[Qemu-devel] [PATCH v5 24/33] nvdimm acpi: build ACPI NFIT table

[Xiao Guangrong]

NFIT is defined in ACPI 6.0: 5.2.25 NVDIMM Firmware Interface Table (NFIT)

Currently, we only support PMEM mode. Each device has 3 structures:
- SPA structure, defines the PMEM region info

- MEM DEV structure, it has the @handle which is used to associate specified
  ACPI NVDIMM  device we will introduce in later patch.
  Also we can happily ignored the memory device's interleave, the real
  nvdimm hardware access is hidden behind host

- DCR structure, it defines vendor ID used to associate specified vendor
  nvdimm driver. Since we only implement PMEM mode this time, Command
  window and Data window are not needed

Signed-off-by: Xiao Guangrong <address@hidden>
---
 hw/acpi/nvdimm.c        | 355 ++++++++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c    |   6 +
 include/hw/mem/nvdimm.h |  10 ++
 3 files changed, 371 insertions(+)

diff --git a/hw/acpi/nvdimm.c b/hw/acpi/nvdimm.c
index 647a5dd..3fc82e1 100644
--- a/hw/acpi/nvdimm.c
+++ b/hw/acpi/nvdimm.c
@@ -26,8 +26,348 @@
  * License along with this library; if not, see <http://www.gnu.org/licenses/>
  */
 
+#include "hw/acpi/acpi.h"
+#include "hw/acpi/aml-build.h"
 #include "hw/mem/nvdimm.h"
 
