Skip to contentSkip to conversation

Add prog_kfunc feature probe

PATCH Superseded Tao Chen 2 participants 7 messages Jan 23, 2025 – Jan 24, 2025

Subsystems (4)

v2 of 7 Superseded Latest is v7 →

lore
Share & tools
lore

Tao Chen

author
#0 lore raw

v2 cover Add prog_kfunc feature probe

More and more kfunc functions are being added to the kernel. Different prog types have different restrictions when using kfunc. Therefore, prog_kfunc probe is added to check whether it is supported, and the use of this api will be added to bpftool later.

Change list:
- v1 -> v2:
- check unsupported prog type like probe_bpf_helper
- add off parameter for module btf
- chenk verifier info when kfunc id invalid

Revisions:
- v1
https://lore.kernel.org/bpf/[email protected]

Tao Chen (2):
libbpf: Add libbpf_probe_bpf_kfunc API selftests/bpf: Add libbpf_probe_bpf_kfunc API selftests

 tools/lib/bpf/libbpf.h                        | 17 ++++++-
 tools/lib/bpf/libbpf.map                      |  1 +
 tools/lib/bpf/libbpf_probes.c                 | 47 +++++++++++++++++++
 .../selftests/bpf/prog_tests/libbpf_probes.c  | 35 ++++++++++++++
 4 files changed, 99 insertions(+), 1 deletion(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH v2 1/2 libbpf: Add libbpf_probe_bpf_kfunc API

Similarly to libbpf_probe_bpf_helper, the libbpf_probe_bpf_kfunc used to test the availability of the different eBPF kfuncs on the current system.

---
 tools/lib/bpf/libbpf.h        | 17 ++++++++++++-
 tools/lib/bpf/libbpf.map      |  1 +
 tools/lib/bpf/libbpf_probes.c | 47 +++++++++++++++++++++++++++++++++++
 3 files changed, 64 insertions(+), 1 deletion(-)
tools/lib/bpf/libbpf.h +16 -1
index 3020ee45303a..ac0b46d8d8f9 100644
@@ -1680,7 +1680,22 @@ LIBBPF_API int libbpf_probe_bpf_map_type(enum bpf_map_type map_type, const void
  */
 LIBBPF_API int libbpf_probe_bpf_helper(enum bpf_prog_type prog_type,
 				       enum bpf_func_id helper_id, const void *opts);
-
+/**
+ * @brief **libbpf_probe_bpf_kfunc()** detects if host kernel supports the
+ * use of a given BPF kfunc from specified BPF program type.
+ * @param prog_type BPF program type used to check the support of BPF kfunc
+ * @param kfunc_id The btf ID of BPF kfunc to check support for
+ * @param off The module BTF FD, 0 for vmlinux
+ * @param opts reserved for future extensibility, should be NULL
+ * @return 1, if given combination of program type and kfunc is supported; 0,
+ * if the combination is not supported; negative error code if feature
+ * detection for provided input arguments failed or can't be performed
+ *
+ * Make sure the process has required set of CAP_* permissions (or runs as
+ * root) when performing feature checking.
+ */
+LIBBPF_API int libbpf_probe_bpf_kfunc(enum bpf_prog_type prog_type,
+				      int kfunc_id, __s16 off, const void *opts);
 /**
  * @brief **libbpf_num_possible_cpus()** is a helper function to get the
  * number of possible CPUs that the host kernel supports and expects.
tools/lib/bpf/libbpf.map +1 -0
index a8b2936a1646..e93fae101efd 100644
@@ -436,4 +436,5 @@ LIBBPF_1.6.0 {
 		bpf_linker__add_buf;
 		bpf_linker__add_fd;
 		bpf_linker__new_fd;
+		libbpf_probe_bpf_kfunc;
 } LIBBPF_1.5.0;
tools/lib/bpf/libbpf_probes.c +46 -0

52 lines not on this page · raw message

Signature · 1 line
2.43.0
maintainer
#2 lore raw

Re: [PATCH bpf-next v2 1/2] libbpf: Add libbpf_probe_bpf_kfunc API

On Fri, 2025-01-24 at 01:05 +0800, Tao Chen wrote:

[...]

tools/lib/bpf/libbpf_probes.c int libbpf_probe_bpf_map_type(enum bpf_map_type map_type, const void *opts) · line 417

Quoted 2 lines
tools/lib/bpf/libbpf_probes.c +2 -0
+int libbpf_probe_bpf_kfunc(enum bpf_prog_type prog_type, int kfunc_id,
+			   __s16 off, const void *opts)
                                 ^^^

Nit: maybe name this btf_fd?

