Skip to contentSkip to conversation

libbpf: support custom SEC() handlers

PATCH Applied Andrii Nakryiko 2 participants 5 messages Mar 5, 2022

Subsystems (4)
lore
Share & tools
lore

Andrii Nakryiko

author maintainer
#0 lore raw

v5 cover libbpf: support custom SEC() handlers

Add ability for user applications and libraries to register custom BPF program SEC() handlers. See patch #2 for examples where this is useful.

Patch #1 does some preliminary refactoring to allow exponsing program init, preload, and attach callbacks as public API. It also establishes a protocol to allow optional auto-attach behavior. This will also help the case of sometimes auto-attachable uprobes.

v4->v5:
- API documentation improvements (Daniel);
v3->v4:
- init_fn -> prog_setup_fn, preload_fn -> prog_prepare_load_fn (Alexei);
v2->v3:
- moved callbacks and cookie into OPTS struct (Alan);
- added more test scenarios (Alan);
- address most of Alan's feedback, but kept API name;
v1->v2:
- resubmitting due to git send-email screw up.

Cc Alan Maguire <[email protected]>

Andrii Nakryiko (3):
libbpf: allow BPF program auto-attach handlers to bail out
libbpf: support custom SEC() handlers selftests/bpf: add custom SEC() handling selftest

 tools/lib/bpf/libbpf.c                        | 332 ++++++++++++------
 tools/lib/bpf/libbpf.h                        | 109 ++++++
 tools/lib/bpf/libbpf.map                      |   6 +
 tools/lib/bpf/libbpf_version.h                |   2 +-
 .../bpf/prog_tests/custom_sec_handlers.c      | 176 ++++++++++
 .../bpf/progs/test_custom_sec_handlers.c      |  63 ++++
 6 files changed, 586 insertions(+), 102 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/custom_sec_handlers.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_custom_sec_handlers.c
Signature · 1 line
2.30.2
author maintainer
#1 lore raw

PATCH v5 1/3 libbpf: allow BPF program auto-attach handlers to bail out

Allow some BPF program types to support auto-attach only in subste of cases. Currently, if some BPF program type specifies attach callback, it is assumed that during skeleton attach operation all such programs either successfully attach or entire skeleton attachment fails. If some program doesn't support auto-attachment from skeleton, such BPF program types shouldn't have attach callback specified.

This is limiting for cases when, depending on how full the SEC("") definition is, there could either be enough details to support auto-attach or there might not be and user has to use some specific API to provide more details at runtime.

One specific example of such desired behavior might be SEC("uprobe"). If it's specified as just uprobe auto-attach isn't possible. But if it's SEC("uprobe/<some_binary>:<some_func>") then there are enough details to support auto-attach. Note that there is a somewhat subtle difference between auto-attach behavior of BPF skeleton and using "generic" bpf_program__attach(prog) (which uses the same attach handlers under the cover). Skeleton allow some programs within bpf_object to not have auto-attach implemented and doesn't treat that as an error. Instead such BPF programs are just skipped during skeleton's (optional) attach step. bpf_program__attach(), on the other hand, is called when user *expects* auto-attach to work, so if specified program doesn't implement or doesn't support auto-attach functionality, that will be treated as an error.

Another improvement to the way libbpf is handling SEC()s would be to not require providing dummy kernel function name for kprobe. Currently, SEC("kprobe/whatever") is necessary even if actual kernel function is determined by user at runtime and bpf_program__attach_kprobe() is used to specify it. With changes in this patch, it's possible to support both SEC("kprobe") and SEC("kprobe/<actual_kernel_function"), while only in the latter case auto-attach will be performed. In the former one, such kprobe will be skipped during skeleton attach operation.

---
 tools/lib/bpf/libbpf.c | 140 +++++++++++++++++++++++++----------------
 1 file changed, 85 insertions(+), 55 deletions(-)
tools/lib/bpf/libbpf.c +85 -55
index 81bf01d67671..1da4a438ba00 100644
@@ -201,11 +201,12 @@ struct reloc_desc {
 	};
 };
 
-struct bpf_sec_def;
-
-typedef int (*init_fn_t)(struct bpf_program *prog, long cookie);
-typedef int (*preload_fn_t)(struct bpf_program *prog, struct bpf_prog_load_opts *opts, long cookie);
-typedef struct bpf_link *(*attach_fn_t)(const struct bpf_program *prog, long cookie);
+typedef int (*libbpf_prog_setup_fn_t)(struct bpf_program *prog, long cookie);
+typedef int (*libbpf_prog_prepare_load_fn_t)(struct bpf_program *prog,
+					     struct bpf_prog_load_opts *opts, long cookie);
+/* If auto-attach is not supported, callback should return 0 and set link to NULL */
+typedef int (*libbpf_prog_attach_fn_t)(const struct bpf_program *prog, long cookie,
+				       struct bpf_link **link);
 
 /* stored as sec_def->cookie for all libbpf-supported SEC()s */
 enum sec_def_flags {
@@ -239,9 +240,9 @@ struct bpf_sec_def {
 	enum bpf_attach_type expected_attach_type;
 	long cookie;
 
