Summary: | sys-apps/groff-1.20.1-r1 fails to build with parallel make | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Nick Soveiko <gentoo-bugzilla> |
Component: | Current packages | Assignee: | Gentoo's Team for Core System packages <base-system> |
Status: | RESOLVED OBSOLETE | ||
Severity: | normal | CC: | 404errorqc, bberberov+gentoo, chaujc, loki_val, ryao, scy-bugs-gentoo, Tanktalus, transacid |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
build log
Another case of fail Build log |
Description
Nick Soveiko
2009-02-09 04:21:51 UTC
Created attachment 181410 [details]
build log
update: it appears that the problem is in parallel make. MAKEOPTS="-j 1" emerge sys-apps/groff does the trick. my default is MAKEOPTS="-j -l 8". i think that incompatible MAKEOPTS should be sanitized in the ebuild. What happens when you try with MAKEOPTS="-j -l 4" and lower values for -l ? (In reply to comment #3) > What happens when you try with MAKEOPTS="-j -l 4" and lower values for -l ? with "-l 8" 2 out of 4 emerges failed (both at the same point). with "-l 4" 6 out of 6 went through. hope this helps. A dev should confirm this, but I do not think -l should be greater than the number of processors on the system that is running the emerge. This should not affect distcc. According to your emerge --info, you have a 9550 which should be recognized as 4 processors. the common wisdom is that -l should be greater than the number of processors (anywhere from n+1 to 2n), up to the point where there is 0 idle cpu time. load average by definition includes number of jobs waiting not only for the cpu, but for i/o as well. that's besides that there could be lower priority jobs that may or may not be contending (depending on the kernel scheduler) with emerge for the cpu, but they still count towards load average. (In reply to comment #6) > load > average by definition includes number of jobs waiting not only for the cpu, but > for i/o as well. from Debian's uptime(1): System load averages is the average number of processes that are either in a runnable or uninterruptable state. A process in a runnable state is either using the CPU or waiting to use the CPU. A process in unin- terruptable state is waiting for some I/O access, eg waiting for disk. for some reason, this paragraph is not in the man page on gentoo. Created attachment 203079 [details]
Another case of fail
Another instance.
Created attachment 227209 [details] Build log This problem also occurs while bootstrapping Gentoo Prefix (http://www.gentoo.org/proj/en/gentoo-alt/prefix/index.xml). I was running the bootstrapping steps on a dual core machine with MAKEOPTS=-j3 in etc/make.conf. Build log attached. *** Bug 362859 has been marked as a duplicate of this bug. *** Is anyone working on this? Would someone suffering from this test the following upstream patch? http://cvs.savannah.gnu.org/viewvc/groff/Makefile.in?root=groff&r1=1.118&r2=1.119&view=patch It looks like it resolves this problem. Closing as obsolete because version bug was filed against is no longer available in Gentoo repository. Please re-open if you are still able to reproduce against latest available version. |