loongarch queue

-----BEGIN PGP SIGNATURE-----
 
 iHUEABYKAB0WIQQNhkKjomWfgLCz0aQfewwSUazn0QUCZ24ohgAKCRAfewwSUazn
 0cH1AP9jJB/UK+tm0E4zAXcVb7uInNa465qUbLk4F1Ep9+pLhwEApNatYEXtOxdP
 zFm0W64/hD0kppDkXG6r3W3uoAjWMAM=
 =IJGD
 -----END PGP SIGNATURE-----

Merge tag 'pull-loongarch-20241227' of https://gitlab.com/bibo-mao/qemu into staging

loongarch queue

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQQNhkKjomWfgLCz0aQfewwSUazn0QUCZ24ohgAKCRAfewwSUazn
# 0cH1AP9jJB/UK+tm0E4zAXcVb7uInNa465qUbLk4F1Ep9+pLhwEApNatYEXtOxdP
# zFm0W64/hD0kppDkXG6r3W3uoAjWMAM=
# =IJGD
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 26 Dec 2024 23:09:42 EST
# gpg:                using EDDSA key 0D8642A3A2659F80B0B3D1A41F7B0C1251ACE7D1
# gpg: Good signature from "bibo mao <maobibo@loongson.cn>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 7044 3A00 19C0 E97A 31C7  13C4 8E86 8FB7 A176 9D4C
#      Subkey fingerprint: 0D86 42A3 A265 9F80 B0B3  D1A4 1F7B 0C12 51AC E7D1

* tag 'pull-loongarch-20241227' of https://gitlab.com/bibo-mao/qemu:
  target/loongarch: Use auto method with LASX feature
  target/loongarch: Use auto method with LSX feature
  hw/loongarch/virt: Improve fdt table creation for CPU object
  hw/loongarch/virt: Create fdt table on machine creation done notification
  target/loongarch: Use actual operand size with vbsrl check
  target/loongarch: Fix vldi inst

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

hw/loongarch/virt.c

@@ -365,26 +365,35 @@ static void create_fdt(LoongArchVirtMachineState *lvms)
 static void fdt_add_cpu_nodes(const LoongArchVirtMachineState *lvms)
 {
     int num;
-    const MachineState *ms = MACHINE(lvms);
+    MachineState *ms = MACHINE(lvms);
-    int smp_cpus = ms->smp.cpus;
+    MachineClass *mc = MACHINE_GET_CLASS(ms);
+    const CPUArchIdList *possible_cpus;
+    LoongArchCPU *cpu;
+    CPUState *cs;
+    char *nodename, *map_path;
 
     qemu_fdt_add_subnode(ms->fdt, "/cpus");
     qemu_fdt_setprop_cell(ms->fdt, "/cpus", "#address-cells", 0x1);
     qemu_fdt_setprop_cell(ms->fdt, "/cpus", "#size-cells", 0x0);
 
     /* cpu nodes */
-    for (num = smp_cpus - 1; num >= 0; num--) {
+    possible_cpus = mc->possible_cpu_arch_ids(ms);
-        char *nodename = g_strdup_printf("/cpus/cpu@%d", num);
+    for (num = 0; num < possible_cpus->len; num++) {
-        LoongArchCPU *cpu = LOONGARCH_CPU(qemu_get_cpu(num));
+        cs = possible_cpus->cpus[num].cpu;
-        CPUState *cs = CPU(cpu);
+        if (cs == NULL) {
+            continue;
+        }
+
+        nodename = g_strdup_printf("/cpus/cpu@%d", num);
+        cpu = LOONGARCH_CPU(cs);
 
         qemu_fdt_add_subnode(ms->fdt, nodename);
         qemu_fdt_setprop_string(ms->fdt, nodename, "device_type", "cpu");
         qemu_fdt_setprop_string(ms->fdt, nodename, "compatible",
                                 cpu->dtb_compatible);
-        if (ms->possible_cpus->cpus[cs->cpu_index].props.has_node_id) {
+        if (possible_cpus->cpus[num].props.has_node_id) {
             qemu_fdt_setprop_cell(ms->fdt, nodename, "numa-node-id",
-                ms->possible_cpus->cpus[cs->cpu_index].props.node_id);
+                possible_cpus->cpus[num].props.node_id);
         }
         qemu_fdt_setprop_cell(ms->fdt, nodename, "reg", num);
         qemu_fdt_setprop_cell(ms->fdt, nodename, "phandle",
@@ -394,11 +403,13 @@ static void fdt_add_cpu_nodes(const LoongArchVirtMachineState *lvms)
 
