* [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
@ 2026-03-12 8:01 Hari Bathini
2026-03-12 18:22 ` Yonghong Song
2026-03-13 16:22 ` Mykyta Yatsenko
0 siblings, 2 replies; 6+ messages in thread
From: Hari Bathini @ 2026-03-12 8:01 UTC (permalink / raw)
To: bpf
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, Shuah Khan,
linux-kselftest
On powerpc, immediate load instructions are sign extended. In case
of unsigned types, arguments should be explicitly zero-extended by
the caller. For kfunc call, this needs to be handled in the JIT code.
In bpf_kfunc_call_test4(), that tests for sign-extension of signed
argument types in kfunc calls, add some additional failure checks.
And add bpf_kfunc_call_test5() to test zero-extension of unsigned
argument types in kfunc calls.
Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
---
Changes in v2:
- Added asm version of the selftest for consistent testing across
different BPF ISA versions.
- Added comments clearly stating the intent of the test cases.
- Updated sign-extension selftest to have additional failure checks.
.../selftests/bpf/prog_tests/kfunc_call.c | 2 +
.../selftests/bpf/progs/kfunc_call_test.c | 98 +++++++++++++++++++
.../selftests/bpf/test_kmods/bpf_testmod.c | 54 +++++++++-
.../bpf/test_kmods/bpf_testmod_kfunc.h | 1 +
4 files changed, 154 insertions(+), 1 deletion(-)
diff --git a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
index f79c8e53cb3e..62f3fb79f5d1 100644
--- a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
+++ b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
@@ -74,6 +74,8 @@ static struct kfunc_test_params kfunc_tests[] = {
TC_TEST(kfunc_call_test1, 12),
TC_TEST(kfunc_call_test2, 3),
TC_TEST(kfunc_call_test4, -1234),
+ TC_TEST(kfunc_call_test5, 0),
+ TC_TEST(kfunc_call_test5_asm, 0),
TC_TEST(kfunc_call_test_ref_btf_id, 0),
TC_TEST(kfunc_call_test_get_mem, 42),
SYSCALL_TEST(kfunc_syscall_test, 0),
diff --git a/tools/testing/selftests/bpf/progs/kfunc_call_test.c b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
index 8b86113a0126..5edc51564f71 100644
--- a/tools/testing/selftests/bpf/progs/kfunc_call_test.c
+++ b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
@@ -2,8 +2,106 @@
/* Copyright (c) 2021 Facebook */
#include <vmlinux.h>
#include <bpf/bpf_helpers.h>
+#include "bpf_misc.h"
#include "../test_kmods/bpf_testmod_kfunc.h"
+SEC("tc")
+int kfunc_call_test5(struct __sk_buff *skb)
+{
+ struct bpf_sock *sk = skb->sk;
+ int ret;
+ u32 val32;
+ u16 val16;
+ u8 val8;
+
+ if (!sk)
+ return -1;
+
+ sk = bpf_sk_fullsock(sk);
+ if (!sk)
+ return -1;
+
+ /*
+ * Test with constant values to verify zero-extension.
+ * ISA-dependent BPF asm:
+ * With ALU32: w1 = 0xFF; w2 = 0xFFFF; w3 = 0xFFFFffff
+ * Without ALU32: r1 = 0xFF; r2 = 0xFFFF; r3 = 0xFFFFffff
+ * Both zero-extend to 64-bit before the kfunc call.
+ */
+ ret = bpf_kfunc_call_test5(0xFF, 0xFFFF, 0xFFFFffffULL);
+ if (ret)
+ return ret;
+
+ val32 = bpf_get_prandom_u32();
+ val16 = val32 & 0xFFFF;
+ val8 = val32 & 0xFF;
+ ret = bpf_kfunc_call_test5(val8, val16, val32);
+ if (ret)
+ return ret;
+
+ /*
+ * Test multiplication with different operand sizes:
+ *
+ * val8 * 0xFF:
+ * - Both operands promote to int (32-bit signed)
+ * - Result: 32-bit multiplication, truncated to u8, then zero-extended
+ *
+ * val16 * 0xFFFF:
+ * - Both operands promote to int (32-bit signed)
+ * - Result: 32-bit multiplication, truncated to u16, then zero-extended
+ *
+ * val32 * 0xFFFFffffULL:
+ * - val32 (u32) promotes to unsigned long long (due to ULL suffix)
+ * - Result: 64-bit unsigned multiplication, truncated to u32, then zero-extended
+ */
+ ret = bpf_kfunc_call_test5(val8 * 0xFF, val16 * 0xFFFF, val32 * 0xFFFFffffULL);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+/*
+ * Assembly version testing the multiplication edge case explicitly.
+ * This ensures consistent testing across different ISA versions.
+ */
+SEC("tc")
+__naked int kfunc_call_test5_asm(void)
+{
+ asm volatile (
+ /* Get a random u32 value */
+ "call %[bpf_get_prandom_u32];"
+ "r6 = r0;" /* Save val32 in r6 */
+
+ /* Prepare first argument: val8 * 0xFF */
+ "r1 = r6;"
+ "r1 &= 0xFF;" /* val8 = val32 & 0xFF */
+ "r7 = 0xFF;"
+ "r1 *= r7;" /* 64-bit mult: r1 = r1 * r7 */
+
+ /* Prepare second argument: val16 * 0xFFFF */
+ "r2 = r6;"
+ "r2 &= 0xFFFF;" /* val16 = val32 & 0xFFFF */
+ "r7 = 0xFFFF;"
+ "r2 *= r7;" /* 64-bit mult: r2 = r2 * r7 */
+
+ /* Prepare third argument: val32 * 0xFFFFffff */
+ "r3 = r6;" /* val32 */
+ "r7 = 0xFFFFffff;"
+ "r3 *= r7;" /* 64-bit mult: r3 = r3 * r7 */
+
+ /* Call kfunc with multiplication results */
+ "call bpf_kfunc_call_test5;"
+
+ /* Check return value */
+ "if r0 != 0 goto exit_%=;"
+ "r0 = 0;"
+ "exit_%=: exit;"
+ :
+ : __imm(bpf_get_prandom_u32)
+ : __clobber_all);
+}
+
SEC("tc")
int kfunc_call_test4(struct __sk_buff *skb)
{
diff --git a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
index e62c6b78657f..94edbd2afa67 100644
--- a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
+++ b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
@@ -760,12 +760,63 @@ __bpf_kfunc struct sock *bpf_kfunc_call_test3(struct sock *sk)
__bpf_kfunc long noinline bpf_kfunc_call_test4(signed char a, short b, int c, long d)
{
- /* Provoke the compiler to assume that the caller has sign-extended a,
+ /*
+ * Make val as volatile to avoid compiler optimizations.
+ * Verify that negative signed values remain negative after
+ * sign-extension (JIT must sign-extend, not zero-extend).
+ */
+ volatile long val;
+
+ /* val will be positive, if JIT does zero-extension instead of sign-extension */
+ val = a;
+ if (val >= 0)
+ return 1;
+
+ val = b;
+ if (val >= 0)
+ return 2;
+
+ val = c;
+ if (val >= 0)
+ return 3;
+
+ /*
+ * Provoke the compiler to assume that the caller has sign-extended a,
* b and c on platforms where this is required (e.g. s390x).
*/
return (long)a + (long)b + (long)c + d;
}
+__bpf_kfunc int bpf_kfunc_call_test5(u8 a, u16 b, u32 c)
+{
+ /*
+ * Make val as volatile to avoid compiler optimizations on the below checks
+ * In C, assigning u8/u16/u32 to long performs zero-extension.
+ */
+ volatile long val = a;
+
+ /* Check zero-extension */
+ if (val != (unsigned long)a)
+ return 1;
+ /* Check no sign-extension */
+ if (val < 0)
+ return 2;
+
+ val = b;
+ if (val != (unsigned long)b)
+ return 3;
+ if (val < 0)
+ return 4;
+
+ val = c;
+ if (val != (unsigned long)c)
+ return 5;
+ if (val < 0)
+ return 6;
+
+ return 0;
+}
+
static struct prog_test_ref_kfunc prog_test_struct = {
.a = 42,
.b = 108,
@@ -1228,6 +1279,7 @@ BTF_ID_FLAGS(func, bpf_kfunc_call_test1)
BTF_ID_FLAGS(func, bpf_kfunc_call_test2)
BTF_ID_FLAGS(func, bpf_kfunc_call_test3)
BTF_ID_FLAGS(func, bpf_kfunc_call_test4)
+BTF_ID_FLAGS(func, bpf_kfunc_call_test5)
BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_pass1)
BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail1)
BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail2)
diff --git a/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h b/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
index b393bf771131..aa0b8d41e71b 100644
--- a/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
+++ b/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
@@ -110,6 +110,7 @@ __u64 bpf_kfunc_call_test1(struct sock *sk, __u32 a, __u64 b,
int bpf_kfunc_call_test2(struct sock *sk, __u32 a, __u32 b) __ksym;
struct sock *bpf_kfunc_call_test3(struct sock *sk) __ksym;
long bpf_kfunc_call_test4(signed char a, short b, int c, long d) __ksym;
+int bpf_kfunc_call_test5(__u8 a, __u16 b, __u32 c) __ksym;
void bpf_kfunc_call_test_pass_ctx(struct __sk_buff *skb) __ksym;
void bpf_kfunc_call_test_pass1(struct prog_test_pass1 *p) __ksym;
--
2.53.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
2026-03-12 8:01 [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call Hari Bathini
@ 2026-03-12 18:22 ` Yonghong Song
2026-03-13 9:29 ` Hari Bathini
2026-03-13 16:22 ` Mykyta Yatsenko
1 sibling, 1 reply; 6+ messages in thread
From: Yonghong Song @ 2026-03-12 18:22 UTC (permalink / raw)
To: Hari Bathini, bpf
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, Shuah Khan,
linux-kselftest
On 3/12/26 1:01 AM, Hari Bathini wrote:
> On powerpc, immediate load instructions are sign extended. In case
> of unsigned types, arguments should be explicitly zero-extended by
> the caller. For kfunc call, this needs to be handled in the JIT code.
