Message ID | 20161121183217.GA1046@waldemar-brodkorb.de |
---|---|
State | Accepted |
Headers | show |
Hello, On Mon, 21 Nov 2016 19:32:18 +0100, Waldemar Brodkorb wrote: > +# GCC bug with Os/O1/O2/O3 > +# internal compiler error: in gen_add2_insn, at optabs.c:4454 > +ifeq ($(BR2_bfin),y) > +PTPD2_CONF_ENV += CFLAGS="-O0" You forgot to pass $(TARGET_CFLAGS) here. With your change, we loose TARGET_CFLAGS. So I've fixed that up and applied. However, I continue to dislike such hacks: when the problem gets fixed by gcc, how will we notice? Thanks, Thomas
Hi Thomas, Thomas Petazzoni wrote, > Hello, > > On Mon, 21 Nov 2016 19:32:18 +0100, Waldemar Brodkorb wrote: > > > +# GCC bug with Os/O1/O2/O3 > > +# internal compiler error: in gen_add2_insn, at optabs.c:4454 > > +ifeq ($(BR2_bfin),y) > > +PTPD2_CONF_ENV += CFLAGS="-O0" > > You forgot to pass $(TARGET_CFLAGS) here. With your change, we loose > TARGET_CFLAGS. So I've fixed that up and applied. > > However, I continue to dislike such hacks: when the problem gets fixed > by gcc, how will we notice? Either when we get notice that the PR is closed, this time I need to create one. But we have similar issues where we already have PR's. Or we check when a new gcc release come out, if the results from "grep bfin package/*/*mk" are still required. I think better then never ending autobuild failures. best regards Waldemar
Hello, On Tue, 22 Nov 2016 11:12:20 +0100, Waldemar Brodkorb wrote: > > However, I continue to dislike such hacks: when the problem gets fixed > > by gcc, how will we notice? > > Either when we get notice that the PR is closed, this time I need to > create one. But we have similar issues where we already have PR's. > Or we check when a new gcc release come out, if the results from > "grep bfin package/*/*mk" are still required. Maybe we need to have some kind of convention of a specific string to include in the comment on such hacks, in order to more easily grep? Like: # Workaround gcc-prXYZ or # Workaround binutils-prXYZ This way, we can grep for "gcc-pr" and "binutils-pr" and easily check the status of those PRs. > I think better then never ending autobuild failures. For sure, I'm not claiming we should keep things as-is. It's good to add temporary hacks, but it's also good to think about how we can at some point remove those hacks. Best regards, Thomas
diff --git a/package/ptpd2/ptpd2.mk b/package/ptpd2/ptpd2.mk index da6a106..43b59dc 100644 --- a/package/ptpd2/ptpd2.mk +++ b/package/ptpd2/ptpd2.mk @@ -24,6 +24,12 @@ else PTPD2_CONF_OPTS += --disable-snmp endif +# GCC bug with Os/O1/O2/O3 +# internal compiler error: in gen_add2_insn, at optabs.c:4454 +ifeq ($(BR2_bfin),y) +PTPD2_CONF_ENV += CFLAGS="-O0" +endif + define PTPD2_INSTALL_INIT_SYSV $(INSTALL) -m 755 -D package/ptpd2/S65ptpd2 \ $(TARGET_DIR)/etc/init.d/S65ptpd2
Fixes: http://autobuild.buildroot.net/results/cb967c8af006caa9272e800968f794ca018a7a27 http://autobuild.buildroot.net/results/9a046dc5ef390e562b89338f0afeaef14b945458 Signed-off-by: Waldemar Brodkorb <wbx@openadk.org> --- package/ptpd2/ptpd2.mk | 6 ++++++ 1 file changed, 6 insertions(+)