Re: [PATCH v5 09/28] btrfs: align device extent allocation to zone boundary

[Naohiro Aota <naohiro.aota@xxxxxxx>]

On Thu, Dec 05, 2019 at 09:56:25AM +0100, Johannes Thumshirn wrote:
> On Wed, Dec 04, 2019 at 05:17:16PM +0900, Naohiro Aota wrote:
> [...]
>
> Only commenting on the code, not the design, sorry. I'll leave that to someone
> with more experience in BTRFS.
>
> >  	 * at an offset of at least 1MB.
> >  	 */
> >  	search_start = max_t(u64, search_start, SZ_1M);
> > +	/*
> > +	 * For a zoned block device, skip the first zone of the device
> > +	 * entirely.
> > +	 */
> > +	if (device->zone_info)
> > +		zone_size = device->zone_info->zone_size;
> > +	search_start = max_t(u64, search_start, zone_size);
> > +	search_start = btrfs_zone_align(device, search_start);
>
> 	if (device->zone_info) {
> 		zone_size = device->zone_info->zone_size;
> 		search_start = max_t(u64, search_start, zone_size);
> 		search_start = btrfs_zone_align(device, search_start);
> 	}
>
> That's the equivalent code, but should make it a bit more clear what's
> happening int the HMZONED and !HMZOED cases.
>
> And I /guess/ we're saving some cycles in the !HMZONED case as we don't have
> to adjust search start there.
>
> [...]
>
> > @@ -4778,6 +4805,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
> >  	int i;
> >  	int j;
> >  	int index;
> > +	int hmzoned = btrfs_fs_incompat(info, HMZONED);
>
> 	bool hmzoned = btrfs_fs_incompat(info, HMZONED);

Thanks, followed all the change.