> In bpf_kfunc_call_test4(), that tests for sign-extension of signed
> argument types in kfunc calls, add some additional failure checks.
> And add bpf_kfunc_call_test5() to test zero-extension of unsigned
> argument types in kfunc calls.
>
> Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
LGTM with a nit below.
Acked-by: Yonghong Song <yonghong.song@linux.dev>
> ---
>
> Changes in v2:
> - Added asm version of the selftest for consistent testing across
> different BPF ISA versions.
> - Added comments clearly stating the intent of the test cases.
> - Updated sign-extension selftest to have additional failure checks.
>
>
> .../selftests/bpf/prog_tests/kfunc_call.c | 2 +
> .../selftests/bpf/progs/kfunc_call_test.c | 98 +++++++++++++++++++
> .../selftests/bpf/test_kmods/bpf_testmod.c | 54 +++++++++-
> .../bpf/test_kmods/bpf_testmod_kfunc.h | 1 +
> 4 files changed, 154 insertions(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> index f79c8e53cb3e..62f3fb79f5d1 100644
> --- a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> +++ b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> @@ -74,6 +74,8 @@ static struct kfunc_test_params kfunc_tests[] = {
> TC_TEST(kfunc_call_test1, 12),
> TC_TEST(kfunc_call_test2, 3),
> TC_TEST(kfunc_call_test4, -1234),
> + TC_TEST(kfunc_call_test5, 0),
> + TC_TEST(kfunc_call_test5_asm, 0),
> TC_TEST(kfunc_call_test_ref_btf_id, 0),
> TC_TEST(kfunc_call_test_get_mem, 42),
> SYSCALL_TEST(kfunc_syscall_test, 0),
> diff --git a/tools/testing/selftests/bpf/progs/kfunc_call_test.c b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> index 8b86113a0126..5edc51564f71 100644
> --- a/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> +++ b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> @@ -2,8 +2,106 @@
> /* Copyright (c) 2021 Facebook */
> #include <vmlinux.h>
> #include <bpf/bpf_helpers.h>
> +#include "bpf_misc.h"
> #include "../test_kmods/bpf_testmod_kfunc.h"
>
> +SEC("tc")
> +int kfunc_call_test5(struct __sk_buff *skb)
> +{
> + struct bpf_sock *sk = skb->sk;
> + int ret;
> + u32 val32;
> + u16 val16;
> + u8 val8;
> +
> + if (!sk)
> + return -1;
> +
> + sk = bpf_sk_fullsock(sk);
> + if (!sk)
> + return -1;
> +
> + /*
> + * Test with constant values to verify zero-extension.
> + * ISA-dependent BPF asm:
> + * With ALU32: w1 = 0xFF; w2 = 0xFFFF; w3 = 0xFFFFffff
> + * Without ALU32: r1 = 0xFF; r2 = 0xFFFF; r3 = 0xFFFFffff
> + * Both zero-extend to 64-bit before the kfunc call.
> + */
> + ret = bpf_kfunc_call_test5(0xFF, 0xFFFF, 0xFFFFffffULL);
Can we just use 0xFFFFffff instead of 0xFFFFffffULL?
> + if (ret)
> + return ret;
> +
> + val32 = bpf_get_prandom_u32();
> + val16 = val32 & 0xFFFF;
> + val8 = val32 & 0xFF;
> + ret = bpf_kfunc_call_test5(val8, val16, val32);
> + if (ret)
> + return ret;
> +
> + /*
> + * Test multiplication with different operand sizes:
> + *
> + * val8 * 0xFF:
> + * - Both operands promote to int (32-bit signed)
> + * - Result: 32-bit multiplication, truncated to u8, then zero-extended
> + *
> + * val16 * 0xFFFF:
> + * - Both operands promote to int (32-bit signed)
> + * - Result: 32-bit multiplication, truncated to u16, then zero-extended
> + *
> + * val32 * 0xFFFFffffULL:
> + * - val32 (u32) promotes to unsigned long long (due to ULL suffix)
> + * - Result: 64-bit unsigned multiplication, truncated to u32, then zero-extended
> + */
> + ret = bpf_kfunc_call_test5(val8 * 0xFF, val16 * 0xFFFF, val32 * 0xFFFFffffULL);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
[...]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
2026-03-12 18:22 ` Yonghong Song
@ 2026-03-13 9:29 ` Hari Bathini
2026-03-13 14:20 ` Alexei Starovoitov
0 siblings, 1 reply; 6+ messages in thread
From: Hari Bathini @ 2026-03-13 9:29 UTC (permalink / raw)
To: Yonghong Song, bpf
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, Shuah Khan,
linux-kselftest
On 12/03/26 11:52 pm, Yonghong Song wrote:
>
>
> On 3/12/26 1:01 AM, Hari Bathini wrote:
>> On powerpc, immediate load instructions are sign extended. In case
>> of unsigned types, arguments should be explicitly zero-extended by
>> the caller. For kfunc call, this needs to be handled in the JIT code.
>> In bpf_kfunc_call_test4(), that tests for sign-extension of signed
>> argument types in kfunc calls, add some additional failure checks.
>> And add bpf_kfunc_call_test5() to test zero-extension of unsigned
>> argument types in kfunc calls.
>>
>> Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
>
> LGTM with a nit below.
>
> Acked-by: Yonghong Song <yonghong.song@linux.dev>
Thanks for the review, Yonghong.
>
>> ---
>>
>> Changes in v2:
>> - Added asm version of the selftest for consistent testing across
>> different BPF ISA versions.
>> - Added comments clearly stating the intent of the test cases.
>> - Updated sign-extension selftest to have additional failure checks.
>>
>>
>> .../selftests/bpf/prog_tests/kfunc_call.c | 2 +
>> .../selftests/bpf/progs/kfunc_call_test.c | 98 +++++++++++++++++++
>> .../selftests/bpf/test_kmods/bpf_testmod.c | 54 +++++++++-
>> .../bpf/test_kmods/bpf_testmod_kfunc.h | 1 +
>> 4 files changed, 154 insertions(+), 1 deletion(-)
>>
>> diff --git a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c b/
>> tools/testing/selftests/bpf/prog_tests/kfunc_call.c
>> index f79c8e53cb3e..62f3fb79f5d1 100644
>> --- a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
>> +++ b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
>> @@ -74,6 +74,8 @@ static struct kfunc_test_params kfunc_tests[] = {
>> TC_TEST(kfunc_call_test1, 12),
>> TC_TEST(kfunc_call_test2, 3),
>> TC_TEST(kfunc_call_test4, -1234),
>> + TC_TEST(kfunc_call_test5, 0),
>> + TC_TEST(kfunc_call_test5_asm, 0),
>> TC_TEST(kfunc_call_test_ref_btf_id, 0),
>> TC_TEST(kfunc_call_test_get_mem, 42),
>> SYSCALL_TEST(kfunc_syscall_test, 0),
>> diff --git a/tools/testing/selftests/bpf/progs/kfunc_call_test.c b/
>> tools/testing/selftests/bpf/progs/kfunc_call_test.c
>> index 8b86113a0126..5edc51564f71 100644
>> --- a/tools/testing/selftests/bpf/progs/kfunc_call_test.c
>> +++ b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
>> @@ -2,8 +2,106 @@
>> /* Copyright (c) 2021 Facebook */
>> #include <vmlinux.h>
>> #include <bpf/bpf_helpers.h>
>> +#include "bpf_misc.h"
>> #include "../test_kmods/bpf_testmod_kfunc.h"
>> +SEC("tc")
>> +int kfunc_call_test5(struct __sk_buff *skb)
>> +{
>> + struct bpf_sock *sk = skb->sk;
>> + int ret;
>> + u32 val32;
>> + u16 val16;
>> + u8 val8;
>> +
>> + if (!sk)
>> + return -1;
>> +
>> + sk = bpf_sk_fullsock(sk);
>> + if (!sk)
>> + return -1;
>> +
>> + /*
>> + * Test with constant values to verify zero-extension.
>> + * ISA-dependent BPF asm:
>> + * With ALU32: w1 = 0xFF; w2 = 0xFFFF; w3 = 0xFFFFffff
>> + * Without ALU32: r1 = 0xFF; r2 = 0xFFFF; r3 = 0xFFFFffff
>> + * Both zero-extend to 64-bit before the kfunc call.
>> + */
>> + ret = bpf_kfunc_call_test5(0xFF, 0xFFFF, 0xFFFFffffULL);
>
> Can we just use 0xFFFFffff instead of 0xFFFFffffULL?
Alexei, can you confirm if I need to respin with this change?
- Hari
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
2026-03-13 9:29 ` Hari Bathini
@ 2026-03-13 14:20 ` Alexei Starovoitov
2026-03-13 16:51 ` Ilya Leoshkevich
0 siblings, 1 reply; 6+ messages in thread
From: Alexei Starovoitov @ 2026-03-13 14:20 UTC (permalink / raw)
To: Hari Bathini, Ilya Leoshkevich
Cc: Yonghong Song, bpf, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, Shuah Khan, open list:KERNEL SELFTEST FRAMEWORK
On Fri, Mar 13, 2026 at 2:30 AM Hari Bathini <hbathini@linux.ibm.com> wrote:
>
>
>
> On 12/03/26 11:52 pm, Yonghong Song wrote:
> >
> >
> > On 3/12/26 1:01 AM, Hari Bathini wrote:
> >> On powerpc, immediate load instructions are sign extended. In case
> >> of unsigned types, arguments should be explicitly zero-extended by
> >> the caller. For kfunc call, this needs to be handled in the JIT code.
> >> In bpf_kfunc_call_test4(), that tests for sign-extension of signed
> >> argument types in kfunc calls, add some additional failure checks.
> >> And add bpf_kfunc_call_test5() to test zero-extension of unsigned
> >> argument types in kfunc calls.
> >>
> >> Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
> >
> > LGTM with a nit below.
> >
> > Acked-by: Yonghong Song <yonghong.song@linux.dev>
>
> Thanks for the review, Yonghong.
>
> >
> >> ---
> >>
> >> Changes in v2:
> >> - Added asm version of the selftest for consistent testing across
> >> different BPF ISA versions.
> >> - Added comments clearly stating the intent of the test cases.
> >> - Updated sign-extension selftest to have additional failure checks.
> >>
> >>
> >> .../selftests/bpf/prog_tests/kfunc_call.c | 2 +
> >> .../selftests/bpf/progs/kfunc_call_test.c | 98 +++++++++++++++++++
> >> .../selftests/bpf/test_kmods/bpf_testmod.c | 54 +++++++++-
> >> .../bpf/test_kmods/bpf_testmod_kfunc.h | 1 +
> >> 4 files changed, 154 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c b/
> >> tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> >> index f79c8e53cb3e..62f3fb79f5d1 100644
> >> --- a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> >> +++ b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> >> @@ -74,6 +74,8 @@ static struct kfunc_test_params kfunc_tests[] = {
> >> TC_TEST(kfunc_call_test1, 12),
> >> TC_TEST(kfunc_call_test2, 3),
> >> TC_TEST(kfunc_call_test4, -1234),
> >> + TC_TEST(kfunc_call_test5, 0),
> >> + TC_TEST(kfunc_call_test5_asm, 0),
> >> TC_TEST(kfunc_call_test_ref_btf_id, 0),
> >> TC_TEST(kfunc_call_test_get_mem, 42),
> >> SYSCALL_TEST(kfunc_syscall_test, 0),
> >> diff --git a/tools/testing/selftests/bpf/progs/kfunc_call_test.c b/
> >> tools/testing/selftests/bpf/progs/kfunc_call_test.c
> >> index 8b86113a0126..5edc51564f71 100644
> >> --- a/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> >> +++ b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> >> @@ -2,8 +2,106 @@
> >> /* Copyright (c) 2021 Facebook */
> >> #include <vmlinux.h>
> >> #include <bpf/bpf_helpers.h>
> >> +#include "bpf_misc.h"
> >> #include "../test_kmods/bpf_testmod_kfunc.h"
> >> +SEC("tc")
> >> +int kfunc_call_test5(struct __sk_buff *skb)
> >> +{
> >> + struct bpf_sock *sk = skb->sk;
> >> + int ret;
> >> + u32 val32;
> >> + u16 val16;
> >> + u8 val8;
> >> +
> >> + if (!sk)
> >> + return -1;
> >> +
> >> + sk = bpf_sk_fullsock(sk);
> >> + if (!sk)
> >> + return -1;
> >> +
> >> + /*
> >> + * Test with constant values to verify zero-extension.
