[PATCH] kselftest/arm64: Don't enable v8.5 for MTE selftest builds

[PATCH] kselftest/arm64: Don't enable v8.5 for MTE selftest builds

[Mark Brown]

Currently we set -march=armv8.5+memtag when building the MTE selftests,
allowing the compiler to emit v8.5 and MTE instructions for anything it
generates. This means that we may get code that will generate SIGILLs when
run on older systems rather than skipping on non-MTE systems as should be
the case. Most toolchains don't select any incompatible instructions but
I have seen some reports which suggest that some may be appearing which do
so. This is also potentially problematic in that if the compiler chooses to
emit any MTE instructions for the C code it may interfere with the MTE
usage we are trying to test.

Since the only reason we are specifying this option is to allow us to
assemble MTE instructions in mte_helper.S we can avoid these issues by
moving to using a .arch directive there and adding the -march explicitly to
the toolchain support check instead of the generic CFLAGS.

Signed-off-by: Mark Brown <broonie@kernel.org>
---
 tools/testing/selftests/arm64/mte/Makefile     | 5 +----
 tools/testing/selftests/arm64/mte/mte_helper.S | 2 ++
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/arm64/mte/Makefile b/tools/testing/selftests/arm64/mte/Makefile
index a5a0744423d8..037046f5784e 100644
--- a/tools/testing/selftests/arm64/mte/Makefile
+++ b/tools/testing/selftests/arm64/mte/Makefile
@@ -11,11 +11,8 @@ LDFLAGS += -pthread
 SRCS := $(filter-out mte_common_util.c,$(wildcard *.c))
 PROGS := $(patsubst %.c,%,$(SRCS))
 
-#Add mte compiler option
-CFLAGS += -march=armv8.5-a+memtag
-
 #check if the compiler works well
-mte_cc_support := $(shell if ($(CC) $(CFLAGS) -E -x c /dev/null -o /dev/null 2>&1) then echo "1"; fi)
+mte_cc_support := $(shell if ($(CC) $(CFLAGS) -march=armv8.5-a+memtag -E -x c /dev/null -o /dev/null 2>&1) then echo "1"; fi)
 
 ifeq ($(mte_cc_support),1)
 # Generated binaries to be installed by top KSFT script
diff --git a/tools/testing/selftests/arm64/mte/mte_helper.S b/tools/testing/selftests/arm64/mte/mte_helper.S
index a02c04cd0aac..a55dbbc56ed1 100644
--- a/tools/testing/selftests/arm64/mte/mte_helper.S
+++ b/tools/testing/selftests/arm64/mte/mte_helper.S
@@ -3,6 +3,8 @@
 
 #include "mte_def.h"
 
+.arch	armv8.5-a+memtag
+
 #define ENTRY(name) \
 	.globl name ;\
 	.p2align 2;\
-- 
2.30.2


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH] kselftest/arm64: Don't enable v8.5 for MTE selftest builds

[Catalin Marinas]

On Wed, 28 Sep 2022 16:45:17 +0100, Mark Brown wrote:
> Currently we set -march=armv8.5+memtag when building the MTE selftests,
> allowing the compiler to emit v8.5 and MTE instructions for anything it
> generates. This means that we may get code that will generate SIGILLs when
> run on older systems rather than skipping on non-MTE systems as should be
> the case. Most toolchains don't select any incompatible instructions but
> I have seen some reports which suggest that some may be appearing which do
> so. This is also potentially problematic in that if the compiler chooses to
> emit any MTE instructions for the C code it may interfere with the MTE
> usage we are trying to test.
> 
> [...]

Applied to arm64 (for-next/kselftest), thanks!

[1/1] kselftest/arm64: Don't enable v8.5 for MTE selftest builds
      https://git.kernel.org/arm64/c/55c8a987dd73

-- 
Catalin


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel