[PATCH bpf-next v6 0/4] Relax tracing prog recursive attach rules

[PATCH bpf-next v6 0/4] Relax tracing prog recursive attach rules

[Dmitrii Dolgov]

Currently, it's not allowed to attach an fentry/fexit prog to another
fentry/fexit. At the same time it's not uncommon to see a tracing
program with lots of logic in use, and the attachment limitation
prevents usage of fentry/fexit for performance analysis (e.g. with
"bpftool prog profile" command) in this case. An example could be
falcosecurity libs project that uses tp_btf tracing programs for
offloading certain part of logic into tail-called programs, but the
use-case is still generic enough -- a tracing program could be
complicated and heavy enough to warrant its profiling, yet frustratingly
it's not possible to do so use best tooling for that.

Following the corresponding discussion [1], the reason for that is to
avoid tracing progs call cycles without introducing more complex
solutions. But currently it seems impossible to load and attach tracing
programs in a way that will form such a cycle. Replace "no same type"
requirement with verification that no more than one level of attachment
nesting is allowed. In this way only one fentry/fexit program could be
attached to another fentry/fexit to cover profiling use case, and still
no cycle could be formed.

The series contains a test for recursive attachment, as well as a fix +
test for an issue in re-attachment branch of bpf_tracing_prog_attach.
When preparing the test for the main change set, I've stumbled upon the
possibility to construct a sequence of events when attach_btf would be
NULL while computing a trampoline key. It doesn't look like this issue
is triggered by the main change, because the reproduces doesn't actually
need to have an fentry attachment chain.

[1]: https://lore.kernel.org/bpf/20191108064039.2041889-16-ast@kernel.org/

Dmitrii Dolgov (3):
  bpf: Relax tracing prog recursive attach rules
  selftests/bpf: Add test for recursive attachment of tracing progs
  selftests/bpf: Test re-attachment fix for bpf_tracing_prog_attach

Jiri Olsa (1):
  bpf: Fix re-attachment branch in bpf_tracing_prog_attach

 include/linux/bpf.h                           |   1 +
 include/uapi/linux/bpf.h                      |   1 +
 kernel/bpf/syscall.c                          |  21 ++++
 kernel/bpf/verifier.c                         |  33 ++---
 tools/include/uapi/linux/bpf.h                |   1 +
 .../bpf/prog_tests/recursive_attach.c         | 117 ++++++++++++++++++
 .../selftests/bpf/progs/fentry_recursive.c    |  19 +++
 .../bpf/progs/fentry_recursive_target.c       |  31 +++++
 8 files changed, 210 insertions(+), 14 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/recursive_attach.c
 create mode 100644 tools/testing/selftests/bpf/progs/fentry_recursive.c
 create mode 100644 tools/testing/selftests/bpf/progs/fentry_recursive_target.c


base-commit: 40d0eb0259ae77ace3e81d7454d1068c38bc95c2
-- 
2.41.0

[PATCH bpf-next v6 1/4] bpf: Relax tracing prog recursive attach rules

[Dmitrii Dolgov]

Currently, it's not allowed to attach an fentry/fexit prog to another
one fentry/fexit. At the same time it's not uncommon to see a tracing
program with lots of logic in use, and the attachment limitation
prevents usage of fentry/fexit for performance analysis (e.g. with
"bpftool prog profile" command) in this case. An example could be
falcosecurity libs project that uses tp_btf tracing programs.

Following the corresponding discussion [1], the reason for that is to
avoid tracing progs call cycles without introducing more complex
solutions. But currently it seems impossible to load and attach tracing
programs in a way that will form such a cycle. The limitation is coming
from the fact that attach_prog_fd is specified at the prog load (thus
making it impossible to attach to a program loaded after it in this
way), as well as tracing progs not implementing link_detach.

Replace "no same type" requirement with verification that no more than
one level of attachment nesting is allowed. In this way only one
fentry/fexit program could be attached to another fentry/fexit to cover
profiling use case, and still no cycle could be formed. To implement,
add a new field into bpf_prog_aux to track the depth of attachment.

[1]: https://lore.kernel.org/bpf/20191108064039.2041889-16-ast@kernel.org/

Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
---
Previous discussion: https://lore.kernel.org/bpf/20231201154734.8545-1-9erthalion6@gmail.com/

Changes in v6:
    - Apply nesting level limitation only to tracing programs, otherwise
      it's possible to apply it in "fentry->extension" case and break it

Changes in v5:
    - Remove follower_cnt and drop unreachable cycle prevention condition
    - Allow only one level of attachment nesting
    - Do not display attach_depth in bpftool, as it doesn't make sense
      anymore

Changes in v3:
    - Fix incorrect decreasing of attach_depth, setting to 0 instead
    - Place bookkeeping later, to not miss a cleanup if needed
    - Display attach_depth in bpftool only if the value is not 0

Changes in v2:
    - Verify tgt_prog is not null
    - Replace boolean followed with number of followers, to handle
      multiple progs attaching/detaching

 include/linux/bpf.h            |  1 +
 include/uapi/linux/bpf.h       |  1 +
 kernel/bpf/syscall.c           | 12 ++++++++++++
 kernel/bpf/verifier.c          | 33 +++++++++++++++++++--------------
 tools/include/uapi/linux/bpf.h |  1 +
 5 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index eb447b0a9423..1588e48fe31c 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1400,6 +1400,7 @@ struct bpf_prog_aux {
 	u32 real_func_cnt; /* includes hidden progs, only used for JIT and freeing progs */
 	u32 func_idx; /* 0 for non-func prog, the index in func array for func prog */
 	u32 attach_btf_id; /* in-kernel BTF type id to attach to */
+	u32 attach_depth; /* for tracing prog, level of attachment nesting */
 	u32 ctx_arg_info_size;
 	u32 max_rdonly_access;
 	u32 max_rdwr_access;
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index e88746ba7d21..9cf45ad914f1 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -6468,6 +6468,7 @@ struct bpf_prog_info {
 	__u32 verified_insns;
 	__u32 attach_btf_obj_id;
 	__u32 attach_btf_id;
+	__u32 attach_depth;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 5e43ddd1b83f..9c56b5970d7e 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3039,6 +3039,7 @@ static void bpf_tracing_link_release(struct bpf_link *link)
 
 	bpf_trampoline_put(tr_link->trampoline);
 
+	link->prog->aux->attach_depth = 0;
 	/* tgt_prog is NULL if target is a kernel function */
 	if (tr_link->tgt_prog)
 		bpf_prog_put(tr_link->tgt_prog);
@@ -3243,6 +3244,16 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 		goto out_unlock;
 	}
 
+	if (tgt_prog) {
+		/* Bookkeeping for managing the prog attachment chain. If it's a
+		 * tracing program, bump the level, otherwise carry it on.
+		 */
+		if (prog->type == BPF_PROG_TYPE_TRACING)
+			prog->aux->attach_depth = tgt_prog->aux->attach_depth + 1;
+		else
+			prog->aux->attach_depth = tgt_prog->aux->attach_depth;
+	}
+
 	link->tgt_prog = tgt_prog;
 	link->trampoline = tr;
 
@@ -4510,6 +4521,7 @@ static int bpf_prog_get_info_by_fd(struct file *file,
 	if (prog->aux->btf)
 		info.btf_id = btf_obj_id(prog->aux->btf);
 	info.attach_btf_id = prog->aux->attach_btf_id;
+	info.attach_depth = prog->aux->attach_depth;
 	if (attach_btf)
 		info.attach_btf_obj_id = btf_obj_id(attach_btf);
 
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 8e7b6072e3f4..a3ea2087a6b2 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -20109,6 +20109,11 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
 	if (tgt_prog) {
 		struct bpf_prog_aux *aux = tgt_prog->aux;
 
+		if (aux->attach_depth >= 1) {
+			bpf_log(log, "Cannot attach with more than one level of nesting\n");
+			return -EINVAL;
+		}
+
 		if (bpf_prog_is_dev_bound(prog->aux) &&
 		    !bpf_prog_dev_bound_match(prog, tgt_prog)) {
 			bpf_log(log, "Target program bound device mismatch");
@@ -20147,10 +20152,11 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
 			bpf_log(log, "Can attach to only JITed progs\n");
 			return -EINVAL;
 		}
-		if (tgt_prog->type == prog->type) {
-			/* Cannot fentry/fexit another fentry/fexit program.
-			 * Cannot attach program extension to another extension.
-			 * It's ok to attach fentry/fexit to extension program.
+		if (tgt_prog->type == prog->type && prog_extension) {
+			/*
+			 * To avoid potential call chain cycles, prevent attaching of a
+			 * program extension to another extension. It's ok to attach
+			 * fentry/fexit to extension program.
 			 */
 			bpf_log(log, "Cannot recursively attach\n");
 			return -EINVAL;
@@ -20163,16 +20169,15 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
 			 * except fentry/fexit. The reason is the following.
 			 * The fentry/fexit programs are used for performance
 			 * analysis, stats and can be attached to any program
-			 * type except themselves. When extension program is
-			 * replacing XDP function it is necessary to allow
-			 * performance analysis of all functions. Both original
-			 * XDP program and its program extension. Hence
-			 * attaching fentry/fexit to BPF_PROG_TYPE_EXT is
-			 * allowed. If extending of fentry/fexit was allowed it
-			 * would be possible to create long call chain
-			 * fentry->extension->fentry->extension beyond
-			 * reasonable stack size. Hence extending fentry is not
-			 * allowed.
+			 * type. When extension program is replacing XDP function
+			 * it is necessary to allow performance analysis of all
+			 * functions. Both original XDP program and its program
+			 * extension. Hence attaching fentry/fexit to
+			 * BPF_PROG_TYPE_EXT is allowed. If extending of
+			 * fentry/fexit was allowed it would be possible to create
+			 * long call chain fentry->extension->fentry->extension
+			 * beyond reasonable stack size. Hence extending fentry
+			 * is not allowed.
 			 */
 			bpf_log(log, "Cannot extend fentry/fexit\n");
 			return -EINVAL;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index e88746ba7d21..9cf45ad914f1 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -6468,6 +6468,7 @@ struct bpf_prog_info {
 	__u32 verified_insns;
 	__u32 attach_btf_obj_id;
 	__u32 attach_btf_id;
+	__u32 attach_depth;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {
-- 
2.41.0

[PATCH bpf-next v6 2/4] selftests/bpf: Add test for recursive attachment of tracing progs

[Dmitrii Dolgov]

Verify the fact that only one fentry prog could be attached to another
fentry, building up an attachment chain of limited size. Use existing
bpf_testmod as a start of the chain.

Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
---
Changes in v5:
    - Test only one level of attachment

 .../bpf/prog_tests/recursive_attach.c         | 69 +++++++++++++++++++
 .../selftests/bpf/progs/fentry_recursive.c    | 19 +++++
 .../bpf/progs/fentry_recursive_target.c       | 20 ++++++
 3 files changed, 108 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/recursive_attach.c
 create mode 100644 tools/testing/selftests/bpf/progs/fentry_recursive.c
 create mode 100644 tools/testing/selftests/bpf/progs/fentry_recursive_target.c