-	init_fn_t init_fn;
-	preload_fn_t preload_fn;
-	attach_fn_t attach_fn;
+	libbpf_prog_setup_fn_t prog_setup_fn;
+	libbpf_prog_prepare_load_fn_t prog_prepare_load_fn;
+	libbpf_prog_attach_fn_t prog_attach_fn;
 };
 
 /*
@@ -6572,9 +6573,9 @@ static int bpf_object__sanitize_prog(struct bpf_object *obj, struct bpf_program
 static int libbpf_find_attach_btf_id(struct bpf_program *prog, const char *attach_name,
 				     int *btf_obj_fd, int *btf_type_id);
 
-/* this is called as prog->sec_def->preload_fn for libbpf-supported sec_defs */
-static int libbpf_preload_prog(struct bpf_program *prog,
-			       struct bpf_prog_load_opts *opts, long cookie)
+/* this is called as prog->sec_def->prog_prepare_load_fn for libbpf-supported sec_defs */
+static int libbpf_prepare_prog_load(struct bpf_program *prog,
+				    struct bpf_prog_load_opts *opts, long cookie)
 {
 	enum sec_def_flags def = cookie;
 
@@ -6670,8 +6671,8 @@ static int bpf_object_load_prog_instance(struct bpf_object *obj, struct bpf_prog
 	load_attr.fd_array = obj->fd_array;
 
 	/* adjust load_attr if sec_def provides custom preload callback */
-	if (prog->sec_def && prog->sec_def->preload_fn) {
-		err = prog->sec_def->preload_fn(prog, &load_attr, prog->sec_def->cookie);
+	if (prog->sec_def && prog->sec_def->prog_prepare_load_fn) {
+		err = prog->sec_def->prog_prepare_load_fn(prog, &load_attr, prog->sec_def->cookie);
 		if (err < 0) {
 			pr_warn("prog '%s': failed to prepare load attributes: %d\n",
 				prog->name, err);
@@ -6971,8 +6972,8 @@ static int bpf_object_init_progs(struct bpf_object *obj, const struct bpf_object
 		/* sec_def can have custom callback which should be called
 		 * after bpf_program is initialized to adjust its properties
 		 */
-		if (prog->sec_def->init_fn) {
-			err = prog->sec_def->init_fn(prog, prog->sec_def->cookie);
+		if (prog->sec_def->prog_setup_fn) {
+			err = prog->sec_def->prog_setup_fn(prog, prog->sec_def->cookie);
 			if (err < 0) {
 				pr_warn("prog '%s': failed to initialize: %d\n",
 					prog->name, err);
@@ -8593,16 +8594,16 @@ int bpf_program__set_log_buf(struct bpf_program *prog, char *log_buf, size_t log
 	.prog_type = BPF_PROG_TYPE_##ptype,				    \
 	.expected_attach_type = atype,					    \
 	.cookie = (long)(flags),					    \
-	.preload_fn = libbpf_preload_prog,				    \
+	.prog_prepare_load_fn = libbpf_prepare_prog_load,		    \
 	__VA_ARGS__							    \
 }
 
-static struct bpf_link *attach_kprobe(const struct bpf_program *prog, long cookie);
-static struct bpf_link *attach_tp(const struct bpf_program *prog, long cookie);
-static struct bpf_link *attach_raw_tp(const struct bpf_program *prog, long cookie);
-static struct bpf_link *attach_trace(const struct bpf_program *prog, long cookie);
-static struct bpf_link *attach_lsm(const struct bpf_program *prog, long cookie);
-static struct bpf_link *attach_iter(const struct bpf_program *prog, long cookie);
+static int attach_kprobe(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_tp(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_raw_tp(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_trace(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_lsm(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_iter(const struct bpf_program *prog, long cookie, struct bpf_link **link);
 
 static const struct bpf_sec_def section_defs[] = {
 	SEC_DEF("socket",		SOCKET_FILTER, 0, SEC_NONE | SEC_SLOPPY_PFX),
@@ -8752,7 +8753,7 @@ static char *libbpf_get_type_names(bool attach_type)
 		const struct bpf_sec_def *sec_def = &section_defs[i];
 
 		if (attach_type) {
-			if (sec_def->preload_fn != libbpf_preload_prog)
+			if (sec_def->prog_prepare_load_fn != libbpf_prepare_prog_load)
 				continue;
 
 			if (!(sec_def->cookie & SEC_ATTACHABLE))
@@ -9135,7 +9136,7 @@ int libbpf_attach_type_by_name(const char *name,
 		return libbpf_err(-EINVAL);
 	}
 
-	if (sec_def->preload_fn != libbpf_preload_prog)
+	if (sec_def->prog_prepare_load_fn != libbpf_prepare_prog_load)
 		return libbpf_err(-EINVAL);
 	if (!(sec_def->cookie & SEC_ATTACHABLE))
 		return libbpf_err(-EINVAL);
@@ -10109,14 +10110,13 @@ struct bpf_link *bpf_program__attach_kprobe(const struct bpf_program *prog,
 	return bpf_program__attach_kprobe_opts(prog, func_name, &opts);
 }
 
