the physical address for the given virtual address,
> >> convert that to a tag physical address, and finally return
> >> the host address that backs it.
> >>
> >> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> >> ---
> >>  target/arm/mte_helper.c | 126 ++++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 126 insertions(+)
> >>
> >
> >> +    /* If not normal memory, there is no tag storage: access unchecked. */
> >> +    if (unlikely(flags & TLB_MMIO)) {
> >
> > Comment says we're checking a memory attribute, but the code
> > is checking for TLB_MMIO, which isn't the same thing.
>
> Comment is not trying to allude to Normal vs Device, but "ram" vs "mmio" in the
> qemu sense.

Oh, I see: maybe "if not backed by host RAM, then" ? I tend to
assume "normal memory" means "Normal memory" :-)

thanks
-- PMM

Re: [PATCH v8 32/45] target/arm: Add arm_tlb_bti_gp

[Richard Henderson]

On 6/25/20 5:29 AM, Peter Maydell wrote:
> On Tue, 23 Jun 2020 at 20:37, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> Introduce an lvalue macro to wrap target_tlb_bit0.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> +/* Helper for the macros below, validating the argument type. */
>> +static inline MemTxAttrs *typecheck_memtxattrs(MemTxAttrs *x)
>> +{
>> +    return x;
>> +}
>> +
>> +/*
>> + * Lvalue macros for ARM TLB bits that we must cache in the TCG TLB.
>> + * Using these should be a bit more self-documenting than using the
>> + * generic target bits directly.
>> + */
>> +#define arm_tlb_bti_gp(x) (typecheck_memtxattrs(x)->target_tlb_bit0)
> 
> It seems unlikely that we wouldn't get a compile failure if
> we passed something other than a MemTxAttrs* to this macro,
> given the specificity of the field name, but I guess it
> gives us a less confusing compiler error ?

Yes, it's a quite clear error with the function.  But you're right, we could
probably get away without it.


r~

> 
> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
> 
> thanks
> -- PMM
> 

Re: [PATCH v8 44/45] target/arm: Add allocation tag storage for system mode

[Richard Henderson]

On 6/25/20 10:09 AM, Peter Maydell wrote:
>>> Comment says we're checking a memory attribute, but the code
>>> is checking for TLB_MMIO, which isn't the same thing.
>>
>> Comment is not trying to allude to Normal vs Device, but "ram" vs "mmio" in the
>> qemu sense.
> 
> Oh, I see: maybe "if not backed by host RAM, then" ? I tend to
> assume "normal memory" means "Normal memory" :-)

Sure, changed.  Also noticed two more places where calling via mte_probe1
should not fault.


r~