diff --git a/tools/testing/selftests/bpf/prog_tests/recursive_attach.c b/tools/testing/selftests/bpf/prog_tests/recursive_attach.c
new file mode 100644
index 000000000000..7248d0661ee9
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/recursive_attach.c
@@ -0,0 +1,69 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2023 Red Hat, Inc. */
+#include <test_progs.h>
+#include "fentry_recursive.skel.h"
+#include "fentry_recursive_target.skel.h"
+#include <bpf/btf.h>
+#include "bpf/libbpf_internal.h"
+
+/*
+ * Test following scenarios:
+ * - attach one fentry progs to another one
+ * - more than one nesting levels are not allowed
+ */
+void test_recursive_fentry_attach(void)
+{
+	struct fentry_recursive_target *target_skel = NULL;
+	struct fentry_recursive *tracing_chain[2] = {};
+	struct bpf_program *prog;
+	int prev_fd, err;
+
+	target_skel = fentry_recursive_target__open_and_load();
+	if (!ASSERT_OK_PTR(target_skel, "fentry_recursive_target__open_and_load"))
+		goto close_prog;
+
+	/* Create an attachment chain with two fentry progs */
+	for (int i = 0; i < 2; i++) {
+		tracing_chain[i] = fentry_recursive__open();
+		if (!ASSERT_OK_PTR(tracing_chain[i], "fentry_recursive__open"))
+			goto close_prog;
+
+		/*
+		 * The first prog in the chain is going to be attached to the target
+		 * fentry program, the second one to the previous in the chain.
+		 */
+		if (i == 0) {
+			prog = tracing_chain[0]->progs.recursive_attach;
+			prev_fd = bpf_program__fd(target_skel->progs.test1);
+			err = bpf_program__set_attach_target(prog, prev_fd, "test1");
+		} else {
+			prog = tracing_chain[i]->progs.recursive_attach;
+			prev_fd = bpf_program__fd(tracing_chain[i-1]->progs.recursive_attach);
+			err = bpf_program__set_attach_target(prog, prev_fd, "recursive_attach");
+		}
+
+		if (!ASSERT_OK(err, "bpf_program__set_attach_target"))
+			goto close_prog;
+
+		err = fentry_recursive__load(tracing_chain[i]);
+		/* The first attach should succeed, the second fail */
+		if (i == 0) {
+			if (!ASSERT_OK(err, "fentry_recursive__load"))
+				goto close_prog;
+
+			err = fentry_recursive__attach(tracing_chain[i]);
+			if (!ASSERT_OK(err, "fentry_recursive__attach"))
+				goto close_prog;
+		} else {
+			if (!ASSERT_ERR(err, "fentry_recursive__load"))
+				goto close_prog;
+		}
+	}
+
+close_prog:
+	fentry_recursive_target__destroy(target_skel);
+	for (int i = 0; i < 2; i++) {
+		if (tracing_chain[i])
+			fentry_recursive__destroy(tracing_chain[i]);
+	}
+}
diff --git a/tools/testing/selftests/bpf/progs/fentry_recursive.c b/tools/testing/selftests/bpf/progs/fentry_recursive.c
new file mode 100644
index 000000000000..1df490230344
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/fentry_recursive.c
@@ -0,0 +1,19 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2023 Red Hat, Inc. */
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__u64 test1_result = 0;
+
+/*
+ * Dummy fentry bpf prog for testing fentry attachment chains
+ */
+SEC("fentry/XXX")
+int BPF_PROG(recursive_attach, int a)
+{
+	test1_result = a == 1;
+	return 0;
+}
diff --git a/tools/testing/selftests/bpf/progs/fentry_recursive_target.c b/tools/testing/selftests/bpf/progs/fentry_recursive_target.c
new file mode 100644
index 000000000000..b6fb8ebd598d
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/fentry_recursive_target.c
@@ -0,0 +1,20 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2023 Red Hat, Inc. */
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__u64 test1_result = 0;
+
+/*
+ * Dummy fentry bpf prog for testing fentry attachment chains. It's going to be
+ * a start of the chain.
+ */
+SEC("fentry/bpf_testmod_fentry_test1")
+int BPF_PROG(test1, int a)
+{
+	test1_result = a == 1;
+	return 0;
+}
-- 
2.41.0

[PATCH bpf-next v6 3/4] bpf: Fix re-attachment branch in bpf_tracing_prog_attach

[Dmitrii Dolgov]

From: Jiri Olsa <olsajiri@gmail.com>

The following case can cause a crash due to missing attach_btf:

1) load rawtp program
2) load fentry program with rawtp as target_fd
3) create tracing link for fentry program with target_fd = 0
4) repeat 3

In the end we have:

- prog->aux->dst_trampoline == NULL
- tgt_prog == NULL (because we did not provide target_fd to link_create)
- prog->aux->attach_btf == NULL (the program was loaded with attach_prog_fd=X)
- the program was loaded for tgt_prog but we have no way to find out which one

    BUG: kernel NULL pointer dereference, address: 0000000000000058
    Call Trace:
     <TASK>
     ? __die+0x20/0x70
     ? page_fault_oops+0x15b/0x430
     ? fixup_exception+0x22/0x330
     ? exc_page_fault+0x6f/0x170
     ? asm_exc_page_fault+0x22/0x30
     ? bpf_tracing_prog_attach+0x279/0x560
     ? btf_obj_id+0x5/0x10
     bpf_tracing_prog_attach+0x439/0x560
     __sys_bpf+0x1cf4/0x2de0
     __x64_sys_bpf+0x1c/0x30
     do_syscall_64+0x41/0xf0
     entry_SYSCALL_64_after_hwframe+0x6e/0x76

Return -EINVAL in this situation.

