From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030314
Description of problem:
Starting in RHL 9, glibc starts printing a warning message:
"Incorrectly built binary which accesses errno, h_errno or _res directly. Needs
to be fixed."
on STDOUT for some programs that were built before RHL 9. One example of this
is Veritas NetBackup. NetBackup's processes are started by xinetd. xinetd of
course takes data to be sent over the wire to the client from the server
processes' STDOUT. So, when glibc adds this uninvited bit of text to the STDOUT
stream, all hell breaks loose if the process is not expecting this to happen,
and seriously, why would it?
Ok, ok, we get the picture. The program is written wrong. Chances are that for
some software it will be a while before it can be fixed, so please give me a way
to stop glibc from cramming uninvited data down my STDOUT. Most things only
expect program output from STDOUT, and errors go to STDERR. Isn't that what
STDERR is for?
Please either give me a way to stop this message from printing (what I would
prefer), perhaps an environment variable like
YES_I_KONW_MY_PROGRAM_IS_BROKEN_SHUT_UP_ALREADY, or send the message to STDERR
where I may have a better chance of it not interfering with my xinetd-based
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Run a program that was built before RHL 9 (i.e. Veritas NetBackup)
2. Observe that a warning is printed by glibc on STDOUT.
This has been changed to stderr in glibc-2.3.2-49 and later in rawhide.
where to find glibc-2.3.2-49 ?
Rawhide is the repository of the latest and greatest (but possible unstable)
packages that Red Hat produces. You can find rawhide in any Red Hat mirror like