ahrens commented on this pull request.
> @@ -226,17 +226,26 @@ dmu_zfetch(zfetch_t *zf, uint64_t blkid, uint64_t
> nblks, boolean_t fetch_data)
for (zs = list_head(&zf->zf_stream); zs != NULL;
zs = list_next(&zf->zf_stream, zs)) {
- if (blkid == zs->zs_blkid) {
+ if (blkid == zs->zs_blkid || blkid + 1 == zs->zs_blkid) {
Please add a comment explaining that this is handling misaligned streams, where
each access overlaps the previous access by exactly one block. You might say
something like, "If the first or second block of this access matches the
expected next block in this stream, count it as part of this stream. We count
the second block as a match to handle the case of misaligned accesses, where
each subsequent access is byte-wise adjacent, but block-wise overlaps by
exactly one block."
It might be easier to understand (and simpler code) if we made this check:
`blkid == zs_blkd || (nblks > 1 && blkid + 1 == zs_blkid)`
And then below. i.e. starting at line 235 it could be:
```
if (blkid == zs_blkd)
break;
if (nblks > 1 && blkid + 1 == zs_blkid) {
blkid++;
nblks--;
break;
}
mutex_exit(zs_lock);
```
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/openzfs/openzfs/pull/497#pullrequestreview-80264709
------------------------------------------
openzfs-developer
Archives:
https://openzfs.topicbox.com/groups/developer/discussions/T7851c2386c192095-Mfac30bf61e2b90767520167b
Powered by Topicbox: https://topicbox.com
