Btrfs Async Discard Support Looks To Be Ready For Linux 5.6

[phoronix]

Phoronix: Btrfs Async Discard Support Looks To Be Ready For Linux 5.6

After months of work by Facebook engineers, it looks like the new async discard support for Btrfs is ready for the upcoming Linux 5.6 cycle as a win for this Linux file-system on solid-state storage making use of TRIM/DISCARD functionality...

http://www.phoronix.com/scan.php?page=news_item&px=Btrfs-Async-Discard

[J.G.]

Do other filesystems support this async discard?

[intelfx]

Do other filesystems support this async discard?

This is not something to be "supported". It is not a protocol or something; "async TRIM" as described here is purely an internal Btrfs-specific algorithm improvement.
Having said all that, not sure if other filesystems do anything similar.

[S.Pam]

It's important that the device does not reorder the discard, as it can lead to data loss. i.e. if you "delete 1, discard 1, write 1" but the device does "delete 1, write 1, discard 1". So discards can't technically be out of order once sent to the device queue.

[Zan Lynx]

When I've read about async TRIM, potential drive command reordering doesn't have any effect on it. Nothing is ever written to the areas which are queued up for TRIM. Those areas are treated as allocated until the TRIM completes.