tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs

Add bpf_iter-based way to find all the processes that hold open FDs against
BPF object (map, prog, link, btf). bpftool always attempts to discover this,
but will silently give up if kernel doesn't yet support bpf_iter BPF programs.
Process name and PID are emitted for each process (task group).

Sample output for each of 4 BPF objects:

$ sudo ./bpftool prog show
2694: cgroup_device  tag 8c42dee26e8cd4c2  gpl
        loaded_at 2020-06-16T15:34:32-0700  uid 0
        xlated 648B  jited 409B  memlock 4096B
        pids systemd(1)
2907: cgroup_skb  name egress  tag 9ad187367cf2b9e8  gpl
        loaded_at 2020-06-16T18:06:54-0700  uid 0
        xlated 48B  jited 59B  memlock 4096B  map_ids 2436
        btf_id 1202
        pids test_progs(2238417), test_progs(22384459)

$ sudo ./bpftool map show
2436: array  name test_cgr.bss  flags 0x400
        key 4B  value 8B  max_entries 1  memlock 8192B
        btf_id 1202
        pids test_progs(2238417), test_progs(22384459)
2445: array  name pid_iter.rodata  flags 0x480
        key 4B  value 4B  max_entries 1  memlock 8192B
        btf_id 1214  frozen
        pids bpftool(2239612)

$ sudo ./bpftool link show
61: cgroup  prog 2908
        cgroup_id 375301  attach_type egress
        pids test_progs(2238417), test_progs(22384459)
62: cgroup  prog 2908
        cgroup_id 375344  attach_type egress
        pids test_progs(2238417), test_progs(22384459)

$ sudo ./bpftool btf show
1202: size 1527B  prog_ids 2908,2907  map_ids 2436
        pids test_progs(2238417), test_progs(22384459)
1242: size 34684B
        pids bpftool(2258892)
Signed-off-by: NAndrii Nakryiko <andriin@fb.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Reviewed-by: NQuentin Monnet <quentin@isovalent.com>
Link: https://lore.kernel.org/bpf/20200619231703.738941-9-andriin@fb.com

tools/bpf/bpftool/Makefile

@@ -150,6 +150,8 @@ $(OUTPUT)%.skel.h: $(OUTPUT)%.bpf.o $(BPFTOOL_BOOTSTRAP)
 
 $(OUTPUT)prog.o: $(OUTPUT)profiler.skel.h
 
+$(OUTPUT)pids.o: $(OUTPUT)pid_iter.skel.h
+
 endif
 endif
 

tools/bpf/bpftool/btf.c

@@ -809,6 +809,7 @@ show_btf_plain(struct bpf_btf_info *info, int fd,
 			printf("%s%u", n++ == 0 ? "  map_ids " : ",",
 			       obj->obj_id);
 	}
+	emit_obj_refs_plain(&refs_table, info->id, "\n\tpids ");
 
 	printf("\n");
 }
@@ -841,6 +842,9 @@ show_btf_json(struct bpf_btf_info *info, int fd,
 			jsonw_uint(json_wtr, obj->obj_id);
 	}
 	jsonw_end_array(json_wtr);	/* map_ids */
+
+	emit_obj_refs_json(&refs_table, info->id, json_wtr); /* pids */
+
 	jsonw_end_object(json_wtr);	/* btf object */
 }
 
@@ -893,6 +897,7 @@ static int do_show(int argc, char **argv)
 			close(fd);
 		return err;
 	}
+	build_obj_refs_table(&refs_table, BPF_OBJ_BTF);
 
 	if (fd >= 0) {
 		err = show_btf(fd, &btf_prog_table, &btf_map_table);
@@ -939,6 +944,7 @@ static int do_show(int argc, char **argv)
 exit_free:
 	delete_btf_table(&btf_prog_table);
 	delete_btf_table(&btf_map_table);
+	delete_obj_refs_table(&refs_table);
 
 	return err;
 }

tools/bpf/bpftool/link.c

@@ -143,6 +143,9 @@ static int show_link_close_json(int fd, struct bpf_link_info *info)
 		}
 		jsonw_end_array(json_wtr);
 	}
+
+	emit_obj_refs_json(&refs_table, info->id, json_wtr);
+
 	jsonw_end_object(json_wtr);
 
 	return 0;
@@ -212,6 +215,7 @@ static int show_link_close_plain(int fd, struct bpf_link_info *info)
 				printf("\n\tpinned %s", obj->path);
 		}
 	}
+	emit_obj_refs_plain(&refs_table, info->id, "\n\tpids ");
 
 	printf("\n");
 
@@ -257,6 +261,7 @@ static int do_show(int argc, char **argv)
 
 	if (show_pinned)
 		build_pinned_obj_table(&link_table, BPF_OBJ_LINK);
+	build_obj_refs_table(&refs_table, BPF_OBJ_LINK);
 
 	if (argc == 2) {
 		fd = link_parse_fd(&argc, &argv);
@@ -296,6 +301,8 @@ static int do_show(int argc, char **argv)
 	if (json_output)
 		jsonw_end_array(json_wtr);
 
+	delete_obj_refs_table(&refs_table);
+
 	return errno == ENOENT ? 0 : -1;
 }
 

tools/bpf/bpftool/main.c

@@ -31,6 +31,7 @@ bool relaxed_maps;
 struct pinned_obj_table prog_table;
 struct pinned_obj_table map_table;
 struct pinned_obj_table link_table;
+struct obj_refs_table refs_table;
 
 static void __noreturn clean_and_exit(int i)
 {

tools/bpf/bpftool/main.h

@@ -127,11 +127,13 @@ static const char * const attach_type_name[__MAX_BPF_ATTACH_TYPE] = {
 extern const char * const map_type_name[];
 extern const size_t map_type_name_size;
 
+/* keep in sync with the definition in skeleton/pid_iter.bpf.c */
 enum bpf_obj_type {
 	BPF_OBJ_UNKNOWN,
 	BPF_OBJ_PROG,
 	BPF_OBJ_MAP,
 	BPF_OBJ_LINK,
+	BPF_OBJ_BTF,
 };
 
 extern const char *bin_name;
@@ -139,12 +141,14 @@ extern const char *bin_name;
 extern json_writer_t *json_wtr;
 extern bool json_output;
 extern bool show_pinned;
+extern bool show_pids;
 extern bool block_mount;
 extern bool verifier_logs;
 extern bool relaxed_maps;
 extern struct pinned_obj_table prog_table;
 extern struct pinned_obj_table map_table;
 extern struct pinned_obj_table link_table;
+extern struct obj_refs_table refs_table;
 
 void __printf(1, 2) p_err(const char *fmt, ...);
 void __printf(1, 2) p_info(const char *fmt, ...);
@@ -168,12 +172,35 @@ struct pinned_obj {
 	struct hlist_node hash;
 };
 
+struct obj_refs_table {
+	DECLARE_HASHTABLE(table, 16);
+};
+
+struct obj_ref {
+	int pid;
+	char comm[16];
+};
+
+struct obj_refs {
+	struct hlist_node node;
+	__u32 id;
+	int ref_cnt;
+	struct obj_ref *refs;
+};
+
 struct btf;
 struct bpf_line_info;
 
 int build_pinned_obj_table(struct pinned_obj_table *table,
 			   enum bpf_obj_type type);
 void delete_pinned_obj_table(struct pinned_obj_table *tab);
+__weak int build_obj_refs_table(struct obj_refs_table *table,
+				enum bpf_obj_type type);
+__weak void delete_obj_refs_table(struct obj_refs_table *table);
+__weak void emit_obj_refs_json(struct obj_refs_table *table, __u32 id,
+			       json_writer_t *json_wtr);
+__weak void emit_obj_refs_plain(struct obj_refs_table *table, __u32 id,
+				const char *prefix);
 void print_dev_plain(__u32 ifindex, __u64 ns_dev, __u64 ns_inode);
 void print_dev_json(__u32 ifindex, __u64 ns_dev, __u64 ns_inode);
 

tools/bpf/bpftool/map.c

@@ -509,6 +509,8 @@ static int show_map_close_json(int fd, struct bpf_map_info *info)
 		jsonw_end_array(json_wtr);
 	}
 
+	emit_obj_refs_json(&refs_table, info->id, json_wtr);
+
 	jsonw_end_object(json_wtr);
 
 	return 0;
@@ -596,6 +598,8 @@ static int show_map_close_plain(int fd, struct bpf_map_info *info)
 	if (frozen)
 		printf("%sfrozen", info->btf_id ? "  " : "");
 
+	emit_obj_refs_plain(&refs_table, info->id, "\n\tpids ");
+
 	printf("\n");
 	return 0;
 }
@@ -654,6 +658,7 @@ static int do_show(int argc, char **argv)
 
 	if (show_pinned)
 		build_pinned_obj_table(&map_table, BPF_OBJ_MAP);
+	build_obj_refs_table(&refs_table, BPF_OBJ_MAP);
 
 	if (argc == 2)
 		return do_show_subset(argc, argv);
@@ -697,6 +702,8 @@ static int do_show(int argc, char **argv)
 	if (json_output)
 		jsonw_end_array(json_wtr);
 
+	delete_obj_refs_table(&refs_table);
+
 	return errno == ENOENT ? 0 : -1;
 }
 

tools/bpf/bpftool/pids.c

+// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+/* Copyright (C) 2020 Facebook */
+#include <errno.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <bpf/bpf.h>
+
+#include "main.h"
+#include "skeleton/pid_iter.h"
+
+#ifdef BPFTOOL_WITHOUT_SKELETONS
+
+int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
+{
+	p_err("bpftool built without PID iterator support");
+	return -ENOTSUP;
+}
+void delete_obj_refs_table(struct obj_refs_table *table) {}
+
+#else /* BPFTOOL_WITHOUT_SKELETONS */
+
+#include "pid_iter.skel.h"
+
+static void add_ref(struct obj_refs_table *table, struct pid_iter_entry *e)
+{
+	struct obj_refs *refs;
+	struct obj_ref *ref;
+	void *tmp;
+	int i;
+
+	hash_for_each_possible(table->table, refs, node, e->id) {
+		if (refs->id != e->id)
+			continue;
+
+		for (i = 0; i < refs->ref_cnt; i++) {
+			if (refs->refs[i].pid == e->pid)
+				return;
+		}
+
+		tmp = realloc(refs->refs, (refs->ref_cnt + 1) * sizeof(*ref));
+		if (!tmp) {
+			p_err("failed to re-alloc memory for ID %u, PID %d, COMM %s...",
+			      e->id, e->pid, e->comm);
+			return;
+		}
+		refs->refs = tmp;
+		ref = &refs->refs[refs->ref_cnt];
+		ref->pid = e->pid;
+		memcpy(ref->comm, e->comm, sizeof(ref->comm));
+		refs->ref_cnt++;
+
+		return;
+	}
+
+	/* new ref */
+	refs = calloc(1, sizeof(*refs));
+	if (!refs) {
+		p_err("failed to alloc memory for ID %u, PID %d, COMM %s...",
+		      e->id, e->pid, e->comm);
+		return;
+	}
+
+	refs->id = e->id;
+	refs->refs = malloc(sizeof(*refs->refs));
+	if (!refs->refs) {
+		free(refs);
+		p_err("failed to alloc memory for ID %u, PID %d, COMM %s...",
+		      e->id, e->pid, e->comm);
+		return;
+	}
+	ref = &refs->refs[0];
+	ref->pid = e->pid;
+	memcpy(ref->comm, e->comm, sizeof(ref->comm));
+	refs->ref_cnt = 1;
+	hash_add(table->table, &refs->node, e->id);
+}
+
+static int __printf(2, 0)
+libbpf_print_none(__maybe_unused enum libbpf_print_level level,
+		  __maybe_unused const char *format,
+		  __maybe_unused va_list args)
+{
+	return 0;
+}
+
+int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
+{
+	char buf[4096];
+	struct pid_iter_bpf *skel;
+	struct pid_iter_entry *e;
+	int err, ret, fd = -1, i;
+	libbpf_print_fn_t default_print;
+
+	hash_init(table->table);
+	set_max_rlimit();
+
+	skel = pid_iter_bpf__open();
+	if (!skel) {
+		p_err("failed to open PID iterator skeleton");
+		return -1;
+	}
+
+	skel->rodata->obj_type = type;
+
+	/* we don't want output polluted with libbpf errors if bpf_iter is not
+	 * supported
+	 */
+	default_print = libbpf_set_print(libbpf_print_none);
+	err = pid_iter_bpf__load(skel);
+	libbpf_set_print(default_print);
+	if (err) {
+		/* too bad, kernel doesn't support BPF iterators yet */
+		err = 0;
+		goto out;
+	}
+	err = pid_iter_bpf__attach(skel);
+	if (err) {
+		/* if we loaded above successfully, attach has to succeed */
+		p_err("failed to attach PID iterator: %d", err);
+		goto out;
+	}
+
+	fd = bpf_iter_create(bpf_link__fd(skel->links.iter));
+	if (fd < 0) {
+		err = -errno;
+		p_err("failed to create PID iterator session: %d", err);
+		goto out;
+	}
+
+	while (true) {
+		ret = read(fd, buf, sizeof(buf));
+		if (ret < 0) {
+			err = -errno;
+			p_err("failed to read PID iterator output: %d", err);
+			goto out;
+		}
+		if (ret == 0)
+			break;
+		if (ret % sizeof(*e)) {
+			err = -EINVAL;
+			p_err("invalid PID iterator output format");
+			goto out;
+		}
+		ret /= sizeof(*e);
+
+		e = (void *)buf;
+		for (i = 0; i < ret; i++, e++) {
+			add_ref(table, e);
+		}
+	}
+	err = 0;
+out:
+	if (fd >= 0)
+		close(fd);
+	pid_iter_bpf__destroy(skel);
+	return err;
+}
+
+void delete_obj_refs_table(struct obj_refs_table *table)
+{
+	struct obj_refs *refs;
+	struct hlist_node *tmp;
+	unsigned int bkt;
+
+	hash_for_each_safe(table->table, bkt, tmp, refs, node) {
+		hash_del(&refs->node);
+		free(refs->refs);
+		free(refs);
+	}
+}
+
+void emit_obj_refs_json(struct obj_refs_table *table, __u32 id, json_writer_t *json_wtr)
+{
+	struct obj_refs *refs;
+	struct obj_ref *ref;
+	int i;
+
+	if (hash_empty(table->table))
+		return;
+
+	hash_for_each_possible(table->table, refs, node, id) {
+		if (refs->id != id)
+			continue;
+		if (refs->ref_cnt == 0)
+			break;
+
+		jsonw_name(json_wtr, "pids");
+		jsonw_start_array(json_wtr);
+		for (i = 0; i < refs->ref_cnt; i++) {
+			ref = &refs->refs[i];
+			jsonw_start_object(json_wtr);
+			jsonw_int_field(json_wtr, "pid", ref->pid);
+			jsonw_string_field(json_wtr, "comm", ref->comm);
+			jsonw_end_object(json_wtr);
+		}
+		jsonw_end_array(json_wtr);
+		break;
+	}
+}
+
+void emit_obj_refs_plain(struct obj_refs_table *table, __u32 id, const char *prefix)
+{
+	struct obj_refs *refs;
+	struct obj_ref *ref;
+	int i;
+
+	if (hash_empty(table->table))
+		return;
+
+	hash_for_each_possible(table->table, refs, node, id) {
+		if (refs->id != id)
+			continue;
+		if (refs->ref_cnt == 0)
+			break;
+
+		printf("%s", prefix);
+		for (i = 0; i < refs->ref_cnt; i++) {
+			ref = &refs->refs[i];
+			printf("%s%s(%d)", i == 0 ? "" : ", ", ref->comm, ref->pid);
+		}
+		break;
+	}
+}
+
+
+#endif

tools/bpf/bpftool/prog.c

@@ -190,6 +190,8 @@ static void print_prog_json(struct bpf_prog_info *info, int fd)
 		jsonw_end_array(json_wtr);
 	}
 
+	emit_obj_refs_json(&refs_table, info->id, json_wtr);
+
 	jsonw_end_object(json_wtr);
 }
 
@@ -256,6 +258,8 @@ static void print_prog_plain(struct bpf_prog_info *info, int fd)
 	if (info->btf_id)
 		printf("\n\tbtf_id %d", info->btf_id);
 
+	emit_obj_refs_plain(&refs_table, info->id, "\n\tpids ");
+
 	printf("\n");
 }
 
@@ -321,6 +325,7 @@ static int do_show(int argc, char **argv)
 
 	if (show_pinned)
 		build_pinned_obj_table(&prog_table, BPF_OBJ_PROG);
+	build_obj_refs_table(&refs_table, BPF_OBJ_PROG);
 
 	if (argc == 2)
 		return do_show_subset(argc, argv);
@@ -362,6 +367,8 @@ static int do_show(int argc, char **argv)
 	if (json_output)
 		jsonw_end_array(json_wtr);
 
+	delete_obj_refs_table(&refs_table);
+
 	return err;
 }
 

tools/bpf/bpftool/skeleton/pid_iter.bpf.c

+// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+/* Copyright (c) 2020 Facebook */
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_core_read.h>
+#include <bpf/bpf_tracing.h>
+#include "pid_iter.h"
+
+/* keep in sync with the definition in main.h */
+enum bpf_obj_type {
+	BPF_OBJ_UNKNOWN,
+	BPF_OBJ_PROG,
+	BPF_OBJ_MAP,
+	BPF_OBJ_LINK,
+	BPF_OBJ_BTF,
+};
+
+extern const void bpf_link_fops __ksym;
+extern const void bpf_map_fops __ksym;
+extern const void bpf_prog_fops __ksym;
+extern const void btf_fops __ksym;
+
+const volatile enum bpf_obj_type obj_type = BPF_OBJ_UNKNOWN;
+
+static __always_inline __u32 get_obj_id(void *ent, enum bpf_obj_type type)
+{
+	switch (type) {
+	case BPF_OBJ_PROG:
+		return BPF_CORE_READ((struct bpf_prog *)ent, aux, id);
+	case BPF_OBJ_MAP:
+		return BPF_CORE_READ((struct bpf_map *)ent, id);
+	case BPF_OBJ_BTF:
+		return BPF_CORE_READ((struct btf *)ent, id);
+	case BPF_OBJ_LINK:
+		return BPF_CORE_READ((struct bpf_link *)ent, id);
+	default:
+		return 0;
+	}
+}
+
+SEC("iter/task_file")
+int iter(struct bpf_iter__task_file *ctx)
+{
+	struct file *file = ctx->file;
+	struct task_struct *task = ctx->task;
+	struct pid_iter_entry e;
+	const void *fops;
+
+	if (!file || !task)
+		return 0;
+
+	switch (obj_type) {
+	case BPF_OBJ_PROG:
+		fops = &bpf_prog_fops;
+		break;
+	case BPF_OBJ_MAP:
+		fops = &bpf_map_fops;
+		break;
+	case BPF_OBJ_BTF:
+		fops = &btf_fops;
+		break;
+	case BPF_OBJ_LINK:
+		fops = &bpf_link_fops;
+		break;
+	default:
+		return 0;
+	}
+
+	if (file->f_op != fops)
+		return 0;
+
+	e.pid = task->tgid;
+	e.id = get_obj_id(file->private_data, obj_type);
+	bpf_probe_read(&e.comm, sizeof(e.comm), task->group_leader->comm);
+	bpf_seq_write(ctx->meta->seq, &e, sizeof(e));
+
+	return 0;
+}
+
+char LICENSE[] SEC("license") = "Dual BSD/GPL";

tools/bpf/bpftool/skeleton/pid_iter.h

+/* SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) */
+/* Copyright (c) 2020 Facebook */
+#ifndef __PID_ITER_H
+#define __PID_ITER_H
+
+struct pid_iter_entry {
+	__u32 id;
+	int pid;
+	char comm[16];
+};
+
+#endif