| Message ID | 20170912203119.24166-12-eblake@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | make dirty-bitmap byte-based | expand |
On 09/12/2017 04:31 PM, Eric Blake wrote: > Thanks to recent cleanups, most callers were scaling a return value > of sectors into bytes (the exception, in qcow2-bitmap, will be > converted to byte-based iteration later). Update the interface to > do the scaling internally instead. > > In qcow2-bitmap, the code was specifically checking for an error > to be -1; it is more robust to treat all negative values as an > error, but at the same time it is also easy enough to ensure we > return -1 (and not -512) on error. > > Signed-off-by: Eric Blake <eblake@redhat.com> > This patch now smells like a bugfix and a separate incremental feature enhancement. Do we need to backport the error-checking to a possible 2.10.1? If no: Reviewed-by: John Snow <jsnow@redhat.com>
On 09/13/2017 07:15 PM, John Snow wrote: > > > On 09/12/2017 04:31 PM, Eric Blake wrote: >> Thanks to recent cleanups, most callers were scaling a return value >> of sectors into bytes (the exception, in qcow2-bitmap, will be >> converted to byte-based iteration later). Update the interface to >> do the scaling internally instead. >> >> In qcow2-bitmap, the code was specifically checking for an error >> to be -1; it is more robust to treat all negative values as an >> error, but at the same time it is also easy enough to ensure we >> return -1 (and not -512) on error. >> >> Signed-off-by: Eric Blake <eblake@redhat.com> >> > > This patch now smells like a bugfix and a separate incremental feature > enhancement. There is no bug without this patch; more of a fix to avoid a latent regression from happening in further changes. In v6, I (accidentally) had bdrv_dirty_iter_next() temporarily returning -512 instead of -1 on failure; changing the qcow2 code to treat all negatives instead of precisely -1 as error is enough to avoid that regression, but so also is fixing bdrv_dirty_iter_next() to always return -1 on failure. This patch does both, rather than either fix in isolation, but that means we don't need a backport. > > Do we need to backport the error-checking to a possible 2.10.1? > > If no: > > Reviewed-by: John Snow <jsnow@redhat.com> >
On 09/14/2017 07:58 AM, Eric Blake wrote: > On 09/13/2017 07:15 PM, John Snow wrote: >> >> >> On 09/12/2017 04:31 PM, Eric Blake wrote: >>> Thanks to recent cleanups, most callers were scaling a return value >>> of sectors into bytes (the exception, in qcow2-bitmap, will be >>> converted to byte-based iteration later). Update the interface to >>> do the scaling internally instead. >>> >>> In qcow2-bitmap, the code was specifically checking for an error >>> to be -1; it is more robust to treat all negative values as an >>> error, but at the same time it is also easy enough to ensure we >>> return -1 (and not -512) on error. >>> >>> Signed-off-by: Eric Blake <eblake@redhat.com> >>> >> >> This patch now smells like a bugfix and a separate incremental feature >> enhancement. > > There is no bug without this patch; more of a fix to avoid a latent Ah, you're right, I see. Please take the RB. > regression from happening in further changes. In v6, I (accidentally) > had bdrv_dirty_iter_next() temporarily returning -512 instead of -1 on > failure; changing the qcow2 code to treat all negatives instead of > precisely -1 as error is enough to avoid that regression, but so also is > fixing bdrv_dirty_iter_next() to always return -1 on failure. This > patch does both, rather than either fix in isolation, but that means we > don't need a backport. > >> >> Do we need to backport the error-checking to a possible 2.10.1? >> >> If no: >> >> Reviewed-by: John Snow <jsnow@redhat.com> >> >
