On Friday 09 December 2022 23:46:11 Paul Eggert wrote:
> On 12/6/22 23:57, Luke Dashjr wrote:
> > getdents64(3, [{d_ino=1179698, d_off=3802031580503460560, d_reclen=24,
> > d_type=DT_DIR, d_name="."}, {d_ino=1179668, d_off=4580148598467312926,
> > d_reclen=32, d_type=DT_REG, d_name="banlist.json"}, ...
>
> This hints at the problem, as it says that "." and its subdirectory
> "banlist.json" have the same inode number. Still, this glitch in your
> filesystem shouldn't cause 'du' to abort (it can happen with mounts).
>
> Perhaps enabling FTS debugging will help. I committed a change to try to
> get that working again. Please build by bootstrapping from the latest
> commit on Savannah, or from the following tarball (which should be
> equivalent):
>
> https://www.cs.ucla.edu/~eggert/coreutils-9.1.104-66ceda.tar.xz
>
> Build with -DGNULIB_FTS_DEBUG in your CFLAGS, and then run with
> fts_debug set to true (either via a debugger, or simply by editing the
> source code before building so that it is initially true). What is the
> output of your problematic 'du' in that case?Unfortunately, my system crashed this morning, and so I can no longer reproduce the issue (short of trying to replay the strace I sent). :/
