arm/kvm: add support for MTE

Extend the 'mte' property for the virt machine to cover KVM as
well. For KVM, we don't allocate tag memory, but instead enable
the capability.

If MTE has been enabled, we need to disable migration, as we do not
yet have a way to migrate the tags as well. Therefore, MTE will stay
off with KVM unless requested explicitly.

[gankulkarni: This patch is rework of commit b320e21c48
which broke TCG since it made the TCG -cpu max
report the presence of MTE to the guest even if the board hadn't
enabled MTE by wiring up the tag RAM. This meant that if the guest
then tried to use MTE QEMU would segfault accessing the
non-existent tag RAM.]

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Gustavo Romero <gustavo.romero@linaro.org>
Signed-off-by: Ganapatrao Kulkarni <gankulkarni@os.amperecomputing.com>
Message-id: 20241008114302.4855-1-gankulkarni@os.amperecomputing.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

diff --git a/hw/arm/virt.c b/hw/arm/virt.c
index 8b2b991d978c7f875fa4b8c445094050f7d6f9fc..1a381e9a2bd71862cb93ea7d9861ced0b0ee79af 100644 (file)
--- a/hw/arm/virt.c
+++ b/hw/arm/virt.c
@@ -2213,7 +2213,7 @@ static void machvirt_init(MachineState *machine)
         exit(1);
     }
 
-    if (vms->mte && (kvm_enabled() || hvf_enabled())) {
+    if (vms->mte && hvf_enabled()) {
         error_report("mach-virt: %s does not support providing "
                      "MTE to the guest CPU",
                      current_accel_name());
@@ -2283,39 +2283,51 @@ static void machvirt_init(MachineState *machine)
         }
 
         if (vms->mte) {
-            /* Create the memory region only once, but link to all cpus. */
-            if (!tag_sysmem) {
-                /*
-                 * The property exists only if MemTag is supported.
-                 * If it is, we must allocate the ram to back that up.
-                 */
-                if (!object_property_find(cpuobj, "tag-memory")) {
-                    error_report("MTE requested, but not supported "
-                                 "by the guest CPU");
-                    exit(1);
+            if (tcg_enabled()) {
+                /* Create the memory region only once, but link to all cpus. */
+                if (!tag_sysmem) {
+                    /*
+                     * The property exists only if MemTag is supported.
+                     * If it is, we must allocate the ram to back that up.
+                     */
+                    if (!object_property_find(cpuobj, "tag-memory")) {
+                        error_report("MTE requested, but not supported "
+                                     "by the guest CPU");
+                        exit(1);
+                    }
+
+                    tag_sysmem = g_new(MemoryRegion, 1);
+                    memory_region_init(tag_sysmem, OBJECT(machine),
+                                       "tag-memory", UINT64_MAX / 32);
+
+                    if (vms->secure) {
+                        secure_tag_sysmem = g_new(MemoryRegion, 1);
+                        memory_region_init(secure_tag_sysmem, OBJECT(machine),
+                                           "secure-tag-memory",
+                                           UINT64_MAX / 32);
+
+                        /* As with ram, secure-tag takes precedence over tag. */
+                        memory_region_add_subregion_overlap(secure_tag_sysmem,
+                                                            0, tag_sysmem, -1);
+                    }
                 }
 
-                tag_sysmem = g_new(MemoryRegion, 1);
-                memory_region_init(tag_sysmem, OBJECT(machine),
-                                   "tag-memory", UINT64_MAX / 32);
-
+                object_property_set_link(cpuobj, "tag-memory",
+                                         OBJECT(tag_sysmem), &error_abort);
                 if (vms->secure) {
-                    secure_tag_sysmem = g_new(MemoryRegion, 1);
-                    memory_region_init(secure_tag_sysmem, OBJECT(machine),
-                                       "secure-tag-memory", UINT64_MAX / 32);
-
-                    /* As with ram, secure-tag takes precedence over tag.  */
-                    memory_region_add_subregion_overlap(secure_tag_sysmem, 0,
-                                                        tag_sysmem, -1);
+                    object_property_set_link(cpuobj, "secure-tag-memory",
+                                             OBJECT(secure_tag_sysmem),
+                                             &error_abort);
                 }
-            }
-
-            object_property_set_link(cpuobj, "tag-memory", OBJECT(tag_sysmem),
-                                     &error_abort);
-            if (vms->secure) {
-                object_property_set_link(cpuobj, "secure-tag-memory",
-                                         OBJECT(secure_tag_sysmem),
-                                         &error_abort);
+            } else if (kvm_enabled()) {
+                if (!kvm_arm_mte_supported()) {
+                    error_report("MTE requested, but not supported by KVM");
+                    exit(1);
+                }
+                kvm_arm_enable_mte(cpuobj, &error_abort);
+            } else {
+                    error_report("MTE requested, but not supported ");
+                    exit(1);
             }
         }
 

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 1320fd8c8fea8c8c3b087e67e9d2b8631787e682..5b751439bdc73b47382df45824b283cee55912dc 100644 (file)
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -2390,14 +2390,22 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
 
 #ifndef CONFIG_USER_ONLY
         /*
-         * If we do not have tag-memory provided by the machine,
-         * reduce MTE support to instructions enabled at EL0.
+         * If we run with TCG and do not have tag-memory provided by
+         * the machine, then reduce MTE support to instructions enabled at EL0.
          * This matches Cortex-A710 BROADCASTMTE input being LOW.
          */
-        if (cpu->tag_memory == NULL) {
+        if (tcg_enabled() && cpu->tag_memory == NULL) {
             cpu->isar.id_aa64pfr1 =
                 FIELD_DP64(cpu->isar.id_aa64pfr1, ID_AA64PFR1, MTE, 1);
         }
+
+        /*
+         * If MTE is supported by the host, however it should not be
+         * enabled on the guest (i.e mte=off), clear guest's MTE bits."
+         */
+        if (kvm_enabled() && !cpu->kvm_mte) {
+                FIELD_DP64(cpu->isar.id_aa64pfr1, ID_AA64PFR1, MTE, 0);
+        }
 #endif
     }
 

diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index f065756c5c7d0e5d8e2001902b1ff812a386abce..8fc8b6398f7f4d45e4f7eb55e6b814c40f9e5ecf 100644 (file)
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -922,6 +922,8 @@ struct ArchCPU {
 
     /* CPU has memory protection unit */
     bool has_mpu;
+    /* CPU has MTE enabled in KVM mode */
+    bool kvm_mte;
     /* PMSAv7 MPU number of supported regions */
     uint32_t pmsav7_dregion;
     /* PMSAv8 MPU number of supported hyp regions */

diff --git a/target/arm/kvm.c b/target/arm/kvm.c
index f1f1b5b375a55ed5fae03b0fe81b85ce4bb9b456..000afa03631c2fadc1fdea6a2dd125f10f48902a 100644 (file)
--- a/target/arm/kvm.c
+++ b/target/arm/kvm.c
@@ -39,6 +39,7 @@
 #include "hw/acpi/acpi.h"
 #include "hw/acpi/ghes.h"
 #include "target/arm/gtimer.h"
+#include "migration/blocker.h"
 
 const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_LAST_INFO
@@ -119,6 +120,21 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try,
     if (vmfd < 0) {
         goto err;
     }
+
+    /*
+     * The MTE capability must be enabled by the VMM before creating
+     * any VCPUs in order to allow the MTE bits of the ID_AA64PFR1
+     * register to be probed correctly, as they are masked if MTE
+     * is not enabled.
+     */
+    if (kvm_arm_mte_supported()) {
+        KVMState kvm_state;
+
+        kvm_state.fd = kvmfd;
+        kvm_state.vmfd = vmfd;
+        kvm_vm_enable_cap(&kvm_state, KVM_CAP_ARM_MTE, 0);
+    }
+
     cpufd = ioctl(vmfd, KVM_CREATE_VCPU, 0);
     if (cpufd < 0) {
         goto err;
@@ -1793,6 +1809,11 @@ bool kvm_arm_sve_supported(void)
     return kvm_check_extension(kvm_state, KVM_CAP_ARM_SVE);
 }
 
