Skip to contentSkip to conversation

bpftool: adding support for BTF program names

PATCH Superseded Raman Shukhau 2 participants 2 messages Jan 18, 2022 – Jan 19, 2022

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bpftool: adding support for BTF program names

bpftool prog list and other bpftool subcommands that show BPF program names currently get them from bpf_prog_info.name. That field is limited by 16 (BPF_OBJ_NAME_LEN) chars what leads to truncated names since many progs have much longer names.

The idea of this change is to improve all bpftool commands that output prog name so that bpftool uses info from BTF to print program names if available.

It tries bpf_prog_info.name first and fall back to btf only if the name is suspected to be truncated (has 15 chars length).

Right now bpftool p show id <id> returns capped prog name

<id>: kprobe name example_cap_cap tag 712e...
...

With this change it would return

<id>: kprobe name example_cap_capable tag 712e...
...

Note, other commands that prints prog names (e.g. "bpftool cgroup tree") are also addressed in this change.

---
 tools/bpf/bpftool/cgroup.c |  6 ++++--
 tools/bpf/bpftool/common.c | 34 ++++++++++++++++++++++++++++++++++
 tools/bpf/bpftool/main.h   |  2 ++
 tools/bpf/bpftool/prog.c   | 17 +++++++++--------
 4 files changed, 49 insertions(+), 10 deletions(-)
tools/bpf/bpftool/cgroup.c +4 -2
index 3571a281c43f..5e098d9772ae 100644
@@ -73,7 +73,8 @@ static int show_bpf_prog(int id, enum bpf_attach_type attach_type,
 			jsonw_uint_field(json_wtr, "attach_type", attach_type);
 		jsonw_string_field(json_wtr, "attach_flags",
 				   attach_flags_str);
-		jsonw_string_field(json_wtr, "name", info.name);
+		jsonw_string_field(json_wtr, "name",
+				   get_prog_full_name(&info, prog_fd));
 		jsonw_end_object(json_wtr);
 	} else {
 		printf("%s%-8u ", level ? "    " : "", info.id);
@@ -81,7 +82,8 @@ static int show_bpf_prog(int id, enum bpf_attach_type attach_type,
 			printf("%-15s", attach_type_name[attach_type]);
 		else
 			printf("type %-10u", attach_type);
-		printf(" %-15s %-15s\n", attach_flags_str, info.name);
+		printf(" %-15s %-15s\n", attach_flags_str,
+		       get_prog_full_name(&info, prog_fd));
 	}
 
 	close(prog_fd);
tools/bpf/bpftool/common.c +34 -0
index fa8eb8134344..b94d0020e4b4 100644
@@ -24,6 +24,7 @@
 #include <bpf/bpf.h>
 #include <bpf/hashmap.h>
 #include <bpf/libbpf.h> /* libbpf_num_possible_cpus */
+#include <bpf/btf.h>
 
 #include "main.h"
 
@@ -304,6 +305,39 @@ const char *get_fd_type_name(enum bpf_obj_type type)
 	return names[type];
 }
 
+const char *get_prog_full_name(const struct bpf_prog_info *prog_info,
+			       int prog_fd)
+{
+	const struct btf_type *func_type;
+	const struct bpf_func_info finfo;
+	struct bpf_prog_info info = {};
+	__u32 info_len = sizeof(info);
+	const struct btf *prog_btf;
+
+	if (strlen(prog_info->name) < BPF_OBJ_NAME_LEN - 1)
+		return prog_info->name;
+
+	if (!prog_info->btf_id || prog_info->nr_func_info == 0)
+		return prog_info->name;
+
+	info.nr_func_info = 1;
+	info.func_info_rec_size = prog_info->func_info_rec_size;
+	info.func_info = ptr_to_u64(&finfo);
+
+	if (bpf_obj_get_info_by_fd(prog_fd, &info, &info_len))
+		return prog_info->name;
+
+	prog_btf = btf__load_from_kernel_by_id(info.btf_id);
+	if (libbpf_get_error(prog_btf))
+		return prog_info->name;
+
+	func_type = btf__type_by_id(prog_btf, finfo.type_id);
+	if (!func_type || !btf_is_func(func_type))
+		return prog_info->name;
+
+	return btf__name_by_offset(prog_btf, func_type->name_off);
+}
+
 int get_fd_type(int fd)
 {
 	char path[PATH_MAX];
tools/bpf/bpftool/main.h +2 -0
index 8d76d937a62b..cbd03cc821b7 100644
@@ -142,6 +142,8 @@ int cmd_select(const struct cmd *cmds, int argc, char **argv,
 
 int get_fd_type(int fd);
 const char *get_fd_type_name(enum bpf_obj_type type);
+const char *get_prog_full_name(const struct bpf_prog_info *prog_info,
+			       int prog_fd);
 char *get_fdinfo(int fd, const char *key);
 int open_obj_pinned(const char *path, bool quiet);
 int open_obj_pinned_any(const char *path, enum bpf_obj_type exp_type);
tools/bpf/bpftool/prog.c +9 -8
index 33ca834d5f51..e54d97ee998c 100644
@@ -424,7 +424,7 @@ static void show_prog_metadata(int fd, __u32 num_maps)
 	free(value);
 }
 
-static void print_prog_header_json(struct bpf_prog_info *info)
+static void print_prog_header_json(struct bpf_prog_info *info, int fd)
 {
 	jsonw_uint_field(json_wtr, "id", info->id);
 	if (info->type < ARRAY_SIZE(prog_type_name))
@@ -434,7 +434,8 @@ static void print_prog_header_json(struct bpf_prog_info *info)
 		jsonw_uint_field(json_wtr, "type", info->type);
 
 	if (*info->name)
-		jsonw_string_field(json_wtr, "name", info->name);
+		jsonw_string_field(json_wtr, "name",
+				   get_prog_full_name(info, fd));
 
 	jsonw_name(json_wtr, "tag");
 	jsonw_printf(json_wtr, "\"" BPF_TAG_FMT "\"",
@@ -455,7 +456,7 @@ static void print_prog_json(struct bpf_prog_info *info, int fd)
 	char *memlock;
 
 	jsonw_start_object(json_wtr);
-	print_prog_header_json(info);
+	print_prog_header_json(info, fd);
 	print_dev_json(info->ifindex, info->netns_dev, info->netns_ino);
 
 	if (info->load_time) {
@@ -507,7 +508,7 @@ static void print_prog_json(struct bpf_prog_info *info, int fd)
 	jsonw_end_object(json_wtr);
 }
 
-static void print_prog_header_plain(struct bpf_prog_info *info)
+static void print_prog_header_plain(struct bpf_prog_info *info, int fd)
 {
 	printf("%u: ", info->id);
 	if (info->type < ARRAY_SIZE(prog_type_name))
@@ -516,7 +517,7 @@ static void print_prog_header_plain(struct bpf_prog_info *info)
 		printf("type %u  ", info->type);
 
 	if (*info->name)
-		printf("name %s  ", info->name);
+		printf("name %s  ", get_prog_full_name(info, fd));
 
 	printf("tag ");
 	fprint_hex(stdout, info->tag, BPF_TAG_SIZE, "");
@@ -534,7 +535,7 @@ static void print_prog_plain(struct bpf_prog_info *info, int fd)
 {
 	char *memlock;
 
-	print_prog_header_plain(info);
+	print_prog_header_plain(info, fd);
 
 	if (info->load_time) {
 		char buf[32];
@@ -972,10 +973,10 @@ static int do_dump(int argc, char **argv)
 
 		if (json_output && nb_fds > 1) {
 			jsonw_start_object(json_wtr);	/* prog object */
-			print_prog_header_json(&info);
+			print_prog_header_json(&info, fds[i]);
 			jsonw_name(json_wtr, "insns");
 		} else if (nb_fds > 1) {
-			print_prog_header_plain(&info);
+			print_prog_header_plain(&info, fds[i]);
 		}
 
 		err = prog_dump(&info, mode, filepath, opcodes, visual, linum);
Signature · 1 line
2.30.2
#1 lore raw
Quoted 4 lines · Raman Shukhau

bpftool prog list and other bpftool subcommands that show BPF program names currently get them from bpf_prog_info.name. That field is limited by 16 (BPF_OBJ_NAME_LEN) chars what leads

typo: limited to? which leads?

Quoted 14 lines

to truncated names since many progs have much longer names.

The idea of this change is to improve all bpftool commands that output prog name so that bpftool uses info from BTF to print program names if available.

It tries bpf_prog_info.name first and fall back to btf only if the name is suspected to be truncated (has 15 chars length).

Right now bpftool p show id <id> returns capped prog name

<id>: kprobe name example_cap_cap tag 712e...
...

Quoted 6 lines

With this change it would return

<id>: kprobe name example_cap_capable tag 712e...
...

Note, other commands that prints prog names (e.g. "bpftool

typo: print

Quoted 47 lines

cgroup tree") are also addressed in this change.

---
 tools/bpf/bpftool/cgroup.c |  6 ++++--
 tools/bpf/bpftool/common.c | 34 ++++++++++++++++++++++++++++++++++
 tools/bpf/bpftool/main.h   |  2 ++
 tools/bpf/bpftool/prog.c   | 17 +++++++++--------
 4 files changed, 49 insertions(+), 10 deletions(-)
tools/bpf/bpftool/cgroup.c +4 -2
index 3571a281c43f..5e098d9772ae 100644
@@ -73,7 +73,8 @@ static int show_bpf_prog(int id, enum bpf_attach_type attach_type,
                        jsonw_uint_field(json_wtr, "attach_type", attach_type);
                jsonw_string_field(json_wtr, "attach_flags",
                                   attach_flags_str);
-               jsonw_string_field(json_wtr, "name", info.name);
+               jsonw_string_field(json_wtr, "name",
+                                  get_prog_full_name(&info, prog_fd));
                jsonw_end_object(json_wtr);
        } else {
                printf("%s%-8u ", level ? "    " : "", info.id);
@@ -81,7 +82,8 @@ static int show_bpf_prog(int id, enum bpf_attach_type attach_type,
                        printf("%-15s", attach_type_name[attach_type]);
                else
                        printf("type %-10u", attach_type);
-               printf(" %-15s %-15s\n", attach_flags_str, info.name);
+               printf(" %-15s %-15s\n", attach_flags_str,
+                      get_prog_full_name(&info, prog_fd));
        }
 
        close(prog_fd);
tools/bpf/bpftool/common.c +1 -0
index fa8eb8134344..b94d0020e4b4 100644
@@ -24,6 +24,7 @@
 #include <bpf/bpf.h>
 #include <bpf/hashmap.h>
 #include <bpf/libbpf.h> /* libbpf_num_possible_cpus */
+#include <bpf/btf.h>
 
 #include "main.h"
 

tools/bpf/bpftool/common.c const char *get_fd_type_name(enum bpf_obj_type type) · line 309

Quoted 5 lines
tools/bpf/bpftool/common.c +2 -0
@@ -304,6 +305,39 @@ const char *get_fd_type_name(enum bpf_obj_type type)
        return names[type];
 }
 
+const char *get_prog_full_name(const struct bpf_prog_info *prog_info,
+                              int prog_fd)

see below about leaks and corrupted pointer, you need to handle string returning more carefully. Either alway strdup() and then free in each caller (which honestly sucks from usability and safety point of view) or let callers pass buffer and buffer size and make sure you put the result into that buffer (and truncate, if necessary).

Quoted 5 lines
tools/bpf/bpftool/common.c +5 -0
+{
+       const struct btf_type *func_type;
+       const struct bpf_func_info finfo;
+       struct bpf_prog_info info = {};
+       __u32 info_len = sizeof(info);

tools/bpf/bpftool/common.c const char *get_fd_type_name(enum bpf_obj_type type) · line 324

Quoted 10 lines
tools/bpf/bpftool/common.c +10 -0
+       const struct btf *prog_btf;
+
+       if (strlen(prog_info->name) < BPF_OBJ_NAME_LEN - 1)
+               return prog_info->name;
+
+       if (!prog_info->btf_id || prog_info->nr_func_info == 0)
+               return prog_info->name;
+
+       info.nr_func_info = 1;
+       info.func_info_rec_size = prog_info->func_info_rec_size;

if prog_info->func_info_rec_size is bigger than sizeof(finfo) you'll let kernel overwrite stack (corrupting info, probably). Should this be min of sizeof(finfo) and prog_info->func_info_rec_size?

tools/bpf/bpftool/common.c const char *get_fd_type_name(enum bpf_obj_type type) · line 331

Quoted 7 lines
tools/bpf/bpftool/common.c +7 -0
+       info.func_info = ptr_to_u64(&finfo);
+
+       if (bpf_obj_get_info_by_fd(prog_fd, &info, &info_len))
+               return prog_info->name;
+
+       prog_btf = btf__load_from_kernel_by_id(info.btf_id);
+       if (libbpf_get_error(prog_btf))

bpftool is using libbpf 1.0 strict mode, so prog_btf will be NULL on error, so just check for NULL instead of libbpf_get_error().

tools/bpf/bpftool/common.c const char *get_fd_type_name(enum bpf_obj_type type) · line 336

Quoted 5 lines
tools/bpf/bpftool/common.c +5 -0
+               return prog_info->name;
+
+       func_type = btf__type_by_id(prog_btf, finfo.type_id);
+       if (!func_type || !btf_is_func(func_type))
+               return prog_info->name;

leaking prog_btf here

tools/bpf/bpftool/common.c const char *get_fd_type_name(enum bpf_obj_type type) · line 338

Quoted 2 lines
tools/bpf/bpftool/common.c +2 -0
+
+       return btf__name_by_offset(prog_btf, func_type->name_off);

and here. Also, here you are returning a pointer to a string inside that prog_btf, so once you fix up the leak you'll be returning invalid pointer.

Quoted 5 lines
tools/bpf/bpftool/common.c +2 -0
+}
+
 int get_fd_type(int fd)
 {
        char path[PATH_MAX];

[...]