> >> + * ISA-dependent BPF asm:
> >> + * With ALU32: w1 = 0xFF; w2 = 0xFFFF; w3 = 0xFFFFffff
> >> + * Without ALU32: r1 = 0xFF; r2 = 0xFFFF; r3 = 0xFFFFffff
> >> + * Both zero-extend to 64-bit before the kfunc call.
> >> + */
> >> + ret = bpf_kfunc_call_test5(0xFF, 0xFFFF, 0xFFFFffffULL);
> >
> > Can we just use 0xFFFFffff instead of 0xFFFFffffULL?
>
> Alexei, can you confirm if I need to respin with this change?
I prefer the explicit type when the constant is on the boundary.
I was about to apply it, but it fails on s390. See CI:
Error: #160 kfunc_call
Error: #160/13 kfunc_call/kfunc_call_test5_asm
Error: #160/13 kfunc_call/kfunc_call_test5_asm
verify_success:PASS:skel 0 nsec
verify_success:PASS:bpf_object__find_program_by_name 0 nsec
verify_success:PASS:kfunc_call_test5_asm 0 nsec
verify_success:FAIL:retval unexpected retval: actual 6 != expected 0
Test Results:
bpftool: PASS
test_progs: FAIL (returned 1)
It doesn't look to be endianness related (which is often
the case for breaking s390).
In this case it could be an s390 JIT issue?
Ilya ?
pw-bot: cr
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
2026-03-12 8:01 [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call Hari Bathini
2026-03-12 18:22 ` Yonghong Song
@ 2026-03-13 16:22 ` Mykyta Yatsenko
1 sibling, 0 replies; 6+ messages in thread
From: Mykyta Yatsenko @ 2026-03-13 16:22 UTC (permalink / raw)
To: Hari Bathini, bpf
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, Shuah Khan,
linux-kselftest
Hari Bathini <hbathini@linux.ibm.com> writes:
> On powerpc, immediate load instructions are sign extended. In case
> of unsigned types, arguments should be explicitly zero-extended by
> the caller. For kfunc call, this needs to be handled in the JIT code.
> In bpf_kfunc_call_test4(), that tests for sign-extension of signed
> argument types in kfunc calls, add some additional failure checks.
> And add bpf_kfunc_call_test5() to test zero-extension of unsigned
> argument types in kfunc calls.
>
> Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
> ---
>
> Changes in v2:
> - Added asm version of the selftest for consistent testing across
> different BPF ISA versions.
> - Added comments clearly stating the intent of the test cases.
> - Updated sign-extension selftest to have additional failure checks.
>
>
> .../selftests/bpf/prog_tests/kfunc_call.c | 2 +
> .../selftests/bpf/progs/kfunc_call_test.c | 98 +++++++++++++++++++
> .../selftests/bpf/test_kmods/bpf_testmod.c | 54 +++++++++-
> .../bpf/test_kmods/bpf_testmod_kfunc.h | 1 +
> 4 files changed, 154 insertions(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> index f79c8e53cb3e..62f3fb79f5d1 100644
> --- a/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> +++ b/tools/testing/selftests/bpf/prog_tests/kfunc_call.c
> @@ -74,6 +74,8 @@ static struct kfunc_test_params kfunc_tests[] = {
> TC_TEST(kfunc_call_test1, 12),
> TC_TEST(kfunc_call_test2, 3),
> TC_TEST(kfunc_call_test4, -1234),
> + TC_TEST(kfunc_call_test5, 0),
> + TC_TEST(kfunc_call_test5_asm, 0),
> TC_TEST(kfunc_call_test_ref_btf_id, 0),
> TC_TEST(kfunc_call_test_get_mem, 42),
> SYSCALL_TEST(kfunc_syscall_test, 0),
> diff --git a/tools/testing/selftests/bpf/progs/kfunc_call_test.c b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> index 8b86113a0126..5edc51564f71 100644
> --- a/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> +++ b/tools/testing/selftests/bpf/progs/kfunc_call_test.c
> @@ -2,8 +2,106 @@
> /* Copyright (c) 2021 Facebook */
> #include <vmlinux.h>
> #include <bpf/bpf_helpers.h>
> +#include "bpf_misc.h"
> #include "../test_kmods/bpf_testmod_kfunc.h"
>
> +SEC("tc")
> +int kfunc_call_test5(struct __sk_buff *skb)
> +{
> + struct bpf_sock *sk = skb->sk;
> + int ret;
> + u32 val32;
> + u16 val16;
> + u8 val8;
> +
> + if (!sk)
> + return -1;
> +
> + sk = bpf_sk_fullsock(sk);
It looks like this socket stuff is not really needed in this test.
