Re: [Qemu-devel] [RFC PATCH v0 14/15] ppc: Add CPU hotplug support for sPAPR guests

[Tyrel Datwyler <turtle.in.the.kernel@gmail.com>]

On 09/03/2014 11:06 PM, Bharata B Rao wrote:
> Add support for cpu-add monitor command. Use the exising EPOW event
> infrastructure to send CPU hotplug notification to the guest.
> 
> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
> ---
>  hw/ppc/spapr.c         | 240 ++++++++++++++++++++++++++++++++++++++++++++++++-
>  hw/ppc/spapr_events.c  |  35 ++++++--
>  include/hw/ppc/spapr.h |   1 +
>  3 files changed, 269 insertions(+), 7 deletions(-)
> 
> diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> index d128834..9a3d1ca 100644
> --- a/hw/ppc/spapr.c
> +++ b/hw/ppc/spapr.c
> @@ -618,6 +618,8 @@ static void add_str(GString *s, const gchar *s1)
>      g_string_append_len(s, s1, strlen(s1) + 1);
>  }
>  
> +uint32_t cpus_per_socket;
> +
>  static void *spapr_create_fdt_skel(hwaddr initrd_base,
>                                     hwaddr initrd_size,
>                                     hwaddr kernel_size,
> @@ -638,9 +640,10 @@ static void *spapr_create_fdt_skel(hwaddr initrd_base,
>      unsigned char vec5[] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x80};
>      QemuOpts *opts = qemu_opts_find(qemu_find_opts("smp-opts"), NULL);
>      unsigned sockets = opts ? qemu_opt_get_number(opts, "sockets", 0) : 0;
> -    uint32_t cpus_per_socket = sockets ? (smp_cpus / sockets) : 1;
>      char *buf;
>  
> +    cpus_per_socket = sockets ? (smp_cpus / sockets) : 1;
> +
>      add_str(hypertas, "hcall-pft");
>      add_str(hypertas, "hcall-term");
>      add_str(hypertas, "hcall-dabr");
> @@ -1603,6 +1606,208 @@ static SaveVMHandlers savevm_htab_handlers = {
>      .load_state = htab_load,
>  };
>  
> +Notifier cpu_added_notifier;
> +
> +/* TODO: Duplicates code from spapr_create_fdt_skel(), Fix this */
> +static int spapr_populate_cpu_dt(CPUState *cs, void *fdt, int offset)
> +{
> +    PowerPCCPU *cpu = POWERPC_CPU(cs);
> +    CPUPPCState *env = &cpu->env;
> +    PowerPCCPUClass *pcc = POWERPC_CPU_GET_CLASS(cs);
> +    int index = ppc_get_vcpu_dt_id(cpu);
> +    uint32_t segs[] = {cpu_to_be32(28), cpu_to_be32(40),
> +                       0xffffffff, 0xffffffff};
> +    uint32_t tbfreq = kvm_enabled() ? kvmppc_get_tbfreq() : TIMEBASE_FREQ;
> +    uint32_t cpufreq = kvm_enabled() ? kvmppc_get_clockfreq() : 1000000000;
> +    uint32_t page_sizes_prop[64];
> +    size_t page_sizes_prop_size;
> +    sPAPRDrcEntry *drc_entry;
> +    int smpt = ppc_get_compat_smt_threads(cpu);
> +    uint32_t servers_prop[smpt];
> +    uint32_t gservers_prop[smpt * 2];
> +    int i;
> +    uint32_t pft_size_prop[] = {0, cpu_to_be32(spapr->htab_shift)};
> +
> +    _FDT((fdt_setprop_cell(fdt, offset, "reg", index)));
> +    _FDT((fdt_setprop_string(fdt, offset, "device_type", "cpu")));
> +
> +    _FDT((fdt_setprop_cell(fdt, offset, "cpu-version", env->spr[SPR_PVR])));
> +    _FDT((fdt_setprop_cell(fdt, offset, "d-cache-block-size",
> +                        env->dcache_line_size)));
> +    _FDT((fdt_setprop_cell(fdt, offset, "d-cache-line-size",
> +                        env->dcache_line_size)));
> +    _FDT((fdt_setprop_cell(fdt, offset, "i-cache-block-size",
> +                        env->icache_line_size)));
> +    _FDT((fdt_setprop_cell(fdt, offset, "i-cache-line-size",
> +                            env->icache_line_size)));
> +
> +    if (pcc->l1_dcache_size) {
> +        _FDT((fdt_setprop_cell(fdt, offset, "d-cache-size",
> +            pcc->l1_dcache_size)));
> +    } else {
> +        fprintf(stderr, "Warning: Unknown L1 dcache size for cpu\n");
> +    }
> +    if (pcc->l1_icache_size) {
> +        _FDT((fdt_setprop_cell(fdt, offset, "i-cache-size",
> +            pcc->l1_icache_size)));
> +    } else {
> +        fprintf(stderr, "Warning: Unknown L1 icache size for cpu\n");
> +    }
