Bug 632 - gcc 4.2.4 (ubuntu) goes into busyloop on building elfutils_0.148
Summary: gcc 4.2.4 (ubuntu) goes into busyloop on building elfutils_0.148
Status: RESOLVED OBSOLETE
Alias: None
Product: OE-Core
Classification: Build System, Metadata & Runtime
Component: devtools / tool chain (show other bugs)
Version: 0.9
Hardware: x86 Multiple
: Medium normal
Target Milestone: 0.9.2
Assignee: Nitin Kamble
QA Contact:
URL:
Whiteboard:
Depends on:
Blocks:
 
Reported: 2011-01-05 07:49 UTC by Aleksandr Koltsoff
Modified: 2011-02-14 20:44 UTC (History)
4 users (show)

See Also:
OS type for building Yocto: ---
Type of Regression: ---
Verified:
Documentation change: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Aleksandr Koltsoff 2011-01-05 07:49:07 UTC
Using hardy (8.04) and building minimal image for qemuarm leads cc into busyloop. Quite likely a bug in gcc-4.2.4 (Ubuntu 4.2.4-1ubuntu4).

Three hosts were tested:
 * x86_64 running 8.04 (mostly clean install)
 * i686 running 8.04 (vm, clean pristine install with all updates)
 * i686 running 10.04 (vm, clean pristine install with all updates)

Both 8.04 end up in cc1 using 100% CPU without making any system calls (busyloop) indefinitely. The directory which gcc is building is libdwfl. 10.04/Lucid works without problems (newer gcc).

Also, on x86_64 tested the same target using master, with same results, so issue is not 0.9/4.0-specific.

Steps to recreate:
 1. Get a clean 8.04 and do all updates
 2. Using the yocto project quick start document, install the recommended packages.
 3. Get and extract laverne-4.0 as per instructions.
 4. Install python2.6 as per https://lists.yoctoproject.org/pipermail/yocto/2010-December/000447.html
 5. source poky-laverne-4.0/poky-init-build-env poky-4.0-build (as per quick start)
 6. The following modifications were done to local.conf:
DL_DIR ?= /auto/oe-sources
MACHINE ?= "qemuarm"
SDKMACHINE ?= "i586" (for non-x86-64 targets)
 6. bitbake elfutils-native-0.148-r1
 7. Fix the 2 remaining issues that bitbake env check complains about (vm.mmap start, default shell)
 8. elfutils-native-0.148-r1
 9. Get a lot of coffee and come back to observe cc1 stuck at 100%

Other things ruled out so far (changing these had no effect):
 * parallel builds / parallel threads
 * ccache
 * psyco
 * using master vs 0.9 tarball
Comment 1 Aleksandr Koltsoff 2011-01-05 07:51:05 UTC
Step 8 should read:
 8. bitbake elfutils-native-0.148-r1
Comment 2 Nitin Kamble 2011-01-19 17:30:35 UTC
is there a hardy (8.04) system for me to use. All the systems I work on are Fedora.
Comment 3 Nitin Kamble 2011-01-19 17:31:25 UTC
is there a hardy (8.04) system for me to use. All the systems I work on are Fedora.
Comment 4 Nitin Kamble 2011-02-01 17:30:06 UTC
I see the latest hardy is 8.04.4 LTS, does it happen on it ? I am about to install ubuntu on a system, just checking which is right ubuntu to install.

Thanks,
Nitin
Comment 5 Nitin Kamble 2011-02-08 14:26:10 UTC
Can not reproduce :( met python error with current poky master branch.

$ bitbake elfutils
Pseudo is not present but is required, building this first before the main build
Sorry, python 2.6.0 or later is required for this version of bitbake
Comment 6 Nitin Kamble 2011-02-08 15:10:50 UTC
updated hardy gave me this:
# python --version
Python 2.5.2
Comment 7 Nitin Kamble 2011-02-14 16:19:25 UTC
So from python's point of view, the hardy distro is too old to run our bitbake. So is it worth going after the older distro for this bug ?
Comment 8 Aleksandr Koltsoff 2011-02-14 20:44:06 UTC
It's possible to get a working python also on Hardy. However, since Hardy on Desktop is coming to its end-of-life soon (end of security updates for desktop), maybe you're right to move this into obsolete (I'm ok with it).