On Tue, Dec 03, 2019 at 09:45:58AM +0800, Yang Xu wrote:
>
>
> on 2019/12/03 7:21, Darrick J. Wong wrote:
> > On Thu, Nov 28, 2019 at 03:10:45PM +0800, Yang Xu wrote:
> > > When I backport fixed patches from Darrick tighten-verifiers tree, this case
> >
> > NAK, why are you backporting out of my development tree and not upstream?
> My description is wrong. I use upstream kernel 5.4-rc8 and 4 attr patches as
> your tighten-verifiers tree do.
> > What are you even backporting, seeing as the patches for this are in 5.5
> > and Linus hasn't even pulled that yet??
> > I see Linus pulled the patches 3 hour ago,
> > Also, why are you sending a patch to fstests that is (a) barely a week
> > old, (b) doesn't actually list me as the author, and (c) I myself
> > haven't even sent to fstests yet?
> >
> I have listed you as author, I think I should add "Darrick J. Wong
> pointed selinux and sort problem" info. I find ls and "a_are_bad/for_you"
> problem.
> Also, in personal mail("Re: question about xfstests test case
> xfs/148(new)"), I said "I think you may busy with other important things. I
> will send this simple patch with correct "a_are_bad/for_you" info and add
> your signed-off-by tag."). Anyway, I am sorry, I should have received your
> reply before sending the patch.
<nod> Apology accepted. I'll send this (and everything else queueing in
my fstest tree) tomorrow like usual.
> > > also fails. As below:
> > > -----------------------------------------------------------
> > > "diff test/xfs/148.out result/xfs/148.out.bad
> > > 7,8d6
> > > < Attribute "a_something" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > < Attribute "a_too_many_beans" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > 9a8
> > > > Attribute "a_too_many_beans" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > 10a10,11
> > > > Attribute "a_something" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > > Attribute "selinux" has a 37 byte value for TEST_DIR/mount-148/testfile
> > > 49,50c50,51
> > > < Attribute "a_are_bad/for_you" had a 3 byte value for TEST_DIR/mount-148/testfile:
> > > < heh
> > > > attr_get: No data available
> > > > Could not get "a_are_bad/for_you" for TEST_DIR/mount-148/testfile"
> > > -------------------------------------------------------------
> > > We should sort attribute list output and filter selinux. Also "a_are_bad/for_you"
> > > doesn't exist, we should correct it in output.
> > >
> > > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> ...
> > > Signed-off-by: Yang Xu <xuyang2018.jy@xxxxxxxxxxxxxx>
> > > ---
> > > tests/xfs/148 | 2 +-
> > > tests/xfs/148.out | 8 ++++----
> > > 2 files changed, 5 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/tests/xfs/148 b/tests/xfs/148
> > > index 42cfdab0..06862faa 100755
> > > --- a/tests/xfs/148
> > > +++ b/tests/xfs/148
> > > @@ -76,7 +76,7 @@ test_names+=("too_many" "are_bad/for_you")
> > > access_stuff() {
> > > ls $testdir
> > > - $ATTR_PROG -l $testfile
> > > + $ATTR_PROG -l $testfile | grep "a_" | sort
> > > for name in "${test_names[@]}"; do
> > > ls "$testdir/f_$name"
> > > diff --git a/tests/xfs/148.out b/tests/xfs/148.out
> > > index c301ecb6..ab3fc25b 100644
> > > --- a/tests/xfs/148.out
> > > +++ b/tests/xfs/148.out
> > > @@ -4,10 +4,10 @@ f_another
> > > f_are_bad_for_you
> > > f_something
> > > f_too_many_beans
> > > +Attribute "a_another" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > +Attribute "a_are_bad_for_you" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > Attribute "a_something" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > Attribute "a_too_many_beans" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > -Attribute "a_are_bad_for_you" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > -Attribute "a_another" has a 3 byte value for TEST_DIR/mount-148/testfile
> > > TEST_DIR/mount-148/testdir/f_something
> > > Attribute "a_something" had a 3 byte value for TEST_DIR/mount-148/testfile:
> > > heh
> > > @@ -46,5 +46,5 @@ ls: cannot access 'TEST_DIR/mount-148/testdir/f_too_many': No such file or direc
> > > attr_get: No data available
> > > Could not get "a_too_many" for TEST_DIR/mount-148/testfile
> > > ls: cannot access 'TEST_DIR/mount-148/testdir/f_are_bad/for_you': No such file or directory
> > > -Attribute "a_are_bad/for_you" had a 3 byte value for TEST_DIR/mount-148/testfile:
> > > -heh
> > > +attr_get: No data available
> > > +Could not get "a_are_bad/for_you" for TEST_DIR/mount-148/testfile
> >
> > Huh? Why???
> >
> > NAK NAK NAK
> a_are_bad_for_you has been corrupted by "/". So it should be nothing such as
It's been changed, yes, but '/' is a valid character in an attribute
name. Note that checksums are deliberately disabled on the loopdev
filesystem so that the only checks we perform are the name checks
themselves.
> a_too_many_beans by beans. am I wrong?( I use 5.4-rc8 and 4 patches
> xfs: check attribute leaf block structure
> xfs: namecheck attribute names before listing them
> xfs: namecheck directory entry names before listing them
> xfs: replace -EIO with -EFSCORRUPTED for corrupt metadata
> ,this four patches from https://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfs-linux.git/commit/?h=tighten-verifiers
This is my development git repo ^^^^^^^^^^ which sometimes contains
older versions of patches that do not end up going upstream.
And this is the official xfs repo:
https://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git/
Anyway, Linus pulled the 5.5 patches now, so you *really* ought to be
backporting from that.
--D
> )>
> >
> > --D
> >
> > > --
> > > 2.18.0
> > >
> > >
> > >
> >
> >
>
>
