[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v2 2/6] block: Fix dirty bitmap in bdrv_co_disca
From: |
John Snow |
Subject: |
Re: [Qemu-devel] [PATCH v2 2/6] block: Fix dirty bitmap in bdrv_co_discard |
Date: |
Mon, 11 May 2015 15:22:44 -0400 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.6.0 |
On 05/06/2015 12:52 AM, Fam Zheng wrote:
> Unsetting dirty globally with discard is not very correct. The discard may
> zero
> out sectors (depending on can_write_zeroes_with_unmap), we should replicate
> this change to destinition side to make sure that the guest sees the same
> data.
>
> Calling bdrv_reset_dirty also troubles mirror job because the hbitmap iterator
> doesn't expect unsetting of bits after current position.
>
> So let's do it the opposite way which fixes both problems: set the dirty bits
> if we are to discard it.
>
> Reported-by: address@hidden
> Signed-off-by: Fam Zheng <address@hidden>
> ---
> block/io.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/block/io.c b/block/io.c
> index 1ce62c4..809688b 100644
> --- a/block/io.c
> +++ b/block/io.c
> @@ -2343,8 +2343,6 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs,
> int64_t sector_num,
> return -EROFS;
> }
>
> - bdrv_reset_dirty(bs, sector_num, nb_sectors);
> -
> /* Do nothing if disabled. */
> if (!(bs->open_flags & BDRV_O_UNMAP)) {
> return 0;
> @@ -2354,6 +2352,8 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs,
> int64_t sector_num,
> return 0;
> }
>
> + bdrv_set_dirty(bs, sector_num, nb_sectors);
> +
> max_discard = MIN_NON_ZERO(bs->bl.max_discard, BDRV_REQUEST_MAX_SECTORS);
> while (nb_sectors > 0) {
> int ret;
>
For the clueless: will discard *always* change the data, or is it
possible that some implementations might do nothing?
Is it possible to just omit a set/reset from this function altogether
and let whatever function that is called later (e.g. a write_zeroes
call) worry about setting the dirty bits?
What I wonder about: Is it possible that we are needlessly marking data
as dirty when it has not changed?
- [Qemu-devel] [PATCH v2 0/6] block: Mirror discarded sectors, Fam Zheng, 2015/05/06
- [Qemu-devel] [PATCH v2 1/6] mirror: Discard target sectors if not allocated at source side, Fam Zheng, 2015/05/06
- [Qemu-devel] [PATCH v2 2/6] block: Fix dirty bitmap in bdrv_co_discard, Fam Zheng, 2015/05/06
- Re: [Qemu-devel] [PATCH v2 2/6] block: Fix dirty bitmap in bdrv_co_discard,
John Snow <=
- [Qemu-devel] [PATCH v2 3/6] block: Remove bdrv_reset_dirty, Fam Zheng, 2015/05/06
- [Qemu-devel] [PATCH v2 4/6] qemu-iotests: Make block job methods common, Fam Zheng, 2015/05/06
- [Qemu-devel] [PATCH v2 5/6] qemu-iotests: Add test case for mirror with unmap, Fam Zheng, 2015/05/06
- [Qemu-devel] [PATCH v2 6/6] iotests: Use event_wait in wait_ready, Fam Zheng, 2015/05/06
- Re: [Qemu-devel] [Qemu-block] [PATCH v2 0/6] block: Mirror discarded sectors, Stefan Hajnoczi, 2015/05/07