Module: kamailio Branch: master Commit: 1f99806fa2db09b9ac6da7fbeb388a93c7638f13 URL: https://github.com/kamailio/kamailio/commit/1f99806fa2db09b9ac6da7fbeb388a93...
Author: Henning Westerholt hw@kamailio.org Committer: Henning Westerholt hw@kamailio.org Date: 2018-10-06T18:04:00+02:00
Makefile.defs: cleanup the CPUTYPE/mtune definitions for i386 and x86_64
- Makefile.defs: cleanup the CPUTYPE/mtune definitions for i386 and x86_64 - revert change c925278b8ee3492f71a from 2011 for gcc SUSE 4.5.x, now not needed - use for current gcc (>= gcc 4.2) the generic mtune, which is now available - From the gcc manual: "Produce code optimized for the most common IA32/AMD64/EM64T processors. If you know the CPU on which your code will run, then you should use the corresponding -mtune or -march option instead of -mtune=generic. But, if you do not know exactly what CPU users of your application will have, then you should use this option. As new processors are deployed in the marketplace, the behavior of this option will change. Therefore, if you upgrade to a newer version of GCC, code generation controlled by this option will change to reflect the processors that are most common at the time that version of GCC is released. - unify opteron usage, use equivalent athlon64 mtune for old compilers instead - simplify x86_64 gcc logic, remove now redundant 4.2.x if case - there should be no performance implications, as we compile without debugging - we prefer right now to have proper debugging, this would break with agressive optimization
---
Modified: src/Makefile.defs
---
Diff: https://github.com/kamailio/kamailio/commit/1f99806fa2db09b9ac6da7fbeb388a93... Patch: https://github.com/kamailio/kamailio/commit/1f99806fa2db09b9ac6da7fbeb388a93...
---
diff --git a/src/Makefile.defs b/src/Makefile.defs index ee5fffa209..fe4dc4ce2f 100644 --- a/src/Makefile.defs +++ b/src/Makefile.defs @@ -888,7 +888,7 @@ ifeq ($(CC_NAME), gcc) CFLAGS=-g $(CC_OPT) -funroll-loops -Wcast-align $(PROFILE) #if gcc 8.0+, 7.0+, 6.0+, 5.0+, 4.5+ or 4.2+ ifeq (,$(strip $(filter-out 4.2+ 4.5+ 5.0+ 6.0+ 7.0+ 8.0+,$(CC_SHORTVER)))) -$(call set_if_empty,CPUTYPE,athlon64) +$(call set_if_empty,CPUTYPE,generic) CFLAGS+=-m32 -minline-all-stringops \ -falign-loops \ -ftree-vectorize \ @@ -969,18 +969,8 @@ ifeq ($(CC_NAME), gcc) #common stuff CFLAGS=-g $(CC_OPT) -funroll-loops -Wcast-align $(PROFILE) #if gcc 4.5+ - # don't add '-mtune=$(CPUTYPE)' - gcc failure -ifeq ($(CC_SHORTVER),$(filter $(CC_SHORTVER),4.5+ 5.0+ 6.0+ 7.0+ 8.0+)) -$(call set_if_empty,CPUTYPE,opteron) - CFLAGS+=-m64 -minline-all-stringops \ - -falign-loops \ - -ftree-vectorize \ - -fno-strict-overflow - LDFLAGS+=-m64 -else - #if gcc 4.2+ -ifeq ($(CC_SHORTVER), 4.2+) -$(call set_if_empty,CPUTYPE,opteron) +ifeq ($(CC_SHORTVER),$(filter $(CC_SHORTVER),4.2+ 4.5+ 5.0+ 6.0+ 7.0+ 8.0+)) +$(call set_if_empty,CPUTYPE,generic) CFLAGS+=-m64 -minline-all-stringops \ -falign-loops \ -ftree-vectorize \ @@ -990,7 +980,7 @@ $(call set_if_empty,CPUTYPE,opteron) else #if gcc 4.0+ ifeq ($(CC_SHORTVER), 4.x) -$(call set_if_empty,CPUTYPE,opteron) +$(call set_if_empty,CPUTYPE,athlon64) CFLAGS+=-m64 -minline-all-stringops \ -falign-loops \ -ftree-vectorize \ @@ -1028,7 +1018,6 @@ endif # CC_SHORTVER, 2.9x endif # CC_SHORTVER, 3.0 endif # CC_SHORTVER, 3.4 endif # CC_SHORTVER, 4.x -endif # CC_SHORTVER, 4.2+ endif # CC_SHORTVER, 8.0+, 7.0+, 6.0+, 5.0+, 4.5+
else # CC_NAME, gcc
Am Samstag, 6. Oktober 2018, 18:08:31 CEST schrieb Henning Westerholt:
[..]
I did a spelling mistake in the commit message:
[..]
- there should be no performance implications, as we compile without
debugging
- there should be no performance implications, as we compile without optimization
Best regards,
Henning