[Qemu-devel] [QEMU RFC PATCH v2 4/6] Migration: migrate QTAILQ

[Jianjun Duan <duanj@linux.vnet.ibm.com>]

A recursive structure has elements of the same type in itself. Currently
we cannot directly transfer a QTAILQ instance, or any recursive
structure such as lists in migration because of the limitation in the
migration code. Here we introduce a general approach to transfer such
structures. In our approach a recursive structure is tagged with VMS_CSTM.
We extend VMStateField with 3 fields: meta_data to store the meta data
about the recursive structure in question, extend_get to load the structure
from migration stream to memory, and extend_put to dump the structure into
the migration stream. This extension mirrors VMStateInfo. We then modify
vmstate_save_state and vmstate_load_state so that when VMS_CSTM is
encountered, extend_put and extend_get are called respectively with the
knowledge of the meta data.

To make it work for QTAILQ in qemu/queue.h, we created the meta data
format, extend_get and extend_put for it. We will use this approach to
transfer pending_events and ccs_list in spapr state.

We also create some macros in qemu/queue.h to get the layout information
about QTAILQ. This ensures that we do not depend on the implementation
details about QTAILQ in the migration code.

Signed-off-by: Jianjun Duan <duanj@linux.vnet.ibm.com>
---
 include/migration/vmstate.h | 59 +++++++++++++++++++++++++++++++
 include/qemu/queue.h        | 38 ++++++++++++++++++++
 migration/vmstate.c         | 84 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 181 insertions(+)

diff --git a/include/migration/vmstate.h b/include/migration/vmstate.h
index 1622638..bf57b25 100644
--- a/include/migration/vmstate.h
+++ b/include/migration/vmstate.h
@@ -183,6 +183,8 @@ enum VMStateFlags {
      * to determine the number of entries in the array. Only valid in
      * combination with one of VMS_VARRAY*. */
     VMS_MULTIPLY_ELEMENTS = 0x4000,
+    /* For fields which need customized handling, such as QTAILQ in queue.h*/
+    VMS_CSTM            = 0x8000,
 };
 
 typedef struct {
@@ -198,6 +200,14 @@ typedef struct {
     const VMStateDescription *vmsd;
     int version_id;
     bool (*field_exists)(void *opaque, int version_id);
+    /*
+     * Following 3 fields are for VMStateField which needs customized handling,
+     * such as QTAILQ in qemu/queue.h, lists, and tree.
+     */
+    const void *meta_data;
+    int (*extend_get)(QEMUFile *f, const void *metadata, void *opaque);
+    void (*extend_put)(QEMUFile *f, const void *metadata, void *opaque,
+                       QJSON *vmdesc);
 } VMStateField;
 
 struct VMStateDescription {
@@ -654,6 +664,18 @@ extern const VMStateInfo vmstate_info_bitmap;
     .offset       = offsetof(_state, _field),                        \
 }
 
+/* For fields that need customized handling, such as queue, list */
+#define VMSTATE_CSTM_V(_field, _state, _version, _meta_data, _get, _put)  \
+{                                                                         \
+    .name         = (stringify(_field)),                                  \
+    .version_id   = (_version),                                           \
+    .flags        = VMS_CSTM,                                             \
+    .offset       = offsetof(_state, _field),                             \
+    .meta_data    = &(_meta_data),                                        \
+    .extend_get   = (_get),                                               \
+    .extend_put   = (_put),                                               \
+}
+
 /* _f : field name
    _f_n : num of elements field_name
    _n : num of elements
@@ -970,4 +992,41 @@ int64_t self_announce_delay(int round)
 
 void dump_vmstate_json_to_file(FILE *out_fp);
 
