Re: test-openat-safer failure on older kernel

[Eric Blake]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

According to Jim Meyering on 9/5/2009 9:29 AM:
>> an empty argument to openat, which is why my test-openat-safer was the
>> first thing to catch this in nearly 3 years of use.
>> +    openat: fail with ENOENT on empty name
>> +    * lib/openat-proc.c (openat_proc_name): Special-case the empty
>> +    buffer.
> 
> Good catch.  Thanks.

Ouch.  glibc has the same bug.  We may need to provide *at wrappers
unconditionally on Linux, if anyone is running against an older kernel but
newer glibc that uses the /proc/self/fd fallbacks without needing the
current state of the gnulib wrappers.  Are there still systems in the wild
like that?
http://sources.redhat.com/bugzilla/show_bug.cgi?id=10609

Oh, and POSIX has a bug - it says that readlinkat returns 0 instead of the
symlink size on success.  The Linux man pages copied the POSIX text, but
implemented things correctly (ie. whether readlink is a call to
readlinkat(AT_FDCWD), or readlinkat is a call to
readlink("/proc/self/fd/n/name"), the return value should be the same).
https://www.opengroup.org/sophocles/show_mail.tpl?CALLER=index.tpl&source=L&listname=austin-group-l&id=12734

- --
Don't work too hard, make some time for fun as well!

Eric Blake             address@hidden
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Public key at home.comcast.net/~ericblake/eblake.gpg
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkqikbYACgkQ84KuGfSFAYCcoACfSkESp2E8HdhpVIxp+RtMcGvI
1LgAn2HLCdlVxm6AWVZ8HxiMfYhkKViN
=HCjL
-----END PGP SIGNATURE-----