     /*cpu map */
     qemu_fdt_add_subnode(ms->fdt, "/cpus/cpu-map");
+    for (num = 0; num < possible_cpus->len; num++) {
+        cs = possible_cpus->cpus[num].cpu;
+        if (cs == NULL) {
+            continue;
+        }
 
-    for (num = smp_cpus - 1; num >= 0; num--) {
+        nodename = g_strdup_printf("/cpus/cpu@%d", num);
-        char *cpu_path = g_strdup_printf("/cpus/cpu@%d", num);
-        char *map_path;
-
         if (ms->smp.threads > 1) {
             map_path = g_strdup_printf(
                 "/cpus/cpu-map/socket%d/core%d/thread%d",
@@ -412,10 +423,10 @@ static void fdt_add_cpu_nodes(const LoongArchVirtMachineState *lvms)
                 num % ms->smp.cores);
         }
         qemu_fdt_add_path(ms->fdt, map_path);
-        qemu_fdt_setprop_phandle(ms->fdt, map_path, "cpu", cpu_path);
+        qemu_fdt_setprop_phandle(ms->fdt, map_path, "cpu", nodename);
 
         g_free(map_path);
-        g_free(cpu_path);
+        g_free(nodename);
     }
 }
 
@@ -615,12 +626,67 @@ static void virt_build_smbios(LoongArchVirtMachineState *lvms)
     }
 }
 
+static void virt_fdt_setup(LoongArchVirtMachineState *lvms)
+{
+    MachineState *machine = MACHINE(lvms);
+    uint32_t cpuintc_phandle, eiointc_phandle, pch_pic_phandle, pch_msi_phandle;
+    int i;
+
+    create_fdt(lvms);
+    fdt_add_cpu_nodes(lvms);
+    fdt_add_memory_nodes(machine);
+    fdt_add_fw_cfg_node(lvms);
+    fdt_add_flash_node(lvms);
+
+    /* Add cpu interrupt-controller */
+    fdt_add_cpuic_node(lvms, &cpuintc_phandle);
+    /* Add Extend I/O Interrupt Controller node */
+    fdt_add_eiointc_node(lvms, &cpuintc_phandle, &eiointc_phandle);
+    /* Add PCH PIC node */
+    fdt_add_pch_pic_node(lvms, &eiointc_phandle, &pch_pic_phandle);
+    /* Add PCH MSI node */
+    fdt_add_pch_msi_node(lvms, &eiointc_phandle, &pch_msi_phandle);
+    /* Add pcie node */
+    fdt_add_pcie_node(lvms, &pch_pic_phandle, &pch_msi_phandle);
+
+    /*
+     * Create uart fdt node in reverse order so that they appear
+     * in the finished device tree lowest address first
+     */
+    for (i = VIRT_UART_COUNT; i-- > 0;) {
+        hwaddr base = VIRT_UART_BASE + i * VIRT_UART_SIZE;
+        int irq = VIRT_UART_IRQ + i - VIRT_GSI_BASE;
+        fdt_add_uart_node(lvms, &pch_pic_phandle, base, irq, i == 0);
+    }
+
+    fdt_add_rtc_node(lvms, &pch_pic_phandle);
+    fdt_add_ged_reset(lvms);
+    platform_bus_add_all_fdt_nodes(machine->fdt, "/platic",
+                                   VIRT_PLATFORM_BUS_BASEADDRESS,
+                                   VIRT_PLATFORM_BUS_SIZE,
+                                   VIRT_PLATFORM_BUS_IRQ);
+
+    /*
+     * Since lowmem region starts from 0 and Linux kernel legacy start address
+     * at 2 MiB, FDT base address is located at 1 MiB to avoid NULL pointer
+     * access. FDT size limit with 1 MiB.
+     * Put the FDT into the memory map as a ROM image: this will ensure
+     * the FDT is copied again upon reset, even if addr points into RAM.
+     */
+    qemu_fdt_dumpdtb(machine->fdt, lvms->fdt_size);
+    rom_add_blob_fixed_as("fdt", machine->fdt, lvms->fdt_size, FDT_BASE,
+                          &address_space_memory);
+    qemu_register_reset_nosnapshotload(qemu_fdt_randomize_seeds,
+            rom_ptr_for_as(&address_space_memory, FDT_BASE, lvms->fdt_size));
+}
+
 static void virt_done(Notifier *notifier, void *data)
 {
     LoongArchVirtMachineState *lvms = container_of(notifier,
                                       LoongArchVirtMachineState, machine_done);
     virt_build_smbios(lvms);
     loongarch_acpi_setup(lvms);
+    virt_fdt_setup(lvms);
 }
 
 static void virt_powerdown_req(Notifier *notifier, void *opaque)
@@ -699,9 +765,7 @@ static DeviceState *create_platform_bus(DeviceState *pch_pic)
 }
 
 static void virt_devices_init(DeviceState *pch_pic,
-                                   LoongArchVirtMachineState *lvms,
+                                   LoongArchVirtMachineState *lvms)
-                                   uint32_t *pch_pic_phandle,
-                                   uint32_t *pch_msi_phandle)
 {
     MachineClass *mc = MACHINE_GET_CLASS(lvms);
     DeviceState *gpex_dev;
@@ -747,9 +811,6 @@ static void virt_devices_init(DeviceState *pch_pic,
         gpex_set_irq_num(GPEX_HOST(gpex_dev), i, 16 + i);
     }
 
-    /* Add pcie node */
-    fdt_add_pcie_node(lvms, pch_pic_phandle, pch_msi_phandle);
-
     /*
      * Create uart fdt node in reverse order so that they appear
      * in the finished device tree lowest address first
@@ -760,7 +821,6 @@ static void virt_devices_init(DeviceState *pch_pic,
         serial_mm_init(get_system_memory(), base, 0,
                        qdev_get_gpio_in(pch_pic, irq),
                        115200, serial_hd(i), DEVICE_LITTLE_ENDIAN);
-        fdt_add_uart_node(lvms, pch_pic_phandle, base, irq, i == 0);
     }
 
     /* Network init */
@@ -774,8 +834,6 @@ static void virt_devices_init(DeviceState *pch_pic,
     sysbus_create_simple("ls7a_rtc", VIRT_RTC_REG_BASE,
                          qdev_get_gpio_in(pch_pic,
                          VIRT_RTC_IRQ - VIRT_GSI_BASE));
-    fdt_add_rtc_node(lvms, pch_pic_phandle);
-    fdt_add_ged_reset(lvms);
 
     /* acpi ged */
     lvms->acpi_ged = create_acpi_ged(pch_pic, lvms);
@@ -793,7 +851,6 @@ static void virt_irq_init(LoongArchVirtMachineState *lvms)
     CPULoongArchState *env;
     CPUState *cpu_state;
     int cpu, pin, i, start, num;
-    uint32_t cpuintc_phandle, eiointc_phandle, pch_pic_phandle, pch_msi_phandle;
 
     /*
      * Extended IRQ model.
@@ -850,9 +907,6 @@ static void virt_irq_init(LoongArchVirtMachineState *lvms)
     memory_region_add_subregion(&lvms->system_iocsr, MAIL_SEND_ADDR,
                    sysbus_mmio_get_region(SYS_BUS_DEVICE(ipi), 1));
 
-    /* Add cpu interrupt-controller */
-    fdt_add_cpuic_node(lvms, &cpuintc_phandle);
-
     for (cpu = 0; cpu < ms->smp.cpus; cpu++) {
         cpu_state = qemu_get_cpu(cpu);
         cpudev = DEVICE(cpu_state);
@@ -891,9 +945,6 @@ static void virt_irq_init(LoongArchVirtMachineState *lvms)
         }
     }
 
-    /* Add Extend I/O Interrupt Controller node */
-    fdt_add_eiointc_node(lvms, &cpuintc_phandle, &eiointc_phandle);
-
     pch_pic = qdev_new(TYPE_LOONGARCH_PIC);
     num = VIRT_PCH_PIC_IRQ_NUM;
     qdev_prop_set_uint32(pch_pic, "pch_pic_irq_num", num);
@@ -913,9 +964,6 @@ static void virt_irq_init(LoongArchVirtMachineState *lvms)
         qdev_connect_gpio_out(DEVICE(d), i, qdev_get_gpio_in(extioi, i));
     }
 
-    /* Add PCH PIC node */
-    fdt_add_pch_pic_node(lvms, &eiointc_phandle, &pch_pic_phandle);
-
     pch_msi = qdev_new(TYPE_LOONGARCH_PCH_MSI);
     start   =  num;
     num = EXTIOI_IRQS - start;
@@ -930,10 +978,7 @@ static void virt_irq_init(LoongArchVirtMachineState *lvms)
                               qdev_get_gpio_in(extioi, i + start));
     }
 
-    /* Add PCH MSI node */
+    virt_devices_init(pch_pic, lvms);
-    fdt_add_pch_msi_node(lvms, &eiointc_phandle, &pch_msi_phandle);
-
-    virt_devices_init(pch_pic, lvms, &pch_pic_phandle, &pch_msi_phandle);
 }
 
 static void virt_firmware_init(LoongArchVirtMachineState *lvms)
@@ -1151,8 +1196,6 @@ static void virt_init(MachineState *machine)
         cpu_model = LOONGARCH_CPU_TYPE_NAME("la464");
     }
 
-    create_fdt(lvms);
-
     /* Create IOCSR space */
     memory_region_init_io(&lvms->system_iocsr, OBJECT(machine), NULL,
                           machine, "iocsr", UINT64_MAX);
@@ -1171,8 +1214,6 @@ static void virt_init(MachineState *machine)
         lacpu = LOONGARCH_CPU(cpu);
         lacpu->phy_id = machine->possible_cpus->cpus[i].arch_id;
     }
-    fdt_add_cpu_nodes(lvms);
-    fdt_add_memory_nodes(machine);
     fw_cfg_add_memory(machine);
 
     /* Node0 memory */
@@ -1224,34 +1265,15 @@ static void virt_init(MachineState *machine)
                         memmap_table,
                         sizeof(struct memmap_entry) * (memmap_entries));
     }
-    fdt_add_fw_cfg_node(lvms);
-    fdt_add_flash_node(lvms);
 
     /* Initialize the IO interrupt subsystem */
     virt_irq_init(lvms);
-    platform_bus_add_all_fdt_nodes(machine->fdt, "/platic",
-                                   VIRT_PLATFORM_BUS_BASEADDRESS,
-                                   VIRT_PLATFORM_BUS_SIZE,
-                                   VIRT_PLATFORM_BUS_IRQ);
     lvms->machine_done.notify = virt_done;
     qemu_add_machine_init_done_notifier(&lvms->machine_done);
      /* connect powerdown request */
     lvms->powerdown_notifier.notify = virt_powerdown_req;
     qemu_register_powerdown_notifier(&lvms->powerdown_notifier);
 
-    /*
-     * Since lowmem region starts from 0 and Linux kernel legacy start address
-     * at 2 MiB, FDT base address is located at 1 MiB to avoid NULL pointer
-     * access. FDT size limit with 1 MiB.
-     * Put the FDT into the memory map as a ROM image: this will ensure
-     * the FDT is copied again upon reset, even if addr points into RAM.
-     */
-    qemu_fdt_dumpdtb(machine->fdt, lvms->fdt_size);
-    rom_add_blob_fixed_as("fdt", machine->fdt, lvms->fdt_size, FDT_BASE,
-                          &address_space_memory);
-    qemu_register_reset_nosnapshotload(qemu_fdt_randomize_seeds,
-            rom_ptr_for_as(&address_space_memory, FDT_BASE, lvms->fdt_size));
-
     lvms->bootinfo.ram_size = ram_size;
     loongarch_load_kernel(machine, &lvms->bootinfo);
 }

target/loongarch/cpu.c

@@ -379,6 +379,7 @@ static void loongarch_la464_initfn(Object *obj)
 {
     LoongArchCPU *cpu = LOONGARCH_CPU(obj);
     CPULoongArchState *env = &cpu->env;
+    uint32_t data = 0;
     int i;
 
     for (i = 0; i < 21; i++) {
@@ -388,7 +389,6 @@ static void loongarch_la464_initfn(Object *obj)
     cpu->dtb_compatible = "loongarch,Loongson-3A5000";
     env->cpucfg[0] = 0x14c010;  /* PRID */
 
