Merge tag 'pull-tcg-20240723' of https://gitlab.com/rth7680/qemu into staging

accel/tcg: Export set/clear_helper_retaddr
target/arm: Use set_helper_retaddr for dc_zva, sve and sme
target/ppc: Tidy dcbz helpers
target/ppc: Use set_helper_retaddr for dcbz
target/s390x: Use set_helper_retaddr in mem_helper.c

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmafJKIdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV+FBAf7Bup+karxeGHZx2rN
# cPeF248bcCWTxBWHK7dsYze4KqzsrlNIJlPeOKErU2bbbRDZGhOp1/N95WVz+P8V
# 6Ny63WTsAYkaFWKxE6Jf0FWJlGw92btk75pTV2x/TNZixg7jg0vzVaYkk0lTYc5T
# m5e4WycYEbzYm0uodxI09i+wFvpd+7WCnl6xWtlJPWZENukvJ36Ss43egFMDtuMk
# vTJuBkS9wpwZ9MSi6EY6M+Raieg8bfaotInZeDvE/yRPNi7CwrA7Dgyc1y626uBA
# joGkYRLzhRgvT19kB3bvFZi1AXa0Pxr+j0xJqwspP239Gq5qezlS5Bv/DrHdmGHA
# jaqSwg==
# =XgUE
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 23 Jul 2024 01:33:54 PM AEST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-tcg-20240723' of https://gitlab.com/rth7680/qemu:
  target/riscv: Simplify probing in vext_ldff
  target/s390x: Use set/clear_helper_retaddr in mem_helper.c
  target/s390x: Use user_or_likely in access_memmove
  target/s390x: Use user_or_likely in do_access_memset
  target/ppc: Improve helper_dcbz for user-only
  target/ppc: Merge helper_{dcbz,dcbzep}
  target/ppc: Split out helper_dbczl for 970
  target/ppc: Hoist dcbz_size out of dcbz_common
  target/ppc/mem_helper.c: Remove a conditional from dcbz_common()
  target/arm: Use set/clear_helper_retaddr in SVE and SME helpers
  target/arm: Use set/clear_helper_retaddr in helper-a64.c
  accel/tcg: Move {set,clear}_helper_retaddr to cpu_ldst.h

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

12 files changed, 225 insertions, 131 deletions

diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 9010dad..8163295 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -41,9 +41,6 @@
 #include "tb-context.h"
 #include "internal-common.h"
 #include "internal-target.h"
-#if defined(CONFIG_USER_ONLY)
-#include "user-retaddr.h"
-#endif
 
 /* -icount align implementation. */
 
diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c
index 80d2454..7ddc47b 100644
--- a/accel/tcg/user-exec.c
+++ b/accel/tcg/user-exec.c
@@ -33,7 +33,6 @@
 #include "tcg/tcg-ldst.h"
 #include "internal-common.h"
 #include "internal-target.h"
-#include "user-retaddr.h"
 
 __thread uintptr_t helper_retaddr;
 
diff --git a/accel/tcg/user-retaddr.h b/accel/tcg/user-retaddr.h
deleted file mode 100644
index e0f57e1..0000000
--- a/accel/tcg/user-retaddr.h
+++ /dev/null
@@ -1,28 +0,0 @@
-#ifndef ACCEL_TCG_USER_RETADDR_H
-#define ACCEL_TCG_USER_RETADDR_H
-
-#include "qemu/atomic.h"
-
-extern __thread uintptr_t helper_retaddr;
-
-static inline void set_helper_retaddr(uintptr_t ra)
-{
-    helper_retaddr = ra;
-    /*
-     * Ensure that this write is visible to the SIGSEGV handler that
-     * may be invoked due to a subsequent invalid memory operation.
-     */
-    signal_barrier();
-}
-
-static inline void clear_helper_retaddr(void)
-{
-    /*
-     * Ensure that previous memory operations have succeeded before
-     * removing the data visible to the signal handler.
-     */
-    signal_barrier();
-    helper_retaddr = 0;
-}
-
-#endif
diff --git a/include/exec/cpu_ldst.h b/include/exec/cpu_ldst.h
index 71009f8..dac12bd 100644
--- a/include/exec/cpu_ldst.h
+++ b/include/exec/cpu_ldst.h
@@ -379,4 +379,38 @@ void *tlb_vaddr_to_host(CPUArchState *env, abi_ptr addr,
                         MMUAccessType access_type, int mmu_idx);
 #endif
 
