On Thu 29-03-18 08:41:23, Darrick J. Wong wrote:
> On Thu, Mar 29, 2018 at 03:31:23PM +0200, Jan Kara wrote:
> > On Fri 23-03-18 09:27:47, Eryu Guan wrote:
> > > On Tue, Feb 20, 2018 at 06:23:07PM +0100, Jan Kara wrote:
> > > > From: Jiri Slaby <jslaby@xxxxxxx>
> > > >
> > > > ext4_iomap_begin() has a bug where offset returned in the iomap
> > > > structure will be truncated to unsigned long size. On 64-bit
> > > > architectures this is fine but on 32-bit architectures obviously not.
> > > > Not many places actually use the offset stored in the iomap structure
> > > > but one of visible failures is in SEEK_HOLE / SEEK_DATA implementation.
> > > > If we create a file like:
> > > >
> > > > dd if=/dev/urandom of=file bs=1k seek=8m count=1
> > > >
> > > > then
> > > >
> > > > lseek64("file", 0x100000000ULL, SEEK_DATA)
> > > >
> > > > wrongly returns 0x100000000 on unfixed kernel while it should return
> > > > 0x200000000. Avoid the overflow by proper type cast.
> > >
> > > It looks like a good candidate for a regression test in fstests :)
> >
> > Actually, one of SEEK_HOLE/SEEK_DATA tests in fstests will fail because of
> > this bug (I've checked that). Just not many people run fstests in fully
> > 32-bit environments.
>
> Which fstest?
It was one of the tests using src/seek_sanity_test.c. I *think*
it was generic/285 and one of huge_file_tests() there (tests 10-12).
Honza
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR
