Re: [PATCH v22 28/28] kselftest/riscv: kselftest for user mode cfi

From: Deepak Gupta <debug@rivosinc.com>
To: Joel Stanley <joel@jms.id.au>
Cc: Paul Walmsley <paul.walmsley@sifive.com>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	Albert Ou <aou@eecs.berkeley.edu>,
	Conor Dooley <conor@kernel.org>, Arnd Bergmann <arnd@arndb.de>,
	Peter Zijlstra <peterz@infradead.org>,
	Kees Cook <kees@kernel.org>,
	linux-kernel@vger.kernel.org, linux-riscv@lists.infradead.org,
	linux-kselftest@vger.kernel.org, alistair.francis@wdc.com,
	richard.henderson@linaro.org, jim.shu@sifive.com,
	andybnac@gmail.com, kito.cheng@sifive.com, charlie@rivosinc.com,
	atishp@rivosinc.com, evan@rivosinc.com, cleger@rivosinc.com,
	alexghiti@rivosinc.com, Michael Ellerman <mpe@tenstorrent.com>
Subject: Re: [PATCH v22 28/28] kselftest/riscv: kselftest for user mode cfi
Date: Wed, 5 Nov 2025 13:44:32 -0800	[thread overview]
Message-ID: <aQvFQM38u0iwSlJ8@debug.ba.rivosinc.com> (raw)
In-Reply-To: <CACPK8Xcm2SeSStk2EBEcUoyiUc0=d6ce_MfnP=ce3z6MMXyuiw@mail.gmail.com>

On Tue, Nov 04, 2025 at 05:42:04PM +1030, Joel Stanley wrote:
>On Fri, 24 Oct 2025 at 03:33, Deepak Gupta via B4 Relay
><devnull+debug.rivosinc.com@kernel.org> wrote:
>
>> diff --git a/tools/testing/selftests/riscv/cfi/Makefile b/tools/testing/selftests/riscv/cfi/Makefile
>> new file mode 100644
>> index 000000000000..55165a93845f
>> --- /dev/null
>> +++ b/tools/testing/selftests/riscv/cfi/Makefile
>> @@ -0,0 +1,16 @@
>> +CFLAGS += -I$(top_srcdir)/tools/include
>> +
>> +CFLAGS += -march=rv64gc_zicfilp_zicfiss -fcf-protection=full
>> +
>> +ifeq ($(shell $(CC) $(CFLAGS) -nostdlib -xc /dev/null -o /dev/null > /dev/null 2>&1; echo $$?),0)
>> +TEST_GEN_PROGS := cfitests|
>
>I hacked this up a bit to get it working for me.
>
>I added this at the top to use the local headers from a `make
>headers`. Without this the selftest lacked the required ptrace
>changes:
>
>+CFLAGS += $(KHDR_INCLUDES)

hmm weird. In my case it picks it from location where headers get placed as part
of build (or make headers_install). I wonder why it's not happening in your
case. In any case, I think this can be added.

>
>I needed this before the ifeq so it used the cross compiler to perform
>the test, otherwise the test was always false:
>
>+# Check for zicfli* extensions needs cross compiler
>+# which is not set until lib.mk is included
>+ifeq ($(LLVM)$(CC),cc)
>+CC := $(CROSS_COMPILE)gcc
>+endif
>+#nedif
>
>Findally, a cosmetic one. If we set the name of the main test file to
>be the same as the test, the rule from lib.mk will give us neat
>output:

Sure, make sense. Will do that.
>
>     make[2]: Entering directory 'tools/testing/selftests/riscv/cfi'
>       CC       cfitests
>
>--- a/tools/testing/selftests/riscv/cfi/Makefile
>+++ b/tools/testing/selftests/riscv/cfi/Makefile
>@@ -12,12 +12,11 @@ endif
> ifeq ($(shell $(CC) $(CFLAGS) -nostdlib -xc /dev/null -o /dev/null >
>/dev/null 2>&1; echo $$?),0)
> TEST_GEN_PROGS := cfitests
>
>-include ../../lib.mk
>+$(OUTPUT)/cfitests: cfitests.c shadowstack.c
>
>-$(OUTPUT)/cfitests: riscv_cfi_test.c shadowstack.c
>-       $(CC) -o$@ $(CFLAGS) $(LDFLAGS) $^
> else
>-include ../../lib.mk
>
> $(shell echo "Toolchain doesn't support CFI, skipping CFI kselftest." >&2)
> endif
>+
>+include ../../lib.mk
>
>I can push up a branch with my changes if you'd like.

Sure, please point me to that and I'll pick it up. Test locally and send it out.