diff --git a/block/backup.c b/block/backup.c index ac9c018717..06ddbfd03d 100644 --- a/block/backup.c +++ b/block/backup.c @@ -375,7 +375,7 @@ static int coroutine_fn backup_run_incremental(BackupBlockJob *job) dbi = bdrv_dirty_iter_new(job->sync_bitmap); /* Find the next dirty sector(s) */ - while ((offset = bdrv_dirty_iter_next(dbi) * BDRV_SECTOR_SIZE) >= 0) { + while ((offset = bdrv_dirty_iter_next(dbi)) >= 0) { cluster = offset / job->cluster_size; /* Fake progress updates for any clusters we skipped */ diff --git a/block/dirty-bitmap.c b/block/dirty-bitmap.c index d50c46621d..49229fd501 100644 --- a/block/dirty-bitmap.c +++ b/block/dirty-bitmap.c @@ -508,7 +508,8 @@ void bdrv_dirty_iter_free(BdrvDirtyBitmapIter *iter) int64_t bdrv_dirty_iter_next(BdrvDirtyBitmapIter *iter) { - return hbitmap_iter_next(&iter->hbi); + int64_t ret = hbitmap_iter_next(&iter->hbi); + return ret < 0 ? -1 : ret * BDRV_SECTOR_SIZE; } /* Called within bdrv_dirty_bitmap_lock..unlock */ diff --git a/block/mirror.c b/block/mirror.c index 0b063b3c20..77bf5aa3a4 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -336,10 +336,10 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s) int max_io_bytes = MAX(s->buf_size / MAX_IN_FLIGHT, MAX_IO_BYTES); bdrv_dirty_bitmap_lock(s->dirty_bitmap); - offset = bdrv_dirty_iter_next(s->dbi) * BDRV_SECTOR_SIZE; + offset = bdrv_dirty_iter_next(s->dbi); if (offset < 0) { bdrv_set_dirty_iter(s->dbi, 0); - offset = bdrv_dirty_iter_next(s->dbi) * BDRV_SECTOR_SIZE; + offset = bdrv_dirty_iter_next(s->dbi); trace_mirror_restart_iter(s, bdrv_get_dirty_count(s->dirty_bitmap) * BDRV_SECTOR_SIZE); assert(offset >= 0); @@ -370,11 +370,11 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s) break; } - next_dirty = bdrv_dirty_iter_next(s->dbi) * BDRV_SECTOR_SIZE; + next_dirty = bdrv_dirty_iter_next(s->dbi); if (next_dirty > next_offset || next_dirty < 0) { /* The bitmap iterator's cache is stale, refresh it */ bdrv_set_dirty_iter(s->dbi, next_offset); - next_dirty = bdrv_dirty_iter_next(s->dbi) * BDRV_SECTOR_SIZE; + next_dirty = bdrv_dirty_iter_next(s->dbi); } assert(next_dirty == next_offset); nb_chunks++; diff --git a/block/qcow2-bitmap.c b/block/qcow2-bitmap.c index 44329fc74f..b09010b1d3 100644 --- a/block/qcow2-bitmap.c +++ b/block/qcow2-bitmap.c @@ -1109,7 +1109,7 @@ static uint64_t *store_bitmap_data(BlockDriverState *bs, sbc = limit >> BDRV_SECTOR_BITS; assert(DIV_ROUND_UP(bm_size, limit) == tb_size); - while ((sector = bdrv_dirty_iter_next(dbi)) != -1) { + while ((sector = bdrv_dirty_iter_next(dbi) >> BDRV_SECTOR_BITS) >= 0) { uint64_t cluster = sector / sbc; uint64_t end, write_size; int64_t off;
Thanks to recent cleanups, most callers were scaling a return value of sectors into bytes (the exception, in qcow2-bitmap, will be converted to byte-based iteration later). Update the interface to do the scaling internally instead. In qcow2-bitmap, the code was specifically checking for an error to be -1; it is more robust to treat all negative values as an error, but at the same time it is also easy enough to ensure we return -1 (and not -512) on error. Signed-off-by: Eric Blake <eblake@redhat.com> --- v7: return -1, not -512; and fix qcow2-bitmap to check all negatives [Kevin] v5-v6: no change v4: rebase to persistent bitmap v3: no change v2: no change --- block/backup.c | 2 +- block/dirty-bitmap.c | 3 ++- block/mirror.c | 8 ++++---- block/qcow2-bitmap.c | 2 +- 4 files changed, 8 insertions(+), 7 deletions(-)