Overview

Request 761566 accepted

- disable LTO to fix gcc9 build errors on x86_64

Loading...

Fabian Vogt's avatar

Does the "%_lto_cflags" way not work? IMO it should be disabled unconditionally.


Ralf Habacker's avatar

Here againg because formatting was broken by obs editor

For the record: _lto_cflags are defined in

/usr/lib/rpm/suse/macros:302:%_lto_cflags -flto=auto

and

/usr/lib/rpm/platform/x86_64-linux/macros:63:%_lto_cflags -flto=auto

-flto=auto let building fail on x86_64 arch.

https://build.opensuse.org/projects/openSUSE:Factory/prjconf already excludes _lto_cflags for ppc64

%ifnarch ppc64 
Optflags: * -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type %%{?_lto_cflags} 
%else
Optflags: * -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type
%endif

which let me think that x86_64 should also not use this flag e.g.

%ifnarch ppc64 x86_64
Optflags: * -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type %%{?_lto_cflags}
%else
Optflags: * -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type
%endif

What do you think ?


Fabian Vogt's avatar

No, enabling LTO for x86_64 was the main point of the effort - it won't be undone. Please read https://en.opensuse.org/openSUSE:LTO#LTO_enablement_in_openSUSE:Factory

Request History
Ralf Habacker's avatar

rhabacker created request

- disable LTO to fix gcc9 build errors on x86_64


Fabian Vogt's avatar

Vogtinator accepted request

LGTM

openSUSE Build Service is sponsored by