-static struct bpf_link *attach_kprobe(const struct bpf_program *prog, long cookie)
+static int attach_kprobe(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
 	DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts);
 	unsigned long offset = 0;
-	struct bpf_link *link;
 	const char *func_name;
 	char *func;
-	int n, err;
+	int n;
 
 	opts.retprobe = str_has_pfx(prog->sec_name, "kretprobe/");
 	if (opts.retprobe)
@@ -10126,21 +10126,19 @@ static struct bpf_link *attach_kprobe(const struct bpf_program *prog, long cooki
 
 	n = sscanf(func_name, "%m[a-zA-Z0-9_.]+%li", &func, &offset);
 	if (n < 1) {
-		err = -EINVAL;
 		pr_warn("kprobe name is invalid: %s\n", func_name);
-		return libbpf_err_ptr(err);
+		return -EINVAL;
 	}
 	if (opts.retprobe && offset != 0) {
 		free(func);
-		err = -EINVAL;
 		pr_warn("kretprobes do not support offset specification\n");
-		return libbpf_err_ptr(err);
+		return -EINVAL;
 	}
 
 	opts.offset = offset;
-	link = bpf_program__attach_kprobe_opts(prog, func, &opts);
+	*link = bpf_program__attach_kprobe_opts(prog, func, &opts);
 	free(func);
-	return link;
+	return libbpf_get_error(*link);
 }
 
 static void gen_uprobe_legacy_event_name(char *buf, size_t buf_sz,
@@ -10395,14 +10393,13 @@ struct bpf_link *bpf_program__attach_tracepoint(const struct bpf_program *prog,
 	return bpf_program__attach_tracepoint_opts(prog, tp_category, tp_name, NULL);
 }
 
-static struct bpf_link *attach_tp(const struct bpf_program *prog, long cookie)
+static int attach_tp(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
 	char *sec_name, *tp_cat, *tp_name;
-	struct bpf_link *link;
 
 	sec_name = strdup(prog->sec_name);
 	if (!sec_name)
-		return libbpf_err_ptr(-ENOMEM);
+		return -ENOMEM;
 
 	/* extract "tp/<category>/<name>" or "tracepoint/<category>/<name>" */
 	if (str_has_pfx(prog->sec_name, "tp/"))
@@ -10412,14 +10409,14 @@ static struct bpf_link *attach_tp(const struct bpf_program *prog, long cookie)
 	tp_name = strchr(tp_cat, '/');
 	if (!tp_name) {
 		free(sec_name);
-		return libbpf_err_ptr(-EINVAL);
+		return -EINVAL;
 	}
 	*tp_name = '\0';
 	tp_name++;
 
-	link = bpf_program__attach_tracepoint(prog, tp_cat, tp_name);
+	*link = bpf_program__attach_tracepoint(prog, tp_cat, tp_name);
 	free(sec_name);
-	return link;
+	return libbpf_get_error(*link);
 }
 
 struct bpf_link *bpf_program__attach_raw_tracepoint(const struct bpf_program *prog,
@@ -10452,7 +10449,7 @@ struct bpf_link *bpf_program__attach_raw_tracepoint(const struct bpf_program *pr
 	return link;
 }
 
-static struct bpf_link *attach_raw_tp(const struct bpf_program *prog, long cookie)
+static int attach_raw_tp(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
 	static const char *const prefixes[] = {
 		"raw_tp/",
@@ -10472,10 +10469,11 @@ static struct bpf_link *attach_raw_tp(const struct bpf_program *prog, long cooki
 	if (!tp_name) {
 		pr_warn("prog '%s': invalid section name '%s'\n",
 			prog->name, prog->sec_name);
-		return libbpf_err_ptr(-EINVAL);
+		return -EINVAL;
 	}
 
-	return bpf_program__attach_raw_tracepoint(prog, tp_name);
+	*link = bpf_program__attach_raw_tracepoint(prog, tp_name);
+	return libbpf_get_error(link);
 }
 
 /* Common logic for all BPF program types that attach to a btf_id */
@@ -10518,14 +10516,16 @@ struct bpf_link *bpf_program__attach_lsm(const struct bpf_program *prog)
 	return bpf_program__attach_btf_id(prog);
 }
 
-static struct bpf_link *attach_trace(const struct bpf_program *prog, long cookie)
+static int attach_trace(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
-	return bpf_program__attach_trace(prog);
+	*link = bpf_program__attach_trace(prog);
+	return libbpf_get_error(*link);
 }
 
-static struct bpf_link *attach_lsm(const struct bpf_program *prog, long cookie)
+static int attach_lsm(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
-	return bpf_program__attach_lsm(prog);
+	*link = bpf_program__attach_lsm(prog);
+	return libbpf_get_error(*link);
 }
 
 static struct bpf_link *
@@ -10654,17 +10654,33 @@ bpf_program__attach_iter(const struct bpf_program *prog,
 	return link;
 }
 
-static struct bpf_link *attach_iter(const struct bpf_program *prog, long cookie)
+static int attach_iter(const struct bpf_program *prog, long cookie, struct bpf_link **link)
 {
-	return bpf_program__attach_iter(prog, NULL);
+	*link = bpf_program__attach_iter(prog, NULL);
+	return libbpf_get_error(*link);
 }
 
 struct bpf_link *bpf_program__attach(const struct bpf_program *prog)
 {
-	if (!prog->sec_def || !prog->sec_def->attach_fn)
-		return libbpf_err_ptr(-ESRCH);
+	struct bpf_link *link = NULL;
+	int err;
+
+	if (!prog->sec_def || !prog->sec_def->prog_attach_fn)
+		return libbpf_err_ptr(-EOPNOTSUPP);
 
-	return prog->sec_def->attach_fn(prog, prog->sec_def->cookie);
+	err = prog->sec_def->prog_attach_fn(prog, prog->sec_def->cookie, &link);
+	if (err)
+		return libbpf_err_ptr(err);
+
+	/* When calling bpf_program__attach() explicitly, auto-attach support
+	 * is expected to work, so NULL returned link is considered an error.
+	 * This is different for skeleton's attach, see comment in
+	 * bpf_object__attach_skeleton().
+	 */
+	if (!link)
+		return libbpf_err_ptr(-EOPNOTSUPP);
+
+	return link;
 }
 
 static int bpf_link__detach_struct_ops(struct bpf_link *link)
@@ -11805,16 +11821,30 @@ int bpf_object__attach_skeleton(struct bpf_object_skeleton *s)
 			continue;
 
 		/* auto-attaching not supported for this program */
-		if (!prog->sec_def || !prog->sec_def->attach_fn)
+		if (!prog->sec_def || !prog->sec_def->prog_attach_fn)
 			continue;
 
-		*link = bpf_program__attach(prog);
-		err = libbpf_get_error(*link);
+		/* if user already set the link manually, don't attempt auto-attach */
+		if (*link)
+			continue;
+
+		err = prog->sec_def->prog_attach_fn(prog, prog->sec_def->cookie, link);
 		if (err) {
-			pr_warn("failed to auto-attach program '%s': %d\n",
+			pr_warn("prog '%s': failed to auto-attach: %d\n",
 				bpf_program__name(prog), err);
 			return libbpf_err(err);
 		}
+
+		/* It's possible that for some SEC() definitions auto-attach
+		 * is supported in some cases (e.g., if definition completely
+		 * specifies target information), but is not in other cases.
+		 * SEC("uprobe") is one such case. If user specified target
+		 * binary and function name, such BPF program can be
+		 * auto-attached. But if not, it shouldn't trigger skeleton's
+		 * attach to fail. It should just be skipped.
+		 * attach_fn signals such case with returning 0 (no error) and
+		 * setting link to NULL.
+		 */
 	}
 
 	return 0;
Signature · 1 line
2.30.2
author maintainer
#2 lore raw

PATCH v5 2/3 libbpf: support custom SEC() handlers

In reply to Andrii Nakryiko

Allow registering and unregistering custom handlers for BPF program. This allows user applications and libraries to plug into libbpf's declarative SEC() definition handling logic. This allows to offload complex and intricate custom logic into external libraries, but still provide a great user experience.

One such example is USDT handling library, which has a lot of code and complexity which doesn't make sense to put into libbpf directly, but it would be really great for users to be able to specify BPF programs with something like SEC("usdt/<path-to-binary>:<usdt_provider>:<usdt_name>") and have correct BPF program type set (BPF_PROGRAM_TYPE_KPROBE, as it is uprobe) and even support BPF skeleton's auto-attach logic.

In some cases, it might be even good idea to override libbpf's default handling, like for SEC("perf_event") programs. With custom library, it's possible to extend logic to support specifying perf event specification right there in SEC() definition without burdening libbpf with lots of custom logic or extra library dependecies (e.g., libpfm4). With current patch it's possible to override libbpf's SEC("perf_event") handling and specify a completely custom ones.

Further, it's possible to specify a generic fallback handling for any SEC() that doesn't match any other custom or standard libbpf handlers. This allows to accommodate whatever legacy use cases there might be, if necessary.

See doc comments for libbpf_register_prog_handler() and libbpf_unregister_prog_handler() for detailed semantics.

This patch also bumps libbpf development version to v0.8 and adds new APIs there.

---
 tools/lib/bpf/libbpf.c         | 204 ++++++++++++++++++++++++---------
 tools/lib/bpf/libbpf.h         | 109 ++++++++++++++++++
 tools/lib/bpf/libbpf.map       |   6 +
 tools/lib/bpf/libbpf_version.h |   2 +-
 4 files changed, 268 insertions(+), 53 deletions(-)
tools/lib/bpf/libbpf.c +152 -52
index 1da4a438ba00..43161fdd44bb 100644
@@ -201,13 +201,6 @@ struct reloc_desc {
 	};
 };
 
-typedef int (*libbpf_prog_setup_fn_t)(struct bpf_program *prog, long cookie);
-typedef int (*libbpf_prog_prepare_load_fn_t)(struct bpf_program *prog,
-					     struct bpf_prog_load_opts *opts, long cookie);
-/* If auto-attach is not supported, callback should return 0 and set link to NULL */
-typedef int (*libbpf_prog_attach_fn_t)(const struct bpf_program *prog, long cookie,
-				       struct bpf_link **link);
-
 /* stored as sec_def->cookie for all libbpf-supported SEC()s */
 enum sec_def_flags {
 	SEC_NONE = 0,
@@ -235,10 +228,11 @@ enum sec_def_flags {
 };
 
 struct bpf_sec_def {
-	const char *sec;
+	char *sec;
 	enum bpf_prog_type prog_type;
 	enum bpf_attach_type expected_attach_type;
 	long cookie;
+	int handler_id;
 
 	libbpf_prog_setup_fn_t prog_setup_fn;
 	libbpf_prog_prepare_load_fn_t prog_prepare_load_fn;
@@ -8590,7 +8584,7 @@ int bpf_program__set_log_buf(struct bpf_program *prog, char *log_buf, size_t log
 }
 
