[Qemu-devel] [RFC PATCH RDMA support v5: 09/12] transmit pc.ram using RDMA

[mrhines@linux.vnet.ibm.com]

From: "Michael R. Hines" <mrhines@us.ibm.com>


Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
---
 arch_init.c |   59 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 56 insertions(+), 3 deletions(-)

diff --git a/arch_init.c b/arch_init.c
index c2cbc71..5cf7509 100644
--- a/arch_init.c
+++ b/arch_init.c
@@ -45,6 +45,7 @@
 #include "exec/address-spaces.h"
 #include "hw/pcspk.h"
 #include "migration/page_cache.h"
+#include "migration/rdma.h"
 #include "qemu/config-file.h"
 #include "qmp-commands.h"
 #include "trace.h"
@@ -115,6 +116,7 @@ const uint32_t arch_type = QEMU_ARCH;
 #define RAM_SAVE_FLAG_EOS      0x10
 #define RAM_SAVE_FLAG_CONTINUE 0x20
 #define RAM_SAVE_FLAG_XBZRLE   0x40
+#define RAM_SAVE_FLAG_RDMA     0x80 /* Do server dynamic RDMA registerations */
 
 
 static struct defconfig_file {
@@ -447,15 +449,23 @@ static int ram_save_block(QEMUFile *f, bool last_stage)
                 ram_bulk_stage = false;
             }
         } else {
+            bool zero;
             uint8_t *p;
             int cont = (block == last_sent_block) ?
                 RAM_SAVE_FLAG_CONTINUE : 0;
 
             p = memory_region_get_ram_ptr(mr) + offset;
 
+            /* use capability now, defaults to true */
+            zero = migrate_check_for_zero() ? is_zero_page(p) : false;
+
             /* In doubt sent page as normal */
             bytes_sent = -1;
-            if (is_zero_page(p)) {
+            if ((bytes_sent = save_rdma_page(f, block->offset, 
+                            offset, cont, TARGET_PAGE_SIZE, zero)) >= 0) {
+                acct_info.norm_pages++;
+                qemu_file_update_position(f, bytes_sent);
+            } else if (zero) {
                 acct_info.dup_pages++;
                 if (!ram_bulk_stage) {
                     bytes_sent = save_block_hdr(f, block, offset, cont,
@@ -476,7 +486,7 @@ static int ram_save_block(QEMUFile *f, bool last_stage)
             }
 
             /* XBZRLE overflow or normal page */
-            if (bytes_sent == -1) {
+            if (bytes_sent == -1 || bytes_sent == -ENOTSUP) {
                 bytes_sent = save_block_hdr(f, block, offset, cont, RAM_SAVE_FLAG_PAGE);
                 qemu_put_buffer_async(f, p, TARGET_PAGE_SIZE);
                 bytes_sent += TARGET_PAGE_SIZE;
@@ -603,6 +613,33 @@ static int ram_save_setup(QEMUFile *f, void *opaque)
     return 0;
 }
 
+/* 
+ * Inform server to begin handling dynamic page registrations
+ */
+static void ram_registration_start(QEMUFile *f)
+{
+    if(qemu_file_ops_are(f, &rdma_write_ops)) {
+        qemu_put_be64(f, RAM_SAVE_FLAG_RDMA);
+    }
+}
+
+/*
+ * Inform server that dynamic registrations are done for now.
+ * First, flush writes, if any.
+ */
+static int ram_registration_stop(QEMUFile *f)
+{
+    int ret = 0;
+
+    if (qemu_file_ops_are(f, &rdma_write_ops)) {
+        ret = qemu_rdma_drain_cq(f);
+        if(ret >= 0)
+            ret = qemu_rdma_finish_registrations(f);
+    }
+
+    return ret;
+}
+
 static int ram_save_iterate(QEMUFile *f, void *opaque)
 {
     int ret;
@@ -616,6 +653,8 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
         reset_ram_globals();
     }
 
+    ram_registration_start(f);
+
     t0 = qemu_get_clock_ns(rt_clock);
     i = 0;
     while ((ret = qemu_file_rate_limit(f)) == 0) {
@@ -646,6 +685,9 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
 
     qemu_mutex_unlock_ramlist();
 
+    if(ret >= 0)
+        ret = ram_registration_stop(f);
+
     if (ret < 0) {
         bytes_transferred += total_sent;
         return ret;
@@ -660,8 +702,11 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
 
 static int ram_save_complete(QEMUFile *f, void *opaque)
 {
+    int ret = 0;
+
     qemu_mutex_lock_ramlist();
     migration_bitmap_sync();
+    ram_registration_start(f);
 
     /* try transferring iterative blocks of memory */
 
@@ -676,12 +721,15 @@ static int ram_save_complete(QEMUFile *f, void *opaque)
         }
         bytes_transferred += bytes_sent;
     }
+
+    ret = ram_registration_stop(f);
+
     migration_end();
 
     qemu_mutex_unlock_ramlist();
     qemu_put_be64(f, RAM_SAVE_FLAG_EOS);
 
-    return 0;
+    return ret;
 }
 
 static uint64_t ram_save_pending(QEMUFile *f, void *opaque, uint64_t max_size)
@@ -864,6 +912,11 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id)
                 ret = -EINVAL;
                 goto done;
             }
+        } else if ((flags & RAM_SAVE_FLAG_RDMA) &&
+                          qemu_file_ops_are(f, &rdma_read_ops)) {
+            ret = qemu_rdma_handle_registrations(f);
+            if(ret < 0)
+                goto done;
         }
         error = qemu_file_get_error(f);
         if (error) {
-- 
1.7.10.4