There is a bug in handling BDRV_REQ_NO_WAIT flag: we still may wait in
wait_serialising_requests() if request is unaligned. And this is
possible for the only user of this flag (preallocate filter) if
underlying file is unaligned to its request_alignment on start.
So, we have to fix preallocate filter to do only aligned preallocate
requests.
Next, we should fix generic block/io.c somehow. Keeping in mind that
preallocate is the only user of BDRV_REQ_NO_WAIT and that we have to
fix its behavior now, it seems more safe to just assert that we never
use BDRV_REQ_NO_WAIT with unaligned requests and add corresponding
comment. Let's do so.
Signed-off-by: Vladimir Sementsov-Ogievskiy <[email protected]>
Reviewed-by: Denis V. Lunev <[email protected]>
Message-Id: <
20220215121609[email protected]>
[hreitz: Rebased on block GS/IO split]
Signed-off-by: Hanna Reitz <[email protected]>
padding = bdrv_init_padding(bs, offset, bytes, &pad);
if (padding) {
+ assert(!(flags & BDRV_REQ_NO_WAIT));
bdrv_make_request_serialising(req, align);
bdrv_padding_rmw_read(child, req, &pad, true);
* serialize the request to prevent interactions of the
* widened region with other transactions.
*/
+ assert(!(flags & BDRV_REQ_NO_WAIT));
bdrv_make_request_serialising(&req, align);
bdrv_padding_rmw_read(child, &req, &pad, false);
}
/* TODO We can support BDRV_REQ_NO_FALLBACK here */
assert(!(read_flags & BDRV_REQ_NO_FALLBACK));
assert(!(write_flags & BDRV_REQ_NO_FALLBACK));
+ assert(!(read_flags & BDRV_REQ_NO_WAIT));
+ assert(!(write_flags & BDRV_REQ_NO_WAIT));
if (!dst || !dst->bs || !bdrv_is_inserted(dst->bs)) {
return -ENOMEDIUM;
int64_t end = offset + bytes;
int64_t prealloc_start, prealloc_end;
int ret;
+ uint32_t file_align = bs->file->bs->bl.request_alignment;
+ uint32_t prealloc_align = MAX(s->opts.prealloc_align, file_align);
+
+ assert(QEMU_IS_ALIGNED(prealloc_align, file_align));
if (!has_prealloc_perms(bs)) {
/* We don't have state neither should try to recover it */
/* Now we want new preallocation, as request writes beyond s->file_end. */
- prealloc_start = want_merge_zero ? MIN(offset, s->file_end) : s->file_end;
- prealloc_end = QEMU_ALIGN_UP(end + s->opts.prealloc_size,
- s->opts.prealloc_align);
+ prealloc_start = QEMU_ALIGN_UP(
+ want_merge_zero ? MIN(offset, s->file_end) : s->file_end,
+ file_align);
+ prealloc_end = QEMU_ALIGN_UP(
+ MAX(prealloc_start, end) + s->opts.prealloc_size,
+ prealloc_align);
+
+ want_merge_zero = want_merge_zero && (prealloc_start <= offset);
ret = bdrv_co_pwrite_zeroes(
bs->file, prealloc_start, prealloc_end - prealloc_start,
/*
* If we need to wait for other requests, just fail immediately. Used
- * only together with BDRV_REQ_SERIALISING.
+ * only together with BDRV_REQ_SERIALISING. Used only with requests aligned
+ * to request_alignment (corresponding assertions are in block/io.c).
*/
BDRV_REQ_NO_WAIT = 0x400,