-    uint32_t data = 0;
     data = FIELD_DP32(data, CPUCFG1, ARCH, 2);
     data = FIELD_DP32(data, CPUCFG1, PGMMU, 1);
     data = FIELD_DP32(data, CPUCFG1, IOCSR, 1);
@@ -477,7 +477,7 @@ static void loongarch_la132_initfn(Object *obj)
 {
     LoongArchCPU *cpu = LOONGARCH_CPU(obj);
     CPULoongArchState *env = &cpu->env;
-
+    uint32_t data = 0;
     int i;
 
     for (i = 0; i < 21; i++) {
@@ -487,7 +487,6 @@ static void loongarch_la132_initfn(Object *obj)
     cpu->dtb_compatible = "loongarch,Loongson-1C103";
     env->cpucfg[0] = 0x148042;  /* PRID */
 
-    uint32_t data = 0;
     data = FIELD_DP32(data, CPUCFG1, ARCH, 1); /* LA32 */
     data = FIELD_DP32(data, CPUCFG1, PGMMU, 1);
     data = FIELD_DP32(data, CPUCFG1, IOCSR, 1);
@@ -615,54 +614,74 @@ static void loongarch_cpu_realizefn(DeviceState *dev, Error **errp)
 
 static bool loongarch_get_lsx(Object *obj, Error **errp)
 {
-    LoongArchCPU *cpu = LOONGARCH_CPU(obj);
+    return LOONGARCH_CPU(obj)->lsx != ON_OFF_AUTO_OFF;
-    bool ret;
-
-    if (FIELD_EX32(cpu->env.cpucfg[2], CPUCFG2, LSX)) {
-        ret = true;
-    } else {
-        ret = false;
-    }
-    return ret;
 }
 
 static void loongarch_set_lsx(Object *obj, bool value, Error **errp)
 {
     LoongArchCPU *cpu = LOONGARCH_CPU(obj);
+    uint32_t val;
 
-    if (value) {
+    cpu->lsx = value ? ON_OFF_AUTO_ON : ON_OFF_AUTO_OFF;
-        cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LSX, 1);
+    if (cpu->lsx == ON_OFF_AUTO_OFF) {
-    } else {
+        cpu->lasx = ON_OFF_AUTO_OFF;
-        cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LSX, 0);
+        if (cpu->lasx == ON_OFF_AUTO_ON) {
-        cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LASX, 0);
+            error_setg(errp, "Failed to disable LSX since LASX is enabled");
+            return;
+        }
     }
+
+    if (kvm_enabled()) {
+        /* kvm feature detection in function kvm_arch_init_vcpu */
+        return;
+    }
+
+    /* LSX feature detection in TCG mode */
+    val = cpu->env.cpucfg[2];
+    if (cpu->lsx == ON_OFF_AUTO_ON) {
+        if (FIELD_EX32(val, CPUCFG2, LSX) == 0) {
+            error_setg(errp, "Failed to enable LSX in TCG mode");
+            return;
+        }
+    } else {
+        cpu->env.cpucfg[2] = FIELD_DP32(val, CPUCFG2, LASX, 0);
+        val = cpu->env.cpucfg[2];
+    }
+
+    cpu->env.cpucfg[2] = FIELD_DP32(val, CPUCFG2, LSX, value);
 }
 
 static bool loongarch_get_lasx(Object *obj, Error **errp)
 {
-    LoongArchCPU *cpu = LOONGARCH_CPU(obj);
+    return LOONGARCH_CPU(obj)->lasx != ON_OFF_AUTO_OFF;
-    bool ret;
-
-    if (FIELD_EX32(cpu->env.cpucfg[2], CPUCFG2, LASX)) {
-        ret = true;
-    } else {
-        ret = false;
-    }
-    return ret;
 }
 
 static void loongarch_set_lasx(Object *obj, bool value, Error **errp)
 {
     LoongArchCPU *cpu = LOONGARCH_CPU(obj);
+    uint32_t val;
 
