From ba1a6723f58640ba281bc952abc255e97c70bad5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20Li=C5=A1ka?= <mliska@suse.cz>
Date: Fri, 24 Jun 2022 15:42:55 +0100
Subject: [PATCH 01/25] sphinx: change default language to 'en'

Fixes the following Sphinx warning (treated as error) starting
with 5.0 release:

Warning, treated as error:
Invalid configuration value found: 'language = None'. Update your configuration to a valid langauge code. Falling back to 'en' (English).

Signed-off-by: Martin Liska <mliska@suse.cz>
Message-id: e91e51ee-48ac-437e-6467-98b56ee40042@suse.cz
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 docs/conf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/conf.py b/docs/conf.py
index 49dab44cca..e33cf3d381 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -126,7 +126,7 @@ finally:
 #
 # This is also used if you do content translation via gettext catalogs.
 # Usually you set "language" from the command line for these cases.
-language = None
+language = 'en'
 
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.

From 55bd445c4195966675236936c0a8642f1965dddb Mon Sep 17 00:00:00 2001
From: Alexander Graf <agraf@csgraf.de>
Date: Fri, 24 Jun 2022 15:42:56 +0100
Subject: [PATCH 02/25] accel: Introduce current_accel_name()

We need to fetch the name of the current accelerator in flexible error
messages more going forward. Let's create a helper that gives it to us
without casting in the target code.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620192242.70573-1-agraf@csgraf.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 accel/accel-common.c | 8 ++++++++
 include/qemu/accel.h | 1 +
 softmmu/vl.c         | 3 +--
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/accel/accel-common.c b/accel/accel-common.c
index 7b8ec7e0f7..50035bda55 100644
--- a/accel/accel-common.c
+++ b/accel/accel-common.c
@@ -49,6 +49,14 @@ AccelClass *accel_find(const char *opt_name)
     return ac;
 }
 
+/* Return the name of the current accelerator */
+const char *current_accel_name(void)
+{
+    AccelClass *ac = ACCEL_GET_CLASS(current_accel());
+
+    return ac->name;
+}
+
 static void accel_init_cpu_int_aux(ObjectClass *klass, void *opaque)
 {
     CPUClass *cc = CPU_CLASS(klass);
diff --git a/include/qemu/accel.h b/include/qemu/accel.h
index 4f4c283f6f..be56da1b99 100644
--- a/include/qemu/accel.h
+++ b/include/qemu/accel.h
@@ -68,6 +68,7 @@ typedef struct AccelClass {
 
 AccelClass *accel_find(const char *opt_name);
 AccelState *current_accel(void);
+const char *current_accel_name(void);
 
 void accel_init_interfaces(AccelClass *ac);
 
diff --git a/softmmu/vl.c b/softmmu/vl.c
index 54e920ada1..3dca5936c7 100644
--- a/softmmu/vl.c
+++ b/softmmu/vl.c
@@ -2271,8 +2271,7 @@ static void configure_accelerators(const char *progname)
     }
 
     if (init_failed && !qtest_chrdev) {
-        AccelClass *ac = ACCEL_GET_CLASS(current_accel());
-        error_report("falling back to %s", ac->name);
+        error_report("falling back to %s", current_accel_name());
     }
 
     if (icount_enabled() && !tcg_enabled()) {

From 045e50641fd17655b02c4af485835bca38577bf3 Mon Sep 17 00:00:00 2001
From: Alexander Graf <agraf@csgraf.de>
Date: Fri, 24 Jun 2022 15:42:56 +0100
Subject: [PATCH 03/25] target/arm: Catch invalid kvm state also for hvf

Some features such as running in EL3 or running M profile code are
incompatible with virtualization as QEMU implements it today. To prevent
users from picking invalid configurations on other virt solutions like
Hvf, let's run the same checks there too.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1073
Signed-off-by: Alexander Graf <agraf@csgraf.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620192242.70573-2-agraf@csgraf.de
[PMM: Allow qtest accelerator too; tweak comment]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 1b5d535788..d9c4a9f56d 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -39,6 +39,7 @@
 #include "hw/boards.h"
 #endif
 #include "sysemu/tcg.h"
+#include "sysemu/qtest.h"
 #include "sysemu/hw_accel.h"
 #include "kvm_arm.h"
 #include "disas/capstone.h"
@@ -1490,25 +1491,32 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
         }
     }
 
-    if (kvm_enabled()) {
+    if (!tcg_enabled() && !qtest_enabled()) {
         /*
+         * We assume that no accelerator except TCG (and the "not really an
+         * accelerator" qtest) can handle these features, because Arm hardware
+         * virtualization can't virtualize them.
+         *
          * Catch all the cases which might cause us to create more than one
          * address space for the CPU (otherwise we will assert() later in
          * cpu_address_space_init()).
          */
         if (arm_feature(env, ARM_FEATURE_M)) {
             error_setg(errp,
-                       "Cannot enable KVM when using an M-profile guest CPU");
+                       "Cannot enable %s when using an M-profile guest CPU",
+                       current_accel_name());
             return;
         }
         if (cpu->has_el3) {
             error_setg(errp,
-                       "Cannot enable KVM when guest CPU has EL3 enabled");
+                       "Cannot enable %s when guest CPU has EL3 enabled",
+                       current_accel_name());
             return;
         }
         if (cpu->tag_memory) {
             error_setg(errp,
-                       "Cannot enable KVM when guest CPUs has MTE enabled");
+                       "Cannot enable %s when guest CPUs has MTE enabled",
+                       current_accel_name());
             return;
         }
     }

From 9e5ec745e33ad6ace7a01653262dcedf1a5676d3 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:45 -0700
Subject: [PATCH 04/25] target/arm: Implement TPIDR2_EL0

This register is part of SME, but isn't closely related to the
rest of the extension.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h    |  1 +
 target/arm/helper.c | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 33 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index df677b2d5d..05d1e2e8dd 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -474,6 +474,7 @@ typedef struct CPUArchState {
             };
             uint64_t tpidr_el[4];
         };
+        uint64_t tpidr2_el0;
         /* The secure banks of these registers don't map anywhere */
         uint64_t tpidrurw_s;
         uint64_t tpidrprw_s;
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 6457e6301c..d21ba7ab83 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6279,6 +6279,35 @@ static const ARMCPRegInfo zcr_reginfo[] = {
       .writefn = zcr_write, .raw_writefn = raw_write },
 };
 
+#ifdef TARGET_AARCH64
+static CPAccessResult access_tpidr2(CPUARMState *env, const ARMCPRegInfo *ri,
+                                    bool isread)
+{
+    int el = arm_current_el(env);
+
+    if (el == 0) {
+        uint64_t sctlr = arm_sctlr(env, el);
+        if (!(sctlr & SCTLR_EnTP2)) {
+            return CP_ACCESS_TRAP;
+        }
+    }
+    /* TODO: FEAT_FGT */
+    if (el < 3
+        && arm_feature(env, ARM_FEATURE_EL3)
+        && !(env->cp15.scr_el3 & SCR_ENTP2)) {
+        return CP_ACCESS_TRAP_EL3;
+    }
+    return CP_ACCESS_OK;
+}
+
+static const ARMCPRegInfo sme_reginfo[] = {
+    { .name = "TPIDR2_EL0", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 3, .crn = 13, .crm = 0, .opc2 = 5,
+      .access = PL0_RW, .accessfn = access_tpidr2,
+      .fieldoffset = offsetof(CPUARMState, cp15.tpidr2_el0) },
+};
+#endif /* TARGET_AARCH64 */
+
 void hw_watchpoint_update(ARMCPU *cpu, int n)
 {
     CPUARMState *env = &cpu->env;
@@ -8440,6 +8469,9 @@ void register_cp_regs_for_features(ARMCPU *cpu)
     }
 
 #ifdef TARGET_AARCH64
+    if (cpu_isar_feature(aa64_sme, cpu)) {
+        define_arm_cp_regs(cpu, sme_reginfo);
+    }
     if (cpu_isar_feature(aa64_pauth, cpu)) {
         define_arm_cp_regs(cpu, pauth_reginfo);
     }

From 6b2ca83e4c7953e6652e897bb4d4fb5280555dbf Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:46 -0700
Subject: [PATCH 05/25] target/arm: Add SMEEXC_EL to TB flags

This is CheckSMEAccess, which is the basis for a set of
related tests for various SME cpregs and instructions.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h           |  2 ++
 target/arm/helper.c        | 52 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c |  1 +
 target/arm/translate.h     |  1 +
 4 files changed, 56 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 05d1e2e8dd..e99de18097 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -1134,6 +1134,7 @@ void aarch64_sync_64_to_32(CPUARMState *env);
 
 int fp_exception_el(CPUARMState *env, int cur_el);
 int sve_exception_el(CPUARMState *env, int cur_el);
+int sme_exception_el(CPUARMState *env, int cur_el);
 
 /**
  * sve_vqm1_for_el:
@@ -3148,6 +3149,7 @@ 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)
+FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2)
 
 /*
  * Helpers for using the above.
diff --git a/target/arm/helper.c b/target/arm/helper.c
index d21ba7ab83..2c080c6cac 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6218,6 +6218,55 @@ int sve_exception_el(CPUARMState *env, int el)
     return 0;
 }
 
+/*
+ * Return the exception level to which exceptions should be taken for SME.
+ * C.f. the ARM pseudocode function CheckSMEAccess.
+ */
+int sme_exception_el(CPUARMState *env, int el)
+{
+#ifndef CONFIG_USER_ONLY
+    if (el <= 1 && !el_is_in_host(env, el)) {
+        switch (FIELD_EX64(env->cp15.cpacr_el1, CPACR_EL1, SMEN)) {
+        case 1:
+            if (el != 0) {
+                break;
+            }
+            /* fall through */
+        case 0:
+        case 2:
+            return 1;
+        }
+    }
+
+    if (el <= 2 && arm_is_el2_enabled(env)) {
+        /* CPTR_EL2 changes format with HCR_EL2.E2H (regardless of TGE). */
+        if (env->cp15.hcr_el2 & HCR_E2H) {
+            switch (FIELD_EX64(env->cp15.cptr_el[2], CPTR_EL2, SMEN)) {
+            case 1:
+                if (el != 0 || !(env->cp15.hcr_el2 & HCR_TGE)) {
+                    break;
+                }
+                /* fall through */
+            case 0:
+            case 2:
+                return 2;
+            }
+        } else {
+            if (FIELD_EX64(env->cp15.cptr_el[2], CPTR_EL2, TSM)) {
+                return 2;
+            }
+        }
+    }
+
+    /* CPTR_EL3.  Since ESM is negative we must check for EL3.  */
+    if (arm_feature(env, ARM_FEATURE_EL3)
+        && !FIELD_EX64(env->cp15.cptr_el[3], CPTR_EL3, ESM)) {
+        return 3;
+    }
+#endif
+    return 0;
+}
+
 /*
  * Given that SVE is enabled, return the vector length for EL.
  */
@@ -11197,6 +11246,9 @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el,
         }
         DP_TBFLAG_A64(flags, SVEEXC_EL, sve_el);
     }
+    if (cpu_isar_feature(aa64_sme, env_archcpu(env))) {
+        DP_TBFLAG_A64(flags, SMEEXC_EL, sme_exception_el(env, el));
+    }
 
     sctlr = regime_sctlr(env, stage1);
 
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 4c64546090..9a285dd177 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14603,6 +14603,7 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->align_mem = EX_TBFLAG_ANY(tb_flags, ALIGN_MEM);
     dc->pstate_il = EX_TBFLAG_ANY(tb_flags, PSTATE__IL);
     dc->sve_excp_el = EX_TBFLAG_A64(tb_flags, SVEEXC_EL);
+    dc->sme_excp_el = EX_TBFLAG_A64(tb_flags, SMEEXC_EL);
     dc->vl = (EX_TBFLAG_A64(tb_flags, VL) + 1) * 16;
     dc->pauth_active = EX_TBFLAG_A64(tb_flags, PAUTH_ACTIVE);
     dc->bt = EX_TBFLAG_A64(tb_flags, BT);
diff --git a/target/arm/translate.h b/target/arm/translate.h
index 88dc18a034..c88c953325 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -42,6 +42,7 @@ typedef struct DisasContext {
     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 */
+    int sme_excp_el; /* SME exception EL or 0 if enabled */
     int vl;          /* current vector length in bytes */
     bool vfp_enabled; /* FP enabled via FPSCR.EN */
     int vec_len;

From 58b2908ee1011c33cc01d7d9341673f8af6d14b7 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:47 -0700
Subject: [PATCH 06/25] target/arm: Add syn_smetrap

This will be used for raising various traps for SME.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/syndrome.h | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/target/arm/syndrome.h b/target/arm/syndrome.h
index c105f9e6ba..73df5e3793 100644
--- a/target/arm/syndrome.h
+++ b/target/arm/syndrome.h
@@ -48,6 +48,7 @@ enum arm_exception_class {
     EC_AA64_SMC               = 0x17,
     EC_SYSTEMREGISTERTRAP     = 0x18,
     EC_SVEACCESSTRAP          = 0x19,
+    EC_SMETRAP                = 0x1d,
     EC_INSNABORT              = 0x20,
     EC_INSNABORT_SAME_EL      = 0x21,
     EC_PCALIGNMENT            = 0x22,
@@ -68,6 +69,13 @@ enum arm_exception_class {
     EC_AA64_BKPT              = 0x3c,
 };
 
+typedef enum {
+    SME_ET_AccessTrap,
+    SME_ET_Streaming,
+    SME_ET_NotStreaming,
+    SME_ET_InactiveZA,
+} SMEExceptionType;
+
 #define ARM_EL_EC_SHIFT 26
 #define ARM_EL_IL_SHIFT 25
 #define ARM_EL_ISV_SHIFT 24
@@ -207,6 +215,12 @@ static inline uint32_t syn_sve_access_trap(void)
     return EC_SVEACCESSTRAP << ARM_EL_EC_SHIFT;
 }
 
+static inline uint32_t syn_smetrap(SMEExceptionType etype, bool is_16bit)
+{
+    return (EC_SMETRAP << ARM_EL_EC_SHIFT)
+        | (is_16bit ? 0 : ARM_EL_IL) | etype;
+}
+
 static inline uint32_t syn_pactrap(void)
 {
     return EC_PACTRAP << ARM_EL_EC_SHIFT;

From bca063d579cbd6075d0bab78cc702131df199d6e Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:48 -0700
Subject: [PATCH 07/25] target/arm: Add ARM_CP_SME

This will be used for controlling access to SME cpregs.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-5-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpregs.h        |  5 +++++
 target/arm/translate-a64.c | 18 ++++++++++++++++++
 2 files changed, 23 insertions(+)

diff --git a/target/arm/cpregs.h b/target/arm/cpregs.h
index d9b678c2f1..d30758ee71 100644
--- a/target/arm/cpregs.h
+++ b/target/arm/cpregs.h
@@ -113,6 +113,11 @@ enum {
     ARM_CP_EL3_NO_EL2_UNDEF      = 1 << 16,
     ARM_CP_EL3_NO_EL2_KEEP       = 1 << 17,
     ARM_CP_EL3_NO_EL2_C_NZ       = 1 << 18,
+    /*
+     * Flag: Access check for this sysreg is constrained by the
+     * ARM pseudocode function CheckSMEAccess().
+     */
+    ARM_CP_SME                   = 1 << 19,
 };
 
 /*
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 9a285dd177..8f609f46b6 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1187,6 +1187,22 @@ bool sve_access_check(DisasContext *s)
     return fp_access_check(s);
 }
 
+/*
+ * Check that SME access is enabled, raise an exception if not.
+ * Note that this function corresponds to CheckSMEAccess and is
+ * only used directly for cpregs.
+ */
+static bool sme_access_check(DisasContext *s)
+{
+    if (s->sme_excp_el) {
+        gen_exception_insn_el(s, s->pc_curr, EXCP_UDEF,
+                              syn_smetrap(SME_ET_AccessTrap, false),
+                              s->sme_excp_el);
+        return false;
+    }
+    return true;
+}
+
 /*
  * This utility function is for doing register extension with an
  * optional shift. You will likely want to pass a temporary for the
@@ -1958,6 +1974,8 @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread,
         return;
     } else if ((ri->type & ARM_CP_SVE) && !sve_access_check(s)) {
         return;
+    } else if ((ri->type & ARM_CP_SME) && !sme_access_check(s)) {
+        return;
     }
 
     if ((tb_cflags(s->base.tb) & CF_USE_ICOUNT) && (ri->type & ARM_CP_IO)) {

From c37e6ac9eb94667d803d0cc1c4cc39ab351a6921 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:49 -0700
Subject: [PATCH 08/25] target/arm: Add SVCR

This cpreg is used to access two new bits of PSTATE
that are not visible via any other mechanism.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-6-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h    |  6 ++++++
 target/arm/helper.c | 13 +++++++++++++
 2 files changed, 19 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index e99de18097..bb8cb959d1 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -258,6 +258,7 @@ typedef struct CPUArchState {
      *  nRW (also known as M[4]) is kept, inverted, in env->aarch64
      *  DAIF (exception masks) are kept in env->daif
      *  BTYPE is kept in env->btype
+     *  SM and ZA are kept in env->svcr
      *  all other bits are stored in their correct places in env->pstate
      */
     uint32_t pstate;
@@ -292,6 +293,7 @@ typedef struct CPUArchState {
     uint32_t condexec_bits; /* IT bits.  cpsr[15:10,26:25].  */
     uint32_t btype;  /* BTI branch type.  spsr[11:10].  */
     uint64_t daif; /* exception masks, in the bits they are in PSTATE */
+    uint64_t svcr; /* PSTATE.{SM,ZA} in the bits they are in SVCR */
 
     uint64_t elr_el[4]; /* AArch64 exception link regs  */
     uint64_t sp_el[4]; /* AArch64 banked stack pointers */
@@ -1428,6 +1430,10 @@ FIELD(CPTR_EL3, TCPAC, 31, 1)
 #define PSTATE_MODE_EL1t 4
 #define PSTATE_MODE_EL0t 0
 
+/* PSTATE bits that are accessed via SVCR and not stored in SPSR_ELx. */
+FIELD(SVCR, SM, 0, 1)
+FIELD(SVCR, ZA, 1, 1)
+
 /* Write a new value to v7m.exception, thus transitioning into or out
  * of Handler mode; this may result in a change of active stack pointer.
  */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 2c080c6cac..3acc1dc378 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6349,11 +6349,24 @@ static CPAccessResult access_tpidr2(CPUARMState *env, const ARMCPRegInfo *ri,
     return CP_ACCESS_OK;
 }
 
+static void svcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
+                       uint64_t value)
+{
+    value &= R_SVCR_SM_MASK | R_SVCR_ZA_MASK;
+    /* TODO: Side effects. */
+    env->svcr = value;
+}
+
 static const ARMCPRegInfo sme_reginfo[] = {
     { .name = "TPIDR2_EL0", .state = ARM_CP_STATE_AA64,
       .opc0 = 3, .opc1 = 3, .crn = 13, .crm = 0, .opc2 = 5,
       .access = PL0_RW, .accessfn = access_tpidr2,
       .fieldoffset = offsetof(CPUARMState, cp15.tpidr2_el0) },
+    { .name = "SVCR", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 3, .crn = 4, .crm = 2, .opc2 = 2,
+      .access = PL0_RW, .type = ARM_CP_SME,
+      .fieldoffset = offsetof(CPUARMState, svcr),
+      .writefn = svcr_write, .raw_writefn = raw_write },
 };
 #endif /* TARGET_AARCH64 */
 

From de5619887cbc7549bfc63bec4993de94626ccf09 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:50 -0700
Subject: [PATCH 09/25] target/arm: Add SMCR_ELx

These cpregs control the streaming vector length and whether the
full a64 instruction set is allowed while in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-7-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h    |  8 ++++++--
 target/arm/helper.c | 41 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 2 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index bb8cb959d1..dec52c6c3b 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -669,8 +669,8 @@ typedef struct CPUArchState {
         float_status standard_fp_status;
         float_status standard_fp_status_f16;
 
-        /* ZCR_EL[1-3] */
-        uint64_t zcr_el[4];
+        uint64_t zcr_el[4];   /* ZCR_EL[1-3] */
+        uint64_t smcr_el[4];  /* SMCR_EL[1-3] */
     } vfp;
     uint64_t exclusive_addr;
     uint64_t exclusive_val;
@@ -1434,6 +1434,10 @@ FIELD(CPTR_EL3, TCPAC, 31, 1)
 FIELD(SVCR, SM, 0, 1)
 FIELD(SVCR, ZA, 1, 1)
 
+/* Fields for SMCR_ELx. */
+FIELD(SMCR, LEN, 0, 4)
+FIELD(SMCR, FA64, 31, 1)
+
 /* Write a new value to v7m.exception, thus transitioning into or out
  * of Handler mode; this may result in a change of active stack pointer.
  */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 3acc1dc378..2072f2a550 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -5879,6 +5879,8 @@ 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,  1, 2, 6), K(3, 4,  1, 2, 6), K(3, 5, 1, 2, 6),
+          "SMCR_EL1", "SMCR_EL2", "SMCR_EL12", isar_feature_aa64_sme },
 
         { 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 },
@@ -6357,6 +6359,30 @@ static void svcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
     env->svcr = value;
 }
 
