Re: [PATCH] [PATCH] xfs/348: add _fixed_by tag

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]



On Thu, Aug 08, 2024 at 08:39:10AM +0000, Xinjian Ma (Fujitsu) wrote:
> > On Tue, Aug 06, 2024 at 09:19:03PM +0800, Zorro Lang wrote:
> > > On Tue, Jul 30, 2024 at 07:47:51AM -0700, Darrick J. Wong wrote:
> > > > On Tue, Jul 30, 2024 at 03:56:53PM +0800, Ma Xinjian wrote:
> > > > > This test requires a kernel patch since 3bf963a6c6 ("xfs/348:
> > > > > partially revert dbcc549317"), so note that in the test.
> > > > >
> > > > > Signed-off-by: Ma Xinjian <maxj.fnst@xxxxxxxxxxx>
> > > > > ---
> > > > >  tests/xfs/348 | 3 +++
> > > > >  1 file changed, 3 insertions(+)
> > > > >
> > > > > diff --git a/tests/xfs/348 b/tests/xfs/348 index
> > > > > 3502605c..e4bc1328 100755
> > > > > --- a/tests/xfs/348
> > > > > +++ b/tests/xfs/348
> > > > > @@ -12,6 +12,9 @@
> > > > >  . ./common/preamble
> > > > >  _begin_fstest auto quick fuzzers repair
> > > > >
> > > > > +_fixed_by_git_commit kernel 38de567906d95 \
> > > > > +	"xfs: allow symlinks with short remote targets"
> > > >
> > > > Considering that 38de567906d95 is itself a fix for 1eb70f54c445f, do
> > > > we want a _broken_by_git_commit to warn people not to apply 1eb70
> > > > without also applying 38de5?
> > >
> > > We already have _wants_xxxx_commit and _fixed_by_xxxx_commit, for now,
> > > I don't think we need a new one. Maybe:
> > >
> > >   _fixed_by_kernel_commit 38de567906d95 ..............
> > >   _wants_kernel_commit 1eb70f54c445f .............
> > >
> > > make sense? And use some comments to explain why 1eb70 is wanted?
> > 
> > Oh!  I didn't realize we had _wants_kernel_commit.  Yeah, that's fine.
> 
> 
> Hi Darrick
> 
> Sorry, I still don't quite understand your intention.
> Considering that 38de567906d95 is a fix for 1eb70f54c445f, I think the current xfs/348 test should have the following 3 situations:
> 1. Neither 1eb70f54c445f nor 38de567906d95 are applied in the kernel: xfs/348 passes
> 2. Only 1eb70f54c445f is applied in the kernel without 38de567906d95: xfs/348 fails
> 3. Both 1eb70f54c445f and 1eb70f54c445f are applied in the kernel: xfs/348 passes
> The situation of " Only 38de567906d95 is applied in the kernel without 1eb70f54c445f " should not exist.
> 
> Since only the second case fails, I think it's sufficient to just point out that 38de567906d95 might be missing using "_fixed_by_kernel_commit ".
> If my understanding is wrong, feel free to correct me, thank you very much.

1eb70f54c445f was a bugfix for a null pointer dereference due to
insufficient validation, so we really /do/ want to nudge kernel
distributors to add it (and 38de567906d95) to their kernels if they
don't have either.

But I see your point that xfs/348 will pass without either applied, so
that's not much of a nudge.  In the end, I'd rather this went in with
annotations for both commits, but if Zorro decides that only noting
38de567906d95 is ok, then I'll go along with that too.

--D

> Best regards
> Ma
> > 
> > --D
> > 
> > > Thanks,
> > > Zorro
> > >
> > > >
> > > > --D
> > > >
> > > > > +
> > > > >  # Import common functions.
> > > > >  . ./common/filter
> > > > >  . ./common/repair
> > > > > --
> > > > > 2.42.0
> > > > >
> > > > >
> > > >
> > >
> > >




[Index of Archives]     [Linux Filesystems Development]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux