Bug 26517

Summary: build scripts that fail inside parentheses don't cause rpm building to fail
Product: [Retired] Red Hat Linux Reporter: Need Real Name <bradley>
Component: rpmAssignee: Jeff Johnson <jbj>
Status: CLOSED NOTABUG QA Contact: David Lawrence <dkl>
Severity: medium Docs Contact:
Priority: medium    
Version: 6.2   
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: 2001-02-07 19:25:02 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:
Attachments:
Description Flags
Here is a source rpm that improperly continues after the build.
none
And here is the rpm with the spec fixed. But rpm -bb should behave the same for both none

Description Need Real Name 2001-02-07 19:04:32 UTC 
From Bugzilla Helper:
User-Agent: Mozilla/4.61 [en] (X11; I; Linux 2.2.12-20 i686)


In an rpm spec, if you have

%build
(cd foo;make foo)

and the make fails, then rpm goes ahead and tries to finish making the
rpms.  If you do this instead

%build
cd foo;make foo

then a failure in the make causes the whole rpm --bb to stop.

The presence of parentheses should not change the behavior.  I've seen this
in RPM 3.0.5-9.6x
The correct behavior is that when the make fails, the rpm should not be
built.

Reproducible: Always
Steps to Reproduce:
Make an rpm out of a makefile that fails.  (I built a foo.tar.gz that
contained a file foo/foo.c which doesn't compile) and then have the %build
step be
 (cd foo; make foo)
the make fails, but the rpm system will continue by running the %install
script.

This is the bad behavior in which the problem isn't noticed until the
install script runs.

$ rpm -bb foo.spec
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.89140
+ umask 022
+ cd /u0/bradley/redhat/BUILD
+ cd /u0/bradley/redhat/BUILD
+ rm -rf foo
+ /bin/mkdir -p foo
+ cd foo
+ /bin/gzip -dc /u0/bradley/redhat/SOURCES/foo.tar.gz
+ tar -xvvf -
drwxrwxr-x bradley/users     0 2001-02-07 10:01:47 foo/
-rw-rw-r-- bradley/users    17 2001-02-07 10:01:47 foo/foo.c
-rw-rw-r-- bradley/users     0 2001-02-07 10:01:34 foo/foo.c.~1~
+ STATUS=0
+ [ 0 -ne 0 ]
++ /usr/bin/id -u
+ [ 20202 = 0 ]
++ /usr/bin/id -u
+ [ 20202 = 0 ]
+ /bin/chmod -Rf a+rX,g-w,o-w .
+ exit 0
Executing(%build): /bin/sh -e /var/tmp/rpm-tmp.89140
+ umask 022
+ cd /u0/bradley/redhat/BUILD
+ cd foo
+ cd foo
+ make foo
gcc -g -O2    foo.c   -o foo
foo.c:1: parse error before `)'
make: *** [foo] Error 1
+ exit 0
Executing(%install): /bin/sh -e /var/tmp/rpm-tmp.89140
+ umask 022
+ cd /u0/bradley/redhat/BUILD
+ cd foo
+ rm -rf /u0/bradley/redhat/buildroot
+ mkdir --parents /u0/bradley/redhat/buildroot/usr/local
+ cp foo /u0/bradley/redhat/buildroot/usr/local/foo
cp: foo: omitting directory
Bad exit status from /var/tmp/rpm-tmp.89140 (%install)

This is the good behavior
$ rpm -bb foo.spec
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.3115
+ umask 022
+ cd /u0/bradley/redhat/BUILD
+ cd /u0/bradley/redhat/BUILD
+ rm -rf foo
+ /bin/mkdir -p foo
+ cd foo
+ /bin/gzip -dc /u0/bradley/redhat/SOURCES/foo.tar.gz
+ tar -xvvf -
drwxrwxr-x bradley/users     0 2001-02-07 10:01:47 foo/
-rw-rw-r-- bradley/users    17 2001-02-07 10:01:47 foo/foo.c
-rw-rw-r-- bradley/users     0 2001-02-07 10:01:34 foo/foo.c.~1~
+ STATUS=0
+ [ 0 -ne 0 ]
++ /usr/bin/id -u
+ [ 20202 = 0 ]
++ /usr/bin/id -u
+ [ 20202 = 0 ]
+ /bin/chmod -Rf a+rX,g-w,o-w .
+ exit 0
Executing(%build): /bin/sh -e /var/tmp/rpm-tmp.3115
+ umask 022
+ cd /u0/bradley/redhat/BUILD
+ cd foo
+ cd foo
+ make foo
gcc -g -O2    foo.c   -o foo
foo.c:1: parse error before `)'
make: *** [foo] Error 1
Bad exit status from /var/tmp/rpm-tmp.3115 (%build)
Comment 1 Need Real Name 2001-02-07 19:22:19 UTC 
Created attachment 9280 [details]
Here is a source rpm that improperly continues after the build.
Comment 2 Need Real Name 2001-02-07 19:24:58 UTC 
Created attachment 9281 [details]
And here is the rpm with the spec fixed.  But rpm -bb should behave the same for both
Comment 3 Jeff Johnson 2001-02-07 19:36:19 UTC 
This is a packaging problem, not an rpm problem, and has to do with passing
a -e option from parent shell to sub-shell. Try

	(cd foo && exit 1
	make foo && exit 1
	)

and other variant shell scripts to fix.