Override hook_ram_load to receive the pipe file descriptor
passed by source process and page address which will be
extracted to vmsplice the page data from pipe.
Signed-off-by: Lei Li <li...@linux.vnet.ibm.com>
---
migration-local.c | 85 +++++++++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 85 insertions(+), 0 deletions(-)
diff --git a/migration-local.c b/migration-local.c
index 7ccec30..083a24f 100644
--- a/migration-local.c
+++ b/migration-local.c
@@ -32,6 +32,8 @@
#include "trace.h"
#include "qemu/osdep.h"
+#define BEFORE_PIPE_FD 199
+
//#define DEBUG_MIGRATION_LOCAL
#ifdef DEBUG_MIGRATION_LOCAL
@@ -50,6 +52,8 @@ typedef struct QEMUFileLocal {
bool unix_page_flipping;
} QEMUFileLocal;
+static bool pipefd_passed;
+
static int qemu_local_get_sockfd(void *opaque)
{
QEMUFileLocal *s = opaque;
@@ -64,6 +68,22 @@ static int qemu_local_get_buffer(void *opaque, uint8_t *buf,
ssize_t len;
for (;;) {
+ /*
+ * FIX ME: BEFORE_PIPE_FD is hard-coded and checked temporarily here
+ * because that the control message of passed pipe file descriptor
+ * might be 'eaten' to stream file by qemu_recv(), which would lead
+ * to the failure of recv_pipefd(), as it should stay in the socket
+ * and received by the real receiver recvmsg().
+ *
+ * Although this message is followed by the first load_hook flags
+ * RAM_SAVE_FLAG_HOOK, the incoming side is hardly to avoid this
+ * as it would fill it into the stream file before any check action
+ * taken. Need to find a way out to fix this.
+ */
+ if (size > BEFORE_PIPE_FD && !pipefd_passed) {
+ size = BEFORE_PIPE_FD;
+ }
+
len = qemu_recv(s->sockfd, buf, size, 0);
if (len != -1) {
break;
@@ -115,6 +135,7 @@ static int qemu_local_close(void *opaque)
}
static int send_pipefd(int sockfd, int pipefd);
+static int recv_pipefd(int sockfd);
static int qemu_local_send_pipefd(QEMUFile *f, void *opaque,
uint64_t flags)
@@ -192,10 +213,74 @@ static size_t qemu_local_save_ram(QEMUFile *f, void
*opaque,
return RAM_SAVE_CONTROL_NOT_SUPP;
}
+static int qemu_local_ram_load(QEMUFile *f, void *opaque,
+ uint64_t flags)
+{
+ QEMUFileLocal *s = opaque;
+ ram_addr_t addr;
+ struct iovec iov;
+ ssize_t ret;
+
+ /* Receive the pipe file descripter passed from source process */
+ if (!pipefd_passed) {
+ s->pipefd[0] = recv_pipefd(s->sockfd);
+ if (s->pipefd[0] <= 0) {
+ fprintf(stderr, "failed to receive pipe fd: %d\n", s->pipefd[0]);
+ } else {
+ pipefd_passed = 1;
+ DPRINTF(stderr, "succeed\n");
+ }
+
+ return s->pipefd[0];
+ }
+
+ if (pipefd_passed) {
+ void *host;
+
+ /*
+ * Extract the page address from the 8-byte record and
+ * read the page data from the pipe.
+ */
+ addr = qemu_get_be64(s->file);
+ host = qemu_get_ram_ptr(addr);
+
+ iov.iov_base = host;
+ iov.iov_len = TARGET_PAGE_SIZE;
+
+ /* The flag SPLICE_F_MOVE is introduced in kernel for the page
+ * flipping feature in QEMU, which will movie pages rather than
+ * copying, previously unused.
+ *
+ * If a move is not possible the kernel will transparently falls
+ * back to copying data.
+ *
+ * For older kernels the SPLICE_F_MOVE would be ignored and a copy
+ * would occur.
+ */
+ ret = vmsplice(s->pipefd[0], &iov, 1, SPLICE_F_MOVE);
+ if (ret == -1) {
+ if (errno != EAGAIN && errno != EINTR) {
+ fprintf(stderr, "vmsplice() load error: %s", strerror(errno));
+ return ret;
+ }
+ DPRINTF("vmsplice load error\n");
+ } else if (ret == 0) {
+ DPRINTF(stderr, "load_page: zero read\n");
+ }
+
+ DPRINTF("vmsplice (read): %zu\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+
static const QEMUFileOps pipe_read_ops = {
.get_fd = qemu_local_get_sockfd,
.get_buffer = qemu_local_get_buffer,
.close = qemu_local_close,
+ .hook_ram_load = qemu_local_ram_load
};
static const QEMUFileOps pipe_write_ops = {
--
1.7.7.6
