On 14.10.2020 15:01, Max Reitz wrote:
On 12.10.20 19:43, Andrey Shinkevich wrote:Limit COR operations by the base node in the backing chain when the overlay base node name is given. It will be useful for a block stream job when the COR-filter is applied. The overlay base node is passed as the base itself may change due to concurrent commit jobs on the same backing chain. Signed-off-by: Andrey Shinkevich <andrey.shinkevich@xxxxxxxxxxxxx> --- block/copy-on-read.c | 39 +++++++++++++++++++++++++++++++++++++-- 1 file changed, 37 insertions(+), 2 deletions(-) diff --git a/block/copy-on-read.c b/block/copy-on-read.c index c578b1b..dfbd6ad 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -122,8 +122,43 @@ static int coroutine_fn cor_co_preadv_part(BlockDriverState *bs, size_t qiov_offset, int flags) {
[...]
+ ret = bdrv_is_allocated_above(bdrv_cow_bs(bs->file->bs), + state->base_overlay, true, offset, + n, &n); + if (ret) { + local_flags |= BDRV_REQ_COPY_ON_READ; + } + }Furthermore, I just noticed – can the is_allocated functions not return 0 in @n, when @offset is a the EOF? Is that something to look out for? (I’m not sure.) Max
The check for EOF is managed earlier in the stream_run() for a block-stream job. For other cases of using the COR-filter, the check for EOF can be added to the cor_co_preadv_part(). I would be more than happy if we can escape the duplicated checking for is_allocated in the block-stream. But how the stream_run() can stop calling the blk_co_preadv() when EOF is reached if is_allocated removed from it? May the cor_co_preadv_part() return EOF (or other error code) to be handled by a caller if (ret == 0 && n == 0 && (flags & BDRV_REQ_PREFETCH)?
Andrey