+static void smcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
+                       uint64_t value)
+{
+    int cur_el = arm_current_el(env);
+    int old_len = sve_vqm1_for_el(env, cur_el);
+    int new_len;
+
+    QEMU_BUILD_BUG_ON(ARM_MAX_VQ > R_SMCR_LEN_MASK + 1);
+    value &= R_SMCR_LEN_MASK | R_SMCR_FA64_MASK;
+    raw_write(env, ri, value);
+
+    /*
+     * Note that it is CONSTRAINED UNPREDICTABLE what happens to ZA storage
+     * when SVL is widened (old values kept, or zeros).  Choose to keep the
+     * current values for simplicity.  But for QEMU internals, we must still
+     * apply the narrower SVL to the Zregs and Pregs -- see the comment
+     * above aarch64_sve_narrow_vq.
+     */
+    new_len = sve_vqm1_for_el(env, cur_el);
+    if (new_len < old_len) {
+        aarch64_sve_narrow_vq(env, new_len + 1);
+    }
+}
+
 static const ARMCPRegInfo sme_reginfo[] = {
     { .name = "TPIDR2_EL0", .state = ARM_CP_STATE_AA64,
       .opc0 = 3, .opc1 = 3, .crn = 13, .crm = 0, .opc2 = 5,
@@ -6367,6 +6393,21 @@ static const ARMCPRegInfo sme_reginfo[] = {
       .access = PL0_RW, .type = ARM_CP_SME,
       .fieldoffset = offsetof(CPUARMState, svcr),
       .writefn = svcr_write, .raw_writefn = raw_write },
+    { .name = "SMCR_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 1, .crm = 2, .opc2 = 6,
+      .access = PL1_RW, .type = ARM_CP_SME,
+      .fieldoffset = offsetof(CPUARMState, vfp.smcr_el[1]),
+      .writefn = smcr_write, .raw_writefn = raw_write },
+    { .name = "SMCR_EL2", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 4, .crn = 1, .crm = 2, .opc2 = 6,
+      .access = PL2_RW, .type = ARM_CP_SME,
+      .fieldoffset = offsetof(CPUARMState, vfp.smcr_el[2]),
+      .writefn = smcr_write, .raw_writefn = raw_write },
+    { .name = "SMCR_EL3", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 6, .crn = 1, .crm = 2, .opc2 = 6,
+      .access = PL3_RW, .type = ARM_CP_SME,
+      .fieldoffset = offsetof(CPUARMState, vfp.smcr_el[3]),
+      .writefn = smcr_write, .raw_writefn = raw_write },
 };
 #endif /* TARGET_AARCH64 */
 

From d5b1223ac1ddf8f706f5e6feaaa526df8287f8b1 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:51 -0700
Subject: [PATCH 10/25] target/arm: Add SMIDR_EL1, SMPRI_EL1, SMPRIMAP_EL2

Implement the streaming mode identification register, and the
two streaming priority registers.  For QEMU, they are all RES0.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-8-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/helper.c | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 2072f2a550..bbd04fbd67 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6351,6 +6351,18 @@ static CPAccessResult access_tpidr2(CPUARMState *env, const ARMCPRegInfo *ri,
     return CP_ACCESS_OK;
 }
 
+static CPAccessResult access_esm(CPUARMState *env, const ARMCPRegInfo *ri,
+                                 bool isread)
+{
+    /* TODO: FEAT_FGT for SMPRI_EL1 but not SMPRIMAP_EL2 */
+    if (arm_current_el(env) < 3
+        && arm_feature(env, ARM_FEATURE_EL3)
+        && !FIELD_EX64(env->cp15.cptr_el[3], CPTR_EL3, ESM)) {
+        return CP_ACCESS_TRAP_EL3;
+    }
+    return CP_ACCESS_OK;
+}
+
 static void svcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
 {
@@ -6408,6 +6420,27 @@ static const ARMCPRegInfo sme_reginfo[] = {
       .access = PL3_RW, .type = ARM_CP_SME,
       .fieldoffset = offsetof(CPUARMState, vfp.smcr_el[3]),
       .writefn = smcr_write, .raw_writefn = raw_write },
+    { .name = "SMIDR_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 1, .crn = 0, .crm = 0, .opc2 = 6,
+      .access = PL1_R, .accessfn = access_aa64_tid1,
+      /*
+       * IMPLEMENTOR = 0 (software)
+       * REVISION    = 0 (implementation defined)
+       * SMPS        = 0 (no streaming execution priority in QEMU)
+       * AFFINITY    = 0 (streaming sve mode not shared with other PEs)
+       */
+      .type = ARM_CP_CONST, .resetvalue = 0, },
+    /*
+     * Because SMIDR_EL1.SMPS is 0, SMPRI_EL1 and SMPRIMAP_EL2 are RES 0.
+     */
+    { .name = "SMPRI_EL1", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 0, .crn = 1, .crm = 2, .opc2 = 4,
+      .access = PL1_RW, .accessfn = access_esm,
+      .type = ARM_CP_CONST, .resetvalue = 0 },
+    { .name = "SMPRIMAP_EL2", .state = ARM_CP_STATE_AA64,
+      .opc0 = 3, .opc1 = 4, .crn = 1, .crm = 2, .opc2 = 5,
+      .access = PL2_RW, .accessfn = access_esm,
+      .type = ARM_CP_CONST, .resetvalue = 0 },
 };
 #endif /* TARGET_AARCH64 */
 

From a3637e8882f9dbb00036ff77a88b841bd2580900 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:52 -0700
Subject: [PATCH 11/25] target/arm: Add PSTATE.{SM,ZA} to TB flags

These are required to determine if various insns
are allowed to issue.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-9-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h           | 2 ++
 target/arm/helper.c        | 4 ++++
 target/arm/translate-a64.c | 2 ++
 target/arm/translate.h     | 4 ++++
 4 files changed, 12 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index dec52c6c3b..05d369e690 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3160,6 +3160,8 @@ FIELD(TBFLAG_A64, TCMA, 16, 2)
 FIELD(TBFLAG_A64, MTE_ACTIVE, 18, 1)
 FIELD(TBFLAG_A64, MTE0_ACTIVE, 19, 1)
 FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2)
+FIELD(TBFLAG_A64, PSTATE_SM, 22, 1)
+FIELD(TBFLAG_A64, PSTATE_ZA, 23, 1)
 
 /*
  * Helpers for using the above.
diff --git a/target/arm/helper.c b/target/arm/helper.c
index bbd04fbd67..e06c054c3d 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -11335,6 +11335,10 @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el,
     }
     if (cpu_isar_feature(aa64_sme, env_archcpu(env))) {
         DP_TBFLAG_A64(flags, SMEEXC_EL, sme_exception_el(env, el));
+        if (FIELD_EX64(env->svcr, SVCR, SM)) {
+            DP_TBFLAG_A64(flags, PSTATE_SM, 1);
+        }
+        DP_TBFLAG_A64(flags, PSTATE_ZA, FIELD_EX64(env->svcr, SVCR, ZA));
     }
 
     sctlr = regime_sctlr(env, stage1);
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 8f609f46b6..5cf4a283ba 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14630,6 +14630,8 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->ata = EX_TBFLAG_A64(tb_flags, ATA);
     dc->mte_active[0] = EX_TBFLAG_A64(tb_flags, MTE_ACTIVE);
     dc->mte_active[1] = EX_TBFLAG_A64(tb_flags, MTE0_ACTIVE);
+    dc->pstate_sm = EX_TBFLAG_A64(tb_flags, PSTATE_SM);
+    dc->pstate_za = EX_TBFLAG_A64(tb_flags, PSTATE_ZA);
     dc->vec_len = 0;
     dc->vec_stride = 0;
     dc->cp_regs = arm_cpu->cp_regs;
diff --git a/target/arm/translate.h b/target/arm/translate.h
index c88c953325..93766649f7 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -97,6 +97,10 @@ typedef struct DisasContext {
     bool align_mem;
     /* True if PSTATE.IL is set */
     bool pstate_il;
+    /* True if PSTATE.SM is set. */
+    bool pstate_sm;
+    /* True if PSTATE.ZA is set. */
+    bool pstate_za;
     /* True if MVE insns are definitely not predicated by VPR or LTPSIZE */
     bool mve_no_pred;
     /*

From dc993a01a75295c505ef1ff8764c68f31089fcc7 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:53 -0700
Subject: [PATCH 12/25] target/arm: Add the SME ZA storage to CPUARMState

Place this late in the resettable section of the structure,
to keep the most common element offsets from being > 64k.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-10-richard.henderson@linaro.org
[PMM: expanded comment on zarray[] format]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h     | 22 ++++++++++++++++++++++
 target/arm/machine.c | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 05d369e690..52ab6f9bb9 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -694,6 +694,28 @@ typedef struct CPUArchState {
     } keys;
 
     uint64_t scxtnum_el[4];
+
+    /*
+     * SME ZA storage -- 256 x 256 byte array, with bytes in host word order,
+     * as we do with vfp.zregs[].  This corresponds to the architectural ZA
+     * array, where ZA[N] is in the least-significant bytes of env->zarray[N].
+     * When SVL is less than the architectural maximum, the accessible
+     * storage is restricted, such that if the SVL is X bytes the guest can
+     * see only the bottom X elements of zarray[], and only the least
+     * significant X bytes of each element of the array. (In other words,
+     * the observable part is always square.)
+     *
+     * The ZA storage can also be considered as a set of square tiles of
+     * elements of different sizes. The mapping from tiles to the ZA array
+     * is architecturally defined, such that for tiles of elements of esz
+     * bytes, the Nth row (or "horizontal slice") of tile T is in
+     * ZA[T + N * esz]. Note that this means that each tile is not contiguous
+     * in the ZA storage, because its rows are striped through the ZA array.
+     *
+     * Because this is so large, keep this toward the end of the reset area,
+     * to keep the offsets into the rest of the structure smaller.
+     */
+    ARMVectorReg zarray[ARM_MAX_VQ * 16];
 #endif
 
 #if defined(CONFIG_USER_ONLY)
diff --git a/target/arm/machine.c b/target/arm/machine.c
index 285e387d2c..54c5c62433 100644
--- a/target/arm/machine.c
+++ b/target/arm/machine.c
@@ -167,6 +167,39 @@ static const VMStateDescription vmstate_sve = {
         VMSTATE_END_OF_LIST()
     }
 };
+
+static const VMStateDescription vmstate_vreg = {
+    .name = "vreg",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64_ARRAY(d, ARMVectorReg, ARM_MAX_VQ * 2),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool za_needed(void *opaque)
+{
+    ARMCPU *cpu = opaque;
+
+    /*
+     * When ZA storage is disabled, its contents are discarded.
+     * It will be zeroed when ZA storage is re-enabled.
+     */
+    return FIELD_EX64(cpu->env.svcr, SVCR, ZA);
+}
+
+static const VMStateDescription vmstate_za = {
+    .name = "cpu/sme",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = za_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_STRUCT_ARRAY(env.zarray, ARMCPU, ARM_MAX_VQ * 16, 0,
+                             vmstate_vreg, ARMVectorReg),
+        VMSTATE_END_OF_LIST()
+    }
+};
 #endif /* AARCH64 */
 
 static bool serror_needed(void *opaque)
@@ -884,6 +917,7 @@ const VMStateDescription vmstate_arm_cpu = {
         &vmstate_m_security,
 #ifdef TARGET_AARCH64
         &vmstate_sve,
+        &vmstate_za,
 #endif
         &vmstate_serror,
         &vmstate_irq_line_state,