+static int nvdimm_plugged_device_list(Object *obj, void *opaque)
+{
+    GSList **list = opaque;
+
+    if (object_dynamic_cast(obj, TYPE_NVDIMM)) {
+        NVDIMMDevice *nvdimm = NVDIMM(obj);
+
+        if (memory_region_is_mapped(&nvdimm->nvdimm_mr)) {
+            *list = g_slist_append(*list, DEVICE(obj));
+        }
+    }
+
+    object_child_foreach(obj, nvdimm_plugged_device_list, opaque);
+    return 0;
+}
+
+/*
+ * inquire plugged NVDIMM devices and link them into the list which is
+ * returned to the caller.
+ *
+ * Note: it is the caller's responsibility to free the list to avoid
+ * memory leak.
+ */
+static GSList *nvdimm_get_plugged_device_list(void)
+{
+    GSList *list = NULL;
+
+    object_child_foreach(qdev_get_machine(), nvdimm_plugged_device_list,
+                         &list);
+    return list;
+}
+
+#define NVDIMM_UUID_LE(a, b, c, d0, d1, d2, d3, d4, d5, d6, d7)             \
+   { (a) & 0xff, ((a) >> 8) & 0xff, ((a) >> 16) & 0xff, ((a) >> 24) & 0xff, \
+     (b) & 0xff, ((b) >> 8) & 0xff, (c) & 0xff, ((c) >> 8) & 0xff,          \
+     (d0), (d1), (d2), (d3), (d4), (d5), (d6), (d7) }
+/*
+ * define Byte Addressable Persistent Memory (PM) Region according to
+ * ACPI 6.0: 5.2.25.1 System Physical Address Range Structure.
+ */
+static const uint8_t nvdimm_nfit_spa_uuid_pm[] =
+      NVDIMM_UUID_LE(0x66f0d379, 0xb4f3, 0x4074, 0xac, 0x43, 0x0d, 0x33,
+                     0x18, 0xb7, 0x8c, 0xdb);
+
+/*
+ * NVDIMM Firmware Interface Table
+ * @signature: "NFIT"
+ *
+ * It provides information that allows OSPM to enumerate NVDIMM present in
+ * the platform and associate system physical address ranges created by the
+ * NVDIMMs.
+ *
+ * It is defined in ACPI 6.0: 5.2.25 NVDIMM Firmware Interface Table (NFIT)
+ */
+struct nvdimm_nfit {
+    ACPI_TABLE_HEADER_DEF
+    uint32_t reserved;
+} QEMU_PACKED;
+typedef struct nvdimm_nfit nvdimm_nfit;
+
+/*
+ * define NFIT structures according to ACPI 6.0: 5.2.25 NVDIMM Firmware
+ * Interface Table (NFIT).
+ */
+
+/*
+ * System Physical Address Range Structure
+ *
+ * It describes the system physical address ranges occupied by NVDIMMs and
+ * the types of the regions.
+ */
+struct nvdimm_nfit_spa {
+    uint16_t type;
+    uint16_t length;
+    uint16_t spa_index;
+    uint16_t flags;
+    uint32_t reserved;
+    uint32_t proximity_domain;
+    uint8_t type_guid[16];
+    uint64_t spa_base;
+    uint64_t spa_length;
+    uint64_t mem_attr;
+} QEMU_PACKED;
+typedef struct nvdimm_nfit_spa nvdimm_nfit_spa;
+
+/*
+ * Memory Device to System Physical Address Range Mapping Structure
+ *
+ * It enables identifying each NVDIMM region and the corresponding SPA
+ * describing the memory interleave
+ */
+struct nvdimm_nfit_memdev {
+    uint16_t type;
+    uint16_t length;
+    uint32_t nfit_handle;
+    uint16_t phys_id;
+    uint16_t region_id;
+    uint16_t spa_index;
+    uint16_t dcr_index;
+    uint64_t region_len;
+    uint64_t region_offset;
+    uint64_t region_dpa;
+    uint16_t interleave_index;
+    uint16_t interleave_ways;
+    uint16_t flags;
+    uint16_t reserved;
+} QEMU_PACKED;
+typedef struct nvdimm_nfit_memdev nvdimm_nfit_memdev;
+
+/*
+ * NVDIMM Control Region Structure
+ *
+ * It describes the NVDIMM and if applicable, Block Control Window.
+ */
+struct nvdimm_nfit_dcr {
+    uint16_t type;
+    uint16_t length;
+    uint16_t dcr_index;
+    uint16_t vendor_id;
+    uint16_t device_id;
+    uint16_t revision_id;
+    uint16_t sub_vendor_id;
+    uint16_t sub_device_id;
+    uint16_t sub_revision_id;
+    uint8_t reserved[6];
+    uint32_t serial_number;
+    uint16_t fic;
+    uint16_t num_bcw;
+    uint64_t bcw_size;
+    uint64_t cmd_offset;
+    uint64_t cmd_size;
+    uint64_t status_offset;
+    uint64_t status_size;
+    uint16_t flags;
+    uint8_t reserved2[6];
+} QEMU_PACKED;
+typedef struct nvdimm_nfit_dcr nvdimm_nfit_dcr;
+
+/*
+ * Module serial number is a unique number for each device. We use the
+ * slot id of NVDIMM device to generate this number so that each device
+ * associates with a different number.
+ *
+ * 0x123456 is a magic number we arbitrarily chose.
+ */
+static uint32_t nvdimm_slot_to_sn(int slot)
+{
+    return 0x123456 + slot;
+}
+
+/*
+ * handle is used to uniquely associate nfit_memdev structure with NVDIMM
+ * ACPI device - nfit_memdev.nfit_handle matches with the value returned
+ * by ACPI device _ADR method.
+ *
+ * We generate the handle with the slot id of NVDIMM device and reserve
+ * 0 for NVDIMM root device.
+ */
+static uint32_t nvdimm_slot_to_handle(int slot)
+{
+    return slot + 1;
+}
+
+/*
+ * index uniquely identifies the structure, 0 is reserved which indicates
+ * that the structure is not valid or the associated structure is not
+ * present.
+ *
+ * Each NVDIMM device needs two indexes, one for nfit_spa and another for
+ * nfit_dc which are generated by the slot id of NVDIMM device.
+ */
+static uint16_t nvdimm_slot_to_spa_index(int slot)
+{
+    return (slot + 1) << 1;
+}
+
+/* See the comments of nvdimm_slot_to_spa_index(). */
+static uint32_t nvdimm_slot_to_dcr_index(int slot)
+{
+    return nvdimm_slot_to_spa_index(slot) + 1;
+}
+
+/* ACPI 6.0: 5.2.25.1 System Physical Address Range Structure */
+static void
+nvdimm_build_structure_spa(GArray *structures, NVDIMMDevice *nvdimm)
+{
+    nvdimm_nfit_spa *nfit_spa;
+    uint64_t addr = object_property_get_int(OBJECT(nvdimm), DIMM_ADDR_PROP,
+                                            NULL);
+    uint64_t size = object_property_get_int(OBJECT(nvdimm), DIMM_SIZE_PROP,
+                                            NULL);
+    uint32_t node = object_property_get_int(OBJECT(nvdimm), DIMM_NODE_PROP,
+                                            NULL);
+    int slot = object_property_get_int(OBJECT(nvdimm), DIMM_SLOT_PROP,
+                                            NULL);
+
+    nfit_spa = acpi_data_push(structures, sizeof(*nfit_spa));
+
+    nfit_spa->type = cpu_to_le16(0 /* System Physical Address Range
+                                      Structure */);
+    nfit_spa->length = cpu_to_le16(sizeof(*nfit_spa));
+    nfit_spa->spa_index = cpu_to_le16(nvdimm_slot_to_spa_index(slot));
+
+    /*
+     * Control region is strict as all the device info, such as SN, index,
+     * is associated with slot id.
+     */
+    nfit_spa->flags = cpu_to_le16(1 /* Control region is strictly for
+                                       management during hot add/online
+                                       operation */ |
+                                  2 /* Data in Proximity Domain field is
+                                       valid*/);
+
+    /* NUMA node. */
+    nfit_spa->proximity_domain = cpu_to_le32(node);
+    /* the region reported as PMEM. */
+    memcpy(nfit_spa->type_guid, nvdimm_nfit_spa_uuid_pm,
+           sizeof(nvdimm_nfit_spa_uuid_pm));
+
+    nfit_spa->spa_base = cpu_to_le64(addr);
+    nfit_spa->spa_length = cpu_to_le64(size);
+
+    /* It is the PMEM and can be cached as writeback. */
+    nfit_spa->mem_attr = cpu_to_le64(0x8ULL /* EFI_MEMORY_WB */ |
+                                     0x8000ULL /* EFI_MEMORY_NV */);
+}
+
+/*
+ * ACPI 6.0: 5.2.25.2 Memory Device to System Physical Address Range Mapping
+ * Structure
+ */
+static void
+nvdimm_build_structure_memdev(GArray *structures, NVDIMMDevice *nvdimm)
+{
+    nvdimm_nfit_memdev *nfit_memdev;
+    uint64_t addr = object_property_get_int(OBJECT(nvdimm), DIMM_ADDR_PROP,
+                                            NULL);
+    uint64_t size = object_property_get_int(OBJECT(nvdimm), DIMM_SIZE_PROP,
+                                            NULL);
+    int slot = object_property_get_int(OBJECT(nvdimm), DIMM_SLOT_PROP,
+                                            NULL);
+    uint32_t handle = nvdimm_slot_to_handle(slot);
+
+    nfit_memdev = acpi_data_push(structures, sizeof(*nfit_memdev));
+
+    nfit_memdev->type = cpu_to_le16(1 /* Memory Device to System Address
+                                         Range Map Structure*/);
+    nfit_memdev->length = cpu_to_le16(sizeof(*nfit_memdev));
+    nfit_memdev->nfit_handle = cpu_to_le32(handle);
+
+    /*
+     * associate memory device with System Physical Address Range
+     * Structure.
+     */
+    nfit_memdev->spa_index = cpu_to_le16(nvdimm_slot_to_spa_index(slot));
+    /* associate memory device with Control Region Structure. */
+    nfit_memdev->dcr_index = cpu_to_le16(nvdimm_slot_to_dcr_index(slot));
+
+    /* The memory region on the device. */
+    nfit_memdev->region_len = cpu_to_le64(size);
+    nfit_memdev->region_dpa = cpu_to_le64(addr);
+
+    /* Only one interleave for PMEM. */
+    nfit_memdev->interleave_ways = cpu_to_le16(1);
+}
+
+/*
+ * ACPI 6.0: 5.2.25.5 NVDIMM Control Region Structure.
+ */
+static void nvdimm_build_structure_dcr(GArray *structures,
+                                       NVDIMMDevice *nvdimm)
+{
+    nvdimm_nfit_dcr *nfit_dcr;
+    int slot = object_property_get_int(OBJECT(nvdimm), DIMM_SLOT_PROP,
+                                       NULL);
+    uint32_t sn = nvdimm_slot_to_sn(slot);
+
+    nfit_dcr = acpi_data_push(structures, sizeof(*nfit_dcr));
+
+    nfit_dcr->type = cpu_to_le16(4 /* NVDIMM Control Region Structure */);
+    nfit_dcr->length = cpu_to_le16(sizeof(*nfit_dcr));
+    nfit_dcr->dcr_index = cpu_to_le16(nvdimm_slot_to_dcr_index(slot));
+
+    /* vendor: Intel. */
+    nfit_dcr->vendor_id = cpu_to_le16(0x8086);
+    nfit_dcr->device_id = cpu_to_le16(1);
+
+    /* The _DSM method is following Intel's DSM specification. */
+    nfit_dcr->revision_id = cpu_to_le16(1 /* Current Revision supported
+                                             in ACPI 6.0 is 1. */);
+    nfit_dcr->serial_number = cpu_to_le32(sn);
+    nfit_dcr->fic = cpu_to_le16(0x201 /* Format Interface Code. See Chapter
+                                         2: NVDIMM Device Specific Method
+                                         (DSM) in DSM Spec Rev1.*/);
+}
+
+static GArray *nvdimm_build_device_structure(GSList *device_list)
+{
+    GArray *structures = g_array_new(false, true /* clear */, 1);
+
+    for (; device_list; device_list = device_list->next) {
+        NVDIMMDevice *nvdimm = device_list->data;
+
+        /* build System Physical Address Range Structure. */
+        nvdimm_build_structure_spa(structures, nvdimm);
+
+        /*
+         * build Memory Device to System Physical Address Range Mapping
+         * Structure.
+         */
+        nvdimm_build_structure_memdev(structures, nvdimm);
+
+        /* build NVDIMM Control Region Structure. */
+        nvdimm_build_structure_dcr(structures, nvdimm);
+    }
+
+    return structures;
+}
+
+static void nvdimm_build_nfit(GSList *device_list, GArray *table_offsets,
+                              GArray *table_data, GArray *linker)
+{
+    GArray *structures = nvdimm_build_device_structure(device_list);
+    void *header;
+
+    acpi_add_table(table_offsets, table_data);
+
+    /* NFIT header. */
+    header = acpi_data_push(table_data, sizeof(nvdimm_nfit));
+
+    /* NVDIMM device structures. */
+    g_array_append_vals(table_data, structures->data, structures->len);
+
+    build_header(linker, table_data, header, "NFIT",
+                 sizeof(nvdimm_nfit) + structures->len, 1);
+    g_array_free(structures, true);
+}
+
 static uint64_t
 nvdimm_dsm_read(void *opaque, hwaddr addr, unsigned size)
 {
@@ -61,3 +401,18 @@ void nvdimm_init_acpi_state(MemoryRegion *memory, 
MemoryRegion *io,
                           "nvdimm-acpi-io", NVDIMM_ACPI_IO_LEN);
     memory_region_add_subregion(io, NVDIMM_ACPI_IO_BASE, &state->io_mr);
 }
+
+void nvdimm_build_acpi(GArray *table_offsets, GArray *table_data,
+                       GArray *linker)
+{
+    GSList *device_list;
+
+    /* no NVDIMM device is plugged. */
+    device_list = nvdimm_get_plugged_device_list();
+    if (!device_list) {
+        return;
+    }
+
+    nvdimm_build_nfit(device_list, table_offsets, table_data, linker);
+    g_slist_free(device_list);
+}
diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
index 95e0c65..27b2854 100644
--- a/hw/i386/acpi-build.c
+++ b/hw/i386/acpi-build.c
@@ -90,6 +90,7 @@ typedef struct AcpiPmInfo {
     bool s3_disabled;
     bool s4_disabled;
     bool pcihp_bridge_en;
+    bool nvdimm_support;
     uint8_t s4_val;
     uint16_t sci_int;
     uint8_t acpi_enable_cmd;
@@ -231,6 +232,7 @@ static void acpi_get_pm_info(AcpiPmInfo *pm)
     pm->pcihp_bridge_en =
         object_property_get_bool(obj, "acpi-pci-hotplug-with-bridge-support",
                                  NULL);
+    pm->nvdimm_support = object_property_get_bool(obj, "nvdimm-support", NULL);
 }
 