+/*
+ * For user-only, helpers that use guest to host address translation
+ * must protect the actual host memory access by recording 'retaddr'
+ * for the signal handler.  This is required for a race condition in
+ * which another thread unmaps the page between a probe and the
+ * actual access.
+ */
+#ifdef CONFIG_USER_ONLY
+extern __thread uintptr_t helper_retaddr;
+
+static inline void set_helper_retaddr(uintptr_t ra)
+{
+    helper_retaddr = ra;
+    /*
+     * Ensure that this write is visible to the SIGSEGV handler that
+     * may be invoked due to a subsequent invalid memory operation.
+     */
+    signal_barrier();
+}
+
+static inline void clear_helper_retaddr(void)
+{
+    /*
+     * Ensure that previous memory operations have succeeded before
+     * removing the data visible to the signal handler.
+     */
+    signal_barrier();
+    helper_retaddr = 0;
+}
+#else
+#define set_helper_retaddr(ra)   do { } while (0)
+#define clear_helper_retaddr()   do { } while (0)
+#endif
+
 #endif /* CPU_LDST_H */
diff --git a/target/arm/tcg/helper-a64.c b/target/arm/tcg/helper-a64.c
index 0ea8668..c60d2a7 100644
--- a/target/arm/tcg/helper-a64.c
+++ b/target/arm/tcg/helper-a64.c
@@ -928,6 +928,8 @@ uint32_t HELPER(sqrt_f16)(uint32_t a, void *fpstp)
 
 void HELPER(dc_zva)(CPUARMState *env, uint64_t vaddr_in)
 {
+    uintptr_t ra = GETPC();
+
     /*
      * Implement DC ZVA, which zeroes a fixed-length block of memory.
      * Note that we do not implement the (architecturally mandated)
@@ -948,8 +950,6 @@ void HELPER(dc_zva)(CPUARMState *env, uint64_t vaddr_in)
 
 #ifndef CONFIG_USER_ONLY
     if (unlikely(!mem)) {
-        uintptr_t ra = GETPC();
-
         /*
          * Trap if accessing an invalid page.  DC_ZVA requires that we supply
          * the original pointer for an invalid page.  But watchpoints require
@@ -971,7 +971,9 @@ void HELPER(dc_zva)(CPUARMState *env, uint64_t vaddr_in)
     }
 #endif
 
+    set_helper_retaddr(ra);
     memset(mem, 0, blocklen);
+    clear_helper_retaddr();
 }
 
 void HELPER(unaligned_access)(CPUARMState *env, uint64_t addr,
@@ -1120,7 +1122,9 @@ static uint64_t set_step(CPUARMState *env, uint64_t toaddr,
     }
 #endif
     /* Easy case: just memset the host memory */
+    set_helper_retaddr(ra);
     memset(mem, data, setsize);
+    clear_helper_retaddr();
     return setsize;
 }
 
@@ -1163,7 +1167,9 @@ static uint64_t set_step_tags(CPUARMState *env, uint64_t toaddr,
     }
 #endif
     /* Easy case: just memset the host memory */
+    set_helper_retaddr(ra);
     memset(mem, data, setsize);
+    clear_helper_retaddr();
     mte_mops_set_tags(env, toaddr, setsize, *mtedesc);
     return setsize;
 }
@@ -1497,7 +1503,9 @@ static uint64_t copy_step(CPUARMState *env, uint64_t toaddr, uint64_t fromaddr,
     }
 #endif
     /* Easy case: just memmove the host memory */
+    set_helper_retaddr(ra);
     memmove(wmem, rmem, copysize);
+    clear_helper_retaddr();
     return copysize;
 }
 
@@ -1572,7 +1580,9 @@ static uint64_t copy_step_rev(CPUARMState *env, uint64_t toaddr,
      * Easy case: just memmove the host memory. Note that wmem and
      * rmem here point to the *last* byte to copy.
      */
+    set_helper_retaddr(ra);
     memmove(wmem - (copysize - 1), rmem - (copysize - 1), copysize);
+    clear_helper_retaddr();
     return copysize;
 }
 
diff --git a/target/arm/tcg/sme_helper.c b/target/arm/tcg/sme_helper.c
index 5a6dd76..50bb088 100644
--- a/target/arm/tcg/sme_helper.c
+++ b/target/arm/tcg/sme_helper.c
@@ -517,6 +517,8 @@ void sme_ld1(CPUARMState *env, void *za, uint64_t *vg,
         clr_fn(za, 0, reg_off);
     }
 
+    set_helper_retaddr(ra);
+
     while (reg_off <= reg_last) {
         uint64_t pg = vg[reg_off >> 6];
         do {
@@ -529,6 +531,8 @@ void sme_ld1(CPUARMState *env, void *za, uint64_t *vg,
         } while (reg_off <= reg_last && (reg_off & 63));
     }
 
+    clear_helper_retaddr();
+
     /*
      * Use the slow path to manage the cross-page misalignment.
      * But we know this is RAM and cannot trap.
@@ -543,6 +547,8 @@ void sme_ld1(CPUARMState *env, void *za, uint64_t *vg,
         reg_last = info.reg_off_last[1];
         host = info.page[1].host;
 
+        set_helper_retaddr(ra);
+
         do {
             uint64_t pg = vg[reg_off >> 6];
             do {
@@ -554,6 +560,8 @@ void sme_ld1(CPUARMState *env, void *za, uint64_t *vg,
                 reg_off += esize;
             } while (reg_off & 63);
         } while (reg_off <= reg_last);
+
+        clear_helper_retaddr();
     }
 }
 
@@ -701,6 +709,8 @@ void sme_st1(CPUARMState *env, void *za, uint64_t *vg,
     reg_last = info.reg_off_last[0];
     host = info.page[0].host;
 
+    set_helper_retaddr(ra);
+
     while (reg_off <= reg_last) {
         uint64_t pg = vg[reg_off >> 6];
         do {
@@ -711,6 +721,8 @@ void sme_st1(CPUARMState *env, void *za, uint64_t *vg,
         } while (reg_off <= reg_last && (reg_off & 63));
     }
 
+    clear_helper_retaddr();
+
     /*
      * Use the slow path to manage the cross-page misalignment.
      * But we know this is RAM and cannot trap.
@@ -725,6 +737,8 @@ void sme_st1(CPUARMState *env, void *za, uint64_t *vg,
         reg_last = info.reg_off_last[1];
         host = info.page[1].host;
 
+        set_helper_retaddr(ra);
+
         do {
             uint64_t pg = vg[reg_off >> 6];
             do {
@@ -734,6 +748,8 @@ void sme_st1(CPUARMState *env, void *za, uint64_t *vg,
                 reg_off += 1 << esz;
             } while (reg_off & 63);
         } while (reg_off <= reg_last);
+
+        clear_helper_retaddr();
     }
 }
 
diff --git a/target/arm/tcg/sve_helper.c b/target/arm/tcg/sve_helper.c
index dd49e67..f1ee0e0 100644
--- a/target/arm/tcg/sve_helper.c
+++ b/target/arm/tcg/sve_helper.c
@@ -5738,6 +5738,8 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
     reg_last = info.reg_off_last[0];
     host = info.page[0].host;
 
+    set_helper_retaddr(retaddr);
+
     while (reg_off <= reg_last) {
         uint64_t pg = vg[reg_off >> 6];
         do {
@@ -5752,6 +5754,8 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
         } while (reg_off <= reg_last && (reg_off & 63));
     }
 
+    clear_helper_retaddr();
+
     /*
      * Use the slow path to manage the cross-page misalignment.
      * But we know this is RAM and cannot trap.
@@ -5771,6 +5775,8 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
         reg_last = info.reg_off_last[1];
         host = info.page[1].host;
 
+        set_helper_retaddr(retaddr);
+
         do {
             uint64_t pg = vg[reg_off >> 6];
             do {
@@ -5784,6 +5790,8 @@ void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr,
                 mem_off += N << msz;
             } while (reg_off & 63);
         } while (reg_off <= reg_last);
+
+        clear_helper_retaddr();
     }
 }
 
@@ -5934,15 +5942,11 @@ DO_LDN_2(4, dd, MO_64)
 /*
  * Load contiguous data, first-fault and no-fault.
  *
- * For user-only, one could argue that we should hold the mmap_lock during
- * the operation so that there is no race between page_check_range and the
- * load operation.  However, unmapping pages out from under a running thread
- * is extraordinarily unlikely.  This theoretical race condition also affects
- * linux-user/ in its get_user/put_user macros.
- *
- * TODO: Construct some helpers, written in assembly, that interact with
- * host_signal_handler to produce memory ops which can properly report errors
- * without racing.
+ * For user-only, we control the race between page_check_range and
+ * another thread's munmap by using set/clear_helper_retaddr.  Any
+ * SEGV that occurs between those markers is assumed to be because
+ * the guest page vanished.  Keep that block as small as possible
+ * so that unrelated QEMU bugs are not blamed on the guest.
  */
 
 /* Fault on byte I.  All bits in FFR from I are cleared.  The vector
@@ -6093,6 +6097,8 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
     reg_last = info.reg_off_last[0];
     host = info.page[0].host;
 
+    set_helper_retaddr(retaddr);
+
     do {
         uint64_t pg = *(uint64_t *)(vg + (reg_off >> 3));
         do {
@@ -6101,9 +6107,11 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
                     (cpu_watchpoint_address_matches
                      (env_cpu(env), addr + mem_off, 1 << msz)
                      & BP_MEM_READ)) {
+                    clear_helper_retaddr();
                     goto do_fault;
                 }
                 if (mtedesc && !mte_probe(env, mtedesc, addr + mem_off)) {
+                    clear_helper_retaddr();
                     goto do_fault;
                 }
                 host_fn(vd, reg_off, host + mem_off);
@@ -6113,6 +6121,8 @@ void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr,
         } while (reg_off <= reg_last && (reg_off & 63));
     } while (reg_off <= reg_last);
 
+    clear_helper_retaddr();
+
     /*
      * MemSingleNF is allowed to fail for any reason.  We have special
      * code above to handle the first element crossing a page boundary.
@@ -6348,6 +6358,8 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr,
     reg_last = info.reg_off_last[0];
     host = info.page[0].host;
 
+    set_helper_retaddr(retaddr);
+
     while (reg_off <= reg_last) {
         uint64_t pg = vg[reg_off >> 6];
         do {
@@ -6362,6 +6374,8 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr,
         } while (reg_off <= reg_last && (reg_off & 63));
     }
 
+    clear_helper_retaddr();
+
     /*
      * Use the slow path to manage the cross-page misalignment.
      * But we know this is RAM and cannot trap.
@@ -6381,6 +6395,8 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr,
         reg_last = info.reg_off_last[1];
         host = info.page[1].host;
 
+        set_helper_retaddr(retaddr);
+
         do {
             uint64_t pg = vg[reg_off >> 6];
             do {
@@ -6394,6 +6410,8 @@ void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr,
                 mem_off += N << msz;
             } while (reg_off & 63);
         } while (reg_off <= reg_last);
+
+        clear_helper_retaddr();
     }
 }
 
@@ -6560,7 +6578,9 @@ void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                     if (unlikely(info.flags & TLB_MMIO)) {
                         tlb_fn(env, &scratch, reg_off, addr, retaddr);
                     } else {
+                        set_helper_retaddr(retaddr);
                         host_fn(&scratch, reg_off, info.host);
+                        clear_helper_retaddr();
                     }
                 } else {
                     /* Element crosses the page boundary. */
@@ -6782,7 +6802,9 @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
                     goto fault;
                 }
 
+                set_helper_retaddr(retaddr);
                 host_fn(vd, reg_off, info.host);
+                clear_helper_retaddr();
             }
             reg_off += esize;
         } while (reg_off & 63);
@@ -6986,7 +7008,9 @@ void sve_st1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm,
     do {
         void *h = host[i];
         if (likely(h != NULL)) {
+            set_helper_retaddr(retaddr);
             host_fn(vd, reg_off, h);
+            clear_helper_retaddr();
         } else if ((vg[reg_off >> 6] >> (reg_off & 63)) & 1) {
             target_ulong addr = base + (off_fn(vm, reg_off) << scale);
             tlb_fn(env, vd, reg_off, addr, retaddr);
diff --git a/target/ppc/helper.h b/target/ppc/helper.h
index 76b8f25..4fa089c 100644
--- a/target/ppc/helper.h
+++ b/target/ppc/helper.h
@@ -46,8 +46,10 @@ DEF_HELPER_FLAGS_3(stmw, TCG_CALL_NO_WG, void, env, tl, i32)
 DEF_HELPER_4(lsw, void, env, tl, i32, i32)
 DEF_HELPER_5(lswx, void, env, tl, i32, i32, i32)
 DEF_HELPER_FLAGS_4(stsw, TCG_CALL_NO_WG, void, env, tl, i32, i32)
-DEF_HELPER_FLAGS_3(dcbz, TCG_CALL_NO_WG, void, env, tl, i32)
-DEF_HELPER_FLAGS_3(dcbzep, TCG_CALL_NO_WG, void, env, tl, i32)
+DEF_HELPER_FLAGS_3(dcbz, TCG_CALL_NO_WG, void, env, tl, int)
+#ifdef TARGET_PPC64
+DEF_HELPER_FLAGS_2(dcbzl, TCG_CALL_NO_WG, void, env, tl)
+#endif
 DEF_HELPER_FLAGS_2(icbi, TCG_CALL_NO_WG, void, env, tl)
 DEF_HELPER_FLAGS_2(icbiep, TCG_CALL_NO_WG, void, env, tl)
 DEF_HELPER_5(lscbx, tl, env, tl, i32, i32, i32)
diff --git a/target/ppc/mem_helper.c b/target/ppc/mem_helper.c
index f88155a..953dd08 100644
--- a/target/ppc/mem_helper.c
+++ b/target/ppc/mem_helper.c
@@ -271,51 +271,59 @@ void helper_stsw(CPUPPCState *env, target_ulong addr, uint32_t nb,
 }
 
 static void dcbz_common(CPUPPCState *env, target_ulong addr,
-                        uint32_t opcode, bool epid, uintptr_t retaddr)
+                        int mmu_idx, int dcbz_size, uintptr_t retaddr)
 {
-    target_ulong mask, dcbz_size = env->dcache_line_size;
-    uint32_t i;
+    target_ulong mask = ~(target_ulong)(dcbz_size - 1);
     void *haddr;
-    int mmu_idx = epid ? PPC_TLB_EPID_STORE : ppc_env_mmu_index(env, false);
-
-#if defined(TARGET_PPC64)
-    /* Check for dcbz vs dcbzl on 970 */
-    if (env->excp_model == POWERPC_EXCP_970 &&
-        !(opcode & 0x00200000) && ((env->spr[SPR_970_HID5] >> 7) & 0x3) == 1) {
-        dcbz_size = 32;
-    }
-#endif
 
     /* Align address */
-    mask = ~(dcbz_size - 1);
     addr &= mask;
 