+
+/* Meta data for QTAILQ */
+typedef struct QTAILQMetaData {
+    /* the offset of tqh_first in QTAILQ_HEAD */
+    size_t first;
+    /* the offset of tqh_last in QTAILQ_HEAD */
+    size_t last;
+    /* the offset of tqe_next in QTAILQ_ENTRY */
+    size_t next;
+    /* the offset of tqe_prev in QTAILQ_ENTRY */
+    size_t prev;
+    /* the offset of QTAILQ_ENTRY in a QTAILQ element*/
+    size_t entry;
+    /* size of a QTAILQ element */
+    size_t size;
+    /* vmsd of a QTAILQ element */
+    const VMStateDescription *vmsd;
+} QTAILQMetaData;
+
+#define VMSTATE_QTAILQ_METADATA(_field, _state, _type, _next, _vmsd) { \
+    .first = QTAILQ_FIRST_OFFSET(typeof_field(_state, _field)),        \
+    .last =  QTAILQ_LAST_OFFSET(typeof_field(_state, _field)),         \
+    .next = QTAILQ_NEXT_OFFSET(_type, _next),                          \
+    .prev = QTAILQ_PREV_OFFSET(_type, _next),                          \
+    .entry = offsetof(_type, _next),                                   \
+    .size = sizeof(_type),                                             \
+    .vmsd = &(_vmsd),                                                  \
+}
+
+int vmstate_get_qtailq(QEMUFile *f, const void *metadata, void *opaque);
+void vmstate_put_qtailq(QEMUFile *f, const void *metadata,
+                         void *opaque, QJSON *vmdesc);
+
+/* VMStateField for QTAILQ field */
+#define VMSTATE_QTAILQ_V(_field, _state, _version, _meta_data)               \
+    VMSTATE_CSTM_V(_field, _state, _version, _meta_data, vmstate_get_qtailq, \
+                   vmstate_put_qtailq)
 #endif
diff --git a/include/qemu/queue.h b/include/qemu/queue.h
index f781aa2..46962d7 100644
--- a/include/qemu/queue.h
+++ b/include/qemu/queue.h
@@ -437,3 +437,41 @@ struct {                                                                \
         (*(((struct headname *)((elm)->field.tqe_prev))->tqh_last))
 
 #endif  /* !QEMU_SYS_QUEUE_H_ */
+
+/*
+ * Offsets of layout of a tail queue head.
+ */
+#define QTAILQ_FIRST_OFFSET(head_type) \
+        ((size_t) ((char *) &((head_type *)0)->tqh_first - (char *)0))
+
+#define QTAILQ_LAST_OFFSET(head_type) \
+        ((size_t) ((char *) &((head_type *)0)->tqh_last - (char *)0))
+
+/*
+ * Offsets of layout of a tail queue element.
+ */
+#define QTAILQ_NEXT_OFFSET(ele_type, field)                            \
+        ((size_t) ((char *) &((ele_type *)0)->field.tqe_next -         \
+                   (char *) &((ele_type *)0)->field))
+
+#define QTAILQ_PREV_OFFSET(ele_type, field) \
+        ((size_t) ((char *) &((ele_type *)0)->field.tqe_prev -         \
+                   (char *) &((ele_type *)0)->field))
+/*
+ * Tail queue tranversal using pointer arithmetic.
+ */
+#define QTAILQ_RAW_FOREACH(elm, head, entry, first, next)              \
+        for ((elm) = *((void **) ((char *) (head) + (first)));         \
+             (elm);                                                    \
+             (elm) = *((void **) ((char *) (elm) + (entry) + (next))))
+/*
+ * Tail queue insertion using pointer arithmetic.
+ */
+#define QTAILQ_RAW_INSERT_TAIL(head, elm, entry, first, last, next, prev) do { \
+            *((void **) ((char *) (elm) + (entry) + (next))) = NULL;           \
+            *((void **) ((char *) (elm) + (entry) + (prev))) =                 \
+                *((void **) ((char *) (head) + (last)));                       \
+            **((void ***)((char *) (head) + (last))) = (elm);                  \
+            *((void **) ((char *) (head) + (last))) =                          \
+                (void *) ((char *) (elm) + (entry) + (next));                  \
+} while (/*CONSTCOND*/0)
diff --git a/migration/vmstate.c b/migration/vmstate.c
index bf3d5db..47cd052 100644
--- a/migration/vmstate.c
+++ b/migration/vmstate.c
@@ -5,6 +5,7 @@
 #include "migration/vmstate.h"
 #include "qemu/bitops.h"
 #include "qemu/error-report.h"
+#include "qemu/queue.h"
 #include "trace.h"
 #include "qjson.h"
 
