pcp
[Top] [All Lists]

Re: RHEL6 build failure

To: Nathan Scott <nathans@xxxxxxxxxx>
Subject: Re: RHEL6 build failure
From: Max Matveev <makc@xxxxxxxxx>
Date: Mon, 23 May 2011 16:55:06 +1000
Cc: pcp@xxxxxxxxxxx
In-reply-to: <1041909808.67241.1305961343947.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx>
References: <1371871737.67239.1305961318931.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx> <1041909808.67241.1305961343947.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx>
On Sat, 21 May 2011 17:02:23 +1000 (EST), Nathan Scott wrote:

 nathans> Guess you're getting pretty sick of these by now :) but here's an
 nathans> exciting new build failure in rpm/perl packaging land - this time
 nathans> on a fresh RHEL6 install...

How do you enable debuginfo builds?

 nathans> extracting debug info from 
/source/git/pcp/BUILDROOT/pcp-3.5.1-1.x86_64/usr/lib64/perl5/vendor_perl/auto/PCP/PMDA/PMDA.so
 nathans> /usr/lib/rpm/debugedit: -b arg has to be either the same length as -d 
arg, or more than 1 char shorter
 nathans> error: Bad exit status from /var/tmp/rpm-tmp.jCirw0 (%install)

 nathans> Any ideas?
Yes - make the path to the top of the tree longer,
e.g. /source/git/pcp-3.5.1 will be fine.

The reason for the build failure is the bogus check in debugedit -
it's trying to re-write absolute paths in the debug info section of
ELF files and assumes that it does not have enough room to fit new
prefix.

This only affects Perl files so far because it's the only place we're
using full paths in the include search path. We can change them to be
relative - I don't see a reason for using pwd in setting PCP_TOPDIR
but maybe you can remember what was the problem with a8bf0879.

max

<Prev in Thread] Current Thread [Next in Thread>