Comment 4 for bug 1667407

> -----Original Message-----
> From: <email address hidden> [mailto:<email address hidden>] On Behalf Of
> Lukasz Zemczak
> Sent: quinta-feira, 2 de março de 2017 10:31
> To: Gustavo Serra Scalet <email address hidden>
> Subject: [Bug 1667407] Re: improve 2x-3x sha256sum performance on
> ppc64le due to current gcc optimization bug
>
> Could we get the bug details updated to follow the SRU template?

Done

> Also, checking the SRU upload, one thing I don't particularly like is
> that the attached patch has no description. Yes, the change is self-
> explanatory, but some context in patch descriptions is always welcome.
> Won't block on that though as I see some of the existing patches in the
> package have empty 'Description:' fields anyway...

Then I'd add to Michael Stone's patch:

diff --git i/Makefile.in w/Makefile.in
index bdf9a43..7a43897 100644
--- i/Makefile.in
+++ w/Makefile.in
@@ -14434,6 +14434,13 @@ $(factor_tests): $(tf)/run.sh $(tf)/create-test.sh

 $(TEST_LOGS): $(PROGRAMS)

+# REMOVE THIS HANDLING FOR PPC64 IF GCC > 7.0 IS USED
+# Shutting off -fschedule-insns optimization due to huge performance drop
+# (around 2 to 3 times).
+ifeq ($(DEB_TARGET_ARCH), ppc64el)
+lib/sha256.o: CFLAGS+=-fno-schedule-insns
+endif
+
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
 .NOEXPORT: