rpms/kernel
3
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ensure modules_sign runs even if debuginfo is not produced

Browse Source 
tested most of the possible with_* options... hopefully i didn't
miss anything.
This commit is contained in:
Kyle McMartin 2013-01-08 15:32:03 -05:00
parent 9198b1c856
commit 92559e2963
1 changed files with 38 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
kernel.spec
View File

@ -1853,13 +1853,6 @@ chmod -R a=rX Documentation
find Documentation -type d | xargs chmod u+w
%endif
###
### Special hacks for debuginfo subpackages.
###
# This macro is used by %%install, so we must redefine it before that.
%define debug_package %{nil}
# In the modsign case, we do 3 things. 1) We check the "flavour" and hard
# code the value in the following invocations. This is somewhat sub-optimal
# but we're doing this inside of an RPM macro and it isn't as easy as it
@ -1872,13 +1865,9 @@ find Documentation -type d | xargs chmod u+w
# We have to do all of those things _after_ find-debuginfo runs, otherwise
# that will strip the signature off of the modules.
%if %{with_debuginfo}
%define __debug_install_post \
/usr/lib/rpm/find-debuginfo.sh %{debuginfo_args} %{_builddir}/%{?buildsubdir}\
if [ "%{signmodules}" == "1" ]; \
then \
if [ "%{with_pae}" != "0" ]; \
then \
%define __modsign_install_post \
if [ "%{signmodules}" == "1" ]; then \
if [ "%{with_pae}" != "0" ]; then \
Arch=`head -1 configs/kernel-%{version}-%{_target_cpu}-PAE.config | cut -b 3-` \
rm -rf .tmp_versions \
mv .tmp_versions.sign.PAE .tmp_versions \
@ -1887,8 +1876,7 @@ find Documentation -type d | xargs chmod u+w
make -s ARCH=$Arch V=1 INSTALL_MOD_PATH=$RPM_BUILD_ROOT modules_sign KERNELRELEASE=%{KVERREL}.PAE \
%{SOURCE18} $RPM_BUILD_ROOT/lib/modules/%{KVERREL}.PAE/extra/ \
fi \
if [ "%{with_debug}" != "0" ]; \
then \
if [ "%{with_debug}" != "0" ]; then \
Arch=`head -1 configs/kernel-%{version}-%{_target_cpu}-debug.config | cut -b 3-` \
rm -rf .tmp_versions \
mv .tmp_versions.sign.debug .tmp_versions \
@ -1897,8 +1885,7 @@ find Documentation -type d | xargs chmod u+w
make -s ARCH=$Arch V=1 INSTALL_MOD_PATH=$RPM_BUILD_ROOT modules_sign KERNELRELEASE=%{KVERREL}.debug \
%{SOURCE18} $RPM_BUILD_ROOT/lib/modules/%{KVERREL}.debug/extra/ \
fi \
if [ "%{with_pae_debug}" != "0" ]; \
then \
if [ "%{with_pae_debug}" != "0" ]; then \
Arch=`head -1 configs/kernel-%{version}-%{_target_cpu}-PAEdebug.config | cut -b 3-` \
rm -rf .tmp_versions \
mv .tmp_versions.sign.PAEdebug .tmp_versions \
@ -1907,8 +1894,7 @@ find Documentation -type d | xargs chmod u+w
make -s ARCH=$Arch V=1 INSTALL_MOD_PATH=$RPM_BUILD_ROOT modules_sign KERNELRELEASE=%{KVERREL}.PAEdebug \
%{SOURCE18} $RPM_BUILD_ROOT/lib/modules/%{KVERREL}.PAEdebug/extra/ \
fi \
if [ "%{with_up}" != "0" ]; \
then \
if [ "%{with_up}" != "0" ]; then \
Arch=`head -1 configs/kernel-%{version}-%{_target_cpu}.config | cut -b 3-` \
rm -rf .tmp_versions \
mv .tmp_versions.sign .tmp_versions \
@ -1920,13 +1906,38 @@ find Documentation -type d | xargs chmod u+w
fi \
%{nil}
###
### Special hacks for debuginfo subpackages.
###
# This macro is used by %%install, so we must redefine it before that.
%define debug_package %{nil}
%if %{with_debuginfo}
%define __debug_install_post \
/usr/lib/rpm/find-debuginfo.sh %{debuginfo_args} %{_builddir}/%{?buildsubdir}\
%{nil}
%ifnarch noarch
%global __debug_package 1
%files -f debugfiles.list debuginfo-common-%{_target_cpu}
%defattr(-,root,root)
%endif
%endif
#
# Disgusting hack alert! We need to ensure we sign modules *after* all
# invocations of strip occur, which is in __debug_install_post if
# find-debuginfo.sh runs, and __os_install_post if not.
#
%define __spec_install_post \
%{?__debug_package:%{__debug_install_post}}\
%{__arch_install_post}\
%{__os_install_post}\
%{__modsign_install_post}
###
### install
###
@ -2295,6 +2306,13 @@ fi
# ||----w |
# || ||
%changelog
* Tue Jan 08 2013 Kyle McMartin <kmcmarti@redhat.com>
- Ensure modules are signed even if *-debuginfo rpms are not produced by
re-defining __spec_install_post and adding a hook after all strip
invocations. Ideally, in the future, we could patch the rpm macro and
remove the re-define from kernel.spec, but that's another windmill to tilt
at.
* Tue Jan 08 2013 Justin M. Forbes <jforbes@redhat.com> - 3.8.0-0.rc2.git3.1
- Linux v3.8-rc2-222-g2a893f9