> + if (!sk)
> + return -1;
> +
> + /*
> + * Test with constant values to verify zero-extension.
> + * ISA-dependent BPF asm:
> + * With ALU32: w1 = 0xFF; w2 = 0xFFFF; w3 = 0xFFFFffff
> + * Without ALU32: r1 = 0xFF; r2 = 0xFFFF; r3 = 0xFFFFffff
> + * Both zero-extend to 64-bit before the kfunc call.
> + */
> + ret = bpf_kfunc_call_test5(0xFF, 0xFFFF, 0xFFFFffffULL);
> + if (ret)
> + return ret;
> +
> + val32 = bpf_get_prandom_u32();
> + val16 = val32 & 0xFFFF;
> + val8 = val32 & 0xFF;
> + ret = bpf_kfunc_call_test5(val8, val16, val32);
> + if (ret)
> + return ret;
> +
> + /*
> + * Test multiplication with different operand sizes:
> + *
> + * val8 * 0xFF:
> + * - Both operands promote to int (32-bit signed)
> + * - Result: 32-bit multiplication, truncated to u8, then zero-extended
> + *
> + * val16 * 0xFFFF:
> + * - Both operands promote to int (32-bit signed)
> + * - Result: 32-bit multiplication, truncated to u16, then zero-extended
> + *
> + * val32 * 0xFFFFffffULL:
> + * - val32 (u32) promotes to unsigned long long (due to ULL suffix)
> + * - Result: 64-bit unsigned multiplication, truncated to u32, then zero-extended
> + */
> + ret = bpf_kfunc_call_test5(val8 * 0xFF, val16 * 0xFFFF, val32 * 0xFFFFffffULL);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
> +
> +/*
> + * Assembly version testing the multiplication edge case explicitly.
> + * This ensures consistent testing across different ISA versions.
> + */
> +SEC("tc")
> +__naked int kfunc_call_test5_asm(void)
> +{
> + asm volatile (
> + /* Get a random u32 value */
> + "call %[bpf_get_prandom_u32];"
> + "r6 = r0;" /* Save val32 in r6 */
> +
> + /* Prepare first argument: val8 * 0xFF */
> + "r1 = r6;"
> + "r1 &= 0xFF;" /* val8 = val32 & 0xFF */
> + "r7 = 0xFF;"
> + "r1 *= r7;" /* 64-bit mult: r1 = r1 * r7 */
> +
> + /* Prepare second argument: val16 * 0xFFFF */
> + "r2 = r6;"
> + "r2 &= 0xFFFF;" /* val16 = val32 & 0xFFFF */
> + "r7 = 0xFFFF;"
> + "r2 *= r7;" /* 64-bit mult: r2 = r2 * r7 */
> +
> + /* Prepare third argument: val32 * 0xFFFFffff */
> + "r3 = r6;" /* val32 */
> + "r7 = 0xFFFFffff;"
> + "r3 *= r7;" /* 64-bit mult: r3 = r3 * r7 */
> +
> + /* Call kfunc with multiplication results */
> + "call bpf_kfunc_call_test5;"
> +
> + /* Check return value */
> + "if r0 != 0 goto exit_%=;"
> + "r0 = 0;"
> + "exit_%=: exit;"
> + :
> + : __imm(bpf_get_prandom_u32)
> + : __clobber_all);
> +}
> +
> SEC("tc")
> int kfunc_call_test4(struct __sk_buff *skb)
> {
> diff --git a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
> index e62c6b78657f..94edbd2afa67 100644
> --- a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
> +++ b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
> @@ -760,12 +760,63 @@ __bpf_kfunc struct sock *bpf_kfunc_call_test3(struct sock *sk)
>
> __bpf_kfunc long noinline bpf_kfunc_call_test4(signed char a, short b, int c, long d)
> {
> - /* Provoke the compiler to assume that the caller has sign-extended a,
> + /*
> + * Make val as volatile to avoid compiler optimizations.
> + * Verify that negative signed values remain negative after
> + * sign-extension (JIT must sign-extend, not zero-extend).
