From 8a8a8479be704a413233f27c7f13b2abc489f5b5 Mon Sep 17 00:00:00 2001
From: Ian McInerney <i.mcinerney17@imperial.ac.uk>
Date: Mon, 14 Aug 2023 15:41:28 +0100
Subject: [PATCH] Fix cooperlake and sapphire rapids march flags on clang

The march=cooperlake and march=sapphirerapids flags were never getting
added when building with Clang targetting those architectures. Instead
it was falling back to the skylake AVX512 implementation.

Clang added support for these two architectures in Clang 9 and Clang 12,
so introduce new checks for those versions to enable the appropriate
march flag, and fallback to skylake otherwise.
---
 Makefile.system |  5 ++++
 Makefile.x86_64 | 74 +++++++++++++++++++++++++++++++++----------------
 kernel/Makefile |  4 +--
 3 files changed, 57 insertions(+), 26 deletions(-)

diff --git a/Makefile.system b/Makefile.system
index 3c1648dc7..b3968d739 100644
--- a/Makefile.system
+++ b/Makefile.system
@@ -384,6 +384,11 @@ GCCMINORVERSIONGTEQ4 := $(shell expr `$(CC) $(GCCDUMPVERSION_PARAM) | cut -f2 -d
 GCCMINORVERSIONGTEQ7 := $(shell expr `$(CC) $(GCCDUMPVERSION_PARAM) | cut -f2 -d.` \>= 7)
 endif
 
+ifeq ($(C_COMPILER), CLANG)
+CLANGVERSIONGTEQ9 := $(shell expr `$(CC) -dumpversion | cut -f1 -d.` \>= 9)
+CLANGVERSIONGTEQ12 := $(shell expr `$(CC) -dumpversion | cut -f1 -d.` \>= 12)
+endif
+
 #
 #  OS dependent settings
 #
diff --git a/Makefile.x86_64 b/Makefile.x86_64
index 7ab331b1f..702447ace 100644
--- a/Makefile.x86_64
+++ b/Makefile.x86_64
@@ -75,18 +75,31 @@ endif
 ifeq ($(CORE), COOPERLAKE)
 ifndef NO_AVX512
 ifeq ($(C_COMPILER), GCC)
-# cooperlake support was added in 10.1
-ifeq ($(GCCVERSIONGTEQ10)$(GCCMINORVERSIONGTEQ1), 11)
-CCOMMON_OPT += -march=cooperlake
-ifneq ($(F_COMPILER), NAG)
-FCOMMON_OPT += -march=cooperlake
-endif
-else  # gcc not support, fallback to avx512
-CCOMMON_OPT += -march=skylake-avx512
-ifneq ($(F_COMPILER), NAG)
-FCOMMON_OPT += -march=skylake-avx512
-endif
-endif
+ # cooperlake support was added in 10.1
+ ifeq ($(GCCVERSIONGTEQ10)$(GCCMINORVERSIONGTEQ1), 11)
+  CCOMMON_OPT += -march=cooperlake
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=cooperlake
+  endif
+ else  # gcc not support, fallback to avx512
+  CCOMMON_OPT += -march=skylake-avx512
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=skylake-avx512
+  endif
+ endif
+else ifeq ($(C_COMPILER), CLANG)
+ # cooperlake support was added in clang 9
+ ifeq ($(CLANGVERSIONGTEQ9), 1)
+  CCOMMON_OPT += -march=cooperlake
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=cooperlake
+  endif
+ else  # not supported in clang, fallback to avx512
+  CCOMMON_OPT += -march=skylake-avx512
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=skylake-avx512
+  endif
+ endif
 endif
 ifeq ($(OSNAME), CYGWIN_NT)
 CCOMMON_OPT += -fno-asynchronous-unwind-tables
@@ -104,18 +117,31 @@ endif
 ifeq ($(CORE), SAPPHIRERAPIDS)
 ifndef NO_AVX512
 ifeq ($(C_COMPILER), GCC)
-# sapphire rapids support was added in 11
-ifeq ($(GCCVERSIONGTEQ11), 1)
-CCOMMON_OPT += -march=sapphirerapids
-ifneq ($(F_COMPILER), NAG)
-FCOMMON_OPT += -march=sapphirerapids
-endif
-else  # gcc not support, fallback to avx512
-CCOMMON_OPT += -march=skylake-avx512
-ifneq ($(F_COMPILER), NAG)
-FCOMMON_OPT += -march=skylake-avx512
-endif
-endif
+ # sapphire rapids support was added in 11
+ ifeq ($(GCCVERSIONGTEQ11), 1)
+  CCOMMON_OPT += -march=sapphirerapids
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=sapphirerapids
+  endif
+ else  # gcc not support, fallback to avx512
+  CCOMMON_OPT += -march=skylake-avx512
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=skylake-avx512
+  endif
+ endif
+else ifeq ($(C_COMPILER), CLANG)
+ # cooperlake support was added in clang 12
+ ifeq ($(CLANGVERSIONGTEQ12), 1)
+  CCOMMON_OPT += -march=cooperlake
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=cooperlake
+  endif
+ else  # not supported in clang, fallback to avx512
+  CCOMMON_OPT += -march=skylake-avx512
+  ifneq ($(F_COMPILER), NAG)
+   FCOMMON_OPT += -march=skylake-avx512
+  endif
+ endif
 endif
 ifeq ($(OSNAME), CYGWIN_NT)
 CCOMMON_OPT += -fno-asynchronous-unwind-tables
diff --git a/kernel/Makefile b/kernel/Makefile
index d426a1bdb..795f25eec 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -33,7 +33,7 @@ endif
 ifdef TARGET_CORE
 ifeq ($(TARGET_CORE), SAPPHIRERAPIDS)
  override CFLAGS += -DBUILD_KERNEL -DTABLE_NAME=gotoblas_$(TARGET_CORE)
- ifeq ($(GCCVERSIONGTEQ11), 1) 
+ ifeq (1, $(filter 1,$(GCCVERSIONGTEQ11) $(CLANGVERSIONGTEQ12)))
   override CFLAGS += -march=sapphirerapids
  else 
   override CFLAGS += -march=skylake-avx512 -mavx512f
@@ -48,7 +48,7 @@ ifeq ($(TARGET_CORE), SAPPHIRERAPIDS)
  endif
 else ifeq ($(TARGET_CORE), COOPERLAKE)
  override CFLAGS += -DBUILD_KERNEL -DTABLE_NAME=gotoblas_$(TARGET_CORE)
- ifeq ($(GCCVERSIONGTEQ10), 1) 
+ ifeq (1, $(filter 1,$(GCCVERSIONGTEQ10) $(CLANGVERSIONGTEQ9)))
   override CFLAGS += -march=cooperlake
  else 
   override CFLAGS += -march=skylake-avx512 -mavx512f