[PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions

public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed

* [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
@ 2024-09-20 15:44 Vitaly Kuznetsov
  2024-09-22 16:17 ` Sean Christopherson
                   ` (2 more replies)
  0 siblings, 3 replies; 6+ messages in thread
From: Vitaly Kuznetsov @ 2024-09-20 15:44 UTC (permalink / raw)
  To: kvm, Paolo Bonzini, Sean Christopherson; +Cc: Jan Richter, linux-kernel

Some distros switched gcc to '-march=x86-64-v3' by default and while it's
hard to find a CPU which doesn't support it today, many KVM selftests fail
with

  ==== Test Assertion Failure ====
    lib/x86_64/processor.c:570: Unhandled exception in guest
    pid=72747 tid=72747 errno=4 - Interrupted system call
    Unhandled exception '0x6' at guest RIP '0x4104f7'

The failure is easy to reproduce elsewhere with

   $ make clean && CFLAGS='-march=x86-64-v3' make -j && ./x86_64/kvm_pv_test

The root cause of the problem seems to be that with '-march=x86-64-v3' GCC
uses AVX* instructions (VMOVQ in the example above) and without prior
XSETBV() in the guest this results in #UD. It is certainly possible to add
it there, e.g. the following saves the day as well:

diff --git a/tools/testing/selftests/kvm/x86_64/kvm_pv_test.c b/tools/testing/selftests/kvm/x86_64/kvm_pv_test.c
index 78878b3a2725..704668adb3bd 100644
--- a/tools/testing/selftests/kvm/x86_64/kvm_pv_test.c
+++ b/tools/testing/selftests/kvm/x86_64/kvm_pv_test.c
@@ -82,8 +82,17 @@ static void test_hcall(struct hcall_data *hc)

 static void guest_main(void)
 {
+	uint64_t cr4, xcr0;
        int i;

+	cr4 = get_cr4();
+	cr4 |= X86_CR4_OSXSAVE;
+	set_cr4(cr4);
+
+	xcr0 = xgetbv(0);
+	xcr0 |= XFEATURE_MASK_SSE | XFEATURE_MASK_YMM;
+	xsetbv(0x0, xcr0);
+
        for (i = 0; i < ARRAY_SIZE(msrs_to_test); i++) {
                test_msr(&msrs_to_test[i]);
        }

but this needs to be made conditional depending on the compilation target
and added to all selftests. Slap a band-aid on the problem by forcing
'-march=x86-64-v2' in Makefile.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 tools/testing/selftests/kvm/Makefile | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
index 48d32c5aa3eb..3f1b24ed7245 100644
--- a/tools/testing/selftests/kvm/Makefile
+++ b/tools/testing/selftests/kvm/Makefile
@@ -238,6 +238,7 @@ CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
 	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
 	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
 	-I$(<D) -Iinclude/$(ARCH_DIR) -I ../rseq -I.. $(EXTRA_CFLAGS) \
+	-march=x86-64-v2 \
 	$(KHDR_INCLUDES)
 ifeq ($(ARCH),s390)
 	CFLAGS += -march=z10
-- 
2.46.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
  2024-09-20 15:44 [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions Vitaly Kuznetsov
@ 2024-09-22 16:17 ` Sean Christopherson
  2024-09-23 11:16 ` David Laight
  2024-10-21 18:32 ` Mark Brown
  2 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2024-09-22 16:17 UTC (permalink / raw)
  To: Vitaly Kuznetsov; +Cc: kvm, Paolo Bonzini, Jan Richter, linux-kernel

On Fri, Sep 20, 2024, Vitaly Kuznetsov wrote:
> diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
> index 48d32c5aa3eb..3f1b24ed7245 100644
> --- a/tools/testing/selftests/kvm/Makefile
> +++ b/tools/testing/selftests/kvm/Makefile
> @@ -238,6 +238,7 @@ CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
>  	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
>  	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
>  	-I$(<D) -Iinclude/$(ARCH_DIR) -I ../rseq -I.. $(EXTRA_CFLAGS) \
> +	-march=x86-64-v2 \

I would rather go straight to playing nice with AVX.  Not because I care about
being able to use AVX, but because pretty much every instance where KVM selftests
punts setup to individual tests eventually leads to gross copy+paste code.