Quoted 1 line

+{

In v2 function looks identical to libbpf_probe_bpf_helper, do we want to copy-paste it or introduce a utility:

static int probe_insn(enum bpf_prog_type prog_type, struct bpf_insn insn,
const char **accepted_msgs)

And call it from both libbpf_probe_bpf_{helper,kfunc}?

Quoted 46 lines
tools/lib/bpf/libbpf_probes.c +43 -0
+	struct bpf_insn insns[] = {
+		BPF_EXIT_INSN(),
+		BPF_EXIT_INSN(),
+	};
+	const size_t insn_cnt = ARRAY_SIZE(insns);
+	int err;
+	char buf[4096];
+
+	if (opts)
+		return libbpf_err(-EINVAL);
+
+	/* Same logic as probe_bpf_helper check */
+	switch (prog_type) {
+	case BPF_PROG_TYPE_TRACING:
+	case BPF_PROG_TYPE_EXT:
+	case BPF_PROG_TYPE_LSM:
+	case BPF_PROG_TYPE_STRUCT_OPS:
+		return -EOPNOTSUPP;
+	default:
+		break;
+	}
+
+	insns[0].code = BPF_JMP | BPF_CALL;
+	insns[0].src_reg = BPF_PSEUDO_KFUNC_CALL;
+	insns[0].imm = kfunc_id;
+	insns[0].off = off;
+
+	buf[0] = '\0';
+	err = probe_prog_load(prog_type, insns, insn_cnt, buf, sizeof(buf));
+	if (err < 0)
+		return libbpf_err(err);
+
+	/* If BPF verifier recognizes BPF kfunc but it's not supported for
+	 * given BPF program type, it will emit "calling kernel function
+	 * bpf_cpumask_create is not allowed", if the kfunc id is invalid,
+	 * it will emit "kernel btf_id 4294967295 is not a function"
+	 */
+	if (err == 0 && (strstr(buf, "not allowed") || strstr(buf, "not a function")))
+		return 0;
+
+	return 1; /* assume supported */
+}
+
 int libbpf_probe_bpf_helper(enum bpf_prog_type prog_type, enum bpf_func_id helper_id,
 			    const void *opts)
 {
author
#3 lore raw

Re: [PATCH bpf-next v2 1/2] libbpf: Add libbpf_probe_bpf_kfunc API

Quoted 8 lines · Eduard Zingerman

On Fri, 2025-01-24 at 01:05 +0800, Tao Chen wrote:

[...]

tools/lib/bpf/libbpf_probes.c int libbpf_probe_bpf_map_type(enum bpf_map_type map_type, const void *opts) · line 417

Quoted 2 lines
tools/lib/bpf/libbpf_probes.c +2 -0
+int libbpf_probe_bpf_kfunc(enum bpf_prog_type prog_type, int kfunc_id,
+			   __s16 off, const void *opts)
                                  ^^^

Nit: maybe name this btf_fd?

Hi Eduard, thank you for your guidance, ack, i will send it in v3.

Quoted 10 lines
Quoted 2 lines

+{

In v2 function looks identical to libbpf_probe_bpf_helper, do we want to copy-paste it or introduce a utility:

static int probe_insn(enum bpf_prog_type prog_type, struct bpf_insn insn,
const char **accepted_msgs)

And call it from both libbpf_probe_bpf_{helper,kfunc}?

Yes, it seems much more concise. I will send it in v3. Thanks.

