From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:39052) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZyJIn-0006Fl-Ir for qemu-devel@nongnu.org; Mon, 16 Nov 2015 07:54:34 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZyJIj-0007xM-AB for qemu-devel@nongnu.org; Mon, 16 Nov 2015 07:54:33 -0500 Received: from szxga02-in.huawei.com ([119.145.14.65]:32774) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZyJIh-0007wT-MR for qemu-devel@nongnu.org; Mon, 16 Nov 2015 07:54:29 -0500 References: <1446551816-15768-1-git-send-email-zhang.zhanghailiang@huawei.com> <1446551816-15768-17-git-send-email-zhang.zhanghailiang@huawei.com> <20151113163834.GM2456@work-vm> From: zhanghailiang Message-ID: <5649D01E.5080807@huawei.com> Date: Mon, 16 Nov 2015 20:46:22 +0800 MIME-Version: 1.0 In-Reply-To: <20151113163834.GM2456@work-vm> Content-Type: text/plain; charset="windows-1252"; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH COLO-Frame v10 16/38] COLO: Flush PVM's cached RAM into SVM's memory List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Dr. David Alan Gilbert" 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 On 2015/11/14 0:38, Dr. David Alan Gilbert wrote: > * zhanghailiang (zhang.zhanghailiang@huawei.com) wrote: >> During the time of VM's running, PVM may dirty some pages, we will transfer >> PVM's dirty pages to SVM and store them into SVM's RAM cache at next checkpoint >> time. So, the content of SVM's RAM cache will always be some with PVM's memory >> after checkpoint. >> >> Instead of flushing all content of PVM's RAM cache into SVM's MEMORY, >> we do this in a more efficient way: >> Only flush any page that dirtied by PVM since last checkpoint. >> In this way, we can ensure SVM's memory same with PVM's. >> >> Besides, we must ensure flush RAM cache before load device state. > > Yes, just a couple of minor comments below; mostly OK. > >> Signed-off-by: zhanghailiang >> Signed-off-by: Li Zhijian >> Signed-off-by: Gonglei >> --- >> v10: trace the number of dirty pages that be received. >> --- >> include/migration/colo.h | 1 + >> migration/colo.c | 2 -- >> migration/ram.c | 40 ++++++++++++++++++++++++++++++++++++++++ >> trace-events | 1 + >> 4 files changed, 42 insertions(+), 2 deletions(-) >> >> diff --git a/include/migration/colo.h b/include/migration/colo.h >> index 8edd5f1..be2890b 100644 >> --- a/include/migration/colo.h >> +++ b/include/migration/colo.h >> @@ -32,4 +32,5 @@ bool migration_incoming_in_colo_state(void); >> /* ram cache */ >> int colo_init_ram_cache(void); >> void colo_release_ram_cache(void); >> +void colo_flush_ram_cache(void); >> #endif >> diff --git a/migration/colo.c b/migration/colo.c >> index 1339774..0efab21 100644 >> --- a/migration/colo.c >> +++ b/migration/colo.c >> @@ -381,8 +381,6 @@ void *colo_process_incoming_thread(void *opaque) >> } >> qemu_mutex_unlock_iothread(); >> >> - /* TODO: flush vm state */ >> - >> ret = colo_ctl_put(mis->to_src_file, COLO_COMMAND_VMSTATE_LOADED, 0); >> if (ret < 0) { >> goto out; >> diff --git a/migration/ram.c b/migration/ram.c >> index 70879bd..d7e0e37 100644 >> --- a/migration/ram.c >> +++ b/migration/ram.c >> @@ -1601,6 +1601,7 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) >> int flags = 0, ret = 0; >> static uint64_t seq_iter; >> int len = 0; >> + bool need_flush = false; >> >> seq_iter++; >> >> @@ -1669,6 +1670,8 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) >> ret = -EINVAL; >> break; >> } >> + >> + need_flush = true; >> ch = qemu_get_byte(f); >> ram_handle_compressed(host, ch, TARGET_PAGE_SIZE); >> break; >> @@ -1679,6 +1682,8 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) >> ret = -EINVAL; >> break; >> } >> + >> + need_flush = true; >> qemu_get_buffer(f, host, TARGET_PAGE_SIZE); >> break; >> case RAM_SAVE_FLAG_COMPRESS_PAGE: >> @@ -1711,6 +1716,7 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) >> ret = -EINVAL; >> break; >> } >> + need_flush = true; > > You can probably move the 'need_flush' to the big if near the top of the loop in the > current version. > Good catch, i will fix it in next version. >> break; >> case RAM_SAVE_FLAG_EOS: >> /* normal exit */ >> @@ -1730,6 +1736,11 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) >> } >> >> rcu_read_unlock(); >> + >> + if (!ret && ram_cache_enable && need_flush) { >> + DPRINTF("Flush ram_cache\n"); > > trace_ Got it. > >> + colo_flush_ram_cache(); >> + } >> DPRINTF("Completed load of VM with exit code %d seq iteration " >> "%" PRIu64 "\n", ret, seq_iter); >> return ret; >> @@ -1799,6 +1810,35 @@ void colo_release_ram_cache(void) >> rcu_read_unlock(); >> } >> >> +/* >> + * Flush content of RAM cache into SVM's memory. >> + * Only flush the pages that be dirtied by PVM or SVM or both. >> + */ >> +void colo_flush_ram_cache(void) >> +{ >> + RAMBlock *block = NULL; >> + void *dst_host; >> + void *src_host; >> + ram_addr_t offset = 0; >> + >> + trace_colo_flush_ram_cache(migration_dirty_pages); >> + rcu_read_lock(); >> + block = QLIST_FIRST_RCU(&ram_list.blocks); >> + while (block) { >> + offset = migration_bitmap_find_and_reset_dirty(block, offset); > > You'll need to rework that a little (I split that into > migration_bitmap_find_dirty and migration_bitmap_clear_dirty) > Yes, i have rebase it in my private branch after your post-copy merged. ;) Thanks, zhanghailiang >> + if (offset >= block->used_length) { >> + offset = 0; >> + block = QLIST_NEXT_RCU(block, next); >> + } else { >> + dst_host = block->host + offset; >> + src_host = block->host_cache + offset; >> + memcpy(dst_host, src_host, TARGET_PAGE_SIZE); >> + } >> + } >> + rcu_read_unlock(); >> + assert(migration_dirty_pages == 0); >> +} >> + >> static SaveVMHandlers savevm_ram_handlers = { >> .save_live_setup = ram_save_setup, >> .save_live_iterate = ram_save_iterate, >> diff --git a/trace-events b/trace-events >> index ee4679c..c98bc13 100644 >> --- a/trace-events >> +++ b/trace-events >> @@ -1232,6 +1232,7 @@ qemu_file_fclose(void) "" >> migration_bitmap_sync_start(void) "" >> migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64"" >> migration_throttle(void) "" >> +colo_flush_ram_cache(uint64_t dirty_pages) "dirty_pages %" PRIu64"" >> >> # hw/display/qxl.c >> disable qxl_interface_set_mm_time(int qid, uint32_t mm_time) "%d %d" >> -- >> 1.8.3.1 >> >> > -- > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK > > . >