rpms/java-1.8.0-openjdk
7
0
Fork 0
Code Issues Pull Requests Releases Activity

Actually disable LTO

Browse Source 
Previously, the define for _lto_cflags was too far down in the
spec file. optflags is being evaluated before it, thus, it
still contained -flto -ffat-lto-objects cflags which is then
passed to the build. Fix is to move the define further up.
This commit is contained in:
Severin Gehwolf 2020-07-27 12:50:02 +02:00
parent 6d3163d3ed
commit 54c894ca10
1 changed files with 10 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
java-1.8.0-openjdk.spec
View File

@ -119,6 +119,12 @@
# No docs nor bootcycle for debug builds # No docs nor bootcycle for debug builds
%global debug_targets images %global debug_targets images
# This package fails to build with LTO due to undefined symbols. LTO
# was disabled in OpenSuSE as well, but with no real explanation why
# beyond the undefined symbols. It really should be investigated further.
# Disable LTO
%define _lto_cflags %{nil}
# Filter out flags from the optflags macro that cause problems with the OpenJDK build # Filter out flags from the optflags macro that cause problems with the OpenJDK build
# We filter out -Wall which will otherwise cause HotSpot to produce hundreds of thousands of warnings (100+mb logs) # We filter out -Wall which will otherwise cause HotSpot to produce hundreds of thousands of warnings (100+mb logs)
# We filter out -O flags so that the optimization of HotSpot is not lowered from O3 to O2 # We filter out -O flags so that the optimization of HotSpot is not lowered from O3 to O2
@ -128,12 +134,6 @@
%global ourcppflags %(echo %ourflags | sed -e 's|-fexceptions||') %global ourcppflags %(echo %ourflags | sed -e 's|-fexceptions||')
%global ourldflags %{__global_ldflags} %global ourldflags %{__global_ldflags}
# This package fails to build with LTO due to undefined symbols. LTO
# was disabled in OpenSuSE as well, but with no real explanation why
# beyond the undefined symbols. It really should be investigated further.
# Disable LTO
%define _lto_cflags %{nil}
# With disabled nss is NSS deactivated, so NSS_LIBDIR can contain the wrong path # With disabled nss is NSS deactivated, so NSS_LIBDIR can contain the wrong path
# the initialization must be here. Later the pkg-config have buggy behavior # the initialization must be here. Later the pkg-config have buggy behavior
# looks like openjdk RPM specific bug # looks like openjdk RPM specific bug
@ -260,7 +260,7 @@
%global updatever %(VERSION=%{whole_update}; echo ${VERSION##*u}) %global updatever %(VERSION=%{whole_update}; echo ${VERSION##*u})
# eg jdk8u60-b27 -> b27 # eg jdk8u60-b27 -> b27
%global buildver %(VERSION=%{version_tag}; echo ${VERSION##*-}) %global buildver %(VERSION=%{version_tag}; echo ${VERSION##*-})
%global rpmrelease 0 %global rpmrelease 1
# Define milestone (EA for pre-releases, GA ("fcs") for releases) # Define milestone (EA for pre-releases, GA ("fcs") for releases)
# Release will be (where N is usually a number starting at 1): # Release will be (where N is usually a number starting at 1):
# - 0.N%%{?extraver}%%{?dist} for EA releases, # - 0.N%%{?extraver}%%{?dist} for EA releases,
@ -2400,6 +2400,9 @@ require "copy_jdk_configs.lua"
%endif %endif
%changelog %changelog
* Mon Jul 27 2020 Severin Gehwolf <sgehwolf@redhat.com> - 1:1.8.0.265.b01-1
- Reorder _lto_cflags define so that it gets picked up via optflags
* Mon Jul 27 2020 Andrew Hughes <gnu.andrew@redhat.com> - 1:1.8.0.265.b01-1 * Mon Jul 27 2020 Andrew Hughes <gnu.andrew@redhat.com> - 1:1.8.0.265.b01-1
- Update to aarch64-shenandoah-jdk8u265-b01. - Update to aarch64-shenandoah-jdk8u265-b01.
- Update release notes for 8u265 release. - Update release notes for 8u265 release.