Note: This is a beta release of Red Hat Bugzilla 5.0. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Also email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback here.
Bug 170986 - ppc64 debuginfo rpms not generated
Summary: ppc64 debuginfo rpms not generated
Alias: None
Product: Fedora
Classification: Fedora
Component: distribution
Version: 5
Hardware: powerpc
OS: Linux
Target Milestone: ---
Assignee: David Cantrell
QA Contact: Bill Nottingham
Depends On:
Blocks: FC5Target
TreeView+ depends on / blocked
Reported: 2005-10-17 02:24 UTC by Richard Henderson
Modified: 2014-03-17 02:56 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 2006-08-08 13:54:36 UTC

Attachments (Terms of Use)

Description Richard Henderson 2005-10-17 02:24:29 UTC
Assigning this to glibc is wrong, since *none* of the ppc64 packages
have debuginfo rpms built.  But since Jakub had a hand in creating them
in the first place, perhaps he knows the right categorization.

Anyway, the net effect is that valid backtraces through glibc are hard
to come by, unless you happen to be on one of the code paths that have
eh unwind info associated.  Which isn't 100% coverage by far.

Comment 1 Jakub Jelinek 2005-10-17 07:24:59 UTC
ls -l dist/fc4/{glibc,gcc,binutils,coreutils,bash}/*/ppc64/*debuginfo*
-rw-r--r--    1 root     root      1323632 May 20 13:13
-rw-r--r--    1 root     root      4761790 May 25 15:11
-rw-r--r--    1 root     root      2645897 May 25 15:12
dist/fc4/coreutils/5.2.1-48/ppc64/coreutils-debuginfo-5.2.1-48.ppc64.rpm-rw-r--r--    1 root     root     57529743 May 20 13:19
-rw-r--r--    1 root     root     20753528 May 30 17:17

I certainly see them in FC4 (on porkchop and likewise in FC5).
And in rawhide too:

So, are you not seeing them when you build new ppc64.rpm's, or are you just
missing them on ftp in fedora/linux/core/4/ppc/debug/
(where they clearly are missing, guess the compose scripts would need tweaking)?

Comment 2 Richard Henderson 2005-10-17 17:01:57 UTC
I meant the later, ftp in fedora/linux/core/4/ppc/debug/.  I didn't think to
look on our internal servers; I hadn't considered that we might be building
these and forgetting to copy them over.

Comment 3 Jakub Jelinek 2005-10-17 17:05:08 UTC
We aren't copying i386 debuginfo rpms either in the x86_64 composes, etc.
But the difference that matters here is that unlike x86_64 where there is
x86_64 distro and i386 distro, on ppc there is no ppc64 distro.

Anyway, reassigning to distribution, this is a releng issue.

Comment 4 Roland McGrath 2005-11-22 00:34:33 UTC
Is this going to be fixed in FC5?
Should it be on a blocker list?

Comment 5 Bill Nottingham 2005-11-23 17:29:58 UTC
 I can put it on the target list. Note that this is a mess, as the ppc/pp64 or
x86/x86_64 debuginfo RPMS will conflict with each other.

Comment 6 Jesse Keating 2006-08-07 18:34:42 UTC
Eeew, yes this is a major mess, especially since yum if asked to install a
package, will attempt to install all the arches available for said package.

yum install foo-debug 

will try to install


This is particularly bad on say pure ppc systems.  Even non-pure ppc systems, if
you don't already have the 64bit version of an application, when you ask for a
debug package, by nature of bringing in both arches, it'll try to depsolve the
64bit package and pull even more 64bit stuff in.

I'm not sure there is a clean way to fix this, given the broken nature of how we
publish ppc(64). :/

Comment 7 Jesse Keating 2006-08-07 18:36:30 UTC
Hrm, I just looked at FC6 and we're doing exactly what I was afraid of, both ppc
and ppc64 debug info packages are in the ppc/debug/ dir.  We'll see if this
causes problems.

I'm not sure if we'd want to do anything in FC5 land for this.

Comment 8 Jesse Keating 2006-08-08 13:54:36 UTC
I'm not comfortable with changing anything for Fc5.  This is resolved in FC6.

Note You need to log in before you can comment on or make changes to this bug.