The diff ends up being bigger than I was hoping, but that's largely because tests
are already manually enabling stuff in XCR0 (see above copy+paste complaint).

I can post the below later this week (probably as multiple patches).

Note, -march=x86-64-v3 is there just to make it easy to test, I won't actually
include that in the patches :-)

---
 tools/testing/selftests/kvm/Makefile          |  1 +
 .../selftests/kvm/include/x86_64/processor.h  |  5 ++++
 .../selftests/kvm/lib/x86_64/processor.c      | 24 +++++++++++++++++++
 tools/testing/selftests/kvm/x86_64/amx_test.c | 23 ++++--------------
 .../testing/selftests/kvm/x86_64/cpuid_test.c |  6 ++++-
 .../selftests/kvm/x86_64/sev_smoke_test.c     | 11 ---------
 .../testing/selftests/kvm/x86_64/state_test.c |  5 ----
 .../selftests/kvm/x86_64/xcr0_cpuid_test.c    | 11 ++++++---
 8 files changed, 47 insertions(+), 39 deletions(-)

diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
index 960cf6a77198..7ef4b3cc403d 100644
--- a/tools/testing/selftests/kvm/Makefile
+++ b/tools/testing/selftests/kvm/Makefile
@@ -238,6 +238,7 @@ else
 LINUX_TOOL_ARCH_INCLUDE = $(top_srcdir)/tools/arch/$(ARCH)/include
 endif
 CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
+	-march=x86-64-v3 \
 	-Wno-gnu-variable-sized-type-not-at-end -MD -MP -DCONFIG_64BIT \
 	-fno-builtin-memcmp -fno-builtin-memcpy \
 	-fno-builtin-memset -fno-builtin-strnlen \
diff --git a/tools/testing/selftests/kvm/include/x86_64/processor.h b/tools/testing/selftests/kvm/include/x86_64/processor.h
index e247f99e0473..645200e95f89 100644
--- a/tools/testing/selftests/kvm/include/x86_64/processor.h
+++ b/tools/testing/selftests/kvm/include/x86_64/processor.h
@@ -1049,6 +1049,11 @@ static inline void vcpu_set_cpuid(struct kvm_vcpu *vcpu)
 	vcpu_ioctl(vcpu, KVM_GET_CPUID2, vcpu->cpuid);
 }
 
+static inline void vcpu_get_cpuid(struct kvm_vcpu *vcpu)
+{
+	vcpu_ioctl(vcpu, KVM_GET_CPUID2, vcpu->cpuid);
+}
+
 void vcpu_set_cpuid_property(struct kvm_vcpu *vcpu,
 			     struct kvm_x86_cpu_property property,
 			     uint32_t value);
diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index 974bcd2df6d7..636b29ba8985 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -506,6 +506,8 @@ static void vcpu_init_sregs(struct kvm_vm *vm, struct kvm_vcpu *vcpu)
 
 	sregs.cr0 = X86_CR0_PE | X86_CR0_NE | X86_CR0_PG;
 	sregs.cr4 |= X86_CR4_PAE | X86_CR4_OSFXSR;
+	if (kvm_cpu_has(X86_FEATURE_XSAVE))
+		sregs.cr4 |= X86_CR4_OSXSAVE;
 	sregs.efer |= (EFER_LME | EFER_LMA | EFER_NX);
 
 	kvm_seg_set_unusable(&sregs.ldt);
@@ -519,6 +521,20 @@ static void vcpu_init_sregs(struct kvm_vm *vm, struct kvm_vcpu *vcpu)
 	vcpu_sregs_set(vcpu, &sregs);
 }
 
+static void vcpu_init_xcrs(struct kvm_vm *vm, struct kvm_vcpu *vcpu)
+{
+	struct kvm_xcrs xcrs = {
+		.nr_xcrs = 1,
+		.xcrs[0].xcr = 0,
+		.xcrs[0].value = kvm_cpu_supported_xcr0(),
+	};
+
+	if (!kvm_cpu_has(X86_FEATURE_XSAVE))
+		return;
+
+	vcpu_xcrs_set(vcpu, &xcrs);
+}
+
 static void set_idt_entry(struct kvm_vm *vm, int vector, unsigned long addr,
 			  int dpl, unsigned short selector)
 {
@@ -675,6 +691,7 @@ struct kvm_vcpu *vm_arch_vcpu_add(struct kvm_vm *vm, uint32_t vcpu_id)
 	vcpu = __vm_vcpu_add(vm, vcpu_id);
 	vcpu_init_cpuid(vcpu, kvm_get_supported_cpuid());
 	vcpu_init_sregs(vm, vcpu);
+	vcpu_init_xcrs(vm, vcpu);
 
 	/* Setup guest general purpose registers */
 	vcpu_regs_get(vcpu, &regs);
@@ -686,6 +703,13 @@ struct kvm_vcpu *vm_arch_vcpu_add(struct kvm_vm *vm, uint32_t vcpu_id)
 	mp_state.mp_state = 0;
 	vcpu_mp_state_set(vcpu, &mp_state);
 
+	/*
+	 * Refresh CPUID after setting SREGS and XCR0, so that KVM's "runtime"
+	 * updates to guest CPUID, e.g. for OSXSAVE and XSAVE state size, are
+	 * reflected into selftests' vCPU CPUID cache, i.e. so that the cache
+	 * is consistent with vCPU state.
+	 */
+	vcpu_get_cpuid(vcpu);
 	return vcpu;
 }
 
diff --git a/tools/testing/selftests/kvm/x86_64/amx_test.c b/tools/testing/selftests/kvm/x86_64/amx_test.c
index 903940c54d2d..f4ce5a185a7d 100644
--- a/tools/testing/selftests/kvm/x86_64/amx_test.c
+++ b/tools/testing/selftests/kvm/x86_64/amx_test.c
@@ -86,6 +86,8 @@ static inline void __xsavec(struct xstate *xstate, uint64_t rfbm)
 
 static void check_xtile_info(void)
 {
+	GUEST_ASSERT((xgetbv(0) & XFEATURE_MASK_XTILE) == XFEATURE_MASK_XTILE);
+
 	GUEST_ASSERT(this_cpu_has_p(X86_PROPERTY_XSTATE_MAX_SIZE_XCR0));
 	GUEST_ASSERT(this_cpu_property(X86_PROPERTY_XSTATE_MAX_SIZE_XCR0) <= XSAVE_SIZE);
 
@@ -122,29 +124,12 @@ static void set_tilecfg(struct tile_config *cfg)
 	}
 }
 
-static void init_regs(void)
-{
-	uint64_t cr4, xcr0;
-
-	GUEST_ASSERT(this_cpu_has(X86_FEATURE_XSAVE));
-
-	/* turn on CR4.OSXSAVE */
-	cr4 = get_cr4();
-	cr4 |= X86_CR4_OSXSAVE;
-	set_cr4(cr4);
-	GUEST_ASSERT(this_cpu_has(X86_FEATURE_OSXSAVE));
-
-	xcr0 = xgetbv(0);
-	xcr0 |= XFEATURE_MASK_XTILE;
-	xsetbv(0x0, xcr0);
-	GUEST_ASSERT((xgetbv(0) & XFEATURE_MASK_XTILE) == XFEATURE_MASK_XTILE);
-}
-
 static void __attribute__((__flatten__)) guest_code(struct tile_config *amx_cfg,
 						    struct tile_data *tiledata,
 						    struct xstate *xstate)
 {
-	init_regs();
+	GUEST_ASSERT(this_cpu_has(X86_FEATURE_XSAVE) &&
+		     this_cpu_has(X86_FEATURE_OSXSAVE));
 	check_xtile_info();
 	GUEST_SYNC(1);
 
diff --git a/tools/testing/selftests/kvm/x86_64/cpuid_test.c b/tools/testing/selftests/kvm/x86_64/cpuid_test.c
index 8c579ce714e9..e79a6577254f 100644
--- a/tools/testing/selftests/kvm/x86_64/cpuid_test.c
+++ b/tools/testing/selftests/kvm/x86_64/cpuid_test.c
@@ -37,7 +37,11 @@ static void test_guest_cpuids(struct kvm_cpuid2 *guest_cpuid)
 
 		GUEST_ASSERT_EQ(eax, guest_cpuid->entries[i].eax);
 		GUEST_ASSERT_EQ(ebx, guest_cpuid->entries[i].ebx);
-		GUEST_ASSERT_EQ(ecx, guest_cpuid->entries[i].ecx);
+		__GUEST_ASSERT(ecx == guest_cpuid->entries[i].ecx,
+			       "CPUID.0x%x.0x%x.ECX 0%x != 0x%x",
+			       guest_cpuid->entries[i].function,
+			       guest_cpuid->entries[i].index,
+			       ecx, guest_cpuid->entries[i].ecx);
 		GUEST_ASSERT_EQ(edx, guest_cpuid->entries[i].edx);
 	}
 
diff --git a/tools/testing/selftests/kvm/x86_64/sev_smoke_test.c b/tools/testing/selftests/kvm/x86_64/sev_smoke_test.c
index 2e9197eb1652..59a5a2227944 100644
--- a/tools/testing/selftests/kvm/x86_64/sev_smoke_test.c
+++ b/tools/testing/selftests/kvm/x86_64/sev_smoke_test.c
@@ -70,12 +70,6 @@ static void test_sync_vmsa(uint32_t policy)
 
 	double x87val = M_PI;
 	struct kvm_xsave __attribute__((aligned(64))) xsave = { 0 };
-	struct kvm_sregs sregs;
-	struct kvm_xcrs xcrs = {
-		.nr_xcrs = 1,
-		.xcrs[0].xcr = 0,
-		.xcrs[0].value = XFEATURE_MASK_X87_AVX,
-	};
 
 	vm = vm_sev_create_with_one_vcpu(KVM_X86_SEV_ES_VM, guest_code_xsave, &vcpu);
 	gva = vm_vaddr_alloc_shared(vm, PAGE_SIZE, KVM_UTIL_MIN_VADDR,
@@ -84,11 +78,6 @@ static void test_sync_vmsa(uint32_t policy)
 
 	vcpu_args_set(vcpu, 1, gva);
 
-	vcpu_sregs_get(vcpu, &sregs);
-	sregs.cr4 |= X86_CR4_OSFXSR | X86_CR4_OSXSAVE;
-	vcpu_sregs_set(vcpu, &sregs);
-
-	vcpu_xcrs_set(vcpu, &xcrs);
 	asm("fninit\n"
 	    "vpcmpeqb %%ymm4, %%ymm4, %%ymm4\n"
 	    "fldl %3\n"
diff --git a/tools/testing/selftests/kvm/x86_64/state_test.c b/tools/testing/selftests/kvm/x86_64/state_test.c
index 1c756db329e5..141b7fc0c965 100644
--- a/tools/testing/selftests/kvm/x86_64/state_test.c
+++ b/tools/testing/selftests/kvm/x86_64/state_test.c
@@ -145,11 +145,6 @@ static void __attribute__((__flatten__)) guest_code(void *arg)
 
 		memset(buffer, 0xcc, sizeof(buffer));
 