@@ -121,6 +122,8 @@ int vmstate_load_state(QEMUFile *f, const VMStateDescription *vmsd,
                 if (field->flags & VMS_STRUCT) {
                     ret = vmstate_load_state(f, field->vmsd, addr,
                                              field->vmsd->version_id);
+                } else if (field->flags & VMS_CSTM) {
+                    ret = field->extend_get(f, field->meta_data, addr);
                 } else {
                     ret = field->info->get(f, addr, size);
 
@@ -193,6 +196,8 @@ static const char *vmfield_get_type_name(VMStateField *field)
 
     if (field->flags & VMS_STRUCT) {
         type = "struct";
+    } else if (field->flags & VMS_CSTM) {
+        type = "customized";
     } else if (field->info->name) {
         type = field->info->name;
     }
@@ -327,6 +332,8 @@ void vmstate_save_state(QEMUFile *f, const VMStateDescription *vmsd,
                 }
                 if (field->flags & VMS_STRUCT) {
                     vmstate_save_state(f, field->vmsd, addr, vmdesc_loop);
+                } else if  (field->flags & VMS_CSTM) {
+                    field->extend_put(f, field->meta_data, addr, vmdesc_loop);
                 } else {
                     field->info->put(f, addr, size);
                 }
@@ -916,3 +923,80 @@ const VMStateInfo vmstate_info_bitmap = {
     .get = get_bitmap,
     .put = put_bitmap,
 };
+
+/* extend_get for QTAILQ */
+int vmstate_get_qtailq(QEMUFile *f, const void *metadata, void *opaque)
+{
+    bool link;
+    int ret = 0;
+    size_t first, last, next, prev, entry, size;
+    QTAILQMetaData *data = (QTAILQMetaData *)metadata;
+    const VMStateDescription *vmsd = data->vmsd;
+    int version_id = vmsd->version_id;
+    void *elm;
+
+    first = data->first;
+    last = data->last;
+    next = data->next;
+    prev = data->prev;
+    entry = data->entry;
+    size = data->size;
+
+    trace_vmstate_load_state(vmsd->name, version_id);
+    if (version_id > vmsd->version_id) {
+        trace_vmstate_load_state_end(vmsd->name, "too new", -EINVAL);
+        return -EINVAL;
+    }
+    if (version_id < vmsd->minimum_version_id) {
+        trace_vmstate_load_state_end(vmsd->name, "too old", -EINVAL);
+        return -EINVAL;
+    }
+
+    while (true) {
+        vmstate_info_bool.get(f, &link, sizeof(bool));
+        if (!link) {
+            break;
+        }
+
+        elm =  g_malloc(size);
+        ret = vmstate_load_state(f, vmsd, elm, version_id);
+        if (ret) {
+            return ret;
+        }
+        QTAILQ_RAW_INSERT_TAIL(opaque, elm, entry, first, last, next, prev);
+    }
+
+    trace_vmstate_load_state_end(vmsd->name, "end", ret);
+    return ret;
+}
+
+/* extend_get for QTAILQ */
+void vmstate_put_qtailq(QEMUFile *f, const void *metadata, void *opaque,
+                         QJSON *vmdesc)
+{
+    bool link = true;
+    size_t first, next, entry;
+    QTAILQMetaData *data = (QTAILQMetaData *)metadata;
+    const VMStateDescription *vmsd = data->vmsd;
+    void *elm;
+
+    first = data->first;
+    next = data->next;
+    entry = data->entry;
+
+    if (vmdesc) {
+        json_prop_str(vmdesc, "vmsd_name", vmsd->name);
+        json_prop_int(vmdesc, "version", vmsd->version_id);
+        json_start_array(vmdesc, "fields");
+    }
+
+    QTAILQ_RAW_FOREACH(elm, opaque, entry, first, next) {
+        vmstate_info_bool.put(f, &link, sizeof(bool));
+        vmstate_save_state(f, vmsd, elm, vmdesc);
+    }
+    link = false;
+    vmstate_info_bool.put(f, &link, sizeof(bool));
+    if (vmdesc) {
+        json_end_array(vmdesc);
+    }
+}
-- 
1.9.1