Signed-off-by: Jiri Olsa <olsajiri@gmail.com>
Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
---
 kernel/bpf/syscall.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 9c56b5970d7e..82bb1be4dca2 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3181,6 +3181,10 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 	 *
 	 * - if prog->aux->dst_trampoline and tgt_prog is NULL, the program
 	 *   was detached and is going for re-attachment.
+	 *
+	 * - if prog->aux->dst_trampoline is NULL and tgt_prog and prog->aux->attach_btf
+	 *   are NULL, then program was already attached and user did not provide
+	 *   tgt_prog_fd so we have no way to find out or create trampoline
 	 */
 	if (!prog->aux->dst_trampoline && !tgt_prog) {
 		/*
@@ -3194,6 +3198,11 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 			err = -EINVAL;
 			goto out_unlock;
 		}
+		/* We can allow re-attach only if we have valid attach_btf. */
+		if (!prog->aux->attach_btf) {
+			err = -EINVAL;
+			goto out_unlock;
+		}
 		btf_id = prog->aux->attach_btf_id;
 		key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf, btf_id);
 	}
-- 
2.41.0

[PATCH bpf-next v6 4/4] selftests/bpf: Test re-attachment fix for bpf_tracing_prog_attach

[Dmitrii Dolgov]

Add a test case to verify the fix for "prog->aux->dst_trampoline and
tgt_prog is NULL" branch in bpf_tracing_prog_attach. The sequence of
events:

1. load rawtp program
2. load fentry program with rawtp as target_fd
3. create tracing link for fentry program with target_fd = 0
4. repeat 3

Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
---
 .../bpf/prog_tests/recursive_attach.c         | 48 +++++++++++++++++++
 .../bpf/progs/fentry_recursive_target.c       | 11 +++++
 2 files changed, 59 insertions(+)

diff --git a/tools/testing/selftests/bpf/prog_tests/recursive_attach.c b/tools/testing/selftests/bpf/prog_tests/recursive_attach.c
index 7248d0661ee9..6296bcf95481 100644
--- a/tools/testing/selftests/bpf/prog_tests/recursive_attach.c
+++ b/tools/testing/selftests/bpf/prog_tests/recursive_attach.c
@@ -67,3 +67,51 @@ void test_recursive_fentry_attach(void)
 			fentry_recursive__destroy(tracing_chain[i]);
 	}
 }
+
+/*
+ * Test that a tracing prog reattachment (when we land in
+ * "prog->aux->dst_trampoline and tgt_prog is NULL" branch in
+ * bpf_tracing_prog_attach) does not lead to a crash due to missing attach_btf
+ */
+void test_fentry_attach_btf_presence(void)
+{
+	struct fentry_recursive_target *target_skel = NULL;
+	struct fentry_recursive *tracing_skel = NULL;
+	struct bpf_program *prog;
+	int err, link_fd, tgt_prog_fd;
+
+	target_skel = fentry_recursive_target__open_and_load();
+	if (!ASSERT_OK_PTR(target_skel, "fentry_recursive_target__open_and_load"))
+		goto close_prog;
+
+	tracing_skel = fentry_recursive__open();
+	if (!ASSERT_OK_PTR(tracing_skel, "fentry_recursive__open"))
+		goto close_prog;
+
+	prog = tracing_skel->progs.recursive_attach;
+	tgt_prog_fd = bpf_program__fd(target_skel->progs.fentry_target);
+	err = bpf_program__set_attach_target(prog, tgt_prog_fd, "fentry_target");
+	if (!ASSERT_OK(err, "bpf_program__set_attach_target"))
+		goto close_prog;
+
+	err = fentry_recursive__load(tracing_skel);
+	if (!ASSERT_OK(err, "fentry_recursive__load"))
+		goto close_prog;
+
+	LIBBPF_OPTS(bpf_link_create_opts, link_opts);
+
+	link_fd = bpf_link_create(bpf_program__fd(tracing_skel->progs.recursive_attach),
+							  0, BPF_TRACE_FENTRY, &link_opts);
+	if (!ASSERT_GE(link_fd, 0, "link_fd"))
+		goto close_prog;
+
+	fentry_recursive__detach(tracing_skel);
+
+	err = fentry_recursive__attach(tracing_skel);
+	if (!ASSERT_ERR(err, "fentry_recursive__attach"))
+		goto close_prog;
+
+close_prog:
+	fentry_recursive_target__destroy(target_skel);
+	fentry_recursive__destroy(tracing_skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/fentry_recursive_target.c b/tools/testing/selftests/bpf/progs/fentry_recursive_target.c
index b6fb8ebd598d..f812d2de0c3c 100644
--- a/tools/testing/selftests/bpf/progs/fentry_recursive_target.c
+++ b/tools/testing/selftests/bpf/progs/fentry_recursive_target.c
@@ -18,3 +18,14 @@ int BPF_PROG(test1, int a)
 	test1_result = a == 1;
 	return 0;
 }
+
+/*
+ * Dummy bpf prog for testing attach_btf presence when attaching an fentry
+ * program.
+ */
+SEC("raw_tp/sys_enter")
+int BPF_PROG(fentry_target, struct pt_regs *regs, long id)
+{
+	test1_result = id == 1;
+	return 0;
+}
-- 
2.41.0