     /* Check reservation */
-    if ((env->reserve_addr & mask) == addr)  {
+    if (unlikely((env->reserve_addr & mask) == addr))  {
         env->reserve_addr = (target_ulong)-1ULL;
     }
 
     /* Try fast path translate */
+#ifdef CONFIG_USER_ONLY
+    haddr = tlb_vaddr_to_host(env, addr, MMU_DATA_STORE, mmu_idx);
+#else
     haddr = probe_write(env, addr, dcbz_size, mmu_idx, retaddr);
-    if (haddr) {
-        memset(haddr, 0, dcbz_size);
-    } else {
+    if (unlikely(!haddr)) {
         /* Slow path */
-        for (i = 0; i < dcbz_size; i += 8) {
+        for (int i = 0; i < dcbz_size; i += 8) {
             cpu_stq_mmuidx_ra(env, addr + i, 0, mmu_idx, retaddr);
         }
+        return;
     }
+#endif
+
+    set_helper_retaddr(retaddr);
+    memset(haddr, 0, dcbz_size);
+    clear_helper_retaddr();
 }
 
-void helper_dcbz(CPUPPCState *env, target_ulong addr, uint32_t opcode)
+void helper_dcbz(CPUPPCState *env, target_ulong addr, int mmu_idx)
 {
-    dcbz_common(env, addr, opcode, false, GETPC());
+    dcbz_common(env, addr, mmu_idx, env->dcache_line_size, GETPC());
 }
 
-void helper_dcbzep(CPUPPCState *env, target_ulong addr, uint32_t opcode)
+#ifdef TARGET_PPC64
+void helper_dcbzl(CPUPPCState *env, target_ulong addr)
 {
-    dcbz_common(env, addr, opcode, true, GETPC());
+    int dcbz_size = env->dcache_line_size;
+
+    /*
+     * The translator checked for POWERPC_EXCP_970.
+     * All that's left is to check HID5.
+     */
+    if (((env->spr[SPR_970_HID5] >> 7) & 0x3) == 1) {
+        dcbz_size = 32;
+    }
+
+    dcbz_common(env, addr, ppc_env_mmu_index(env, false), dcbz_size, GETPC());
 }
+#endif
 
 void helper_icbi(CPUPPCState *env, target_ulong addr)
 {
diff --git a/target/ppc/translate.c b/target/ppc/translate.c
index 0bc16d7..cba943a 100644
--- a/target/ppc/translate.c
+++ b/target/ppc/translate.c
@@ -178,6 +178,7 @@ struct DisasContext {
     /* Translation flags */
     MemOp default_tcg_memop_mask;
 #if defined(TARGET_PPC64)
+    powerpc_excp_t excp_model;
     bool sf_mode;
     bool has_cfar;
     bool has_bhrb;
@@ -4445,27 +4446,29 @@ static void gen_dcblc(DisasContext *ctx)
 /* dcbz */
 static void gen_dcbz(DisasContext *ctx)
 {
-    TCGv tcgv_addr;
-    TCGv_i32 tcgv_op;
+    TCGv tcgv_addr = tcg_temp_new();
 
     gen_set_access_type(ctx, ACCESS_CACHE);
-    tcgv_addr = tcg_temp_new();
-    tcgv_op = tcg_constant_i32(ctx->opcode & 0x03FF000);
     gen_addr_reg_index(ctx, tcgv_addr);
-    gen_helper_dcbz(tcg_env, tcgv_addr, tcgv_op);
+
+#ifdef TARGET_PPC64
+    if (ctx->excp_model == POWERPC_EXCP_970 && !(ctx->opcode & 0x00200000)) {
+        gen_helper_dcbzl(tcg_env, tcgv_addr);
+        return;
+    }
+#endif
+
+    gen_helper_dcbz(tcg_env, tcgv_addr, tcg_constant_i32(ctx->mem_idx));
 }
 
 /* dcbzep */
 static void gen_dcbzep(DisasContext *ctx)
 {
-    TCGv tcgv_addr;
-    TCGv_i32 tcgv_op;
+    TCGv tcgv_addr = tcg_temp_new();
 
     gen_set_access_type(ctx, ACCESS_CACHE);
-    tcgv_addr = tcg_temp_new();
-    tcgv_op = tcg_constant_i32(ctx->opcode & 0x03FF000);
     gen_addr_reg_index(ctx, tcgv_addr);
-    gen_helper_dcbzep(tcg_env, tcgv_addr, tcgv_op);
+    gen_helper_dcbz(tcg_env, tcgv_addr, tcg_constant_i32(PPC_TLB_EPID_STORE));
 }
 
 /* dst / dstt */
@@ -6486,6 +6489,7 @@ static void ppc_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
     ctx->default_tcg_memop_mask = ctx->le_mode ? MO_LE : MO_BE;
     ctx->flags = env->flags;
 #if defined(TARGET_PPC64)
+    ctx->excp_model = env->excp_model;
     ctx->sf_mode = (hflags >> HFLAGS_64) & 1;
     ctx->has_cfar = !!(env->flags & POWERPC_FLAG_CFAR);
     ctx->has_bhrb = !!(env->flags & POWERPC_FLAG_BHRB);
diff --git a/target/riscv/vector_helper.c b/target/riscv/vector_helper.c
index 1b4d5a8..10a52ce 100644
--- a/target/riscv/vector_helper.c
+++ b/target/riscv/vector_helper.c
@@ -474,7 +474,6 @@ vext_ldff(void *vd, void *v0, target_ulong base,
           vext_ldst_elem_fn *ldst_elem,
           uint32_t log2_esz, uintptr_t ra)
 {
-    void *host;
     uint32_t i, k, vl = 0;
     uint32_t nf = vext_nf(desc);
     uint32_t vm = vext_vm(desc);
@@ -493,27 +492,31 @@ vext_ldff(void *vd, void *v0, target_ulong base,
         }
         addr = adjust_addr(env, base + i * (nf << log2_esz));
         if (i == 0) {
+            /* Allow fault on first element. */
             probe_pages(env, addr, nf << log2_esz, ra, MMU_DATA_LOAD);
         } else {
-            /* if it triggers an exception, no need to check watchpoint */
             remain = nf << log2_esz;
             while (remain > 0) {
+                void *host;
+                int flags;
+
                 offset = -(addr | TARGET_PAGE_MASK);
-                host = tlb_vaddr_to_host(env, addr, MMU_DATA_LOAD, mmu_index);
-                if (host) {
-#ifdef CONFIG_USER_ONLY
-                    if (!page_check_range(addr, offset, PAGE_READ)) {
-                        vl = i;
-                        goto ProbeSuccess;
-                    }
-#else
-                    probe_pages(env, addr, offset, ra, MMU_DATA_LOAD);
-#endif
-                } else {
+
+                /* Probe nonfault on subsequent elements. */
+                flags = probe_access_flags(env, addr, offset, MMU_DATA_LOAD,
+                                           mmu_index, true, &host, 0);
+
+                /*
+                 * Stop if invalid (unmapped) or mmio (transaction may fail).
+                 * Do not stop if watchpoint, as the spec says that
+                 * first-fault should continue to access the same
+                 * elements regardless of any watchpoint.
+                 */
+                if (flags & ~TLB_WATCHPOINT) {
                     vl = i;
                     goto ProbeSuccess;
                 }
-                if (remain <=  offset) {
+                if (remain <= offset) {
                     break;
                 }
                 remain -= offset;
diff --git a/target/s390x/tcg/mem_helper.c b/target/s390x/tcg/mem_helper.c
index 6cdbc34..0e12dae 100644
--- a/target/s390x/tcg/mem_helper.c
+++ b/target/s390x/tcg/mem_helper.c
@@ -225,10 +225,7 @@ static void do_access_memset(CPUS390XState *env, vaddr vaddr, char *haddr,
                              uint8_t byte, uint16_t size, int mmu_idx,
                              uintptr_t ra)
 {
-#ifdef CONFIG_USER_ONLY
-    memset(haddr, byte, size);
-#else
-    if (likely(haddr)) {
+    if (user_or_likely(haddr)) {
         memset(haddr, byte, size);
     } else {
         MemOpIdx oi = make_memop_idx(MO_UB, mmu_idx);
@@ -236,20 +233,19 @@ static void do_access_memset(CPUS390XState *env, vaddr vaddr, char *haddr,
             cpu_stb_mmu(env, vaddr + i, byte, oi, ra);
         }
     }
-#endif
 }
 
 static void access_memset(CPUS390XState *env, S390Access *desta,
                           uint8_t byte, uintptr_t ra)
 {
-
+    set_helper_retaddr(ra);
     do_access_memset(env, desta->vaddr1, desta->haddr1, byte, desta->size1,
                      desta->mmu_idx, ra);
-    if (likely(!desta->size2)) {
-        return;
+    if (unlikely(desta->size2)) {
+        do_access_memset(env, desta->vaddr2, desta->haddr2, byte,
+                         desta->size2, desta->mmu_idx, ra);
     }
-    do_access_memset(env, desta->vaddr2, desta->haddr2, byte, desta->size2,
-                     desta->mmu_idx, ra);
+    clear_helper_retaddr();
 }
 
 static uint8_t access_get_byte(CPUS390XState *env, S390Access *access,
@@ -300,41 +296,39 @@ static void access_memmove(CPUS390XState *env, S390Access *desta,
                            S390Access *srca, uintptr_t ra)
 {
     int len = desta->size1 + desta->size2;
-    int diff;
 
     assert(len == srca->size1 + srca->size2);
 
     /* Fallback to slow access in case we don't have access to all host pages */
-    if (unlikely(!desta->haddr1 || (desta->size2 && !desta->haddr2) ||
-                 !srca->haddr1 || (srca->size2 && !srca->haddr2))) {
-        int i;
-
-        for (i = 0; i < len; i++) {
-            uint8_t byte = access_get_byte(env, srca, i, ra);
-
-            access_set_byte(env, desta, i, byte, ra);
-        }
-        return;
-    }
-
-    diff = desta->size1 - srca->size1;
-    if (likely(diff == 0)) {
-        memmove(desta->haddr1, srca->haddr1, srca->size1);
-        if (unlikely(srca->size2)) {
-            memmove(desta->haddr2, srca->haddr2, srca->size2);
-        }
-    } else if (diff > 0) {
-        memmove(desta->haddr1, srca->haddr1, srca->size1);
-        memmove(desta->haddr1 + srca->size1, srca->haddr2, diff);
-        if (likely(desta->size2)) {
-            memmove(desta->haddr2, srca->haddr2 + diff, desta->size2);
+    if (user_or_likely(desta->haddr1 &&
+                       srca->haddr1 &&
+                       (!desta->size2 || desta->haddr2) &&
+                       (!srca->size2 || srca->haddr2))) {
+        int diff = desta->size1 - srca->size1;
+
+        if (likely(diff == 0)) {
+            memmove(desta->haddr1, srca->haddr1, srca->size1);
+            if (unlikely(srca->size2)) {
+                memmove(desta->haddr2, srca->haddr2, srca->size2);
+            }
+        } else if (diff > 0) {
+            memmove(desta->haddr1, srca->haddr1, srca->size1);
+            memmove(desta->haddr1 + srca->size1, srca->haddr2, diff);
+            if (likely(desta->size2)) {
+                memmove(desta->haddr2, srca->haddr2 + diff, desta->size2);
+            }
+        } else {
+            diff = -diff;
+            memmove(desta->haddr1, srca->haddr1, desta->size1);
+            memmove(desta->haddr2, srca->haddr1 + desta->size1, diff);
+            if (likely(srca->size2)) {
+                memmove(desta->haddr2 + diff, srca->haddr2, srca->size2);
+            }
         }
     } else {
-        diff = -diff;
-        memmove(desta->haddr1, srca->haddr1, desta->size1);
-        memmove(desta->haddr2, srca->haddr1 + desta->size1, diff);
-        if (likely(srca->size2)) {
-            memmove(desta->haddr2 + diff, srca->haddr2, srca->size2);
+        for (int i = 0; i < len; i++) {
+            uint8_t byte = access_get_byte(env, srca, i, ra);
+            access_set_byte(env, desta, i, byte, ra);
         }
     }
 }
@@ -372,6 +366,8 @@ static uint32_t do_helper_nc(CPUS390XState *env, uint32_t l, uint64_t dest,
     access_prepare(&srca1, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&srca2, env, dest, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
+    set_helper_retaddr(ra);
+
     for (i = 0; i < l; i++) {
         const uint8_t x = access_get_byte(env, &srca1, i, ra) &
                           access_get_byte(env, &srca2, i, ra);
@@ -379,6 +375,8 @@ static uint32_t do_helper_nc(CPUS390XState *env, uint32_t l, uint64_t dest,
         c |= x;
         access_set_byte(env, &desta, i, x, ra);
     }
+
+    clear_helper_retaddr();
     return c != 0;
 }
 
@@ -413,6 +411,7 @@ static uint32_t do_helper_xc(CPUS390XState *env, uint32_t l, uint64_t dest,
         return 0;
     }
 
+    set_helper_retaddr(ra);
     for (i = 0; i < l; i++) {
         const uint8_t x = access_get_byte(env, &srca1, i, ra) ^
                           access_get_byte(env, &srca2, i, ra);
@@ -420,6 +419,7 @@ static uint32_t do_helper_xc(CPUS390XState *env, uint32_t l, uint64_t dest,
         c |= x;
         access_set_byte(env, &desta, i, x, ra);
     }
+    clear_helper_retaddr();
     return c != 0;
 }
 
@@ -447,6 +447,8 @@ static uint32_t do_helper_oc(CPUS390XState *env, uint32_t l, uint64_t dest,
     access_prepare(&srca1, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&srca2, env, dest, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
+    set_helper_retaddr(ra);
+
     for (i = 0; i < l; i++) {
         const uint8_t x = access_get_byte(env, &srca1, i, ra) |
                           access_get_byte(env, &srca2, i, ra);
@@ -454,6 +456,8 @@ static uint32_t do_helper_oc(CPUS390XState *env, uint32_t l, uint64_t dest,
         c |= x;
         access_set_byte(env, &desta, i, x, ra);
     }
+
+    clear_helper_retaddr();
     return c != 0;
 }
 
@@ -490,11 +494,13 @@ static uint32_t do_helper_mvc(CPUS390XState *env, uint32_t l, uint64_t dest,
     } else if (!is_destructive_overlap(env, dest, src, l)) {
         access_memmove(env, &desta, &srca, ra);
     } else {
+        set_helper_retaddr(ra);
         for (i = 0; i < l; i++) {
             uint8_t byte = access_get_byte(env, &srca, i, ra);
 
             access_set_byte(env, &desta, i, byte, ra);
         }
+        clear_helper_retaddr();
     }
 
     return env->cc_op;
@@ -520,10 +526,12 @@ void HELPER(mvcrl)(CPUS390XState *env, uint64_t l, uint64_t dest, uint64_t src)
     access_prepare(&srca, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
 
+    set_helper_retaddr(ra);
     for (i = l - 1; i >= 0; i--) {
         uint8_t byte = access_get_byte(env, &srca, i, ra);
         access_set_byte(env, &desta, i, byte, ra);
     }
+    clear_helper_retaddr();
 }
 
 /* move inverse  */
@@ -540,11 +548,13 @@ void HELPER(mvcin)(CPUS390XState *env, uint32_t l, uint64_t dest, uint64_t src)
     src = wrap_address(env, src - l + 1);
     access_prepare(&srca, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
+
+    set_helper_retaddr(ra);
     for (i = 0; i < l; i++) {
         const uint8_t x = access_get_byte(env, &srca, l - i - 1, ra);
-
         access_set_byte(env, &desta, i, x, ra);
     }
+    clear_helper_retaddr();
 }
 
 /* move numerics  */
@@ -561,12 +571,15 @@ void HELPER(mvn)(CPUS390XState *env, uint32_t l, uint64_t dest, uint64_t src)
     access_prepare(&srca1, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&srca2, env, dest, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
+
+    set_helper_retaddr(ra);
     for (i = 0; i < l; i++) {
         const uint8_t x = (access_get_byte(env, &srca1, i, ra) & 0x0f) |
                           (access_get_byte(env, &srca2, i, ra) & 0xf0);
 
         access_set_byte(env, &desta, i, x, ra);
     }
+    clear_helper_retaddr();
 }
 
 /* move with offset  */
@@ -586,6 +599,8 @@ void HELPER(mvo)(CPUS390XState *env, uint32_t l, uint64_t dest, uint64_t src)
 
     /* Handle rightmost byte */
     byte_dest = cpu_ldub_data_ra(env, dest + len_dest - 1, ra);
+
+    set_helper_retaddr(ra);
     byte_src = access_get_byte(env, &srca, len_src - 1, ra);
     byte_dest = (byte_dest & 0x0f) | (byte_src << 4);
     access_set_byte(env, &desta, len_dest - 1, byte_dest, ra);
@@ -601,6 +616,7 @@ void HELPER(mvo)(CPUS390XState *env, uint32_t l, uint64_t dest, uint64_t src)
         byte_dest |= byte_src << 4;
         access_set_byte(env, &desta, i, byte_dest, ra);
     }
+    clear_helper_retaddr();
 }
 
 /* move zones  */
@@ -617,12 +633,15 @@ void HELPER(mvz)(CPUS390XState *env, uint32_t l, uint64_t dest, uint64_t src)
     access_prepare(&srca1, env, src, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&srca2, env, dest, l, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, dest, l, MMU_DATA_STORE, mmu_idx, ra);
+
+    set_helper_retaddr(ra);
     for (i = 0; i < l; i++) {
         const uint8_t x = (access_get_byte(env, &srca1, i, ra) & 0xf0) |
                           (access_get_byte(env, &srca2, i, ra) & 0x0f);
 
         access_set_byte(env, &desta, i, x, ra);
     }
+    clear_helper_retaddr();
 }
 
 /* compare unsigned byte arrays */
@@ -967,15 +986,19 @@ uint32_t HELPER(mvst)(CPUS390XState *env, uint32_t r1, uint32_t r2)
      */
     access_prepare(&srca, env, s, len, MMU_DATA_LOAD, mmu_idx, ra);
     access_prepare(&desta, env, d, len, MMU_DATA_STORE, mmu_idx, ra);
+
+    set_helper_retaddr(ra);
     for (i = 0; i < len; i++) {
         const uint8_t v = access_get_byte(env, &srca, i, ra);
 
         access_set_byte(env, &desta, i, v, ra);
         if (v == c) {
+            clear_helper_retaddr();
             set_address_zero(env, r1, d + i);
             return 1;
         }
     }
+    clear_helper_retaddr();
     set_address_zero(env, r1, d + len);
     set_address_zero(env, r2, s + len);
     return 3;
@@ -1066,6 +1089,7 @@ static inline uint32_t do_mvcl(CPUS390XState *env,
         *dest = wrap_address(env, *dest + len);
     } else {
         access_prepare(&desta, env, *dest, len, MMU_DATA_STORE, mmu_idx, ra);
+        set_helper_retaddr(ra);
 
         /* The remaining length selects the padding byte. */
         for (i = 0; i < len; (*destlen)--, i++) {
@@ -1075,6 +1099,7 @@ static inline uint32_t do_mvcl(CPUS390XState *env,
                 access_set_byte(env, &desta, i, pad >> 8, ra);
             }
         }
+        clear_helper_retaddr();
         *dest = wrap_address(env, *dest + len);
     }