-		set_cr4(get_cr4() | X86_CR4_OSXSAVE);
-		GUEST_ASSERT(this_cpu_has(X86_FEATURE_OSXSAVE));
-
-		xsetbv(0, xgetbv(0) | supported_xcr0);
-
 		/*
 		 * Modify state for all supported xfeatures to take them out of
 		 * their "init" state, i.e. to make them show up in XSTATE_BV.
diff --git a/tools/testing/selftests/kvm/x86_64/xcr0_cpuid_test.c b/tools/testing/selftests/kvm/x86_64/xcr0_cpuid_test.c
index 95ce192d0753..c8a5c5e51661 100644
--- a/tools/testing/selftests/kvm/x86_64/xcr0_cpuid_test.c
+++ b/tools/testing/selftests/kvm/x86_64/xcr0_cpuid_test.c
@@ -48,16 +48,16 @@ do {									\
 
 static void guest_code(void)
 {
-	uint64_t xcr0_reset;
+	uint64_t initial_xcr0;
 	uint64_t supported_xcr0;
 	int i, vector;
 
 	set_cr4(get_cr4() | X86_CR4_OSXSAVE);
 
-	xcr0_reset = xgetbv(0);
+	initial_xcr0 = xgetbv(0);
 	supported_xcr0 = this_cpu_supported_xcr0();
 
-	GUEST_ASSERT(xcr0_reset == XFEATURE_MASK_FP);
+	GUEST_ASSERT(initial_xcr0 == supported_xcr0);
 
 	/* Check AVX */
 	ASSERT_XFEATURE_DEPENDENCIES(supported_xcr0,
@@ -79,6 +79,11 @@ static void guest_code(void)
 	ASSERT_ALL_OR_NONE_XFEATURE(supported_xcr0,
 				    XFEATURE_MASK_XTILE);
 
+	vector = xsetbv_safe(0, XFEATURE_MASK_FP);
+	__GUEST_ASSERT(!vector,
+		       "Expected success on XSETBV(FP), got vector '0x%x'",
+		       vector);
+
 	vector = xsetbv_safe(0, supported_xcr0);
 	__GUEST_ASSERT(!vector,
 		       "Expected success on XSETBV(0x%lx), got vector '0x%x'",

base-commit: 3f8df6285271d9d8f17d733433e5213a63b83a0b
-- 


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* RE: [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
  2024-09-20 15:44 [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions Vitaly Kuznetsov
  2024-09-22 16:17 ` Sean Christopherson
@ 2024-09-23 11:16 ` David Laight
  2024-09-24 12:04   ` Vitaly Kuznetsov
  2024-10-21 18:32 ` Mark Brown
  2 siblings, 1 reply; 6+ messages in thread
From: David Laight @ 2024-09-23 11:16 UTC (permalink / raw)
  To: 'Vitaly Kuznetsov', kvm@vger.kernel.org, Paolo Bonzini,
	Sean Christopherson
  Cc: Jan Richter, linux-kernel@vger.kernel.org

From: Vitaly Kuznetsov <vkuznets@redhat.com>
> Sent: 20 September 2024 16:44
> 
> Some distros switched gcc to '-march=x86-64-v3' by default and while it's
> hard to find a CPU which doesn't support it today,

I didn't think that any of the Atom based cpu supported AVX.
I'm pretty sure one we use that are still in production as
server motherboards don't support it.

Doesn't -v3 also require support for the VEX encoding.
Which removes a lot of perfectly reasonable cpu?