+bool kvm_arm_mte_supported(void)
+{
+    return kvm_check_extension(kvm_state, KVM_CAP_ARM_MTE);
+}
+
 QEMU_BUILD_BUG_ON(KVM_ARM64_SVE_VQ_MIN != 1);
 
 uint32_t kvm_arm_sve_get_vls(ARMCPU *cpu)
@@ -2417,3 +2438,40 @@ int kvm_arch_remove_sw_breakpoint(CPUState *cs, struct kvm_sw_breakpoint *bp)
     }
     return 0;
 }
+
+void kvm_arm_enable_mte(Object *cpuobj, Error **errp)
+{
+    static bool tried_to_enable;
+    static bool succeeded_to_enable;
+    Error *mte_migration_blocker = NULL;
+    ARMCPU *cpu = ARM_CPU(cpuobj);
+    int ret;
+
+    if (!tried_to_enable) {
+        /*
+         * MTE on KVM is enabled on a per-VM basis (and retrying doesn't make
+         * sense), and we only want a single migration blocker as well.
+         */
+        tried_to_enable = true;
+
+        ret = kvm_vm_enable_cap(kvm_state, KVM_CAP_ARM_MTE, 0);
+        if (ret) {
+            error_setg_errno(errp, -ret, "Failed to enable KVM_CAP_ARM_MTE");
+            return;
+        }
+
+        /* TODO: Add migration support with MTE enabled */
+        error_setg(&mte_migration_blocker,
+                   "Live migration disabled due to MTE enabled");
+        if (migrate_add_blocker(&mte_migration_blocker, errp)) {
+            error_free(mte_migration_blocker);
+            return;
+        }
+
+        succeeded_to_enable = true;
+    }
+
+    if (succeeded_to_enable) {
+        cpu->kvm_mte = true;
+    }
+}

diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h
index cfaa0d9bc71a91dd9cc5b2072f80a93395c94f50..4d293618a78f29c37a0e85927bd66cdfcbeb34dc 100644 (file)
--- a/target/arm/kvm_arm.h
+++ b/target/arm/kvm_arm.h
@@ -188,6 +188,13 @@ bool kvm_arm_pmu_supported(void);
  */
 bool kvm_arm_sve_supported(void);
 
+/**
+ * kvm_arm_mte_supported:
+ *
+ * Returns: true if KVM can enable MTE, and false otherwise.
+ */
+bool kvm_arm_mte_supported(void);
+
 /**
  * kvm_arm_get_max_vm_ipa_size:
  * @ms: Machine state handle
@@ -214,6 +221,8 @@ void kvm_arm_pvtime_init(ARMCPU *cpu, uint64_t ipa);
 
 int kvm_arm_set_irq(int cpu, int irqtype, int irq, int level);
 
+void kvm_arm_enable_mte(Object *cpuobj, Error **errp);
+
 #else
 
 /*
@@ -235,6 +244,11 @@ static inline bool kvm_arm_sve_supported(void)
     return false;
 }
 
+static inline bool kvm_arm_mte_supported(void)
+{
+    return false;
+}
+
 /*
  * These functions should never actually be called without KVM support.
  */
@@ -283,6 +297,11 @@ static inline uint32_t kvm_arm_sve_get_vls(ARMCPU *cpu)
     g_assert_not_reached();
 }
 
+static inline void kvm_arm_enable_mte(Object *cpuobj, Error **errp)
+{
+    g_assert_not_reached();
+}
+
 #endif
 
 #endif