From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:46050) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZxJBU-0006zB-NG for qemu-devel@nongnu.org; Fri, 13 Nov 2015 13:34:53 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZxJBO-0003vp-T6 for qemu-devel@nongnu.org; Fri, 13 Nov 2015 13:34:52 -0500 Received: from mx1.redhat.com ([209.132.183.28]:35445) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZxJBO-0003vb-M3 for qemu-devel@nongnu.org; Fri, 13 Nov 2015 13:34:46 -0500 Date: Fri, 13 Nov 2015 18:34:40 +0000 From: "Dr. David Alan Gilbert" Message-ID: <20151113183440.GO2456@work-vm> References: <1446551816-15768-1-git-send-email-zhang.zhanghailiang@huawei.com> <1446551816-15768-18-git-send-email-zhang.zhanghailiang@huawei.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1446551816-15768-18-git-send-email-zhang.zhanghailiang@huawei.com> Subject: Re: [Qemu-devel] [PATCH COLO-Frame v10 17/38] COLO: synchronize PVM's state to SVM periodically List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: zhanghailiang Cc: lizhijian@cn.fujitsu.com, quintela@redhat.com, yunhong.jiang@intel.com, eddie.dong@intel.com, peter.huangpeng@huawei.com, qemu-devel@nongnu.org, arei.gonglei@huawei.com, stefanha@redhat.com, amit.shah@redhat.com * zhanghailiang (zhang.zhanghailiang@huawei.com) wrote: > Do checkpoint periodically, the default interval is 200ms. > > Signed-off-by: zhanghailiang > Signed-off-by: Li Zhijian > --- > migration/colo.c | 14 ++++++++++++++ > 1 file changed, 14 insertions(+) > > diff --git a/migration/colo.c b/migration/colo.c > index 0efab21..a6791f4 100644 > --- a/migration/colo.c > +++ b/migration/colo.c > @@ -11,12 +11,19 @@ > */ > > #include > +#include "qemu/timer.h" > #include "sysemu/sysemu.h" > #include "migration/colo.h" > #include "trace.h" > #include "qemu/error-report.h" > #include "qemu/sockets.h" > > +/* > + * checkpoint interval: unit ms > + * Note: Please change this default value to 10000 when we support hybrid mode. > + */ > +#define CHECKPOINT_MAX_PEROID 200 Why not put the patch that makes this a configurable parameter before this, and then we can use it straight away? > /* colo buffer */ > #define COLO_BUFFER_BASE_SIZE (4 * 1024 * 1024) > > @@ -183,6 +190,7 @@ out: > static void colo_process_checkpoint(MigrationState *s) > { > QEMUSizedBuffer *buffer = NULL; > + int64_t current_time, checkpoint_time = qemu_clock_get_ms(QEMU_CLOCK_HOST); > int fd, ret = 0; > > /* Dup the fd of to_dst_file */ > @@ -220,11 +228,17 @@ static void colo_process_checkpoint(MigrationState *s) > trace_colo_vm_state_change("stop", "run"); > > while (s->state == MIGRATION_STATUS_COLO) { > + current_time = qemu_clock_get_ms(QEMU_CLOCK_HOST); > + if (current_time - checkpoint_time < CHECKPOINT_MAX_PEROID) { > + g_usleep(100000); > + continue; > + } I'm a bit concerned at the 100ms wait, when the period is 200ms; depending how the times work out, couldn't we end up waiting for just under 300ms? - that's a big error - and it's even more weird when we make it configurable later. I don't think we've got a sleep-until, which is a shame; but how about something like: if (current_time - checkpoint_time < CHECKPOINT_MAX_PEROID) { int64_t delay_ms; delay_ms = CHECKPOINT_MAX_PERIOD - (current_time - checkpoint_time); g_usleep (delay_ms * 1000); } Dave > /* start a colo checkpoint */ > ret = colo_do_checkpoint_transaction(s, buffer); > if (ret < 0) { > goto out; > } > + checkpoint_time = qemu_clock_get_ms(QEMU_CLOCK_HOST); > } > > out: > -- > 1.8.3.1 > > -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK