Re: coreutils-7.1: make check fails under RHEL4

[Jim Meyering]

Christophe LYON wrote:

> Hi all,
>
> I have just tried to build coreutils-7.1 on a RHEL4 (32bits) machine
> (building with GCC-4.1.2).
>
> "make check" fails on mv/part-symlink.log
> The end of the log contains:
> ===============================================================
> + diff -u expected-7010 actual-7010
> --- expected-7010     2009-03-16 15:33:22.000000000 +0100
> +++ actual-7010       2009-03-16 15:33:22.000000000 +0100
> @@ -30,8 +30,8 @@
>  0 cp -bd rem_reg loc_sl (loc_sl loc_sl~ -> rem_reg) (rem_reg)
>  1 cp -d rem_reg loc_sl [cp: `rem_reg' and `loc_sl' are the same file
> ](loc_sl -> rem_reg) (rem_reg)
>
> -0 mv loc_reg rem_sl () (rem_sl)
> -0 mv -b loc_reg rem_sl () (rem_sl rem_sl~ -> dir/loc_reg)
> +0 mv loc_reg rem_sl [mv: setting attribute `security.selinux' for
> security.selinux': Operation not supported ]() (rem_sl)
> +0 mv -b loc_reg rem_sl [mv: setting attribute `security.selinux' for
> security.selinux': Operation not supported ]() (rem_sl rem_sl~ ->
> dir/loc_reg)
>
>  1 mv rem_sl loc_reg [mv: `rem_sl' and `loc_reg' are the same file
> ](loc_reg) (rem_sl -> dir/loc_reg)
>  0 mv -b rem_sl loc_reg (loc_reg -> dir/loc_reg loc_reg~) ()
> ===============================================================
>
> so it seems the problem has to do with selinux, but I am not familiar
> with this. How can I figure out for sure what is wrong?

Thanks for the report.
You may safely ignore that failure.
It's saying that when moving a local regular file to a remote (different
partition) symlink, with or without -b, you get an unexpected diagnostic
(about the selinux attribute-setting failure).  If you choose a partition
for which selinux support works, the test will pass.

You can do that by setting this envvar

   export CANDIDATE_TMP_DIRS=/some/other/partition/tmp

and rerunning the test.