On 07/27/2017 09:11 PM, Duncan wrote:
> Hans van Kranenburg posted on Wed, 26 Jul 2017 21:59:25 +0200 as
> excerpted:
>
>> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
>> index 4f8f75d9e839..b091dd3f5b38 100644
>> --- a/fs/btrfs/ctree.h
>> +++ b/fs/btrfs/ctree.h
>> @@ -472,7 +472,7 @@ struct btrfs_block_rsv {
>> /*
>> * free clusters are used to claim free space in relatively large chunks,
>> * allowing us to do less seeky writes. They are used for all metadata
>> - * allocations and data allocations in ssd mode.
>> + * allocations and data allocations in ssd_spread mode.
>> */
>> struct btrfs_free_cluster {
>> spinlock_t lock;
>
> That (post-patch) comment says all metadata and data allocations in
> ssd_spread mode, but if my understanding is correct, it'll also use
> free clusters for metadata (only, not data, the patch only changing)
> the data behavior) in normal ssd mode.
>
> Perhaps that (metadata only use of free clusers in ssd mode) should be
> mentioned as well, since post-patch there's now a distinction to be made?
The text is confusing yes.
It can be read as...
(all metadata allocations and data allocations) in ssd_spread mode
or...
(all metadata allocations) and (data allocations in ssd_spread mode)
...the second one being correct.
Good catch.
I changed it in: "They are used for all metadata allocations. In
ssd_spread mode they are also used for data allocations."
--
Hans van Kranenburg
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