-    if (value) {
+    cpu->lasx = value ? ON_OFF_AUTO_ON : ON_OFF_AUTO_OFF;
-	if (!FIELD_EX32(cpu->env.cpucfg[2], CPUCFG2, LSX)) {
+    if ((cpu->lsx == ON_OFF_AUTO_OFF) && (cpu->lasx == ON_OFF_AUTO_ON)) {
-            cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LSX, 1);
+        error_setg(errp, "Failed to enable LASX since lSX is disabled");
-	}
+        return;
-        cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LASX, 1);
-    } else {
-        cpu->env.cpucfg[2] = FIELD_DP32(cpu->env.cpucfg[2], CPUCFG2, LASX, 0);
     }
+
+    if (kvm_enabled()) {
+        /* kvm feature detection in function kvm_arch_init_vcpu */
+        return;
+    }
+
+    /* LASX feature detection in TCG mode */
+    val = cpu->env.cpucfg[2];
+    if (cpu->lasx == ON_OFF_AUTO_ON) {
+        if (FIELD_EX32(val, CPUCFG2, LASX) == 0) {
+            error_setg(errp, "Failed to enable LASX in TCG mode");
+            return;
+        }
+    }
+
+    cpu->env.cpucfg[2] = FIELD_DP32(val, CPUCFG2, LASX, value);
 }
 
 static bool loongarch_get_lbt(Object *obj, Error **errp)
@@ -693,6 +712,8 @@ void loongarch_cpu_post_init(Object *obj)
 {
     LoongArchCPU *cpu = LOONGARCH_CPU(obj);
 
+    cpu->lsx = ON_OFF_AUTO_AUTO;
+    cpu->lasx = ON_OFF_AUTO_AUTO;
     object_property_add_bool(obj, "lsx", loongarch_get_lsx,
                              loongarch_set_lsx);
     object_property_add_bool(obj, "lasx", loongarch_get_lasx,
@@ -713,6 +734,7 @@ void loongarch_cpu_post_init(Object *obj)
 
     } else {
         cpu->lbt = ON_OFF_AUTO_OFF;
+        cpu->pmu = ON_OFF_AUTO_OFF;
     }
 }
 

target/loongarch/cpu.h

@@ -283,6 +283,8 @@ typedef struct LoongArchTLB LoongArchTLB;
 #endif
 
 enum loongarch_features {
+    LOONGARCH_FEATURE_LSX,
+    LOONGARCH_FEATURE_LASX,
     LOONGARCH_FEATURE_LBT, /* loongson binary translation extension */
     LOONGARCH_FEATURE_PMU,
 };
@@ -404,6 +406,8 @@ struct ArchCPU {
     uint32_t  phy_id;
     OnOffAuto lbt;
     OnOffAuto pmu;
+    OnOffAuto lsx;
+    OnOffAuto lasx;
 
     /* 'compatible' string for this CPU for Linux device trees */
     const char *dtb_compatible;

target/loongarch/kvm/kvm.c

@@ -798,8 +798,61 @@ static bool kvm_feature_supported(CPUState *cs, enum loongarch_features feature)
 {
     int ret;
     struct kvm_device_attr attr;
+    uint64_t val;
 
     switch (feature) {
+    case LOONGARCH_FEATURE_LSX:
+        attr.group = KVM_LOONGARCH_VM_FEAT_CTRL;
+        attr.attr = KVM_LOONGARCH_VM_FEAT_LSX;
+        ret = kvm_vm_ioctl(kvm_state, KVM_HAS_DEVICE_ATTR, &attr);
+        if (ret == 0) {
+            return true;
+        }
+
+        /* Fallback to old kernel detect interface */
+        val = 0;
+        attr.group = KVM_LOONGARCH_VCPU_CPUCFG;
+        /* Cpucfg2 */
+        attr.attr  = 2;
+        attr.addr = (uint64_t)&val;
+        ret = kvm_vcpu_ioctl(cs, KVM_HAS_DEVICE_ATTR, &attr);
+        if (!ret) {
+            ret = kvm_vcpu_ioctl(cs, KVM_GET_DEVICE_ATTR, &attr);
+            if (ret) {
+                return false;
+            }
+
+            ret = FIELD_EX32((uint32_t)val, CPUCFG2, LSX);
+            return (ret != 0);
+        }
+        return false;
+
+    case LOONGARCH_FEATURE_LASX:
+        attr.group = KVM_LOONGARCH_VM_FEAT_CTRL;
+        attr.attr = KVM_LOONGARCH_VM_FEAT_LASX;
+        ret = kvm_vm_ioctl(kvm_state, KVM_HAS_DEVICE_ATTR, &attr);
+        if (ret == 0) {
+            return true;
+        }
+
+        /* Fallback to old kernel detect interface */
+        val = 0;
+        attr.group = KVM_LOONGARCH_VCPU_CPUCFG;
+        /* Cpucfg2 */
+        attr.attr  = 2;
+        attr.addr = (uint64_t)&val;
+        ret = kvm_vcpu_ioctl(cs, KVM_HAS_DEVICE_ATTR, &attr);
+        if (!ret) {
+            ret = kvm_vcpu_ioctl(cs, KVM_GET_DEVICE_ATTR, &attr);
+            if (ret) {
+                return false;
+            }
+
+            ret = FIELD_EX32((uint32_t)val, CPUCFG2, LASX);
+            return (ret != 0);
+        }
+        return false;
+
     case LOONGARCH_FEATURE_LBT:
         /*
          * Return all if all the LBT features are supported such as:
@@ -829,6 +882,50 @@ static bool kvm_feature_supported(CPUState *cs, enum loongarch_features feature)
     return false;
 }
 
