Bug 1025143

Summary: Internal error: .size expression not evaluate to a constent
Product: [Fedora] Fedora Reporter: Lameire Alexis <alexis.lameire>
Component: gccAssignee: Jakub Jelinek <jakub>
Status: CLOSED NOTABUG QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: high Docs Contact:
Priority: unspecified    
Version: rawhideCC: alexis.lameire, jakub, law
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2013-11-01 12:54:42 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Lameire Alexis 2013-10-31 05:57:26 UTC
Description of problem:
I try to update my package minion, but build fail with :

{standard input}: Assembler messages:
{standard input}:1384863: Warning: end of file not at end of a line; newline inserted
{standard input}: Error: .size expression for _ZN22ReifiedEqualConstraintI20BoundVarRef_internalIiE10VarRefTypeI26SparseBoundVarRef_internalIiEE15QuickVarRefTypeI19GetBoolVarContainer19BoolVarRef_internalELb1EE8get_varsEv does not evaluate to a constant

This bug look like to be architecture specific, I don't have it on i686

Version-Release number of selected component (if applicable):
I try to compile into rawhide and f20


How reproducible:
rebuild last version of minon package into koji infrastructure

Additional info:
https://koji.fedoraproject.org/koji/taskinfo?taskID=6117479 (build for rawhide)
https://koji.fedoraproject.org/koji/taskinfo?taskID=6117596 (build for f20)

Comment 1 Jakub Jelinek 2013-10-31 07:53:20 UTC
You are misreading the problem.  The assembler messages are clearly just because you are using -pipe and the compiler has been killed by the kernel:
c++: internal compiler error: Killed (program cc1plus)
so it stopped writing more assembler into the pipe.
Now, the question is how much memory did cc1plus need to compile that, how many compilers and other potentially memory hungry processes were running at the same time and if the memory requirements are corresponding to the complexity of the source you are compiling.  If you attach preprocessed source of the file on which cc1plus eats most memory, I can have a look at where the memory is spent on and if it is a g++ bug or not.

Comment 2 Lameire Alexis 2013-11-01 12:54:42 UTC
Your information is exact. The koji buildsystem limit ressources on build task. When I disable parallel build, this solve the bug. 

Thenks for help.