Quoted 49 lines
Quoted 47 lines
tools/lib/bpf/libbpf_probes.c +43 -0
+	struct bpf_insn insns[] = {
+		BPF_EXIT_INSN(),
+		BPF_EXIT_INSN(),
+	};
+	const size_t insn_cnt = ARRAY_SIZE(insns);
+	int err;
+	char buf[4096];
+
+	if (opts)
+		return libbpf_err(-EINVAL);
+
+	/* Same logic as probe_bpf_helper check */
+	switch (prog_type) {
+	case BPF_PROG_TYPE_TRACING:
+	case BPF_PROG_TYPE_EXT:
+	case BPF_PROG_TYPE_LSM:
+	case BPF_PROG_TYPE_STRUCT_OPS:
+		return -EOPNOTSUPP;
+	default:
+		break;
+	}
+
+	insns[0].code = BPF_JMP | BPF_CALL;
+	insns[0].src_reg = BPF_PSEUDO_KFUNC_CALL;
+	insns[0].imm = kfunc_id;
+	insns[0].off = off;
+
+	buf[0] = '\0';
+	err = probe_prog_load(prog_type, insns, insn_cnt, buf, sizeof(buf));
+	if (err < 0)
+		return libbpf_err(err);
+
+	/* If BPF verifier recognizes BPF kfunc but it's not supported for
+	 * given BPF program type, it will emit "calling kernel function
+	 * bpf_cpumask_create is not allowed", if the kfunc id is invalid,
+	 * it will emit "kernel btf_id 4294967295 is not a function"
+	 */
+	if (err == 0 && (strstr(buf, "not allowed") || strstr(buf, "not a function")))
+		return 0;
+
+	return 1; /* assume supported */
+}
+
  int libbpf_probe_bpf_helper(enum bpf_prog_type prog_type, enum bpf_func_id helper_id,
  			    const void *opts)
  {
Signature · 2 lines
Best Regards
Dylane Chen
author
#4 lore raw

PATCH v2 2/2 selftests/bpf: Add libbpf_probe_bpf_kfunc API selftests

In reply to Tao Chen

Add selftests for prog_kfunc feature probing.
./test_progs -t libbpf_probe_kfuncs
#153 libbpf_probe_kfuncs:OK
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED

---
 .../selftests/bpf/prog_tests/libbpf_probes.c  | 35 +++++++++++++++++++
 1 file changed, 35 insertions(+)
tools/testing/selftests/bpf/prog_tests/libbpf_probes.c +35 -0
index 4ed46ed58a7b..d9d69941f694 100644
@@ -126,3 +126,38 @@ void test_libbpf_probe_helpers(void)
 		ASSERT_EQ(res, d->supported, buf);
 	}
 }
+
+void test_libbpf_probe_kfuncs(void)
+{
+	int ret, kfunc_id;
+	char *kfunc = "bpf_cpumask_create";
+	struct btf *btf;
+
+	btf = btf__parse("/sys/kernel/btf/vmlinux", NULL);
+	if (!ASSERT_OK_PTR(btf, "btf_parse"))
+		return;
+
+	kfunc_id = btf__find_by_name_kind(btf, kfunc, BTF_KIND_FUNC);
+	if (!ASSERT_GT(kfunc_id, 0, kfunc))
+		goto cleanup;
+
+	/* prog BPF_PROG_TYPE_SYSCALL supports kfunc bpf_cpumask_create */
+	ret = libbpf_probe_bpf_kfunc(BPF_PROG_TYPE_SYSCALL, kfunc_id, 0, NULL);
+	ASSERT_EQ(ret, 1, kfunc);
+
+	/* prog BPF_PROG_TYPE_KPROBE does not support kfunc bpf_cpumask_create */
+	ret = libbpf_probe_bpf_kfunc(BPF_PROG_TYPE_KPROBE, kfunc_id, 0, NULL);
+	ASSERT_EQ(ret, 0, kfunc);
+
+	/* invalid kfunc id */
+	ret = libbpf_probe_bpf_kfunc(BPF_PROG_TYPE_KPROBE, -1, 0, NULL);
+	ASSERT_EQ(ret, 0, "invalid kfunc id:-1");
+
+	/* invalid prog type */
+	ret = libbpf_probe_bpf_kfunc(100000, kfunc_id, 0, NULL);
+	if (!ASSERT_LE(ret, 0, "invalid prog type:100000"))
+		goto cleanup;
+
+cleanup:
+	btf__free(btf);
+}
Signature · 1 line
2.43.0
maintainer
#5 lore raw
In reply to Tao Chen
Quoted 19 lines · Tao Chen

More and more kfunc functions are being added to the kernel. Different prog types have different restrictions when using kfunc. Therefore, prog_kfunc probe is added to check whether it is supported, and the use of this api will be added to bpftool later.