> +
> +    _FDT((fdt_setprop_cell(fdt, offset, "timebase-frequency", tbfreq)));
> +    _FDT((fdt_setprop_cell(fdt, offset, "clock-frequency", cpufreq)));
> +    _FDT((fdt_setprop_cell(fdt, offset, "ibm,slb-size", env->slb_nr)));
> +    _FDT((fdt_setprop_string(fdt, offset, "status", "okay")));
> +    _FDT((fdt_setprop(fdt, offset, "64-bit", NULL, 0)));
> +
> +    if (env->spr_cb[SPR_PURR].oea_read) {
> +        _FDT((fdt_setprop(fdt, offset, "ibm,purr", NULL, 0)));
> +    }
> +
> +    if (env->mmu_model & POWERPC_MMU_1TSEG) {
> +        _FDT((fdt_setprop(fdt, offset, "ibm,processor-segment-sizes",
> +                           segs, sizeof(segs))));
> +    }
> +
> +    /* Advertise VMX/VSX (vector extensions) if available
> +     *   0 / no property == no vector extensions
> +     *   1               == VMX / Altivec available
> +     *   2               == VSX available */
> +    if (env->insns_flags & PPC_ALTIVEC) {
> +        uint32_t vmx = (env->insns_flags2 & PPC2_VSX) ? 2 : 1;
> +
> +        _FDT((fdt_setprop_cell(fdt, offset, "ibm,vmx", vmx)));
> +    }
> +
> +    /* Advertise DFP (Decimal Floating Point) if available
> +     *   0 / no property == no DFP
> +     *   1               == DFP available */
> +    if (env->insns_flags2 & PPC2_DFP) {
> +        _FDT((fdt_setprop_cell(fdt, offset, "ibm,dfp", 1)));
> +    }
> +
> +    page_sizes_prop_size = create_page_sizes_prop(env, page_sizes_prop,
> +                                                  sizeof(page_sizes_prop));
> +    if (page_sizes_prop_size) {
> +        _FDT((fdt_setprop(fdt, offset, "ibm,segment-page-sizes",
> +                           page_sizes_prop, page_sizes_prop_size)));
> +    }
> +
> +    _FDT((fdt_setprop_cell(fdt, offset, "ibm,chip-id",
> +                                cs->cpu_index / cpus_per_socket)));
> +
> +    drc_entry = spapr_cpu_to_drc_entry(cpu->cpu_dt_id + SPAPR_DRC_CPU_ID_BASE);
> +    g_assert(drc_entry);
> +    _FDT((fdt_setprop_cell(fdt, offset, "ibm,my-drc-index",
> +        drc_entry->drc_index)));
> +
> +    /* Build interrupt servers and gservers properties */
> +    for (i = 0; i < smpt; i++) {
> +        servers_prop[i] = cpu_to_be32(index + i);
> +        /* Hack, direct the group queues back to cpu 0 */
> +        gservers_prop[i*2] = cpu_to_be32(index + i);
> +        gservers_prop[i*2 + 1] = 0;
> +    }
> +    _FDT(fdt_setprop(fdt, offset, "ibm,ppc-interrupt-server#s",
> +                       servers_prop, sizeof(servers_prop)));
> +    _FDT(fdt_setprop(fdt, offset, "ibm,ppc-interrupt-gserver#s",
> +                      gservers_prop, sizeof(gservers_prop)));
> +    _FDT(fdt_setprop(fdt, offset, "ibm,pft-size",
> +                          pft_size_prop, sizeof(pft_size_prop)));
> +    return 0;
> +}
> +
> +static void spapr_cpu_hotplug_add(CPUState *cs)
> +{
> +    int i, j;
> +    sPAPRDrcEntry *drc_entry;
> +    sPAPRConfigureConnectorState *ccs;
> +    int offset, ret;
> +    void *fdt_orig, *fdt;
> +
> +    /*
> +     * TODO: Unlike PCI hotplug, we aren't marking the state as PRESENT
> +     * here since CPU hotplug code in the guest kernel expects the
> +     * state to be UNUSABLE.
> +     */

PCI hotplug is considered a physical entity and the state transitions
are slightly different from that of cpus which are considered a logical
entity. The state needs to start out as UNUSABLE as stated in your
comment. The guest will call get-sensor-state to check the
dr-entity-sensor to ensure that it is in UNUSABLE state (ie. not yet
owned by the OS). Once this is confirmed the guest calls set-indicator
to change the allocation state to USABLE. If that call is successful the
dr-entity-sensor state for the cpu should now be PRESENT indicating that
it is now owned by the OS. The guest then calls set-indicator to change
the isolation state to unisolate which is then followed by the
configure-connector call.