 #define SEC_DEF(sec_pfx, ptype, atype, flags, ...) {			    \
-	.sec = sec_pfx,							    \
+	.sec = (char *)sec_pfx,						    \
 	.prog_type = BPF_PROG_TYPE_##ptype,				    \
 	.expected_attach_type = atype,					    \
 	.cookie = (long)(flags),					    \
@@ -8683,61 +8677,167 @@ static const struct bpf_sec_def section_defs[] = {
 	SEC_DEF("sk_lookup",		SK_LOOKUP, BPF_SK_LOOKUP, SEC_ATTACHABLE | SEC_SLOPPY_PFX),
 };
 
-#define MAX_TYPE_NAME_SIZE 32
+static size_t custom_sec_def_cnt;
+static struct bpf_sec_def *custom_sec_defs;
+static struct bpf_sec_def custom_fallback_def;
+static bool has_custom_fallback_def;
 
-static const struct bpf_sec_def *find_sec_def(const char *sec_name)
+static int last_custom_sec_def_handler_id;
+
+int libbpf_register_prog_handler(const char *sec,
+				 enum bpf_prog_type prog_type,
+				 enum bpf_attach_type exp_attach_type,
+				 const struct libbpf_prog_handler_opts *opts)
 {
-	const struct bpf_sec_def *sec_def;
-	enum sec_def_flags sec_flags;
-	int i, n = ARRAY_SIZE(section_defs), len;
-	bool strict = libbpf_mode & LIBBPF_STRICT_SEC_NAME;
+	struct bpf_sec_def *sec_def;
 
-	for (i = 0; i < n; i++) {
-		sec_def = &section_defs[i];
-		sec_flags = sec_def->cookie;
-		len = strlen(sec_def->sec);
+	if (!OPTS_VALID(opts, libbpf_prog_handler_opts))
+		return libbpf_err(-EINVAL);
 
-		/* "type/" always has to have proper SEC("type/extras") form */
-		if (sec_def->sec[len - 1] == '/') {
-			if (str_has_pfx(sec_name, sec_def->sec))
-				return sec_def;
-			continue;
-		}
+	if (last_custom_sec_def_handler_id == INT_MAX) /* prevent overflow */
+		return libbpf_err(-E2BIG);
 
-		/* "type+" means it can be either exact SEC("type") or
-		 * well-formed SEC("type/extras") with proper '/' separator
-		 */
-		if (sec_def->sec[len - 1] == '+') {
-			len--;
-			/* not even a prefix */
-			if (strncmp(sec_name, sec_def->sec, len) != 0)
-				continue;
-			/* exact match or has '/' separator */
-			if (sec_name[len] == '\0' || sec_name[len] == '/')
-				return sec_def;
-			continue;
-		}
+	if (sec) {
+		sec_def = libbpf_reallocarray(custom_sec_defs, custom_sec_def_cnt + 1,
+					      sizeof(*sec_def));
+		if (!sec_def)
+			return libbpf_err(-ENOMEM);
 
-		/* SEC_SLOPPY_PFX definitions are allowed to be just prefix
-		 * matches, unless strict section name mode
-		 * (LIBBPF_STRICT_SEC_NAME) is enabled, in which case the
-		 * match has to be exact.
-		 */
-		if ((sec_flags & SEC_SLOPPY_PFX) && !strict)  {
-			if (str_has_pfx(sec_name, sec_def->sec))
-				return sec_def;
-			continue;
-		}
+		custom_sec_defs = sec_def;
+		sec_def = &custom_sec_defs[custom_sec_def_cnt];
+	} else {
+		if (has_custom_fallback_def)
+			return libbpf_err(-EBUSY);
 
-		/* Definitions not marked SEC_SLOPPY_PFX (e.g.,
-		 * SEC("syscall")) are exact matches in both modes.
-		 */
-		if (strcmp(sec_name, sec_def->sec) == 0)
+		sec_def = &custom_fallback_def;
+	}
+
+	sec_def->sec = sec ? strdup(sec) : NULL;
+	if (sec && !sec_def->sec)
+		return libbpf_err(-ENOMEM);
+
+	sec_def->prog_type = prog_type;
+	sec_def->expected_attach_type = exp_attach_type;
+	sec_def->cookie = OPTS_GET(opts, cookie, 0);
+
+	sec_def->prog_setup_fn = OPTS_GET(opts, prog_setup_fn, NULL);
+	sec_def->prog_prepare_load_fn = OPTS_GET(opts, prog_prepare_load_fn, NULL);
+	sec_def->prog_attach_fn = OPTS_GET(opts, prog_attach_fn, NULL);
+
+	sec_def->handler_id = ++last_custom_sec_def_handler_id;
+
+	if (sec)
+		custom_sec_def_cnt++;
+	else
+		has_custom_fallback_def = true;
+
+	return sec_def->handler_id;
+}
+
+int libbpf_unregister_prog_handler(int handler_id)
+{
+	struct bpf_sec_def *sec_defs;
+	int i;
+
+	if (handler_id <= 0)
+		return libbpf_err(-EINVAL);
+
+	if (has_custom_fallback_def && custom_fallback_def.handler_id == handler_id) {
+		memset(&custom_fallback_def, 0, sizeof(custom_fallback_def));
+		has_custom_fallback_def = false;
+		return 0;
+	}
+
+	for (i = 0; i < custom_sec_def_cnt; i++) {
+		if (custom_sec_defs[i].handler_id == handler_id)
+			break;
+	}
+
+	if (i == custom_sec_def_cnt)
+		return libbpf_err(-ENOENT);
+
+	free(custom_sec_defs[i].sec);
+	for (i = i + 1; i < custom_sec_def_cnt; i++)
+		custom_sec_defs[i - 1] = custom_sec_defs[i];
+	custom_sec_def_cnt--;
+
+	/* try to shrink the array, but it's ok if we couldn't */
+	sec_defs = libbpf_reallocarray(custom_sec_defs, custom_sec_def_cnt, sizeof(*sec_defs));
+	if (sec_defs)
+		custom_sec_defs = sec_defs;
+
+	return 0;
+}
+
+static bool sec_def_matches(const struct bpf_sec_def *sec_def, const char *sec_name,
+			    bool allow_sloppy)
+{
+	size_t len = strlen(sec_def->sec);
+
+	/* "type/" always has to have proper SEC("type/extras") form */
+	if (sec_def->sec[len - 1] == '/') {
+		if (str_has_pfx(sec_name, sec_def->sec))
+			return true;
+		return false;
+	}
+
+	/* "type+" means it can be either exact SEC("type") or
+	 * well-formed SEC("type/extras") with proper '/' separator
+	 */
+	if (sec_def->sec[len - 1] == '+') {
+		len--;
+		/* not even a prefix */
+		if (strncmp(sec_name, sec_def->sec, len) != 0)
+			return false;
+		/* exact match or has '/' separator */
+		if (sec_name[len] == '\0' || sec_name[len] == '/')
+			return true;
+		return false;
+	}
+
+	/* SEC_SLOPPY_PFX definitions are allowed to be just prefix
+	 * matches, unless strict section name mode
+	 * (LIBBPF_STRICT_SEC_NAME) is enabled, in which case the
+	 * match has to be exact.
+	 */
+	if (allow_sloppy && str_has_pfx(sec_name, sec_def->sec))
+		return true;
+
+	/* Definitions not marked SEC_SLOPPY_PFX (e.g.,
+	 * SEC("syscall")) are exact matches in both modes.
+	 */
+	return strcmp(sec_name, sec_def->sec) == 0;
+}
+
+static const struct bpf_sec_def *find_sec_def(const char *sec_name)
+{
+	const struct bpf_sec_def *sec_def;
+	int i, n;
+	bool strict = libbpf_mode & LIBBPF_STRICT_SEC_NAME, allow_sloppy;
+
+	n = custom_sec_def_cnt;
+	for (i = 0; i < n; i++) {
+		sec_def = &custom_sec_defs[i];
+		if (sec_def_matches(sec_def, sec_name, false))
+			return sec_def;
+	}
+
+	n = ARRAY_SIZE(section_defs);
+	for (i = 0; i < n; i++) {
+		sec_def = &section_defs[i];
+		allow_sloppy = (sec_def->cookie & SEC_SLOPPY_PFX) && !strict;
+		if (sec_def_matches(sec_def, sec_name, allow_sloppy))
 			return sec_def;
 	}
+
+	if (has_custom_fallback_def)
+		return &custom_fallback_def;
+
 	return NULL;
 }
 
+#define MAX_TYPE_NAME_SIZE 32
+
 static char *libbpf_get_type_names(bool attach_type)
 {
 	int i, len = ARRAY_SIZE(section_defs) * MAX_TYPE_NAME_SIZE;
tools/lib/bpf/libbpf.h +109 -0

115 lines not on this page · raw message

tools/lib/bpf/libbpf.map +6 -0
index 47e70c9058d9..df1b947792c8 100644
@@ -439,3 +439,9 @@ LIBBPF_0.7.0 {
 		libbpf_probe_bpf_prog_type;
 		libbpf_set_memlock_rlim_max;
 } LIBBPF_0.6.0;
+
+LIBBPF_0.8.0 {
+	global:
+		libbpf_register_prog_handler;
+		libbpf_unregister_prog_handler;
+} LIBBPF_0.7.0;
tools/lib/bpf/libbpf_version.h +1 -1
index 0fefefc3500b..61f2039404b6 100644
@@ -4,6 +4,6 @@
 #define __LIBBPF_VERSION_H
 
 #define LIBBPF_MAJOR_VERSION 0
-#define LIBBPF_MINOR_VERSION 7
+#define LIBBPF_MINOR_VERSION 8
 
 #endif /* __LIBBPF_VERSION_H */
Signature · 1 line
2.30.2
author maintainer
#3 lore raw

PATCH v5 3/3 selftests/bpf: add custom SEC() handling selftest

In reply to Andrii Nakryiko

Add a selftest validating various aspects of libbpf's handling of custom SEC() handlers. It also demonstrates how libraries can ensure very early callbacks registration and unregistration using __attribute__((constructor))/__attribute__((destructor)) functions.

---
 .../bpf/prog_tests/custom_sec_handlers.c      | 176 ++++++++++++++++++
 .../bpf/progs/test_custom_sec_handlers.c      |  63 +++++++
 2 files changed, 239 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/custom_sec_handlers.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_custom_sec_handlers.c
tools/testing/selftests/bpf/prog_tests/custom_sec_handlers.c new file +176 -0
new file mode 100644
index 000000000000..b2dfc5954aea
@@ -0,0 +1,176 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2022 Facebook */
+
+#include <test_progs.h>
+#include "test_custom_sec_handlers.skel.h"
+
+#define COOKIE_ABC1 1
+#define COOKIE_ABC2 2
+#define COOKIE_CUSTOM 3
+#define COOKIE_FALLBACK 4
+#define COOKIE_KPROBE 5
+
+static int custom_setup_prog(struct bpf_program *prog, long cookie)
+{
+	if (cookie == COOKIE_ABC1)
+		bpf_program__set_autoload(prog, false);
+
+	return 0;
+}
+
+static int custom_prepare_load_prog(struct bpf_program *prog,
+				    struct bpf_prog_load_opts *opts, long cookie)
+{
+	if (cookie == COOKIE_FALLBACK)
+		opts->prog_flags |= BPF_F_SLEEPABLE;
+	else if (cookie == COOKIE_ABC1)
+		ASSERT_FALSE(true, "unexpected preload for abc");
+
+	return 0;
+}
+
+static int custom_attach_prog(const struct bpf_program *prog, long cookie,
+			      struct bpf_link **link)
+{
+	switch (cookie) {
+	case COOKIE_ABC2:
+		*link = bpf_program__attach_raw_tracepoint(prog, "sys_enter");
+		return libbpf_get_error(*link);
+	case COOKIE_CUSTOM:
+		*link = bpf_program__attach_tracepoint(prog, "syscalls", "sys_enter_nanosleep");
+		return libbpf_get_error(*link);
+	case COOKIE_KPROBE:
+	case COOKIE_FALLBACK:
+		/* no auto-attach for SEC("xyz") and SEC("kprobe") */
+		*link = NULL;
+		return 0;
+	default:
+		ASSERT_FALSE(true, "unexpected cookie");
+		return -EINVAL;
+	}
+}
+
+static int abc1_id;
+static int abc2_id;
+static int custom_id;
+static int fallback_id;
+static int kprobe_id;
+
+__attribute__((constructor))
+static void register_sec_handlers(void)
+{
+	LIBBPF_OPTS(libbpf_prog_handler_opts, abc1_opts,
+		.cookie = COOKIE_ABC1,
+		.prog_setup_fn = custom_setup_prog,
+		.prog_prepare_load_fn = custom_prepare_load_prog,
+		.prog_attach_fn = NULL,
+	);
+	LIBBPF_OPTS(libbpf_prog_handler_opts, abc2_opts,
+		.cookie = COOKIE_ABC2,
+		.prog_setup_fn = custom_setup_prog,
+		.prog_prepare_load_fn = custom_prepare_load_prog,
+		.prog_attach_fn = custom_attach_prog,
+	);
+	LIBBPF_OPTS(libbpf_prog_handler_opts, custom_opts,
+		.cookie = COOKIE_CUSTOM,
+		.prog_setup_fn = NULL,
+		.prog_prepare_load_fn = NULL,
+		.prog_attach_fn = custom_attach_prog,
+	);
+
+	abc1_id = libbpf_register_prog_handler("abc", BPF_PROG_TYPE_RAW_TRACEPOINT, 0, &abc1_opts);
+	abc2_id = libbpf_register_prog_handler("abc/", BPF_PROG_TYPE_RAW_TRACEPOINT, 0, &abc2_opts);
+	custom_id = libbpf_register_prog_handler("custom+", BPF_PROG_TYPE_TRACEPOINT, 0, &custom_opts);
+}
+
+__attribute__((destructor))
+static void unregister_sec_handlers(void)
+{
+	libbpf_unregister_prog_handler(abc1_id);
+	libbpf_unregister_prog_handler(abc2_id);
+	libbpf_unregister_prog_handler(custom_id);
+}
+
+void test_custom_sec_handlers(void)
+{
+	LIBBPF_OPTS(libbpf_prog_handler_opts, opts,
+		.prog_setup_fn = custom_setup_prog,
+		.prog_prepare_load_fn = custom_prepare_load_prog,
+		.prog_attach_fn = custom_attach_prog,
+	);
+	struct test_custom_sec_handlers* skel;
+	int err;
+
+	ASSERT_GT(abc1_id, 0, "abc1_id");
+	ASSERT_GT(abc2_id, 0, "abc2_id");
+	ASSERT_GT(custom_id, 0, "custom_id");
+
+	/* override libbpf's handle of SEC("kprobe/...") but also allow pure
+	 * SEC("kprobe") due to "kprobe+" specifier. Register it as
+	 * TRACEPOINT, just for fun.
+	 */
+	opts.cookie = COOKIE_KPROBE;
+	kprobe_id = libbpf_register_prog_handler("kprobe+", BPF_PROG_TYPE_TRACEPOINT, 0, &opts);
