From: mrhines@linux.vnet.ibm.com
To: qemu-devel@nongnu.org
Cc: aliguori@us.ibm.com, mst@redhat.com, owasserm@redhat.com,
abali@us.ibm.com, mrhines@us.ibm.com, gokul@us.ibm.com,
pbonzini@redhat.com
Subject: [Qemu-devel] [RFC PATCH RDMA support v4: 10/10] send pc.ram over RDMA
Date: Sun, 17 Mar 2013 23:19:03 -0400 [thread overview]
Message-ID: <1363576743-6146-11-git-send-email-mrhines@linux.vnet.ibm.com> (raw)
In-Reply-To: <1363576743-6146-1-git-send-email-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 | 28 +++++++++++++++++++++++++++-
1 file changed, 27 insertions(+), 1 deletion(-)
diff --git a/arch_init.c b/arch_init.c
index 98e2bc6..b013cc8 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"
@@ -225,6 +226,18 @@ static void acct_clear(void)
memset(&acct_info, 0, sizeof(acct_info));
}
+/*
+ * RDMA pc.ram doesn't go through QEMUFile directly,
+ * but still needs to be accounted for...
+ */
+uint64_t delta_norm_mig_bytes_transferred(void)
+{
+ static uint64_t last_norm_pages = 0;
+ uint64_t delta_bytes = (acct_info.norm_pages - last_norm_pages) * TARGET_PAGE_SIZE;
+ last_norm_pages = acct_info.norm_pages;
+ return delta_bytes;
+}
+
uint64_t dup_mig_bytes_transferred(void)
{
return acct_info.dup_pages * TARGET_PAGE_SIZE;
@@ -463,7 +476,11 @@ static int ram_save_block(QEMUFile *f, bool last_stage)
/* In doubt sent page as normal */
bytes_sent = -1;
- if (is_dup_page(p)) {
+ if (migrate_use_rdma(f)) {
+ /* for now, mapping the page is slower than RDMA */
+ acct_info.norm_pages++;
+ bytes_sent = save_rdma_page(f, block->offset, offset, cont, TARGET_PAGE_SIZE);
+ } else if (is_dup_page(p)) {
acct_info.dup_pages++;
bytes_sent = save_block_hdr(f, block, offset, cont,
RAM_SAVE_FLAG_COMPRESS);
@@ -648,6 +665,15 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
qemu_mutex_unlock_ramlist();
+ /*
+ * Don't go to the next iteration without
+ * ensuring RDMA transfers have completed.
+ */
+ if ((ret = qemu_drain(f)) < 0) {
+ fprintf(stderr, "failed to drain RDMA first!\n");
+ return ret;
+ }
+
if (ret < 0) {
bytes_transferred += total_sent;
return ret;
--
1.7.10.4
prev parent reply other threads:[~2013-03-18 3:19 UTC|newest]
Thread overview: 73+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-03-18 3:18 [Qemu-devel] [RFC PATCH RDMA support v4: 00/10] cleaner ramblocks and documentation mrhines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 01/10] ./configure --enable-rdma mrhines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 02/10] check for CONFIG_RDMA mrhines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 03/10] more verbose documentation of the RDMA transport mrhines
2013-03-18 10:40 ` Michael S. Tsirkin
2013-03-18 20:24 ` Michael R. Hines
2013-03-18 21:26 ` Michael S. Tsirkin
2013-03-18 23:23 ` Michael R. Hines
2013-03-19 8:19 ` Michael S. Tsirkin
2013-03-19 13:21 ` Michael R. Hines
2013-03-19 15:08 ` Michael R. Hines
2013-03-19 15:16 ` Michael S. Tsirkin
2013-03-19 15:32 ` Michael R. Hines
2013-03-19 15:36 ` Michael S. Tsirkin
