Bug 1037118

Summary:	hardening-check FTBFS if "-Werror=format-security" flag is used
Product:	[Fedora] Fedora	Reporter:	Dhiru Kholia <dkholia>
Component:	hardening-check	Assignee:	Björn 'besser82' Esser <besser82>
Status:	CLOSED NOTABUG	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	rawhide	CC:	besser82, bressers, dhiru
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2014-05-24 14:35:11 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Bug Depends On:
Bug Blocks:	1038083

Description Dhiru Kholia 2013-12-03 03:10:43 UTC

Description of problem
----------------------

hardening-check fails to build if "-Werror=format-security" flag is used.

...

format.c:11:5: error: format not a string literal and no format arguments [-Werror=format-security]

...

We are working on a proposal to enable "-Werror=format-security" for all
packages. Once this flag is enabled, GCC will refuse to compile code that could
be vulnerable to a string format security flaw. For more details, please see
https://fedorahosted.org/fesco/ticket/1185 page.

To understand why it is important to fix this, please see
https://fedoraproject.org/wiki/Format-Security-FAQ page.

How to fix this
---------------

The fix for these errors is quite simple. It's a matter of changing a
line like,

printf(foo);

to read,

printf("%s", foo);

That's it.

Please fix this issue in rawhide with a patch (which you should submit
to upstream to merge moving forward). Please do a new build with the
fix in rawhide. Other releases do not need to be directly fixed, but
there should be no harm in pushing out this fix/patch with other needed
changes to those branches.

In the event you don't fix this bug before the next mass rebuild,
provenpackagers may step in and update your package(s) to fix this
issue.

How reproducible
----------------

Build hardening-check-2.4-1.fc21.src.rpm with "-Werror=format-security" flag to reproduce the problem.

To make this process easier, you can use a modified "redhat-rpm-config" package
from http://people.fedoraproject.org/~halfie/artifacts/redhat-rpm-config/ URL.

$ sha256sum redhat-rpm-config-9.1.0-56.fc20.*
faad7594b2080fe76497d0ce50808c905a93dd7b41c1defdde5ca57e3833d3d2 redhat-rpm-config-9.1.0-56.fc20.noarch.rpm
5aa9357174305c7285ffdbc92d7ffe1c07a8a95d5459b930461308f5aad75413 redhat-rpm-config-9.1.0-56.fc20.src.rpm

Comment 2 Björn 'besser82' Esser 2014-05-24 14:35:11 UTC

This is part of the testsuite from hardening-wrapper.  ;)


Excerpt from build.log:
# Make sure build fails due to -Werror=format-security
! cc    -fstack-protector --param ssp-buffer-size=4  -D_FORTIFY_SOURCE=2  -Wformat -Wformat-security -Werror=format-security  -O2    -Wl,-z,relro  -Wl,-z,now  -o ../build-tree/includes-test-format-security format.c
format.c: In function 'main':
format.c:11:5: error: format not a string literal and no format arguments [-Werror=format-security]
     return fprintf(stderr, argv[0]);
     ^
cc1: some warnings being treated as errors
# Make sure build succeeds with -Wno-format-security
cc    -fstack-protector --param ssp-buffer-size=4  -D_FORTIFY_SOURCE=2  -Wformat -Wformat-security -Werror=format-security  -O2 -Wno-format-security    -Wl,-z,relro  -Wl,-z,now  -o ../build-tree/includes-test-format-security format.c
# Make sure build stack-protects a small ssp buffer
cc    -fstack-protector --param ssp-buffer-size=4  -D_FORTIFY_SOURCE=2  -Wformat -Wformat-security -Werror=format-security  -O2    -Wl,-z,relro  -Wl,-z,now  -o ../build-tree/includes-test-ssp-buffer-size-protect ssp-buffer-size-protect.c


It is intentional to have a build-fail during compilation, because the testsuite is meant to give output which should be checked by a human to make sure everything is working as expected.  One point of that human check is to see 'format.c' *MUST* fail with 'format not a string literal and no format arguments' when '-Werror=format-security' is used during compilation.

Closing this, since this is a false positive and NOT a bug.  ;)