On Mon, Dec 23, 2019 at 09:24:56AM +0800, Yang Xu wrote: > > > on 2019/12/23 5:27, Theodore Ts'o wrote: > > The generic/386 test was checking the "Available" field when it should > > have been checking the "1k-blocks" field, which represents the project > > quota's hard limit. On xfs, an empty directory takes no space, so it > > doesn't matter. But for ext4, an empty directory still takes 4k (or > > whatever the file system's block size happens to be): > > > > Filesystem 1K-blocks Used Available Use% Pathname > > /dev/vdc 512000 4 511996 0% /vdc/test > > > > This causes generic/386 to falsely fail. There was a confusing > > comment claiming that for a very long device name, the df output would > > have a line break, and for that reason, the test would extract the > > field using $(NF-2). However, looking at xfsprogs's quota command, I > > see no evidence that there is any line breaking logic. Since we now > > want to use the second field, even if there was some line breaking > > going on, using $2 should be a better choice. > > > > This fix is needed to fix generic/386 from failing on ext4: > Hi Theodore > > I have a same fix but I use $(NF-4) field of df command. Also I wrongly > think project quota uses a block size. > url: > https://patchwork.kernel.org/patch/11296657/ Yes, $(NF-4) works, as Eric suggested as well. I applied Ted's patch (with $2->$(NF-4) fix), as the commit log is explaining well what's the problem and why taking "1k blocks" is correct. And thank you for the fix all the same! Thanks, Eryu > > Best Regards > Yang Xu > > > > hard limit 524283904 bytes, expected 524288000 > > > > Signed-off-by: Theodore Ts'o <tytso@xxxxxxx> > > --- > > tests/generic/386 | 4 +--- > > 1 file changed, 1 insertion(+), 3 deletions(-) > > > > diff --git a/tests/generic/386 b/tests/generic/386 > > index 0c44c80e..37e9b943 100755 > > --- a/tests/generic/386 > > +++ b/tests/generic/386 > > @@ -62,8 +62,6 @@ _require_scratch > > # both the "df" and the "report" output. For "report", the line we're > > # interested in contains our project name in the first field. For "df" > > # it contains our project directory in the last field. > > -# But if the device name is too long, the "df" output is broke into two > > -# lines, the fourth field is not correct, so take $(NF-2) of "df" > > _filter_quota_rpt() { > > awk ' > > BEGIN { > > @@ -89,7 +87,7 @@ _filter_quota_rpt() { > > bsize = byte_size($4); > > } else if ($NF ~ proj_dir) { > > # this is the "df" output > > - bsize = byte_size($(NF-2)); > > + bsize = byte_size($2)); > > } else { > > next; > > } > > > >
