Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf

Alexei Starovoitov says: ==================== pull-request: bpf 2020-08-21 The following pull-request contains BPF updates for your *net* tree. We've added 11 non-merge commits during the last 5 day(s) which contain a total of 12 files changed, 78 insertions(+), 24 deletions(-). The main changes are: 1) three fixes in BPF task iterator logic, from Yonghong. 2) fix for compressed dwarf sections in vmlinux, from Jiri. 3) fix xdp attach regression, from Andrii. ====================

David S. Miller committed 6 years ago

Commit 4af7b322 parents 774d977 b16fc0912 files · +78 −24Browse files

Unified · Split

12 files

include/uapi/linux/bpf.h +5 −5
kernel/bpf/bpf_iter.c +14 −1
kernel/bpf/task_iter.c +4 −2
net/core/dev.c +8 −6
tools/bpf/bpftool/pids.c +2 −0
tools/bpf/resolve_btfids/main.c +36 −0
tools/include/uapi/linux/bpf.h +5 −5
tools/lib/bpf/btf_dump.c +1 −1
tools/lib/bpf/libbpf.c +1 −1
tools/testing/selftests/bpf/.gitignore +0 −1
tools/testing/selftests/bpf/Makefile +1 −1
tools/testing/selftests/bpf/prog_tests/test_global_funcs.c +1 −1

include/uapi/linux/bpf.h+5−5

kernel/bpf/bpf_iter.c+14−1

 	iter_priv->done_stop = true;
 }
+/* maximum visited objects before bailing out */
+#define MAX_ITER_OBJECTS	1000000
 /* bpf_seq_read, a customized and simpler version for bpf iterator.
  * no_llseek is assumed for this file.
  * The following are differences from seq_read():
 {
 	struct seq_file *seq = file->private_data;
 	size_t n, offs, copied = 0;
-	int err = 0;
+	int err = 0, num_objs = 0;
 	void *p;
 	mutex_lock(&seq->lock);
 	while (1) {
 		loff_t pos = seq->index;
+		num_objs++;
 		offs = seq->count;
 		p = seq->op->next(seq, p, &seq->index);
 		if (pos == seq->index) {
 		if (seq->count >= size)
 			break;
+		if (num_objs >= MAX_ITER_OBJECTS) {
+			if (offs == 0) {
+				err = -EAGAIN;
+				seq->op->stop(seq, p);
+				goto done;
+			}
+			break;
+		}
 		err = seq->op->show(seq, p);
 		if (err > 0) {
 			bpf_iter_dec_seq_num(seq);

kernel/bpf/task_iter.c+4−2

 	rcu_read_lock();
 retry:
-	pid = idr_get_next(&ns->idr, tid);
+	pid = find_ge_pid(*tid, ns);
 	if (pid) {
+		*tid = pid_nr_ns(pid, ns);
 		task = get_pid_task(pid, PIDTYPE_PID);
 		if (!task) {
 			++*tid;
 		f = fcheck_files(curr_files, curr_fd);
 		if (!f)
 			continue;
+		if (!get_file_rcu(f))
+			continue;
 		/* set info->fd */
 		info->fd = curr_fd;
-		get_file(f);
 		rcu_read_unlock();
 		return f;
 	}

net/core/dev.c+8−6

 	int flags;
 };
-static enum bpf_xdp_mode dev_xdp_mode(u32 flags)
+static enum bpf_xdp_mode dev_xdp_mode(struct net_device *dev, u32 flags)
 {
 	if (flags & XDP_FLAGS_HW_MODE)
 		return XDP_MODE_HW;
 	if (flags & XDP_FLAGS_DRV_MODE)
 		return XDP_MODE_DRV;
-	return XDP_MODE_SKB;
+	if (flags & XDP_FLAGS_SKB_MODE)
+		return XDP_MODE_SKB;
+	return dev->netdev_ops->ndo_bpf ? XDP_MODE_DRV : XDP_MODE_SKB;
 }
 static bpf_op_t dev_xdp_bpf_op(struct net_device *dev, enum bpf_xdp_mode mode)
 		return -EINVAL;
 	}
-	mode = dev_xdp_mode(flags);
+	mode = dev_xdp_mode(dev, flags);
 	/* can't replace attached link */
 	if (dev_xdp_link(dev, mode)) {
 		NL_SET_ERR_MSG(extack, "Can't replace active BPF XDP link");
 	ASSERT_RTNL();
-	mode = dev_xdp_mode(link->flags);
+	mode = dev_xdp_mode(dev, link->flags);
 	if (dev_xdp_link(dev, mode) != link)
 		return -EINVAL;
 		goto out_unlock;
 	}
-	mode = dev_xdp_mode(xdp_link->flags);
+	mode = dev_xdp_mode(xdp_link->dev, xdp_link->flags);
 	bpf_op = dev_xdp_bpf_op(xdp_link->dev, mode);
 	err = dev_xdp_install(xdp_link->dev, mode, bpf_op, NULL,
 			      xdp_link->flags, new_prog);
 int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
 		      int fd, int expected_fd, u32 flags)
 {
-	enum bpf_xdp_mode mode = dev_xdp_mode(flags);
+	enum bpf_xdp_mode mode = dev_xdp_mode(dev, flags);
 	struct bpf_prog *new_prog = NULL, *old_prog = NULL;
 	int err;

tools/bpf/bpftool/pids.c+2−0

 	while (true) {
 		ret = read(fd, buf, sizeof(buf));
 		if (ret < 0) {
+			if (errno == EAGAIN)
+				continue;
 			err = -errno;
 			p_err("failed to read PID iterator output: %d", err);
 			goto out;

tools/bpf/resolve_btfids/main.c+36−0

@@ -233,6 +233,39 @@ static struct btf_id add_symbol(struct rb_root root, char *name, size_t size) · 232 unmodified lines
233	return btf_id__add(root, id, false);	233	return btf_id__add(root, id, false);
234	}	234	}
235		235
		236	/*
		237	* The data of compressed section should be aligned to 4
		238	* (for 32bit) or 8 (for 64 bit) bytes. The binutils ld
		239	* sets sh_addralign to 1, which makes libelf fail with
		240	* misaligned section error during the update:
		241	* FAILED elf_update(WRITE): invalid section alignment
		242	*
		243	* While waiting for ld fix, we fix the compressed sections
		244	* sh_addralign value manualy.
		245	*/
		246	static int compressed_section_fix(Elf elf, Elf_Scn scn, GElf_Shdr *sh)
		247	{
		248	int expected = gelf_getclass(elf) == ELFCLASS32 ? 4 : 8;
		249
		250	if (!(sh->sh_flags & SHF_COMPRESSED))
		251	return 0;
		252
		253	if (sh->sh_addralign == expected)
		254	return 0;
		255
		256	pr_debug2(" - fixing wrong alignment sh_addralign %u, expected %u\n",
		257	sh->sh_addralign, expected);
		258
		259	sh->sh_addralign = expected;
		260
		261	if (gelf_update_shdr(scn, sh) == 0) {
		262	printf("FAILED cannot update section header: %s\n",
		263	elf_errmsg(-1));
		264	return -1;
		265	}
		266	return 0;
		267	}
		268
236	static int elf_collect(struct object *obj)	269	static int elf_collect(struct object *obj)
237	{	270	{
238	Elf_Scn *scn = NULL;	271	Elf_Scn *scn = NULL;
@@ -309,6 +342,9 @@ static int elf_collect(struct object *obj) · 70 unmodified lines
309	obj->efile.idlist_shndx = idx;	342	obj->efile.idlist_shndx = idx;
310	obj->efile.idlist_addr = sh.sh_addr;	343	obj->efile.idlist_addr = sh.sh_addr;
311	}	344	}
		345
		346	if (compressed_section_fix(elf, scn, &sh))
		347	return -1;
312	}	348	}
313		349
314	return 0;	350	return 0;

tools/include/uapi/linux/bpf.h+5−5

tools/lib/bpf/btf_dump.c+1−1

@@ -879,7 +879,7 @@ static void btf_dump_emit_struct_def(struct btf_dump *d, · 878 unmodified lines
879	btf_dump_printf(d, ": %d", m_sz);	879	btf_dump_printf(d, ": %d", m_sz);
880	off = m_off + m_sz;	880	off = m_off + m_sz;
881	} else {	881	} else {
882	m_sz = max(0LL, btf__resolve_size(d->btf, m->type));	882	m_sz = max((__s64)0, btf__resolve_size(d->btf, m->type));
883	off = m_off + m_sz * 8;	883	off = m_off + m_sz * 8;
884	}	884	}
885	btf_dump_printf(d, ";");	885	btf_dump_printf(d, ";");

tools/lib/bpf/libbpf.c+1−1

@@ -2264,7 +2264,7 @@ static int bpf_object__init_user_btf_maps(struct bpf_object *obj, bool strict, · 2263 unmodified lines
2264	data = elf_getdata(scn, NULL);	2264	data = elf_getdata(scn, NULL);
2265	if (!scn \|\| !data) {	2265	if (!scn \|\| !data) {
2266	pr_warn("failed to get Elf_Data from map section %d (%s)\n",	2266	pr_warn("failed to get Elf_Data from map section %d (%s)\n",
2267	obj->efile.maps_shndx, MAPS_ELF_SEC);	2267	obj->efile.btf_maps_shndx, MAPS_ELF_SEC);
2268	return -EINVAL;	2268	return -EINVAL;
2269	}	2269	}
2270		2270