> + */
> + volatile long val;
> +
> + /* val will be positive, if JIT does zero-extension instead of sign-extension */
> + val = a;
> + if (val >= 0)
> + return 1;
> +
> + val = b;
> + if (val >= 0)
> + return 2;
> +
> + val = c;
> + if (val >= 0)
> + return 3;
> +
> + /*
> + * Provoke the compiler to assume that the caller has sign-extended a,
> * b and c on platforms where this is required (e.g. s390x).
> */
> return (long)a + (long)b + (long)c + d;
> }
>
> +__bpf_kfunc int bpf_kfunc_call_test5(u8 a, u16 b, u32 c)
> +{
> + /*
> + * Make val as volatile to avoid compiler optimizations on the below checks
> + * In C, assigning u8/u16/u32 to long performs zero-extension.
> + */
> + volatile long val = a;
> +
> + /* Check zero-extension */
> + if (val != (unsigned long)a)
> + return 1;
> + /* Check no sign-extension */
> + if (val < 0)
> + return 2;
> +
> + val = b;
> + if (val != (unsigned long)b)
> + return 3;
> + if (val < 0)
> + return 4;
> +
> + val = c;
> + if (val != (unsigned long)c)
> + return 5;
> + if (val < 0)
> + return 6;
> +
> + return 0;
> +}
> +
> static struct prog_test_ref_kfunc prog_test_struct = {
> .a = 42,
> .b = 108,
> @@ -1228,6 +1279,7 @@ BTF_ID_FLAGS(func, bpf_kfunc_call_test1)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test2)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test3)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test4)
> +BTF_ID_FLAGS(func, bpf_kfunc_call_test5)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_pass1)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail1)
> BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail2)
> diff --git a/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h b/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
> index b393bf771131..aa0b8d41e71b 100644
> --- a/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
> +++ b/tools/testing/selftests/bpf/test_kmods/bpf_testmod_kfunc.h
> @@ -110,6 +110,7 @@ __u64 bpf_kfunc_call_test1(struct sock *sk, __u32 a, __u64 b,
> int bpf_kfunc_call_test2(struct sock *sk, __u32 a, __u32 b) __ksym;
> struct sock *bpf_kfunc_call_test3(struct sock *sk) __ksym;
> long bpf_kfunc_call_test4(signed char a, short b, int c, long d) __ksym;
> +int bpf_kfunc_call_test5(__u8 a, __u16 b, __u32 c) __ksym;
>
> void bpf_kfunc_call_test_pass_ctx(struct __sk_buff *skb) __ksym;
> void bpf_kfunc_call_test_pass1(struct prog_test_pass1 *p) __ksym;
> --
> 2.53.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call
2026-03-13 14:20 ` Alexei Starovoitov
@ 2026-03-13 16:51 ` Ilya Leoshkevich
0 siblings, 0 replies; 6+ messages in thread
From: Ilya Leoshkevich @ 2026-03-13 16:51 UTC (permalink / raw)
To: Alexei Starovoitov, Hari Bathini
Cc: Yonghong Song, bpf, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, Shuah Khan, open list:KERNEL SELFTEST FRAMEWORK
On 3/13/26 15:20, Alexei Starovoitov wrote:
[...]
> I was about to apply it, but it fails on s390. See CI:
>
> Error: #160 kfunc_call
> Error: #160/13 kfunc_call/kfunc_call_test5_asm
> Error: #160/13 kfunc_call/kfunc_call_test5_asm
> verify_success:PASS:skel 0 nsec
> verify_success:PASS:bpf_object__find_program_by_name 0 nsec
> verify_success:PASS:kfunc_call_test5_asm 0 nsec
> verify_success:FAIL:retval unexpected retval: actual 6 != expected 0
> Test Results:
> bpftool: PASS
> test_progs: FAIL (returned 1)
>
> It doesn't look to be endianness related (which is often
> the case for breaking s390).
> In this case it could be an s390 JIT issue?
>
> Ilya ?
>
> pw-bot: cr
I implemented sign-extension required by the s390x ABI, but forgot about
zero-extension.
I will send the fix soon, after ./test_progs completes.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2026-03-13 16:52 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-03-12 8:01 [PATCH bpf-next v2] selftests/bpf: improve test coverage for kfunc call Hari Bathini
2026-03-12 18:22 ` Yonghong Song
2026-03-13 9:29 ` Hari Bathini
2026-03-13 14:20 ` Alexei Starovoitov
2026-03-13 16:51 ` Ilya Leoshkevich
2026-03-13 16:22 ` Mykyta Yatsenko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox