[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH for-2.6 2/2] block/mirror: Refresh stale bitmap
From: |
Kevin Wolf |
Subject: |
Re: [Qemu-devel] [PATCH for-2.6 2/2] block/mirror: Refresh stale bitmap iterator cache |
Date: |
Wed, 20 Apr 2016 14:13:41 +0200 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
Am 20.04.2016 um 00:59 hat Max Reitz geschrieben:
> If the drive's dirty bitmap is dirtied while the mirror operation is
> running, the cache of the iterator used by the mirror code may become
> stale and not contain all dirty bits.
>
> This only becomes an issue if we are looking for contiguously dirty
> chunks on the drive. In that case, we can easily detect the discrepancy
> and just refresh the iterator if one occurs.
>
> Signed-off-by: Max Reitz <address@hidden>
> ---
> block/mirror.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/block/mirror.c b/block/mirror.c
> index 2714a77..9df1fae 100644
> --- a/block/mirror.c
> +++ b/block/mirror.c
> @@ -334,6 +334,11 @@ static uint64_t coroutine_fn
> mirror_iteration(MirrorBlockJob *s)
> }
>
> hbitmap_next = hbitmap_iter_next(&s->hbi);
> + if (hbitmap_next > next_sector || hbitmap_next < 0) {
> + /* The bitmap iterator's cache is stale, refresh it */
> + bdrv_set_dirty_iter(&s->hbi, next_sector);
> + hbitmap_next = hbitmap_iter_next(&s->hbi);
> + }
> assert(hbitmap_next == next_sector);
The iterator doesn't seem to be used afterwards anyway, so why not just
use next_sector and stop using the iterator when we already know what
result we want to get?
Kevin