+static int kvm_cpu_check_lsx(CPUState *cs, Error **errp)
+{
+    CPULoongArchState *env = cpu_env(cs);
+    LoongArchCPU *cpu = LOONGARCH_CPU(cs);
+    bool kvm_supported;
+
+    kvm_supported = kvm_feature_supported(cs, LOONGARCH_FEATURE_LSX);
+    env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LSX, 0);
+    if (cpu->lsx == ON_OFF_AUTO_ON) {
+        if (kvm_supported) {
+            env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LSX, 1);
+        } else {
+            error_setg(errp, "'lsx' feature not supported by KVM on this host");
+            return -ENOTSUP;
+        }
+    } else if ((cpu->lsx == ON_OFF_AUTO_AUTO) && kvm_supported) {
+        env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LSX, 1);
+    }
+
+    return 0;
+}
+
+static int kvm_cpu_check_lasx(CPUState *cs, Error **errp)
+{
+    CPULoongArchState *env = cpu_env(cs);
+    LoongArchCPU *cpu = LOONGARCH_CPU(cs);
+    bool kvm_supported;
+
+    kvm_supported = kvm_feature_supported(cs, LOONGARCH_FEATURE_LASX);
+    env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LASX, 0);
+    if (cpu->lasx == ON_OFF_AUTO_ON) {
+        if (kvm_supported) {
+            env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LASX, 1);
+        } else {
+            error_setg(errp, "'lasx' feature not supported by KVM on host");
+            return -ENOTSUP;
+        }
+    } else if ((cpu->lasx == ON_OFF_AUTO_AUTO) && kvm_supported) {
+        env->cpucfg[2] = FIELD_DP32(env->cpucfg[2], CPUCFG2, LASX, 1);
+    }
+
+    return 0;
+}
+
 static int kvm_cpu_check_lbt(CPUState *cs, Error **errp)
 {
     CPULoongArchState *env = cpu_env(cs);
@@ -889,6 +986,16 @@ int kvm_arch_init_vcpu(CPUState *cs)
         brk_insn = val;
     }
 
+    ret = kvm_cpu_check_lsx(cs, &local_err);
+    if (ret < 0) {
+        error_report_err(local_err);
+    }
+
+    ret = kvm_cpu_check_lasx(cs, &local_err);
+    if (ret < 0) {
+        error_report_err(local_err);
+    }
+
     ret = kvm_cpu_check_lbt(cs, &local_err);
     if (ret < 0) {
         error_report_err(local_err);

target/loongarch/tcg/insn_trans/trans_vec.c.inc

@@ -3480,7 +3480,7 @@ static uint64_t vldi_get_value(DisasContext *ctx, uint32_t imm)
         break;
     case 1:
         /* data: {2{16'0, imm[7:0], 8'0}} */
-        data = (t << 24) | (t << 8);
+        data = (t << 40) | (t << 8);
         break;
     case 2:
         /* data: {2{8'0, imm[7:0], 16'0}} */
@@ -5126,7 +5126,7 @@ static bool do_vbsrl_v(DisasContext *ctx, arg_vv_i *a, uint32_t oprsz)
 {
     int i, ofs;
 
-    if (!check_vec(ctx, 32)) {
+    if (!check_vec(ctx, oprsz)) {
         return true;
     }