On Mon, Jun 08, 2020 at 03:24:10PM +0800, Qu Wenruo wrote:
> On 2020/6/8 下午3:20, Michał Mirosław wrote:
> > On Sun, Jun 07, 2020 at 03:25:12PM +0800, Qu Wenruo wrote:
> >> Signed-off-by: Qu Wenruo <wqu@xxxxxxxx>
> >> ---
> >> fs/btrfs/disk-io.c | 6 ++++++
> >> fs/btrfs/qgroup.c | 43 +++++++++++++++++++++++++++++++++++++++++++
> >> fs/btrfs/qgroup.h | 2 +-
> >> 3 files changed, 50 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
> >> index f8ec2d8606fd..48d047e64461 100644
> >> --- a/fs/btrfs/disk-io.c
> >> +++ b/fs/btrfs/disk-io.c
> >> @@ -4058,6 +4058,12 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
> >> ASSERT(list_empty(&fs_info->delayed_iputs));
> >> set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
> >>
> >> + if (btrfs_qgroup_has_leak(fs_info)) {
> >> + WARN(IS_ENABLED(CONFIG_BTRFS_DEBUG),
> >> + KERN_ERR "BTRFS: qgroup reserved space leaked\n");
> >> + btrfs_err(fs_info, "qgroup reserved space leaked\n");
> >> + }
> >
> > This looks like debugging aid, so:
> >
> > if (IS_ENABLED(CONFIG_BTRFS_DEBUG))
> > btrfs_check_qgroup_leak(fs_info);
> >
> > would be more readable (WARN() pushed to the function).
>
> We want to check to be executed even on production system, but just less
> noisy (no kernel backtrace dump).
> Just like tree-checker and EXTENT_QUOTA_RESERVED check.
In that case I suggest:
btrfs_err(...);
WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
as I expect people look for messages before the Oops for more information.
Best Regards,
Michał Mirosław