Change list:
- v1 -> v2:
- check unsupported prog type like probe_bpf_helper
- add off parameter for module btf
- chenk verifier info when kfunc id invalid

Revisions:
- v1
https://lore.kernel.org/bpf/[email protected]

Tao Chen (2):
libbpf: Add libbpf_probe_bpf_kfunc API selftests/bpf: Add libbpf_probe_bpf_kfunc API selftests

Quoted 6 lines
 tools/lib/bpf/libbpf.h                        | 17 ++++++-
 tools/lib/bpf/libbpf.map                      |  1 +
 tools/lib/bpf/libbpf_probes.c                 | 47 +++++++++++++++++++
 .../selftests/bpf/prog_tests/libbpf_probes.c  | 35 ++++++++++++++
 4 files changed, 99 insertions(+), 1 deletion(-)

Hi Tao,

Looks like something is wrong with the way the patch was generated:
- patchwork link:
https://patchwork.kernel.org/project/netdevbpf/patch/[email protected]/
- error message:
https://github.com/kernel-patches/bpf/pull/8395

Cmd('git') failed due to: exit code(128)
cmdline: git am --3way
stdout: 'Applying: libbpf: Add libbpf_probe_bpf_kfunc API
Patch failed at 0001 libbpf: Add libbpf_probe_bpf_kfunc API When you have resolved this problem, run "git am --continue". If you prefer to skip this patch, run "git am --skip" instead. To restore the original branch and stop patching, run "git am --abort".'
stderr: 'error: corrupt patch at line 103
error: could not build fake ancestor
hint: Use 'git am --show-current-patch=diff' to see the failed patch'

I get the same error when trying to apply locally, could you please double check?

Thanks,
Eduard

author
#6 lore raw
Quoted 46 lines · Eduard Zingerman
Quoted 19 lines · Tao Chen

More and more kfunc functions are being added to the kernel. Different prog types have different restrictions when using kfunc. Therefore, prog_kfunc probe is added to check whether it is supported, and the use of this api will be added to bpftool later.

Change list:
- v1 -> v2:
- check unsupported prog type like probe_bpf_helper
- add off parameter for module btf
- chenk verifier info when kfunc id invalid

Revisions:
- v1
https://lore.kernel.org/bpf/[email protected]

Tao Chen (2):
libbpf: Add libbpf_probe_bpf_kfunc API selftests/bpf: Add libbpf_probe_bpf_kfunc API selftests

Quoted 6 lines
  tools/lib/bpf/libbpf.h                        | 17 ++++++-
  tools/lib/bpf/libbpf.map                      |  1 +
  tools/lib/bpf/libbpf_probes.c                 | 47 +++++++++++++++++++
  .../selftests/bpf/prog_tests/libbpf_probes.c  | 35 ++++++++++++++
  4 files changed, 99 insertions(+), 1 deletion(-)

Hi Tao,

Looks like something is wrong with the way the patch was generated:
- patchwork link:

   https://patchwork.kernel.org/project/netdevbpf/patch/[email protected]/

- error message:
https://github.com/kernel-patches/bpf/pull/8395

Cmd('git') failed due to: exit code(128)
cmdline: git am --3way
stdout: 'Applying: libbpf: Add libbpf_probe_bpf_kfunc API
Patch failed at 0001 libbpf: Add libbpf_probe_bpf_kfunc API When you have resolved this problem, run "git am --continue". If you prefer to skip this patch, run "git am --skip" instead. To restore the original branch and stop patching, run "git am --abort".'
stderr: 'error: corrupt patch at line 103
error: could not build fake ancestor
hint: Use 'git am --show-current-patch=diff' to see the failed patch'

Quoted 6 lines

I get the same error when trying to apply locally, could you please double check?

Thanks,
Eduard

Hi, Eduard, thank you for your reply, i tried it out, and it turns out it's really my problem. I will resend it in v3.

Signature · 2 lines
Best Regards
Dylane Chen