From f84734b87461fbf3ab349399f7936de832e477ed Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:54 -0700
Subject: [PATCH 13/25] target/arm: Implement SMSTART, SMSTOP

These two instructions are aliases of MSR (immediate).
Use the two helpers to properly implement svcr_write.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-11-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h           |  1 +
 target/arm/helper-sme.h    | 21 +++++++++++++
 target/arm/helper.c        |  6 ++--
 target/arm/helper.h        |  1 +
 target/arm/meson.build     |  1 +
 target/arm/sme_helper.c    | 61 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-a64.c | 24 +++++++++++++++
 7 files changed, 112 insertions(+), 3 deletions(-)
 create mode 100644 target/arm/helper-sme.h
 create mode 100644 target/arm/sme_helper.c

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 52ab6f9bb9..5877d76c9f 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -1120,6 +1120,7 @@ void aarch64_sve_change_el(CPUARMState *env, int old_el,
                            int new_el, bool el0_a64);
 void aarch64_add_sve_properties(Object *obj);
 void aarch64_add_pauth_properties(Object *obj);
+void arm_reset_sve_state(CPUARMState *env);
 
 /*
  * SVE registers are encoded in KVM's memory in an endianness-invariant format.
diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h
new file mode 100644
index 0000000000..3bd48c235f
--- /dev/null
+++ b/target/arm/helper-sme.h
@@ -0,0 +1,21 @@
+/*
+ *  AArch64 SME specific helper definitions
+ *
+ *  Copyright (c) 2022 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/>.
+ */
+
+DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32)
+DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32)
diff --git a/target/arm/helper.c b/target/arm/helper.c
index e06c054c3d..88d96f7991 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6366,9 +6366,9 @@ static CPAccessResult access_esm(CPUARMState *env, const ARMCPRegInfo *ri,
 static void svcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
 {
-    value &= R_SVCR_SM_MASK | R_SVCR_ZA_MASK;
-    /* TODO: Side effects. */
-    env->svcr = value;
+    helper_set_pstate_sm(env, FIELD_EX64(value, SVCR, SM));
+    helper_set_pstate_za(env, FIELD_EX64(value, SVCR, ZA));
+    arm_rebuild_hflags(env);
 }
 
 static void smcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
diff --git a/target/arm/helper.h b/target/arm/helper.h
index 07d45faf49..3a8ce42ab0 100644
--- a/target/arm/helper.h
+++ b/target/arm/helper.h
@@ -1022,6 +1022,7 @@ DEF_HELPER_FLAGS_6(gvec_bfmlal_idx, TCG_CALL_NO_RWG,
 #ifdef TARGET_AARCH64
 #include "helper-a64.h"
 #include "helper-sve.h"
+#include "helper-sme.h"
 #endif
 
 #include "helper-mve.h"
diff --git a/target/arm/meson.build b/target/arm/meson.build
index ac571fc45d..43dc600547 100644
--- a/target/arm/meson.build
+++ b/target/arm/meson.build
@@ -47,6 +47,7 @@ arm_ss.add(when: 'TARGET_AARCH64', if_true: files(
   'mte_helper.c',
   'pauth_helper.c',
   'sve_helper.c',
+  'sme_helper.c',
   'translate-a64.c',
   'translate-sve.c',
 ))
diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c
new file mode 100644
index 0000000000..b215725594
--- /dev/null
+++ b/target/arm/sme_helper.c
@@ -0,0 +1,61 @@
+/*
+ * ARM SME Operations
+ *
+ * Copyright (c) 2022 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/helper-proto.h"
+
+/* ResetSVEState */
+void arm_reset_sve_state(CPUARMState *env)
+{
+    memset(env->vfp.zregs, 0, sizeof(env->vfp.zregs));
+    /* Recall that FFR is stored as pregs[16]. */
+    memset(env->vfp.pregs, 0, sizeof(env->vfp.pregs));
+    vfp_set_fpcr(env, 0x0800009f);
+}
+
+void helper_set_pstate_sm(CPUARMState *env, uint32_t i)
+{
+    if (i == FIELD_EX64(env->svcr, SVCR, SM)) {
+        return;
+    }
+    env->svcr ^= R_SVCR_SM_MASK;
+    arm_reset_sve_state(env);
+}
+
+void helper_set_pstate_za(CPUARMState *env, uint32_t i)
+{
+    if (i == FIELD_EX64(env->svcr, SVCR, ZA)) {
+        return;
+    }
+    env->svcr ^= R_SVCR_ZA_MASK;
+
+    /*
+     * ResetSMEState.
+     *
+     * SetPSTATE_ZA zeros on enable and disable.  We can zero this only
+     * on enable: while disabled, the storage is inaccessible and the
+     * value does not matter.  We're not saving the storage in vmstate
+     * when disabled either.
+     */
+    if (i) {
+        memset(env->zarray, 0, sizeof(env->zarray));
+    }
+}
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 5cf4a283ba..c050ebe005 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -1762,6 +1762,30 @@ static void handle_msr_i(DisasContext *s, uint32_t insn,
         }
         break;
 
+    case 0x1b: /* SVCR* */
+        if (!dc_isar_feature(aa64_sme, s) || crm < 2 || crm > 7) {
+            goto do_unallocated;
+        }
+        if (sme_access_check(s)) {
+            bool i = crm & 1;
+            bool changed = false;
+
+            if ((crm & 2) && i != s->pstate_sm) {
+                gen_helper_set_pstate_sm(cpu_env, tcg_constant_i32(i));
+                changed = true;
+            }
+            if ((crm & 4) && i != s->pstate_za) {
+                gen_helper_set_pstate_za(cpu_env, tcg_constant_i32(i));
+                changed = true;
+            }
+            if (changed) {
+                gen_rebuild_hflags(s);
+            } else {
+                s->base.is_jmp = DISAS_NEXT;
+            }
+        }
+        break;
+
     default:
     do_unallocated:
         unallocated_encoding(s);

From 531cc510370eb7f672eaca416b0a3927806b3983 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:55 -0700
Subject: [PATCH 14/25] target/arm: Move error for sve%d property to
 arm_cpu_sve_finalize

Keep all of the error messages together.  This does mean that
when setting many sve length properties we'll only generate
one error, but we only really need one.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-12-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu64.c | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 15665c962b..a46e40f4f2 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -487,8 +487,13 @@ void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp)
                                   "using only sve<N> properties.\n");
             } else {
                 error_setg(errp, "cannot enable sve%d", vq * 128);
-                error_append_hint(errp, "This CPU does not support "
-                                  "the vector length %d-bits.\n", vq * 128);
+                if (vq_supported) {
+                    error_append_hint(errp, "This CPU does not support "
+                                      "the vector length %d-bits.\n", vq * 128);
+                } else {
+                    error_append_hint(errp, "SVE not supported by KVM "
+                                      "on this host\n");
+                }
             }
             return;
         } else {
@@ -606,12 +611,6 @@ static void cpu_arm_set_sve_vq(Object *obj, Visitor *v, const char *name,
         return;
     }
 
-    if (value && kvm_enabled() && !kvm_arm_sve_supported()) {
-        error_setg(errp, "cannot enable %s", name);
-        error_append_hint(errp, "SVE not supported by KVM on this host\n");
-        return;
-    }
-
     cpu->sve_vq_map = deposit32(cpu->sve_vq_map, vq - 1, 1, value);
     cpu->sve_vq_init |= 1 << (vq - 1);
 }

From 7f9e25a6e43356d4a0fb2cc201c1a45c5be5bb6c Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:56 -0700
Subject: [PATCH 15/25] target/arm: Create ARMVQMap

Pull the three sve_vq_* values into a structure.
This will be reused for SME.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-13-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h    | 29 ++++++++++++++---------------
 target/arm/cpu64.c  | 22 +++++++++++-----------
 target/arm/helper.c |  2 +-
 target/arm/kvm64.c  |  2 +-
 4 files changed, 27 insertions(+), 28 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 5877d76c9f..2ce47f8d29 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -807,6 +807,19 @@ typedef enum ARMPSCIState {
 
 typedef struct ARMISARegisters ARMISARegisters;
 
+/*
+ * In map, each set bit is a supported vector length of (bit-number + 1) * 16
+ * bytes, i.e. each bit number + 1 is the vector length in quadwords.
+ *
+ * While processing properties during initialization, corresponding init bits
+ * are set for bits in sve_vq_map that have been set by properties.
+ *
+ * Bits set in supported represent valid vector lengths for the CPU type.
+ */
+typedef struct {
+    uint32_t map, init, supported;
+} ARMVQMap;
+
 /**
  * ARMCPU:
  * @env: #CPUARMState
@@ -1055,21 +1068,7 @@ struct ArchCPU {
     uint32_t sve_default_vq;
 #endif
 
-    /*
-     * In sve_vq_map each set bit is a supported vector length of
-     * (bit-number + 1) * 16 bytes, i.e. each bit number + 1 is the vector
-     * length in quadwords.
-     *
-     * While processing properties during initialization, corresponding
-     * sve_vq_init bits are set for bits in sve_vq_map that have been
-     * set by properties.
-     *
-     * Bits set in sve_vq_supported represent valid vector lengths for
-     * the CPU type.
-     */
-    uint32_t sve_vq_map;
-    uint32_t sve_vq_init;
-    uint32_t sve_vq_supported;
+    ARMVQMap sve_vq;
 
     /* Generic timer counter frequency, in Hz */
     uint64_t gt_cntfrq_hz;
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index a46e40f4f2..cadc401c7e 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -355,8 +355,8 @@ void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp)
      * any of the above.  Finally, if SVE is not disabled, then at least one
      * vector length must be enabled.
      */
-    uint32_t vq_map = cpu->sve_vq_map;
-    uint32_t vq_init = cpu->sve_vq_init;
+    uint32_t vq_map = cpu->sve_vq.map;
+    uint32_t vq_init = cpu->sve_vq.init;
     uint32_t vq_supported;
     uint32_t vq_mask = 0;
     uint32_t tmp, vq, max_vq = 0;
@@ -369,14 +369,14 @@ void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp)
      */
     if (kvm_enabled()) {
         if (kvm_arm_sve_supported()) {
-            cpu->sve_vq_supported = kvm_arm_sve_get_vls(CPU(cpu));
-            vq_supported = cpu->sve_vq_supported;
+            cpu->sve_vq.supported = kvm_arm_sve_get_vls(CPU(cpu));
+            vq_supported = cpu->sve_vq.supported;
         } else {
             assert(!cpu_isar_feature(aa64_sve, cpu));
             vq_supported = 0;
         }
     } else {
-        vq_supported = cpu->sve_vq_supported;
+        vq_supported = cpu->sve_vq.supported;
     }
 
     /*
@@ -534,7 +534,7 @@ void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp)
 
     /* From now on sve_max_vq is the actual maximum supported length. */
     cpu->sve_max_vq = max_vq;
-    cpu->sve_vq_map = vq_map;
+    cpu->sve_vq.map = vq_map;
 }
 
 static void cpu_max_get_sve_max_vq(Object *obj, Visitor *v, const char *name,
@@ -595,7 +595,7 @@ static void cpu_arm_get_sve_vq(Object *obj, Visitor *v, const char *name,
     if (!cpu_isar_feature(aa64_sve, cpu)) {
         value = false;
     } else {
-        value = extract32(cpu->sve_vq_map, vq - 1, 1);
+        value = extract32(cpu->sve_vq.map, vq - 1, 1);
     }
     visit_type_bool(v, name, &value, errp);
 }
@@ -611,8 +611,8 @@ static void cpu_arm_set_sve_vq(Object *obj, Visitor *v, const char *name,
         return;
     }
 
-    cpu->sve_vq_map = deposit32(cpu->sve_vq_map, vq - 1, 1, value);
-    cpu->sve_vq_init |= 1 << (vq - 1);
+    cpu->sve_vq.map = deposit32(cpu->sve_vq.map, vq - 1, 1, value);
+    cpu->sve_vq.init |= 1 << (vq - 1);
 }
 
 static bool cpu_arm_get_sve(Object *obj, Error **errp)
@@ -974,7 +974,7 @@ static void aarch64_max_initfn(Object *obj)
     cpu->dcz_blocksize = 7; /*  512 bytes */
 #endif
 
-    cpu->sve_vq_supported = MAKE_64BIT_MASK(0, ARM_MAX_VQ);
+    cpu->sve_vq.supported = MAKE_64BIT_MASK(0, ARM_MAX_VQ);
 
     aarch64_add_pauth_properties(obj);
     aarch64_add_sve_properties(obj);
@@ -1023,7 +1023,7 @@ static void aarch64_a64fx_initfn(Object *obj)
 
     /* The A64FX supports only 128, 256 and 512 bit vector lengths */
     aarch64_add_sve_properties(obj);
-    cpu->sve_vq_supported = (1 << 0)  /* 128bit */
+    cpu->sve_vq.supported = (1 << 0)  /* 128bit */
                           | (1 << 1)  /* 256bit */
                           | (1 << 3); /* 512bit */
 
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 88d96f7991..a80ca461e5 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6287,7 +6287,7 @@ uint32_t sve_vqm1_for_el(CPUARMState *env, int el)
         len = MIN(len, 0xf & (uint32_t)env->vfp.zcr_el[3]);
     }
 
-    len = 31 - clz32(cpu->sve_vq_map & MAKE_64BIT_MASK(0, len + 1));
+    len = 31 - clz32(cpu->sve_vq.map & MAKE_64BIT_MASK(0, len + 1));
     return len;
 }
 
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
index ff8f65da22..d16d4ea250 100644
--- a/target/arm/kvm64.c
+++ b/target/arm/kvm64.c
@@ -820,7 +820,7 @@ uint32_t kvm_arm_sve_get_vls(CPUState *cs)
 static int kvm_arm_sve_set_vls(CPUState *cs)
 {
     ARMCPU *cpu = ARM_CPU(cs);
-    uint64_t vls[KVM_ARM64_SVE_VLS_WORDS] = { cpu->sve_vq_map };
+    uint64_t vls[KVM_ARM64_SVE_VLS_WORDS] = { cpu->sve_vq.map };
     struct kvm_one_reg reg = {
         .id = KVM_REG_ARM64_SVE_VLS,
         .addr = (uint64_t)&vls[0],

From 0f40784eac0e429e48c26e85f5821120cf35ed79 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:57 -0700
Subject: [PATCH 16/25] target/arm: Generalize cpu_arm_{get,set}_vq

Rename from cpu_arm_{get,set}_sve_vq, and take the
ARMVQMap as the opaque parameter.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-14-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu64.c | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index cadc401c7e..1a3cb953bf 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -579,15 +579,15 @@ static void cpu_max_set_sve_max_vq(Object *obj, Visitor *v, const char *name,
 }
 
 /*
- * Note that cpu_arm_get/set_sve_vq cannot use the simpler
- * object_property_add_bool interface because they make use
- * of the contents of "name" to determine which bit on which
- * to operate.
+ * Note that cpu_arm_{get,set}_vq cannot use the simpler
+ * object_property_add_bool interface because they make use of the
+ * contents of "name" to determine which bit on which to operate.
  */
-static void cpu_arm_get_sve_vq(Object *obj, Visitor *v, const char *name,
-                               void *opaque, Error **errp)
+static void cpu_arm_get_vq(Object *obj, Visitor *v, const char *name,
+                           void *opaque, Error **errp)
 {
     ARMCPU *cpu = ARM_CPU(obj);
+    ARMVQMap *vq_map = opaque;
     uint32_t vq = atoi(&name[3]) / 128;
     bool value;
 
@@ -595,15 +595,15 @@ static void cpu_arm_get_sve_vq(Object *obj, Visitor *v, const char *name,
     if (!cpu_isar_feature(aa64_sve, cpu)) {
         value = false;
     } else {
-        value = extract32(cpu->sve_vq.map, vq - 1, 1);
+        value = extract32(vq_map->map, vq - 1, 1);
     }
     visit_type_bool(v, name, &value, errp);
 }
 
-static void cpu_arm_set_sve_vq(Object *obj, Visitor *v, const char *name,
-                               void *opaque, Error **errp)
+static void cpu_arm_set_vq(Object *obj, Visitor *v, const char *name,
+                           void *opaque, Error **errp)
 {
-    ARMCPU *cpu = ARM_CPU(obj);
+    ARMVQMap *vq_map = opaque;
     uint32_t vq = atoi(&name[3]) / 128;
     bool value;
 
@@ -611,8 +611,8 @@ static void cpu_arm_set_sve_vq(Object *obj, Visitor *v, const char *name,
         return;
     }
 
-    cpu->sve_vq.map = deposit32(cpu->sve_vq.map, vq - 1, 1, value);
-    cpu->sve_vq.init |= 1 << (vq - 1);
+    vq_map->map = deposit32(vq_map->map, vq - 1, 1, value);
+    vq_map->init |= 1 << (vq - 1);
 }
 
 static bool cpu_arm_get_sve(Object *obj, Error **errp)
@@ -691,6 +691,7 @@ static void cpu_arm_get_sve_default_vec_len(Object *obj, Visitor *v,
 
 void aarch64_add_sve_properties(Object *obj)
 {
+    ARMCPU *cpu = ARM_CPU(obj);
     uint32_t vq;
 
     object_property_add_bool(obj, "sve", cpu_arm_get_sve, cpu_arm_set_sve);
@@ -698,8 +699,8 @@ void aarch64_add_sve_properties(Object *obj)
     for (vq = 1; vq <= ARM_MAX_VQ; ++vq) {
         char name[8];
         sprintf(name, "sve%d", vq * 128);
-        object_property_add(obj, name, "bool", cpu_arm_get_sve_vq,
-                            cpu_arm_set_sve_vq, NULL, NULL);
+        object_property_add(obj, name, "bool", cpu_arm_get_vq,
+                            cpu_arm_set_vq, NULL, &cpu->sve_vq);
     }
 
 #ifdef CONFIG_USER_ONLY

From 515816a82c11bbcc48272e1d8d3d61267cb2fd84 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:58 -0700
Subject: [PATCH 17/25] target/arm: Generalize cpu_arm_{get,
 set}_default_vec_len

Rename from cpu_arm_{get,set}_sve_default_vec_len,
and take the pointer to default_vq from opaque.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-15-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu64.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)

diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 1a3cb953bf..b15a0d398a 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -638,11 +638,11 @@ static void cpu_arm_set_sve(Object *obj, bool value, Error **errp)
 
 #ifdef CONFIG_USER_ONLY
 /* Mirror linux /proc/sys/abi/sve_default_vector_length. */
-static void cpu_arm_set_sve_default_vec_len(Object *obj, Visitor *v,
-                                            const char *name, void *opaque,
-                                            Error **errp)
+static void cpu_arm_set_default_vec_len(Object *obj, Visitor *v,
+                                        const char *name, void *opaque,
+                                        Error **errp)
 {
-    ARMCPU *cpu = ARM_CPU(obj);
+    uint32_t *ptr_default_vq = opaque;
     int32_t default_len, default_vq, remainder;
 
     if (!visit_type_int32(v, name, &default_len, errp)) {
@@ -651,7 +651,7 @@ static void cpu_arm_set_sve_default_vec_len(Object *obj, Visitor *v,
 
     /* Undocumented, but the kernel allows -1 to indicate "maximum". */
     if (default_len == -1) {
-        cpu->sve_default_vq = ARM_MAX_VQ;
+        *ptr_default_vq = ARM_MAX_VQ;
         return;
     }
 
@@ -675,15 +675,15 @@ static void cpu_arm_set_sve_default_vec_len(Object *obj, Visitor *v,
         return;
     }
 
-    cpu->sve_default_vq = default_vq;
+    *ptr_default_vq = default_vq;
 }
 
-static void cpu_arm_get_sve_default_vec_len(Object *obj, Visitor *v,
-                                            const char *name, void *opaque,
-                                            Error **errp)
+static void cpu_arm_get_default_vec_len(Object *obj, Visitor *v,
+                                        const char *name, void *opaque,
+                                        Error **errp)
 {
-    ARMCPU *cpu = ARM_CPU(obj);
-    int32_t value = cpu->sve_default_vq * 16;
+    uint32_t *ptr_default_vq = opaque;
+    int32_t value = *ptr_default_vq * 16;
 
     visit_type_int32(v, name, &value, errp);
 }
@@ -706,8 +706,9 @@ void aarch64_add_sve_properties(Object *obj)
 #ifdef CONFIG_USER_ONLY
     /* Mirror linux /proc/sys/abi/sve_default_vector_length. */
     object_property_add(obj, "sve-default-vector-length", "int32",
-                        cpu_arm_get_sve_default_vec_len,
-                        cpu_arm_set_sve_default_vec_len, NULL, NULL);
+                        cpu_arm_get_default_vec_len,
+                        cpu_arm_set_default_vec_len, NULL,
+                        &cpu->sve_default_vq);
 #endif
 }
 

From 073011612b44771190bc091e459d0642d46c69b5 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:51:59 -0700
Subject: [PATCH 18/25] target/arm: Move arm_cpu_*_finalize to internals.h

Drop the aa32-only inline fallbacks,
and just use a couple of ifdefs.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-16-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.c       | 2 ++
 target/arm/cpu.h       | 6 ------
 target/arm/internals.h | 3 +++
 3 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index d9c4a9f56d..660fd8b8b9 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -1422,6 +1422,7 @@ void arm_cpu_finalize_features(ARMCPU *cpu, Error **errp)
 {
     Error *local_err = NULL;
 
+#ifdef TARGET_AARCH64
     if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64)) {
         arm_cpu_sve_finalize(cpu, &local_err);
         if (local_err != NULL) {
@@ -1441,6 +1442,7 @@ void arm_cpu_finalize_features(ARMCPU *cpu, Error **errp)
             return;
         }
     }
+#endif
 
     if (kvm_enabled()) {
         kvm_arm_steal_time_finalize(cpu, &local_err);
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 2ce47f8d29..675c49f93e 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -205,14 +205,8 @@ typedef struct {
 
 #ifdef TARGET_AARCH64
 # define ARM_MAX_VQ    16
-void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp);
-void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp);
-void arm_cpu_lpa2_finalize(ARMCPU *cpu, Error **errp);
 #else
 # define ARM_MAX_VQ    1
-static inline void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp) { }
-static inline void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp) { }
-static inline void arm_cpu_lpa2_finalize(ARMCPU *cpu, Error **errp) { }
 #endif
 
 typedef struct ARMVectorReg {
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 6f94f3019d..aef568adf7 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1288,6 +1288,9 @@ int arm_gdb_get_svereg(CPUARMState *env, GByteArray *buf, int reg);
 int arm_gdb_set_svereg(CPUARMState *env, uint8_t *buf, int reg);
 int aarch64_fpu_gdb_get_reg(CPUARMState *env, GByteArray *buf, int reg);
 int aarch64_fpu_gdb_set_reg(CPUARMState *env, uint8_t *buf, int reg);
+void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp);
+void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp);
+void arm_cpu_lpa2_finalize(ARMCPU *cpu, Error **errp);
 #endif
 
 #ifdef CONFIG_USER_ONLY

From 70cc9ee19e53bc8bc597c5134e294a2ab377c4da Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:52:00 -0700
Subject: [PATCH 19/25] target/arm: Unexport aarch64_add_*_properties

These functions are not used outside cpu64.c,
so make them static.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-17-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h   | 3 ---
 target/arm/cpu64.c | 4 ++--
 2 files changed, 2 insertions(+), 5 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 675c49f93e..d2b005f76c 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -1111,8 +1111,6 @@ int aarch64_cpu_gdb_write_register(CPUState *cpu, uint8_t *buf, int reg);
 void aarch64_sve_narrow_vq(CPUARMState *env, unsigned vq);
 void aarch64_sve_change_el(CPUARMState *env, int old_el,
                            int new_el, bool el0_a64);
-void aarch64_add_sve_properties(Object *obj);
-void aarch64_add_pauth_properties(Object *obj);
 void arm_reset_sve_state(CPUARMState *env);
 
 /*
@@ -1144,7 +1142,6 @@ static inline void aarch64_sve_narrow_vq(CPUARMState *env, unsigned vq) { }
 static inline void aarch64_sve_change_el(CPUARMState *env, int o,
                                          int n, bool a)
 { }
-static inline void aarch64_add_sve_properties(Object *obj) { }
 #endif
 
 void aarch64_sync_32_to_64(CPUARMState *env);
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index b15a0d398a..6f6ee57a91 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -689,7 +689,7 @@ static void cpu_arm_get_default_vec_len(Object *obj, Visitor *v,
 }
 #endif
 
-void aarch64_add_sve_properties(Object *obj)
+static void aarch64_add_sve_properties(Object *obj)
 {
     ARMCPU *cpu = ARM_CPU(obj);
     uint32_t vq;
@@ -752,7 +752,7 @@ static Property arm_cpu_pauth_property =
 static Property arm_cpu_pauth_impdef_property =
     DEFINE_PROP_BOOL("pauth-impdef", ARMCPU, prop_pauth_impdef, false);
 
-void aarch64_add_pauth_properties(Object *obj)
+static void aarch64_add_pauth_properties(Object *obj)
 {
     ARMCPU *cpu = ARM_CPU(obj);
 

From e74c097638d38b46d9c68f11565432034afc0ad0 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:52:01 -0700
Subject: [PATCH 20/25] target/arm: Add cpu properties for SME

Mirror the properties for SVE.  The main difference is
that any arbitrary set of powers of 2 may be supported,
and not the stricter constraints that apply to SVE.

Include a property to control FEAT_SME_FA64, as failing
to restrict the runtime to the proper subset of insns
could be a major point for bugs.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20220620175235.60881-18-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 docs/system/arm/cpu-features.rst |  56 +++++++++++++++
 target/arm/cpu.c                 |  14 +++-
 target/arm/cpu.h                 |   2 +
 target/arm/cpu64.c               | 114 +++++++++++++++++++++++++++++--
 target/arm/internals.h           |   1 +
 5 files changed, 180 insertions(+), 7 deletions(-)

diff --git a/docs/system/arm/cpu-features.rst b/docs/system/arm/cpu-features.rst
index 3e626c4b68..3fd76fa0b4 100644
--- a/docs/system/arm/cpu-features.rst
+++ b/docs/system/arm/cpu-features.rst
@@ -372,6 +372,31 @@ verbose command lines.  However, the recommended way to select vector
 lengths is to explicitly enable each desired length.  Therefore only
 example's (1), (4), and (6) exhibit recommended uses of the properties.
 
+SME CPU Property Examples
+-------------------------
+
+  1) Disable SME::
+
+     $ qemu-system-aarch64 -M virt -cpu max,sme=off
+
+  2) Implicitly enable all vector lengths for the ``max`` CPU type::
+
+     $ qemu-system-aarch64 -M virt -cpu max
+
+  3) Only enable the 256-bit vector length::
+
+     $ qemu-system-aarch64 -M virt -cpu max,sme256=on
+
+  3) Enable the 256-bit and 1024-bit vector lengths::
+
+     $ qemu-system-aarch64 -M virt -cpu max,sme256=on,sme1024=on
+
+  4) Disable the 512-bit vector length.  This results in all the other
+     lengths supported by ``max`` defaulting to enabled
+     (128, 256, 1024 and 2048)::
+
+     $ qemu-system-aarch64 -M virt -cpu max,sve512=off
+
 SVE User-mode Default Vector Length Property
 --------------------------------------------
 
@@ -387,3 +412,34 @@ length supported by QEMU is 256.
 
 If this property is set to ``-1`` then the default vector length
 is set to the maximum possible length.
+
+SME CPU Properties
+==================
+
+The SME CPU properties are much like the SVE properties: ``sme`` is
+used to enable or disable the entire SME feature, and ``sme<N>`` is
+used to enable or disable specific vector lengths.  Finally,
+``sme_fa64`` is used to enable or disable ``FEAT_SME_FA64``, which
+allows execution of the "full a64" instruction set while Streaming
+SVE mode is enabled.
+
+SME is not supported by KVM at this time.
+
+At least one vector length must be enabled when ``sme`` is enabled,
+and all vector lengths must be powers of 2.  The maximum vector
+length supported by qemu is 2048 bits.  Otherwise, there are no
+additional constraints on the set of vector lengths supported by SME.
+
+SME User-mode Default Vector Length Property
+--------------------------------------------
+
+For qemu-aarch64, the cpu propery ``sme-default-vector-length=N`` is
+defined to mirror the Linux kernel parameter file
+``/proc/sys/abi/sme_default_vector_length``.  The default length, ``N``,
+is in units of bytes and must be between 16 and 8192.
+If not specified, the default vector length is 32.
+
+As with ``sve-default-vector-length``, if the default length is larger
+than the maximum vector length enabled, the actual vector length will
+be reduced.  If this property is set to ``-1`` then the default vector
+length is set to the maximum possible length.
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 660fd8b8b9..bb44ad45aa 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -1123,11 +1123,13 @@ static void arm_cpu_initfn(Object *obj)
 #ifdef CONFIG_USER_ONLY
 # ifdef TARGET_AARCH64
     /*
-     * The linux kernel defaults to 512-bit vectors, when sve is supported.
-     * See documentation for /proc/sys/abi/sve_default_vector_length, and
-     * our corresponding sve-default-vector-length cpu property.
+     * The linux kernel defaults to 512-bit for SVE, and 256-bit for SME.
+     * These values were chosen to fit within the default signal frame.
+     * See documentation for /proc/sys/abi/{sve,sme}_default_vector_length,
+     * and our corresponding cpu property.
      */
     cpu->sve_default_vq = 4;
+    cpu->sme_default_vq = 2;
 # endif
 #else
     /* Our inbound IRQ and FIQ lines */
@@ -1430,6 +1432,12 @@ void arm_cpu_finalize_features(ARMCPU *cpu, Error **errp)
             return;
         }
 
+        arm_cpu_sme_finalize(cpu, &local_err);
+        if (local_err != NULL) {
+            error_propagate(errp, local_err);
+            return;
+        }
+
         arm_cpu_pauth_finalize(cpu, &local_err);
         if (local_err != NULL) {
             error_propagate(errp, local_err);
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index d2b005f76c..c018f97b77 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -1060,9 +1060,11 @@ struct ArchCPU {
 #ifdef CONFIG_USER_ONLY
     /* Used to set the default vector length at process start. */
     uint32_t sve_default_vq;
+    uint32_t sme_default_vq;
 #endif
 
     ARMVQMap sve_vq;
+    ARMVQMap sme_vq;
 
     /* Generic timer counter frequency, in Hz */
     uint64_t gt_cntfrq_hz;
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 6f6ee57a91..19188d6cc2 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -589,10 +589,13 @@ static void cpu_arm_get_vq(Object *obj, Visitor *v, const char *name,
     ARMCPU *cpu = ARM_CPU(obj);
     ARMVQMap *vq_map = opaque;
     uint32_t vq = atoi(&name[3]) / 128;
+    bool sve = vq_map == &cpu->sve_vq;
     bool value;
 
-    /* All vector lengths are disabled when SVE is off. */
-    if (!cpu_isar_feature(aa64_sve, cpu)) {
+    /* All vector lengths are disabled when feature is off. */
+    if (sve
+        ? !cpu_isar_feature(aa64_sve, cpu)
+        : !cpu_isar_feature(aa64_sme, cpu)) {
         value = false;
     } else {
         value = extract32(vq_map->map, vq - 1, 1);
@@ -636,8 +639,80 @@ static void cpu_arm_set_sve(Object *obj, bool value, Error **errp)
     cpu->isar.id_aa64pfr0 = t;
 }
 
+void arm_cpu_sme_finalize(ARMCPU *cpu, Error **errp)
+{
+    uint32_t vq_map = cpu->sme_vq.map;
+    uint32_t vq_init = cpu->sme_vq.init;
+    uint32_t vq_supported = cpu->sme_vq.supported;
+    uint32_t vq;
+
+    if (vq_map == 0) {
+        if (!cpu_isar_feature(aa64_sme, cpu)) {
+            cpu->isar.id_aa64smfr0 = 0;
+            return;
+        }
+
+        /* TODO: KVM will require limitations via SMCR_EL2. */
+        vq_map = vq_supported & ~vq_init;
+
+        if (vq_map == 0) {
+            vq = ctz32(vq_supported) + 1;
+            error_setg(errp, "cannot disable sme%d", vq * 128);
+            error_append_hint(errp, "All SME vector lengths are disabled.\n");
+            error_append_hint(errp, "With SME enabled, at least one "
+                              "vector length must be enabled.\n");
+            return;
+        }
+    } else {
+        if (!cpu_isar_feature(aa64_sme, cpu)) {
+            vq = 32 - clz32(vq_map);
+            error_setg(errp, "cannot enable sme%d", vq * 128);
+            error_append_hint(errp, "SME must be enabled to enable "
+                              "vector lengths.\n");
+            error_append_hint(errp, "Add sme=on to the CPU property list.\n");
+            return;
+        }
+        /* TODO: KVM will require limitations via SMCR_EL2. */
+    }
+
+    cpu->sme_vq.map = vq_map;
+}
+
+static bool cpu_arm_get_sme(Object *obj, Error **errp)
+{
+    ARMCPU *cpu = ARM_CPU(obj);
+    return cpu_isar_feature(aa64_sme, cpu);
+}
+
+static void cpu_arm_set_sme(Object *obj, bool value, Error **errp)
+{
+    ARMCPU *cpu = ARM_CPU(obj);
+    uint64_t t;
+
+    t = cpu->isar.id_aa64pfr1;
+    t = FIELD_DP64(t, ID_AA64PFR1, SME, value);
+    cpu->isar.id_aa64pfr1 = t;
+}
+
+static bool cpu_arm_get_sme_fa64(Object *obj, Error **errp)
+{
+    ARMCPU *cpu = ARM_CPU(obj);
+    return cpu_isar_feature(aa64_sme, cpu) &&
+           cpu_isar_feature(aa64_sme_fa64, cpu);
+}
+
+static void cpu_arm_set_sme_fa64(Object *obj, bool value, Error **errp)
+{
+    ARMCPU *cpu = ARM_CPU(obj);
+    uint64_t t;
+
+    t = cpu->isar.id_aa64smfr0;
+    t = FIELD_DP64(t, ID_AA64SMFR0, FA64, value);
+    cpu->isar.id_aa64smfr0 = t;
+}
+
 #ifdef CONFIG_USER_ONLY
-/* Mirror linux /proc/sys/abi/sve_default_vector_length. */
+/* Mirror linux /proc/sys/abi/{sve,sme}_default_vector_length. */
 static void cpu_arm_set_default_vec_len(Object *obj, Visitor *v,
                                         const char *name, void *opaque,
                                         Error **errp)
@@ -663,7 +738,11 @@ static void cpu_arm_set_default_vec_len(Object *obj, Visitor *v,
      * and is the maximum architectural width of ZCR_ELx.LEN.
      */
     if (remainder || default_vq < 1 || default_vq > 512) {
-        error_setg(errp, "cannot set sve-default-vector-length");
+        ARMCPU *cpu = ARM_CPU(obj);
+        const char *which =
+            (ptr_default_vq == &cpu->sve_default_vq ? "sve" : "sme");
+
+        error_setg(errp, "cannot set %s-default-vector-length", which);
         if (remainder) {
             error_append_hint(errp, "Vector length not a multiple of 16\n");
         } else if (default_vq < 1) {
@@ -712,6 +791,31 @@ static void aarch64_add_sve_properties(Object *obj)
 #endif
 }
 
+static void aarch64_add_sme_properties(Object *obj)
+{
+    ARMCPU *cpu = ARM_CPU(obj);
+    uint32_t vq;
+
+    object_property_add_bool(obj, "sme", cpu_arm_get_sme, cpu_arm_set_sme);
+    object_property_add_bool(obj, "sme_fa64", cpu_arm_get_sme_fa64,
+                             cpu_arm_set_sme_fa64);
+
+    for (vq = 1; vq <= ARM_MAX_VQ; vq <<= 1) {
+        char name[8];
+        sprintf(name, "sme%d", vq * 128);
+        object_property_add(obj, name, "bool", cpu_arm_get_vq,
+                            cpu_arm_set_vq, NULL, &cpu->sme_vq);
+    }
+
+#ifdef CONFIG_USER_ONLY
+    /* Mirror linux /proc/sys/abi/sme_default_vector_length. */
+    object_property_add(obj, "sme-default-vector-length", "int32",
+                        cpu_arm_get_default_vec_len,
+                        cpu_arm_set_default_vec_len, NULL,
+                        &cpu->sme_default_vq);
+#endif
+}
+
 void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp)
 {
     int arch_val = 0, impdef_val = 0;
@@ -977,9 +1081,11 @@ static void aarch64_max_initfn(Object *obj)
 #endif
 
     cpu->sve_vq.supported = MAKE_64BIT_MASK(0, ARM_MAX_VQ);
+    cpu->sme_vq.supported = SVE_VQ_POW2_MAP;
 
     aarch64_add_pauth_properties(obj);
     aarch64_add_sve_properties(obj);
+    aarch64_add_sme_properties(obj);
     object_property_add(obj, "sve-max-vq", "uint32", cpu_max_get_sve_max_vq,
                         cpu_max_set_sve_max_vq, NULL, NULL);
     qdev_property_add_static(DEVICE(obj), &arm_cpu_lpa2_property);
diff --git a/target/arm/internals.h b/target/arm/internals.h
index aef568adf7..c66f74a0db 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -1289,6 +1289,7 @@ int arm_gdb_set_svereg(CPUARMState *env, uint8_t *buf, int reg);
 int aarch64_fpu_gdb_get_reg(CPUARMState *env, GByteArray *buf, int reg);
 int aarch64_fpu_gdb_set_reg(CPUARMState *env, uint8_t *buf, int reg);
 void arm_cpu_sve_finalize(ARMCPU *cpu, Error **errp);
+void arm_cpu_sme_finalize(ARMCPU *cpu, Error **errp);
 void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp);
 void arm_cpu_lpa2_finalize(ARMCPU *cpu, Error **errp);
 #endif

From 6ca54aa9a882ece5a6bcf5879f25bdcd7a95331f Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:52:02 -0700
Subject: [PATCH 21/25] target/arm: Introduce sve_vqm1_for_el_sm

When Streaming SVE mode is enabled, the size is taken from
SMCR_ELx instead of ZCR_ELx.  The format is shared, but the
set of vector lengths is not.  Further, Streaming SVE does
not require any particular length to be supported.

Adjust sve_vqm1_for_el to pass the current value of PSTATE.SM
to the new function.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-19-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h    |  9 +++++++--
 target/arm/helper.c | 32 +++++++++++++++++++++++++-------
 2 files changed, 32 insertions(+), 9 deletions(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index c018f97b77..0295e85483 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -1154,13 +1154,18 @@ int sve_exception_el(CPUARMState *env, int cur_el);
 int sme_exception_el(CPUARMState *env, int cur_el);
 
 /**
- * sve_vqm1_for_el:
+ * sve_vqm1_for_el_sm:
  * @env: CPUARMState
  * @el: exception level
+ * @sm: streaming mode
  *
- * Compute the current SVE vector length for @el, in units of
+ * Compute the current vector length for @el & @sm, in units of
  * Quadwords Minus 1 -- the same scale used for ZCR_ELx.LEN.
+ * If @sm, compute for SVL, otherwise NVL.
  */
+uint32_t sve_vqm1_for_el_sm(CPUARMState *env, int el, bool sm);
+
+/* Likewise, but using @sm = PSTATE.SM. */
 uint32_t sve_vqm1_for_el(CPUARMState *env, int el);
 
 static inline bool is_a64(CPUARMState *env)
diff --git a/target/arm/helper.c b/target/arm/helper.c
index a80ca461e5..2e4e739969 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -6272,23 +6272,41 @@ int sme_exception_el(CPUARMState *env, int el)
 /*
  * Given that SVE is enabled, return the vector length for EL.
  */
-uint32_t sve_vqm1_for_el(CPUARMState *env, int el)
+uint32_t sve_vqm1_for_el_sm(CPUARMState *env, int el, bool sm)
 {
     ARMCPU *cpu = env_archcpu(env);
-    uint32_t len = cpu->sve_max_vq - 1;
+    uint64_t *cr = env->vfp.zcr_el;
+    uint32_t map = cpu->sve_vq.map;
+    uint32_t len = ARM_MAX_VQ - 1;
+
+    if (sm) {
+        cr = env->vfp.smcr_el;
+        map = cpu->sme_vq.map;
+    }
 
     if (el <= 1 && !el_is_in_host(env, el)) {
-        len = MIN(len, 0xf & (uint32_t)env->vfp.zcr_el[1]);
+        len = MIN(len, 0xf & (uint32_t)cr[1]);
     }
     if (el <= 2 && arm_feature(env, ARM_FEATURE_EL2)) {
-        len = MIN(len, 0xf & (uint32_t)env->vfp.zcr_el[2]);
+        len = MIN(len, 0xf & (uint32_t)cr[2]);
     }
     if (arm_feature(env, ARM_FEATURE_EL3)) {
-        len = MIN(len, 0xf & (uint32_t)env->vfp.zcr_el[3]);
+        len = MIN(len, 0xf & (uint32_t)cr[3]);
     }
 
-    len = 31 - clz32(cpu->sve_vq.map & MAKE_64BIT_MASK(0, len + 1));
-    return len;
+    map &= MAKE_64BIT_MASK(0, len + 1);
+    if (map != 0) {
+        return 31 - clz32(map);
+    }
+
+    /* Bit 0 is always set for Normal SVE -- not so for Streaming SVE. */
+    assert(sm);
+    return ctz32(cpu->sme_vq.map);
+}
+
+uint32_t sve_vqm1_for_el(CPUARMState *env, int el)
+{
+    return sve_vqm1_for_el_sm(env, el, FIELD_EX64(env->svcr, SVCR, SM));
 }
 
 static void zcr_write(CPUARMState *env, const ARMCPRegInfo *ri,

From 5d7953adcfb30196ba684d3af69271528630367f Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:52:03 -0700
Subject: [PATCH 22/25] target/arm: Add SVL to TB flags

We need SVL separate from VL for RDSVL et al, as well as
ZA storage loads and stores, which do not require PSTATE.SM.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-20-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/cpu.h           | 12 ++++++++++++
 target/arm/helper.c        |  8 +++++++-
 target/arm/translate-a64.c |  1 +
 target/arm/translate.h     |  1 +
 4 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 0295e85483..4a4342f262 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -3182,6 +3182,7 @@ FIELD(TBFLAG_A64, MTE0_ACTIVE, 19, 1)
 FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2)
 FIELD(TBFLAG_A64, PSTATE_SM, 22, 1)
 FIELD(TBFLAG_A64, PSTATE_ZA, 23, 1)
+FIELD(TBFLAG_A64, SVL, 24, 4)
 
 /*
  * Helpers for using the above.
@@ -3227,6 +3228,17 @@ static inline int sve_vq(CPUARMState *env)
     return EX_TBFLAG_A64(env->hflags, VL) + 1;
 }
 
+/**
+ * sme_vq
+ * @env: the cpu context
+ *
+ * Return the SVL cached within env->hflags, in units of quadwords.
+ */
+static inline int sme_vq(CPUARMState *env)
+{
+    return EX_TBFLAG_A64(env->hflags, SVL) + 1;
+}
+
 static inline bool bswap_code(bool sctlr_b)
 {
 #ifdef CONFIG_USER_ONLY
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 2e4e739969..d2886a123a 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -11352,7 +11352,13 @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el,
         DP_TBFLAG_A64(flags, SVEEXC_EL, sve_el);
     }
     if (cpu_isar_feature(aa64_sme, env_archcpu(env))) {
-        DP_TBFLAG_A64(flags, SMEEXC_EL, sme_exception_el(env, el));
+        int sme_el = sme_exception_el(env, el);
+
+        DP_TBFLAG_A64(flags, SMEEXC_EL, sme_el);
+        if (sme_el == 0) {
+            /* Similarly, do not compute SVL if SME is disabled. */
+            DP_TBFLAG_A64(flags, SVL, sve_vqm1_for_el_sm(env, el, true));
+        }
         if (FIELD_EX64(env->svcr, SVCR, SM)) {
             DP_TBFLAG_A64(flags, PSTATE_SM, 1);
         }
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index c050ebe005..c86b97b1d4 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -14647,6 +14647,7 @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase,
     dc->sve_excp_el = EX_TBFLAG_A64(tb_flags, SVEEXC_EL);
     dc->sme_excp_el = EX_TBFLAG_A64(tb_flags, SMEEXC_EL);
     dc->vl = (EX_TBFLAG_A64(tb_flags, VL) + 1) * 16;
+    dc->svl = (EX_TBFLAG_A64(tb_flags, SVL) + 1) * 16;
     dc->pauth_active = EX_TBFLAG_A64(tb_flags, PAUTH_ACTIVE);
     dc->bt = EX_TBFLAG_A64(tb_flags, BT);
     dc->btype = EX_TBFLAG_A64(tb_flags, BTYPE);
diff --git a/target/arm/translate.h b/target/arm/translate.h
index 93766649f7..22fd882368 100644
--- a/target/arm/translate.h
+++ b/target/arm/translate.h
@@ -44,6 +44,7 @@ typedef struct DisasContext {
     int sve_excp_el; /* SVE exception EL or 0 if enabled */
     int sme_excp_el; /* SME exception EL or 0 if enabled */
     int vl;          /* current vector length in bytes */
+    int svl;         /* current streaming vector length in bytes */
     bool vfp_enabled; /* FP enabled via FPSCR.EN */
     int vec_len;
     int vec_stride;

From d61d1b8600caea833c377b31aef61484ccf9e414 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Mon, 20 Jun 2022 10:52:04 -0700
Subject: [PATCH 23/25] target/arm: Move pred_{full, gvec}_reg_{offset, size}
 to translate-a64.h

We will need these functions in translate-sme.c.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-21-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/translate-a64.h | 38 ++++++++++++++++++++++++++++++++++++++
 target/arm/translate-sve.c | 36 ------------------------------------
 2 files changed, 38 insertions(+), 36 deletions(-)

diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h
index dbc917ee65..f0970c6b8c 100644
--- a/target/arm/translate-a64.h
+++ b/target/arm/translate-a64.h
@@ -107,6 +107,44 @@ static inline int vec_full_reg_size(DisasContext *s)
     return s->vl;
 }
 
+/*
+ * Return the offset info CPUARMState of the predicate vector register Pn.
+ * Note for this purpose, FFR is P16.
+ */
+static inline int pred_full_reg_offset(DisasContext *s, int regno)
+{
+    return offsetof(CPUARMState, vfp.pregs[regno]);
+}
+
+/* Return the byte size of the whole predicate register, VL / 64.  */
+static inline int pred_full_reg_size(DisasContext *s)
+{
+    return s->vl >> 3;
+}
+
+/*
+ * Round up the size of a register to a size allowed by
+ * the tcg vector infrastructure.  Any operation which uses this
+ * size may assume that the bits above pred_full_reg_size are zero,
+ * and must leave them the same way.
+ *
+ * Note that this is not needed for the vector registers as they
+ * are always properly sized for tcg vectors.
+ */
+static inline int size_for_gvec(int size)
+{
+    if (size <= 8) {
+        return 8;
+    } else {
+        return QEMU_ALIGN_UP(size, 16);
+    }
+}
+
+static inline int pred_gvec_reg_size(DisasContext *s)
+{
+    return size_for_gvec(pred_full_reg_size(s));
+}
+
 bool disas_sve(DisasContext *, uint32_t);
 
 void gen_gvec_rax1(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 67761bf2cc..62b5f3040c 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -100,42 +100,6 @@ static inline int msz_dtype(DisasContext *s, int msz)
  * Implement all of the translator functions referenced by the decoder.
  */
 
-/* Return the offset info CPUARMState of the predicate vector register Pn.
- * Note for this purpose, FFR is P16.
- */
-static inline int pred_full_reg_offset(DisasContext *s, int regno)
-{
-    return offsetof(CPUARMState, vfp.pregs[regno]);
-}
-
-/* Return the byte size of the whole predicate register, VL / 64.  */
-static inline int pred_full_reg_size(DisasContext *s)
-{
-    return s->vl >> 3;
-}
-
-/* Round up the size of a register to a size allowed by
- * the tcg vector infrastructure.  Any operation which uses this
- * size may assume that the bits above pred_full_reg_size are zero,
- * and must leave them the same way.
- *
- * Note that this is not needed for the vector registers as they
- * are always properly sized for tcg vectors.
- */
-static int size_for_gvec(int size)
-{
-    if (size <= 8) {
-        return 8;
-    } else {
-        return QEMU_ALIGN_UP(size, 16);
-    }
-}
-
-static int pred_gvec_reg_size(DisasContext *s)
-{
-    return size_for_gvec(pred_full_reg_size(s));
-}
-
 /* Invoke an out-of-line helper on 2 Zregs. */
 static bool gen_gvec_ool_zz(DisasContext *s, gen_helper_gvec_2 *fn,
                             int rd, int rn, int data)

From 22536b13247cf041b6dcabf0d708f486058989a9 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Sat, 18 Jun 2022 17:15:40 -0700
Subject: [PATCH 24/25] target/arm: Extend arm_pamax to more than aarch64

Move the code from hw/arm/virt.c that is supposed
to handle v7 into the one function.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reported-by: He Zhe <zhe.he@windriver.com>
Message-id: 20220619001541.131672-2-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 hw/arm/virt.c    | 10 +---------
 target/arm/ptw.c | 24 ++++++++++++++++--------
 2 files changed, 17 insertions(+), 17 deletions(-)

diff --git a/hw/arm/virt.c b/hw/arm/virt.c
index 097238faa7..5502aa60c8 100644
--- a/hw/arm/virt.c
+++ b/hw/arm/virt.c
@@ -2010,15 +2010,7 @@ static void machvirt_init(MachineState *machine)
         cpuobj = object_new(possible_cpus->cpus[0].type);
         armcpu = ARM_CPU(cpuobj);
 
-        if (object_property_get_bool(cpuobj, "aarch64", NULL)) {
-            pa_bits = arm_pamax(armcpu);
-        } else if (arm_feature(&armcpu->env, ARM_FEATURE_LPAE)) {
-            /* v7 with LPAE */
-            pa_bits = 40;
-        } else {
-            /* Anything else */
-            pa_bits = 32;
-        }
+        pa_bits = arm_pamax(armcpu);
 
         object_unref(cpuobj);
 
diff --git a/target/arm/ptw.c b/target/arm/ptw.c
index 4d97a24808..07f7a21861 100644
--- a/target/arm/ptw.c
+++ b/target/arm/ptw.c
@@ -36,15 +36,23 @@ static const uint8_t pamax_map[] = {
 /* The cpu-specific constant value of PAMax; also used by hw/arm/virt. */
 unsigned int arm_pamax(ARMCPU *cpu)
 {
-    unsigned int parange =
-        FIELD_EX64(cpu->isar.id_aa64mmfr0, ID_AA64MMFR0, PARANGE);
+    if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64)) {
+        unsigned int parange =
+            FIELD_EX64(cpu->isar.id_aa64mmfr0, ID_AA64MMFR0, PARANGE);
 
-    /*
-     * id_aa64mmfr0 is a read-only register so values outside of the
-     * supported mappings can be considered an implementation error.
-     */
-    assert(parange < ARRAY_SIZE(pamax_map));
-    return pamax_map[parange];
+        /*
+         * id_aa64mmfr0 is a read-only register so values outside of the
+         * supported mappings can be considered an implementation error.
+         */
+        assert(parange < ARRAY_SIZE(pamax_map));
+        return pamax_map[parange];
+    }
+    if (arm_feature(&cpu->env, ARM_FEATURE_LPAE)) {
+        /* v7 with LPAE */
+        return 40;
+    }
+    /* Anything else */
+    return 32;
 }
 
 /*

From 59e1b8a22ea9f947d038ccac784de1020f266e14 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Sat, 18 Jun 2022 17:15:41 -0700
Subject: [PATCH 25/25] target/arm: Check V7VE as well as LPAE in arm_pamax

In machvirt_init we create a cpu but do not fully initialize it.
Thus the propagation of V7VE to LPAE has not been done, and we
compute the wrong value for some v7 cpus, e.g. cortex-a15.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1078
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reported-by: He Zhe <zhe.he@windriver.com>
Message-id: 20220619001541.131672-3-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/ptw.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/target/arm/ptw.c b/target/arm/ptw.c
index 07f7a21861..da478104f0 100644
--- a/target/arm/ptw.c
+++ b/target/arm/ptw.c
@@ -47,7 +47,13 @@ unsigned int arm_pamax(ARMCPU *cpu)
         assert(parange < ARRAY_SIZE(pamax_map));
         return pamax_map[parange];
     }
-    if (arm_feature(&cpu->env, ARM_FEATURE_LPAE)) {
+
+    /*
+     * In machvirt_init, we call arm_pamax on a cpu that is not fully
+     * initialized, so we can't rely on the propagation done in realize.
+     */
+    if (arm_feature(&cpu->env, ARM_FEATURE_LPAE) ||
+        arm_feature(&cpu->env, ARM_FEATURE_V7VE)) {
         /* v7 with LPAE */
         return 40;
     }