 static void acpi_get_misc_info(AcpiMiscInfo *info)
@@ -1742,6 +1744,10 @@ void acpi_build(PcGuestInfo *guest_info, AcpiBuildTables 
*tables)
         build_dmar_q35(tables_blob, tables->linker);
     }
 
+    if (pm.nvdimm_support) {
+        nvdimm_build_acpi(table_offsets, tables_blob, tables->linker);
+    }
+
     /* Add tables supplied by user (if any) */
     for (u = acpi_table_first(); u; u = acpi_table_next(u)) {
         unsigned len = acpi_table_len(u);
diff --git a/include/hw/mem/nvdimm.h b/include/hw/mem/nvdimm.h
index f0b1dda..7fdf591 100644
--- a/include/hw/mem/nvdimm.h
+++ b/include/hw/mem/nvdimm.h
@@ -25,6 +25,14 @@
 
 #include "hw/mem/dimm.h"
 
+#define NVDIMM_DEBUG 0
+#define nvdimm_debug(fmt, ...)                                \
+    do {                                                      \
+        if (NVDIMM_DEBUG) {                                   \
+            fprintf(stderr, "nvdimm: " fmt, ## __VA_ARGS__);  \
+        }                                                     \
+    } while (0)
+
 /*
  * The minimum label data size is required by NVDIMM Namespace
  * specification, please refer to chapter 2 Namespaces:
@@ -114,4 +122,6 @@ typedef struct AcpiNVDIMMState AcpiNVDIMMState;
 /* Initialize the memory and IO region needed by NVDIMM ACPI emulation.*/
 void nvdimm_init_acpi_state(MemoryRegion *memory, MemoryRegion *io,
                             Object *owner, AcpiNVDIMMState *state);
+void nvdimm_build_acpi(GArray *table_offsets, GArray *table_data,
+                       GArray *linker);
 #endif
-- 
1.8.3.1