Bug 57467

Summary: gcc 3.1-0.10 miscompiles qmake
Product: [Retired] Red Hat Raw Hide Reporter: Bernhard Rosenkraenzer <bero>
Component: gccAssignee: Jakub Jelinek <jakub>
Status: CLOSED RAWHIDE QA Contact:
Severity: medium Docs Contact:
Priority: medium    
Version: 1.0   
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2001-12-21 15:48:19 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Bernhard Rosenkraenzer 2001-12-13 12:55:37 UTC
Description of Problem:
Using qmake (Qt's makefile generator) after compiling it with gcc 
3.1-0.10 produces broken Makefiles. The same version of qt/qmake works 
perfectly with 2.96-101.

Version-Release number of selected component (if applicable):
3.1-0.10


How Reproducible:
100%

Steps to Reproduce:
1. Install qt 3.0.1 sources
2. Attempt to build
3. 

Actual Results:
Build fails because tools/designer/designer/Makefile gets weird rules:
dbconnections\.ui: dbconnections.ui
        $(UIC) dbconnections.ui -o dbconnections\.ui



Expected Results:
The Makefile has the correct rule, such as the one I'm getting when
building qmake with gcc 2.96 or 3.0.2:

dbconnections.h: dbconnections.ui
        $(UIC) dbconnections.ui -o dbconnections.h


Additional Information:
qmake compiles without any warnings even with gcc 3.1 -Wall, so it's 
probably not some newly enforced standards compliance.

Comment 1 Bernhard Rosenkraenzer 2001-12-13 14:19:44 UTC
Happens even at -O0


Comment 2 Bernhard Rosenkraenzer 2001-12-13 15:24:58 UTC
It's even weirder:
qmake is usually compiled with out any -O flags because it has to support 
pretty much all compilers out there.

Adding -O2 to the CFLAGS actually fixes the problem.

So qmake is miscompiled at -O0 only (guess that's the first time I've had to 
_increase_ the optimization level to work around compiler bugs ;) ).



Comment 3 Bernhard Rosenkraenzer 2001-12-13 16:54:36 UTC
Reducing priority after finding a workaround.


Comment 4 Jakub Jelinek 2001-12-20 23:16:05 UTC
Reduced to makefile.o (if it is compiled with -O0, it doesn't work, if -O1
and above, it works). Will need to debug where things change...

Comment 5 Jakub Jelinek 2001-12-21 15:48:14 UTC
As it does go away if makefile.cc is compiled with -fno-inline, I guess this
is just the result of bogus change from 2001-11-15 which caused tree inlining
to happen without optimizations too.
This was fixed on 2001-12-13 in CVS.