tools/testing/selftests/bpf/.gitignore+0−1

@@ -6,7 +6,6 @@ test_lpm_map · 5 unmodified lines
6	test_tag	6	test_tag
7	FEATURE-DUMP.libbpf	7	FEATURE-DUMP.libbpf
8	fixdep	8	fixdep
9	test_align
10	test_dev_cgroup	9	test_dev_cgroup
11	/test_progs*	10	/test_progs*
12	test_tcpbpf_user	11	test_tcpbpf_user

tools/testing/selftests/bpf/Makefile+1−1

@@ -32,7 +32,7 @@ LDLIBS += -lcap -lelf -lz -lrt -lpthread · 31 unmodified lines
32		32
33	# Order correspond to 'make run_tests' order	33	# Order correspond to 'make run_tests' order
34	TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test_progs \	34	TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test_progs \
35	test_align test_verifier_log test_dev_cgroup test_tcpbpf_user \	35	test_verifier_log test_dev_cgroup test_tcpbpf_user \
36	test_sock test_btf test_sockmap get_cgroup_id_user test_socket_cookie \	36	test_sock test_btf test_sockmap get_cgroup_id_user test_socket_cookie \
37	test_cgroup_storage \	37	test_cgroup_storage \
38	test_netcnt test_tcpnotify_user test_sock_fields test_sysctl \	38	test_netcnt test_tcpnotify_user test_sock_fields test_sysctl \

tools/testing/selftests/bpf/prog_tests/test_global_funcs.c+1−1

@@ -19,7 +19,7 @@ static int libbpf_debug_print(enum libbpf_print_level level, · 18 unmodified lines
19	log_buf = va_arg(args, char *);	19	log_buf = va_arg(args, char *);
20	if (!log_buf)	20	if (!log_buf)
21	goto out;	21	goto out;
22	if (strstr(log_buf, err_str) == 0)	22	if (err_str && strstr(log_buf, err_str) == 0)
23	found = true;	23	found = true;
24	out:	24	out:
25	printf(format, log_buf);	25	printf(format, log_buf);

@@ -767,7 +767,7 @@ union bpf_attr { · 766 unmodified lines
767	*	767	*
768	* Also, note that bpf_trace_printk\ () is slow, and should	768	* Also, note that bpf_trace_printk\ () is slow, and should
769	* only be used for debugging purposes. For this reason, a notice	769	* only be used for debugging purposes. For this reason, a notice
770	* bloc (spanning several lines) is printed to kernel logs and	770	* block (spanning several lines) is printed to kernel logs and
771	* states that the helper should not be used "for production use"	771	* states that the helper should not be used "for production use"
772	* the first time this helper is used (or more precisely, when	772	* the first time this helper is used (or more precisely, when
773	* trace_printk\ () buffers are allocated). For passing values	773	* trace_printk\ () buffers are allocated). For passing values
@@ -1033,14 +1033,14 @@ union bpf_attr { · 259 unmodified lines
1033	*	1033	*
1034	* int ret;	1034	* int ret;
1035	* struct bpf_tunnel_key key = {};	1035	* struct bpf_tunnel_key key = {};
1036	*	1036	*
1037	* ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), 0);	1037	* ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), 0);
1038	* if (ret < 0)	1038	* if (ret < 0)
1039	* return TC_ACT_SHOT; // drop packet	1039	* return TC_ACT_SHOT; // drop packet
1040	*	1040	*
1041	* if (key.remote_ipv4 != 0x0a000001)	1041	* if (key.remote_ipv4 != 0x0a000001)
1042	* return TC_ACT_SHOT; // drop packet	1042	* return TC_ACT_SHOT; // drop packet
1043	*	1043	*
1044	* return TC_ACT_OK; // accept packet	1044	* return TC_ACT_OK; // accept packet
1045	*	1045	*
1046	* This interface can also be used with all encapsulation devices	1046	* This interface can also be used with all encapsulation devices
@@ -1147,7 +1147,7 @@ union bpf_attr { · 100 unmodified lines
1147	* Description	1147	* Description
1148	* Retrieve the realm or the route, that is to say the	1148	* Retrieve the realm or the route, that is to say the
1149	* tclassid field of the destination for the skb. The	1149	* tclassid field of the destination for the skb. The
1150	* indentifier retrieved is a user-provided tag, similar to the	1150	* identifier retrieved is a user-provided tag, similar to the
1151	* one used with the net_cls cgroup (see description for	1151	* one used with the net_cls cgroup (see description for
1152	* bpf_get_cgroup_classid\ () helper), but here this tag is	1152	* bpf_get_cgroup_classid\ () helper), but here this tag is
1153	* held by a route (a destination entry), not by a task.	1153	* held by a route (a destination entry), not by a task.