On 08/01/2020, 13:16, "Qu Wenruo" <quwenruo.btrfs@xxxxxxx> wrote:
[...]
The patch itself is OK.
Reviewed-by: Qu Wenruo <wqu@xxxxxxxx>
This means the qgroup get disabled when rescan is still running.
So I'm a little curious, could we just cancel the running rescan and
wait for it before disabling qgroup?
Maybe. I'm still not 100% certain what's the actual trigger. I see it most of the time
with btrfs/117, but running btrfs/117 alone doesn't trigger the memleak report.
I also saw the memleak report once with btrfs/116, but only once out of ~20 runs.
The good thing though is, it's 100% reproducible that we get the memleak report when
running the first 120 btrfs fstests .
P.S.: Sorry for the broken quoting in the reply, I still haven't configured my mail client correctly.
Byte,
Johannes