2013-03-19 17:09 ` Michael R. Hines
2013-03-19 17:14 ` Paolo Bonzini
2013-03-19 17:23 ` Michael S. Tsirkin
2013-03-19 17:40 ` Michael R. Hines
2013-03-19 17:52 ` Paolo Bonzini
2013-03-19 18:04 ` Michael R. Hines
2013-03-20 13:07 ` Michael S. Tsirkin
2013-03-20 15:15 ` Michael R. Hines
2013-03-20 15:22 ` Michael R. Hines
2013-03-20 15:55 ` Michael S. Tsirkin
2013-03-20 16:08 ` Michael R. Hines
2013-03-20 19:06 ` Michael S. Tsirkin
2013-03-20 20:20 ` Michael R. Hines
2013-03-20 20:31 ` Michael S. Tsirkin
2013-03-20 20:39 ` Michael R. Hines
2013-03-20 20:46 ` Michael S. Tsirkin
2013-03-20 20:56 ` Michael R. Hines
2013-03-21 5:20 ` Michael S. Tsirkin
2013-03-20 20:24 ` Michael R. Hines
2013-03-20 20:37 ` Michael S. Tsirkin
2013-03-20 20:45 ` Michael R. Hines
2013-03-20 20:52 ` Michael S. Tsirkin
2013-03-19 17:49 ` Michael R. Hines
2013-03-21 6:11 ` Michael S. Tsirkin
2013-03-21 15:22 ` Michael R. Hines
2013-04-05 20:45 ` Michael R. Hines
2013-04-05 20:46 ` Michael R. Hines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 04/10] iterators for getting the RAMBlocks mrhines
2013-03-18 8:48 ` Paolo Bonzini
2013-03-18 20:25 ` Michael R. Hines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 05/10] reuse function for parsing the QMP 'migrate' string mrhines
2013-03-18 3:18 ` [Qemu-devel] [RFC PATCH RDMA support v4: 06/10] core RDMA migration code (rdma.c) mrhines
2013-03-18 3:19 ` [Qemu-devel] [RFC PATCH RDMA support v4: 07/10] connection-establishment for RDMA mrhines
2013-03-18 8:56 ` Paolo Bonzini
2013-03-18 20:26 ` Michael R. Hines
2013-03-18 3:19 ` [Qemu-devel] [RFC PATCH RDMA support v4: 08/10] introduce QEMUFileRDMA mrhines
2013-03-18 9:09 ` Paolo Bonzini
2013-03-18 20:33 ` Michael R. Hines
2013-03-19 9:18 ` Paolo Bonzini
2013-03-19 13:12 ` Michael R. Hines
2013-03-19 13:25 ` Paolo Bonzini
2013-03-19 13:40 ` Michael R. Hines
2013-03-19 13:45 ` Paolo Bonzini
2013-03-19 14:10 ` Michael R. Hines
2013-03-19 14:22 ` Paolo Bonzini
2013-03-19 15:02 ` [Qemu-devel] [Bug]? (RDMA-related) ballooned memory not consulted during migration? Michael R. Hines
2013-03-19 15:12 ` Michael R. Hines
2013-03-19 15:17 ` Michael S. Tsirkin
2013-03-19 18:27 ` [Qemu-devel] [RFC PATCH RDMA support v4: 08/10] introduce QEMUFileRDMA Michael R. Hines
2013-03-19 18:40 ` Paolo Bonzini
2013-03-20 15:20 ` Paolo Bonzini
2013-03-20 16:09 ` Michael R. Hines
2013-03-18 3:19 ` [Qemu-devel] [RFC PATCH RDMA support v4: 09/10] check for QMP string and bypass nonblock() calls mrhines
2013-03-18 8:47 ` Paolo Bonzini
2013-03-18 20:37 ` Michael R. Hines
2013-03-19 9:23 ` Paolo Bonzini
2013-03-19 13:08 ` Michael R. Hines
2013-03-19 13:20 ` Paolo Bonzini
2013-03-18 3:19 ` mrhines [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1363576743-6146-11-git-send-email-mrhines@linux.vnet.ibm.com \
--to=mrhines@linux.vnet.ibm.com \
--cc=abali@us.ibm.com \
--cc=aliguori@us.ibm.com \
--cc=gokul@us.ibm.com \
--cc=mrhines@us.ibm.com \
--cc=mst@redhat.com \
--cc=owasserm@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).