+	/* fallback treats everything as BPF_PROG_TYPE_SYSCALL program to test
+	 * setting custom BPF_F_SLEEPABLE bit in preload handler
+	 */
+	opts.cookie = COOKIE_FALLBACK;
+	fallback_id = libbpf_register_prog_handler(NULL, BPF_PROG_TYPE_SYSCALL, 0, &opts);
+
+	if (!ASSERT_GT(fallback_id, 0, "fallback_id") /* || !ASSERT_GT(kprobe_id, 0, "kprobe_id")*/) {
+		if (fallback_id > 0)
+			libbpf_unregister_prog_handler(fallback_id);
+		if (kprobe_id > 0)
+			libbpf_unregister_prog_handler(kprobe_id);
+		return;
+	}
+
+	/* open skeleton and validate assumptions */
+	skel = test_custom_sec_handlers__open();
+	if (!ASSERT_OK_PTR(skel, "skel_open"))
+		goto cleanup;
+
+	ASSERT_EQ(bpf_program__type(skel->progs.abc1), BPF_PROG_TYPE_RAW_TRACEPOINT, "abc1_type");
+	ASSERT_FALSE(bpf_program__autoload(skel->progs.abc1), "abc1_autoload");
+
+	ASSERT_EQ(bpf_program__type(skel->progs.abc2), BPF_PROG_TYPE_RAW_TRACEPOINT, "abc2_type");
+	ASSERT_EQ(bpf_program__type(skel->progs.custom1), BPF_PROG_TYPE_TRACEPOINT, "custom1_type");
+	ASSERT_EQ(bpf_program__type(skel->progs.custom2), BPF_PROG_TYPE_TRACEPOINT, "custom2_type");
+	ASSERT_EQ(bpf_program__type(skel->progs.kprobe1), BPF_PROG_TYPE_TRACEPOINT, "kprobe1_type");
+	ASSERT_EQ(bpf_program__type(skel->progs.xyz), BPF_PROG_TYPE_SYSCALL, "xyz_type");
+
+	skel->rodata->my_pid = getpid();
+
+	/* now attempt to load everything */
+	err = test_custom_sec_handlers__load(skel);
+	if (!ASSERT_OK(err, "skel_load"))
+		goto cleanup;
+
+	/* now try to auto-attach everything */
+	err = test_custom_sec_handlers__attach(skel);
+	if (!ASSERT_OK(err, "skel_attach"))
+		goto cleanup;
+
+	skel->links.xyz = bpf_program__attach(skel->progs.kprobe1);
+	ASSERT_EQ(errno, EOPNOTSUPP, "xyz_attach_err");
+	ASSERT_ERR_PTR(skel->links.xyz, "xyz_attach");
+
+	/* trigger programs */
+	usleep(1);
+
+	/* SEC("abc") is set to not auto-loaded */
+	ASSERT_FALSE(skel->bss->abc1_called, "abc1_called");
+	ASSERT_TRUE(skel->bss->abc2_called, "abc2_called");
+	ASSERT_TRUE(skel->bss->custom1_called, "custom1_called");
+	ASSERT_TRUE(skel->bss->custom2_called, "custom2_called");
+	/* SEC("kprobe") shouldn't be auto-attached */
+	ASSERT_FALSE(skel->bss->kprobe1_called, "kprobe1_called");
+	/* SEC("xyz") shouldn't be auto-attached */
+	ASSERT_FALSE(skel->bss->xyz_called, "xyz_called");
+
+cleanup:
+	test_custom_sec_handlers__destroy(skel);
+
+	ASSERT_OK(libbpf_unregister_prog_handler(fallback_id), "unregister_fallback");
+	ASSERT_OK(libbpf_unregister_prog_handler(kprobe_id), "unregister_kprobe");
+}
tools/testing/selftests/bpf/progs/test_custom_sec_handlers.c new file +63 -0

63 lines not on this page · raw message

Signature · 1 line
2.30.2
#4 lore raw
In reply to Andrii Nakryiko
This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Hello:

This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Quoted 9 lines · On Fri, 4 Mar 2022 17:01:26 -0800 you wrote:

Add ability for user applications and libraries to register custom BPF program SEC() handlers. See patch #2 for examples where this is useful.

Patch #1 does some preliminary refactoring to allow exponsing program init, preload, and attach callbacks as public API. It also establishes a protocol to allow optional auto-attach behavior. This will also help the case of sometimes auto-attachable uprobes.

[...]

Here is the summary with links:
- [v5,bpf-next,1/3] libbpf: allow BPF program auto-attach handlers to bail out

    https://git.kernel.org/bpf/bpf-next/c/4fa5bcfe07f7
  - [v5,bpf-next,2/3] libbpf: support custom SEC() handlers
    https://git.kernel.org/bpf/bpf-next/c/697f104db8a6
  - [v5,bpf-next,3/3] selftests/bpf: add custom SEC() handling selftest
    https://git.kernel.org/bpf/bpf-next/c/aa963bcb0adc

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html