[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v3 07/10] migration: do not flush_compressed_dat
From: |
Peter Xu |
Subject: |
Re: [Qemu-devel] [PATCH v3 07/10] migration: do not flush_compressed_data at the end of each iteration |
Date: |
Wed, 8 Aug 2018 12:52:09 +0800 |
User-agent: |
Mutt/1.10.0 (2018-05-17) |
On Tue, Aug 07, 2018 at 05:12:06PM +0800, address@hidden wrote:
> From: Xiao Guangrong <address@hidden>
>
> flush_compressed_data() needs to wait all compression threads to
> finish their work, after that all threads are free until the
> migration feeds new request to them, reducing its call can improve
> the throughput and use CPU resource more effectively
>
> We do not need to flush all threads at the end of iteration, the
> data can be kept locally until the memory block is changed or
> memory migration starts over in that case we will meet a dirtied
> page which may still exists in compression threads's ring
>
> Signed-off-by: Xiao Guangrong <address@hidden>
> ---
> migration/ram.c | 14 +++++++++++++-
> 1 file changed, 13 insertions(+), 1 deletion(-)
>
> diff --git a/migration/ram.c b/migration/ram.c
> index 99ecf9b315..55966bc2c1 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -306,6 +306,8 @@ struct RAMState {
> uint64_t iterations;
> /* number of dirty bits in the bitmap */
> uint64_t migration_dirty_pages;
> + /* last dirty_sync_count we have seen */
> + uint64_t dirty_sync_count_prev;
> /* protects modification of the bitmap */
> QemuMutex bitmap_mutex;
> /* The RAMBlock used in the last src_page_requests */
> @@ -3173,6 +3175,17 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
>
> ram_control_before_iterate(f, RAM_CONTROL_ROUND);
>
> + /*
> + * if memory migration starts over, we will meet a dirtied page which
> + * may still exists in compression threads's ring, so we should flush
> + * the compressed data to make sure the new page is not overwritten by
> + * the old one in the destination.
> + */
> + if (ram_counters.dirty_sync_count != rs->dirty_sync_count_prev) {
> + rs->dirty_sync_count_prev = ram_counters.dirty_sync_count;
> + flush_compressed_data(rs);
AFAIU this only happens when ram_save_pending() calls
migration_bitmap_sync(). Could we just simply flush there? Then we
can avoid that new variable.
> + }
> +
> t0 = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
> i = 0;
> while ((ret = qemu_file_rate_limit(f)) == 0 ||
> @@ -3205,7 +3218,6 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
> }
> i++;
> }
> - flush_compressed_data(rs);
> rcu_read_unlock();
>
> /*
> --
> 2.14.4
>
Regards,
--
Peter Xu
- Re: [Qemu-devel] [PATCH v3 01/10] migration: do not wait for free thread, (continued)
[Qemu-devel] [PATCH v3 02/10] migration: fix counting normal page for compression, guangrong . xiao, 2018/08/07
[Qemu-devel] [PATCH v3 03/10] migration: introduce save_zero_page_to_file, guangrong . xiao, 2018/08/07
[Qemu-devel] [PATCH v3 04/10] migration: drop the return value of do_compress_ram_page, guangrong . xiao, 2018/08/07
[Qemu-devel] [PATCH v3 05/10] migration: move handle of zero page to the thread, guangrong . xiao, 2018/08/07
[Qemu-devel] [PATCH v3 06/10] migration: hold the lock only if it is really needed, guangrong . xiao, 2018/08/07
[Qemu-devel] [PATCH v3 07/10] migration: do not flush_compressed_data at the end of each iteration, guangrong . xiao, 2018/08/07
- Re: [Qemu-devel] [PATCH v3 07/10] migration: do not flush_compressed_data at the end of each iteration,
Peter Xu <=
[Qemu-devel] [PATCH v3 08/10] migration: handle the error condition properly, guangrong . xiao, 2018/08/07
Re: [Qemu-devel] [PATCH v3 08/10] migration: handle the error condition properly, Dr. David Alan Gilbert, 2018/08/08
Re: [Qemu-devel] [PATCH v3 08/10] migration: handle the error condition properly, Xiao Guangrong, 2018/08/08
[Qemu-devel] [PATCH v3 09/10] migration: fix calculating xbzrle_counters.cache_miss_rate, guangrong . xiao, 2018/08/07