Re: [PATCH bpf-next v6 1/4] bpf: Relax tracing prog recursive attach rules

[Jiri Olsa]

On Sat, Dec 02, 2023 at 08:15:47PM +0100, Dmitrii Dolgov wrote:
> Currently, it's not allowed to attach an fentry/fexit prog to another
> one fentry/fexit. At the same time it's not uncommon to see a tracing
> program with lots of logic in use, and the attachment limitation
> prevents usage of fentry/fexit for performance analysis (e.g. with
> "bpftool prog profile" command) in this case. An example could be
> falcosecurity libs project that uses tp_btf tracing programs.
> 
> Following the corresponding discussion [1], the reason for that is to
> avoid tracing progs call cycles without introducing more complex
> solutions. But currently it seems impossible to load and attach tracing
> programs in a way that will form such a cycle. The limitation is coming
> from the fact that attach_prog_fd is specified at the prog load (thus
> making it impossible to attach to a program loaded after it in this
> way), as well as tracing progs not implementing link_detach.
> 
> Replace "no same type" requirement with verification that no more than
> one level of attachment nesting is allowed. In this way only one
> fentry/fexit program could be attached to another fentry/fexit to cover
> profiling use case, and still no cycle could be formed. To implement,
> add a new field into bpf_prog_aux to track the depth of attachment.
> 
> [1]: https://lore.kernel.org/bpf/20191108064039.2041889-16-ast@kernel.org/
> 
> Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
> ---
> Previous discussion: https://lore.kernel.org/bpf/20231201154734.8545-1-9erthalion6@gmail.com/
> 
> Changes in v6:
>     - Apply nesting level limitation only to tracing programs, otherwise
>       it's possible to apply it in "fentry->extension" case and break it
> 
> Changes in v5:
>     - Remove follower_cnt and drop unreachable cycle prevention condition
>     - Allow only one level of attachment nesting
>     - Do not display attach_depth in bpftool, as it doesn't make sense
>       anymore
> 
> Changes in v3:
>     - Fix incorrect decreasing of attach_depth, setting to 0 instead
>     - Place bookkeeping later, to not miss a cleanup if needed
>     - Display attach_depth in bpftool only if the value is not 0
> 
> Changes in v2:
>     - Verify tgt_prog is not null
>     - Replace boolean followed with number of followers, to handle
>       multiple progs attaching/detaching
> 
>  include/linux/bpf.h            |  1 +
>  include/uapi/linux/bpf.h       |  1 +
>  kernel/bpf/syscall.c           | 12 ++++++++++++
>  kernel/bpf/verifier.c          | 33 +++++++++++++++++++--------------
>  tools/include/uapi/linux/bpf.h |  1 +
>  5 files changed, 34 insertions(+), 14 deletions(-)
> 
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index eb447b0a9423..1588e48fe31c 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -1400,6 +1400,7 @@ struct bpf_prog_aux {
>  	u32 real_func_cnt; /* includes hidden progs, only used for JIT and freeing progs */
>  	u32 func_idx; /* 0 for non-func prog, the index in func array for func prog */
>  	u32 attach_btf_id; /* in-kernel BTF type id to attach to */
> +	u32 attach_depth; /* for tracing prog, level of attachment nesting */
>  	u32 ctx_arg_info_size;
>  	u32 max_rdonly_access;
>  	u32 max_rdwr_access;
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index e88746ba7d21..9cf45ad914f1 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -6468,6 +6468,7 @@ struct bpf_prog_info {
>  	__u32 verified_insns;
>  	__u32 attach_btf_obj_id;
>  	__u32 attach_btf_id;
> +	__u32 attach_depth;
>  } __attribute__((aligned(8)));
>  
>  struct bpf_map_info {
> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> index 5e43ddd1b83f..9c56b5970d7e 100644
> --- a/kernel/bpf/syscall.c
> +++ b/kernel/bpf/syscall.c
> @@ -3039,6 +3039,7 @@ static void bpf_tracing_link_release(struct bpf_link *link)
>  
>  	bpf_trampoline_put(tr_link->trampoline);
>  
> +	link->prog->aux->attach_depth = 0;
>  	/* tgt_prog is NULL if target is a kernel function */
>  	if (tr_link->tgt_prog)
>  		bpf_prog_put(tr_link->tgt_prog);
> @@ -3243,6 +3244,16 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
>  		goto out_unlock;
>  	}
>  
> +	if (tgt_prog) {
> +		/* Bookkeeping for managing the prog attachment chain. If it's a
> +		 * tracing program, bump the level, otherwise carry it on.
> +		 */
> +		if (prog->type == BPF_PROG_TYPE_TRACING)
> +			prog->aux->attach_depth = tgt_prog->aux->attach_depth + 1;
> +		else
> +			prog->aux->attach_depth = tgt_prog->aux->attach_depth;
> +	}

I'm not sure why we need to keep the attach_depth for all program types,
I might be missing something but could we perhaps just use flag related
just to tracing programs like in patch below

jirka


---
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 10e5e4d8a00f..2387a29db193 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1425,6 +1425,7 @@ struct bpf_prog_aux {
 	bool dev_bound; /* Program is bound to the netdev. */
 	bool offload_requested; /* Program is bound and offloaded to the netdev. */
 	bool attach_btf_trace; /* true if attaching to BTF-enabled raw tp */
+	bool attach_tracing_prog; /* true if attaching to tracing prog */
 	bool func_proto_unreliable;
 	bool sleepable;
 	bool tail_call_reachable;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 991e7e73ff3c..a3d309197bc0 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3277,6 +3277,9 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 		goto out_unlock;
 	}
 
+	if (tgt_prog && prog->type == BPF_PROG_TYPE_TRACING)
+		prog->aux->attach_tracing_prog = true;
+
 	link->tgt_prog = tgt_prog;
 	link->trampoline = tr;
 
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index e5ce530641ba..899e3b34c058 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -20235,8 +20235,13 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
 			bpf_log(log, "Can attach to only JITed progs\n");
 			return -EINVAL;
 		}
-		if (tgt_prog->type == prog->type) {
-			/* Cannot fentry/fexit another fentry/fexit program.
+		if (prog->type == BPF_PROG_TYPE_TRACING) {
+			if (aux->attach_tracing_prog) {
+				bpf_log(log, "Cannot nest tracing program attach more than once\n");
+				return -EINVAL;
+			}
+		} else if (tgt_prog->type == prog->type) {
+			/*
 			 * Cannot attach program extension to another extension.
 			 * It's ok to attach fentry/fexit to extension program.
 			 */

Re: [PATCH bpf-next v6 1/4] bpf: Relax tracing prog recursive attach rules

[Dmitry Dolgov]

> On Thu, Dec 07, 2023 at 11:07:38AM +0100, Jiri Olsa wrote:
> > +	if (tgt_prog) {
> > +		/* Bookkeeping for managing the prog attachment chain. If it's a
> > +		 * tracing program, bump the level, otherwise carry it on.
> > +		 */
> > +		if (prog->type == BPF_PROG_TYPE_TRACING)
> > +			prog->aux->attach_depth = tgt_prog->aux->attach_depth + 1;
> > +		else
> > +			prog->aux->attach_depth = tgt_prog->aux->attach_depth;
> > +	}
>
> I'm not sure why we need to keep the attach_depth for all program types,
> I might be missing something but could we perhaps just use flag related
> just to tracing programs like in patch below

Originally I was trying to make the change somewhat generic, so that the
same approach could be used e.g. for extensions if needed (say, instead
of preventing extending fentry/fexit). But since we reduce the scope
anyway, I agree we can simply use a boolean flag. It still has to be set
for the target prog if it's fentry/fexit and already attached somewhere,
but otherwise would be similar to your proposal. Let me prepare the new
version.