-Tyrel

> +#if 0
> +    uint32_t encoded = ENCODE_DRC_STATE(INDICATOR_ENTITY_SENSE_PRESENT,
> +                                        INDICATOR_ENTITY_SENSE_MASK,
> +                                        INDICATOR_ENTITY_SENSE_SHIFT);
> +#endif
> +    PowerPCCPU *cpu = POWERPC_CPU(cs);
> +    DeviceClass *dc = DEVICE_GET_CLASS(cs);
> +    PowerPCCPUClass *pcc = POWERPC_CPU_GET_CLASS(cs);
> +    char *nodename;
> +    int index = ppc_get_vcpu_dt_id(cpu);
> +
> +    drc_entry = spapr_cpu_to_drc_entry(cpu->cpu_dt_id + SPAPR_DRC_CPU_ID_BASE);
> +    g_assert(drc_entry);
> +
> +#if 0
> +    drc_entry->state &= ~(uint32_t)INDICATOR_ENTITY_SENSE_MASK;
> +    drc_entry->state |= encoded; /* DR entity present */
> +#endif
> +
> +    /* add OF node for CPU and required OF DT properties */
> +    fdt_orig = g_malloc0(FDT_MAX_SIZE);
> +    offset = fdt_create(fdt_orig, FDT_MAX_SIZE);
> +    fdt_begin_node(fdt_orig, "");
> +    fdt_end_node(fdt_orig);
> +    fdt_finish(fdt_orig);
> +
> +    fdt = g_malloc0(FDT_MAX_SIZE);
> +    fdt_open_into(fdt_orig, fdt, FDT_MAX_SIZE);
> +
> +    if (dc->fw_name == NULL) {
> +        ObjectClass *oc = OBJECT_CLASS(pcc);
> +        const char *typename;
> +
> +        typename = object_class_get_name(oc);
> +        if (kvm_enabled() &&
> +            strcmp(typename, "host-" TYPE_POWERPC_CPU) == 0) {
> +            typename = object_class_get_name(object_class_get_parent(oc));
> +        }
> +        nodename = g_strndup(typename,
> +                             strlen(typename) - strlen("-" TYPE_POWERPC_CPU));
> +        for (i = j = 0; j < strlen(nodename); i++, j++) {
> +            if (nodename[j] == '_') {
> +                j++;
> +            }
> +            if (j > i) {
> +                nodename[i] = nodename[j];
> +            }
> +        }
> +        if (j > i) {
> +            nodename[i] = '\0';
> +        }
> +        dc->fw_name = g_strdup_printf("PowerPC,%s", nodename);
> +        g_free(nodename);
> +    }
> +    nodename = g_strdup_printf("%s@%x", dc->fw_name, index);
> +
> +    offset = fdt_add_subnode(fdt, offset, nodename);
> +    ret = spapr_populate_cpu_dt(cs, fdt, offset);
> +    g_assert(!ret);
> +    g_free(fdt_orig);
> +    g_free(nodename);
> +
> +    /* hold on to node, configure_connector will pass it to the guest later */
> +    ccs = &drc_entry->cc_state;
> +    ccs->fdt = fdt;
> +    ccs->offset_start = offset;
> +    ccs->state = CC_STATE_PENDING;
> +}
> +
> +static void spapr_cpu_added_req(Notifier *n, void *opaque)
> +{
> +    CPUState *cs = CPU(opaque);
> +
> +    spapr_cpu_hotplug_add(cs);
> +    spapr_cpu_hotplug_add_event(cs);
> +    return;
> +}
> +
>  static const char *current_cpu_model;
>  
>  static PowerPCCPU *ppc_new_cpu(const char *cpu_model)
> @@ -1736,6 +1941,10 @@ static void ppc_spapr_init(MachineState *machine)
>          ppc_new_cpu(current_cpu_model);
>      }
>  
> +    /* Register a handler for CPU hotplug */
> +    cpu_added_notifier.notify = spapr_cpu_added_req;
> +    qemu_register_cpu_added_notifier(&cpu_added_notifier);
> +
>      /* allocate RAM */
>      spapr->ram_limit = ram_size;
>      spapr->maxram_limit = machine->maxram_size;
> @@ -1980,6 +2189,34 @@ static void spapr_machine_initfn(Object *obj)
>                              spapr_get_kvm_type, spapr_set_kvm_type, NULL);
>  }
>  
> +static void ppc_hot_add_cpu(const int64_t id, Error **errp)
> +{
> +    CPUState *cs;
> +    PowerPCCPU *cpu;
> +
> +    if (id < 0) {
> +        error_setg(errp, "Invalid CPU id: %" PRIi64, id);
> +        return;
> +    }
> +
> +    CPU_FOREACH(cs) {
> +        if (cs->cpu_index == id) {
> +            error_setg(errp, "Unable to add CPU: %" PRIi64
> +                   ", it already exists", id);
> +            return;
> +        }
> +    }
> +
> +    if (id >= max_cpus) {
> +        error_setg(errp, "Unable to add CPU: %" PRIi64
> +                   ", max allowed: %d", id, max_cpus - 1);
> +        return;
> +    }
> +
> +    cpu = ppc_new_cpu(current_cpu_model);
> +    spapr_cpu_reset(cpu);
> +}
> +
>  static void spapr_machine_class_init(ObjectClass *oc, void *data)
>  {
>      MachineClass *mc = MACHINE_CLASS(oc);
> @@ -1995,6 +2232,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data)
>      mc->no_parallel = 1;
>      mc->default_boot_order = NULL;
>      mc->kvm_type = spapr_kvm_type;
> +    mc->hot_add_cpu = ppc_hot_add_cpu;
>  
>      fwc->get_dev_path = spapr_get_fw_dev_path;
>  }
> diff --git a/hw/ppc/spapr_events.c b/hw/ppc/spapr_events.c
> index bb80080..f772255 100644
> --- a/hw/ppc/spapr_events.c
> +++ b/hw/ppc/spapr_events.c
> @@ -324,8 +324,10 @@ static void spapr_powerdown_req(Notifier *n, void *opaque)
>  }
>  
>  static void spapr_hotplug_req_event(uint8_t hp_type, uint8_t hp_action,
> -                                    sPAPRPHBState *phb, int slot)
> +                                    void *dev, int slot)
>  {
> +    sPAPRPHBState *phb;
> +    PowerPCCPU *cpu;
>      struct rtas_error_log *hdr;
>      struct rtas_event_log_v6 *v6hdr;
>      struct rtas_event_log_v6_maina *maina;
> @@ -372,21 +374,42 @@ static void spapr_hotplug_req_event(uint8_t hp_type, uint8_t hp_action,
>  
>      hp->hotplug_type = hp_type;
>  
> -    drc_entry = spapr_phb_to_drc_entry(phb->buid);
> -    if (!drc_entry) {
> -        drc_entry = spapr_add_phb_to_drc_table(phb->buid, 2 /* Unusable */);
> -    }
> -
>      switch (hp_type) {
>      case RTAS_LOG_V6_HP_TYPE_PCI:
> +        phb = (sPAPRPHBState *)dev;
> +        drc_entry = spapr_phb_to_drc_entry(phb->buid);
> +        if (!drc_entry) {
> +            drc_entry = spapr_add_phb_to_drc_table(phb->buid, 2 /* Unusable */);
> +        }
> +
>          hp->drc.index = drc_entry->child_entries[slot].drc_index;
>          hp->hotplug_identifier = RTAS_LOG_V6_HP_ID_DRC_INDEX;
>          break;
> +
> +    case RTAS_LOG_V6_HP_TYPE_CPU:
> +        cpu = (PowerPCCPU *)dev;
> +        drc_entry = spapr_cpu_to_drc_entry(cpu->cpu_dt_id +
> +            SPAPR_DRC_CPU_ID_BASE);
> +        if (!drc_entry) {
> +            drc_entry = spapr_add_cpu_to_drc_table(cpu->cpu_dt_id +
> +                SPAPR_DRC_CPU_ID_BASE, 2);
> +        }
> +
> +        hp->drc.index = drc_entry->drc_index;
> +        hp->hotplug_identifier = RTAS_LOG_V6_HP_ID_DRC_INDEX;
> +        break;
>      }
>  
>      qemu_irq_pulse(xics_get_qirq(spapr->icp, spapr->check_exception_irq));
>  }
>  
> +void spapr_cpu_hotplug_add_event(CPUState *cs)
> +{
> +    PowerPCCPU *cpu = POWERPC_CPU(cs);
> +    return spapr_hotplug_req_event(RTAS_LOG_V6_HP_TYPE_CPU,
> +                                   RTAS_LOG_V6_HP_ACTION_ADD, cpu, 0);
> +}
> +
>  void spapr_pci_hotplug_add_event(DeviceState *qdev, int slot)
>  {
>      sPAPRPHBState *phb = SPAPR_PCI_HOST_BRIDGE(qdev);
> diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> index d70e5ec..e5fa696 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -567,5 +567,6 @@ sPAPRDrcEntry *spapr_cpu_to_drc_entry(uint64_t cpuid);
>  sPAPRDrcEntry *spapr_find_drc_entry(int drc_index);
>  void spapr_pci_hotplug_add_event(DeviceState *qdev, int slot);
>  void spapr_pci_hotplug_remove_event(DeviceState *qdev, int slot);
> +void spapr_cpu_hotplug_add_event(CPUState *cs);
>  
>  #endif /* !defined (__HW_SPAPR_H__) */
>