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


^ permalink raw reply	[flat|nested] 6+ messages in thread

* RE: [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
  2024-09-23 11:16 ` David Laight
@ 2024-09-24 12:04   ` Vitaly Kuznetsov
  0 siblings, 0 replies; 6+ messages in thread
From: Vitaly Kuznetsov @ 2024-09-24 12:04 UTC (permalink / raw)
  To: David Laight
  Cc: Jan Richter, linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
	Paolo Bonzini, Sean Christopherson

David Laight <David.Laight@ACULAB.COM> writes:

> From: Vitaly Kuznetsov <vkuznets@redhat.com>
>> Sent: 20 September 2024 16:44
>> 
>> Some distros switched gcc to '-march=x86-64-v3' by default and while it's
>> hard to find a CPU which doesn't support it today,
>
> I didn't think that any of the Atom based cpu supported AVX.
> I'm pretty sure one we use that are still in production as
> server motherboards don't support it.
>
> Doesn't -v3 also require support for the VEX encoding.
> Which removes a lot of perfectly reasonable cpu?
>

Well, distros making such decision are obviously not very interested in
running on these CPUs then :-) In this particular case, the distro was
Centos Stream 10 and they are very explicit about the decision:
https://developers.redhat.com/articles/2024/01/02/exploring-x86-64-v3-red-hat-enterprise-linux-10#new_cpu_capabilities_in_x86_64_v3

and 'gcc -v' tells me

Configured with: ../configure ... --with-arch_64=x86-64-v3  ...
...
gcc version 14.2.1 20240801 (Red Hat 14.2.1-1) (GCC) 

-- 
Vitaly


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
  2024-09-20 15:44 [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions Vitaly Kuznetsov
  2024-09-22 16:17 ` Sean Christopherson
  2024-09-23 11:16 ` David Laight
@ 2024-10-21 18:32 ` Mark Brown
  2024-10-21 20:33   ` Mark Brown
  2 siblings, 1 reply; 6+ messages in thread
From: Mark Brown @ 2024-10-21 18:32 UTC (permalink / raw)
  To: Vitaly Kuznetsov
  Cc: kvm, Paolo Bonzini, Sean Christopherson, Jan Richter,
	linux-kernel, Aishwarya.TCV

[-- Attachment #1: Type: text/plain, Size: 2089 bytes --]

On Fri, Sep 20, 2024 at 05:44:22PM +0200, Vitaly Kuznetsov wrote:

> Some distros switched gcc to '-march=x86-64-v3' by default and while it's
> hard to find a CPU which doesn't support it today, many KVM selftests fail
> with

This patch, which is queued in -next as 9a400068a1586bc4 targeted as a
fix, breaks the build on non-x86 architectures:

aarch64-linux-gnu-gcc -D_GNU_SOURCE=  -Wall -Wstrict-prototypes -Wuninitialized 
-O2 -g -std=gnu99 -Wno-gnu-variable-sized-type-not-at-end -MD -MP -DCONFIG_64BIT
 -fno-builtin-memcmp -fno-builtin-memcpy -fno-builtin-memset -fno-builtin-strnlen -fno-stack-protector -fno-PIE -I/build/stage/linux/tools/testing/selftests/../../../tools/include -I/build/stage/linux/tools/testing/selftests/../../../tools/arch/arm64/include -I/build/stage/linux/tools/testing/selftests/../../../usr/include/ -Iinclude -Iaarch64 -Iinclude/aarch64 -I ../rseq -I..  -march=x86-64-v2 -isystem /build/stage/build-work/usr/include -I/build/stage/linux/tools/testing/selftests/../../../tools/arch/arm64/include/generated/   -c aarch64/aarch32_id_regs.c -o /build/stage/build-work/kselftest/kvm/aarch64/aarch32_id_regs.o
cc1: error: unknown value ‘x86-64-v2’ for ‘-march’

This is because:

> diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
> index 48d32c5aa3eb..3f1b24ed7245 100644
> --- a/tools/testing/selftests/kvm/Makefile
> +++ b/tools/testing/selftests/kvm/Makefile
> @@ -238,6 +238,7 @@ CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
>  	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
>  	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
>  	-I$(<D) -Iinclude/$(ARCH_DIR) -I ../rseq -I.. $(EXTRA_CFLAGS) \
> +	-march=x86-64-v2 \
>  	$(KHDR_INCLUDES)
>  ifeq ($(ARCH),s390)
>  	CFLAGS += -march=z10

unconditionally sets an architecture specific flag which is obviously
not going to work on anything except x86.  This should be set under an
architecture check like the similar S/390 flag that can be seen in the
context for the diff.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions
  2024-10-21 18:32 ` Mark Brown
@ 2024-10-21 20:33   ` Mark Brown
  0 siblings, 0 replies; 6+ messages in thread
From: Mark Brown @ 2024-10-21 20:33 UTC (permalink / raw)
  To: Vitaly Kuznetsov
  Cc: kvm, Paolo Bonzini, Sean Christopherson, Jan Richter,
	linux-kernel, Aishwarya.TCV, torvalds

[-- Attachment #1: Type: text/plain, Size: 2331 bytes --]

On Mon, Oct 21, 2024 at 07:32:17PM +0100, Mark Brown wrote:
> On Fri, Sep 20, 2024 at 05:44:22PM +0200, Vitaly Kuznetsov wrote:
> 
> > Some distros switched gcc to '-march=x86-64-v3' by default and while it's
> > hard to find a CPU which doesn't support it today, many KVM selftests fail
> > with
> 
> This patch, which is queued in -next as 9a400068a1586bc4 targeted as a
> fix, breaks the build on non-x86 architectures:

This patch is now in Linus' tree, having been applied on Sunday and as a
result appeared in -next today.

> 
> aarch64-linux-gnu-gcc -D_GNU_SOURCE=  -Wall -Wstrict-prototypes -Wuninitialized 
> -O2 -g -std=gnu99 -Wno-gnu-variable-sized-type-not-at-end -MD -MP -DCONFIG_64BIT
>  -fno-builtin-memcmp -fno-builtin-memcpy -fno-builtin-memset -fno-builtin-strnlen -fno-stack-protector -fno-PIE -I/build/stage/linux/tools/testing/selftests/../../../tools/include -I/build/stage/linux/tools/testing/selftests/../../../tools/arch/arm64/include -I/build/stage/linux/tools/testing/selftests/../../../usr/include/ -Iinclude -Iaarch64 -Iinclude/aarch64 -I ../rseq -I..  -march=x86-64-v2 -isystem /build/stage/build-work/usr/include -I/build/stage/linux/tools/testing/selftests/../../../tools/arch/arm64/include/generated/   -c aarch64/aarch32_id_regs.c -o /build/stage/build-work/kselftest/kvm/aarch64/aarch32_id_regs.o
> cc1: error: unknown value ‘x86-64-v2’ for ‘-march’
> 
> This is because:
> 
> > diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
> > index 48d32c5aa3eb..3f1b24ed7245 100644
> > --- a/tools/testing/selftests/kvm/Makefile
> > +++ b/tools/testing/selftests/kvm/Makefile
> > @@ -238,6 +238,7 @@ CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
> >  	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
> >  	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
> >  	-I$(<D) -Iinclude/$(ARCH_DIR) -I ../rseq -I.. $(EXTRA_CFLAGS) \
> > +	-march=x86-64-v2 \
> >  	$(KHDR_INCLUDES)
> >  ifeq ($(ARCH),s390)
> >  	CFLAGS += -march=z10
> 
> unconditionally sets an architecture specific flag which is obviously
> not going to work on anything except x86.  This should be set under an
> architecture check like the similar S/390 flag that can be seen in the
> context for the diff.



[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2024-10-21 20:33 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-09-20 15:44 [PATCH] KVM: selftests: x86: Avoid using SSE/AVX instructions Vitaly Kuznetsov
2024-09-22 16:17 ` Sean Christopherson
2024-09-23 11:16 ` David Laight
2024-09-24 12:04   ` Vitaly Kuznetsov
2024-10-21 18:32 ` Mark Brown
2024-10-21 20:33   ` Mark Brown

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox