[PATCHv4 bpf-next 0/5] bpf: Tracing and lsm programs re-attach

[PATCHv4 bpf-next 0/5] bpf: Tracing and lsm programs re-attach

[Jiri Olsa]

hi,
while adding test for pinning the module while there's
trampoline attach to it, I noticed that we don't allow
link detach and following re-attach for trampolines.
Adding that for tracing and lsm programs.

You need to have patch [1] from bpf tree for test module
attach test to pass.

v4 changes:
  - fix wrong cleanup goto jump reported by Julia

thanks,
jirka


[1] https://lore.kernel.org/bpf/20210326105900.151466-1-jolsa@kernel.org/
---
Jiri Olsa (5):
      bpf: Allow trampoline re-attach for tracing and lsm programs
      selftests/bpf: Add re-attach test to fentry_test
      selftests/bpf: Add re-attach test to fexit_test
      selftests/bpf: Add re-attach test to lsm test
      selftests/bpf: Test that module can't be unloaded with attached trampoline

 kernel/bpf/syscall.c                                   | 23 +++++++++++++++++------
 kernel/bpf/trampoline.c                                |  2 +-
 tools/testing/selftests/bpf/prog_tests/fentry_test.c   | 51 ++++++++++++++++++++++++++++++++++++---------------
 tools/testing/selftests/bpf/prog_tests/fexit_test.c    | 51 ++++++++++++++++++++++++++++++++++++---------------
 tools/testing/selftests/bpf/prog_tests/module_attach.c | 23 +++++++++++++++++++++++
 tools/testing/selftests/bpf/prog_tests/test_lsm.c      | 48 ++++++++++++++++++++++++++++++++++++++----------
 tools/testing/selftests/bpf/test_progs.h               |  2 +-
 7 files changed, 152 insertions(+), 48 deletions(-)

[PATCHv4 bpf-next 1/5] bpf: Allow trampoline re-attach for tracing and lsm programs

[Jiri Olsa]

Currently we don't allow re-attaching of trampolines. Once
it's detached, it can't be re-attach even when the program
is still loaded.

Adding the possibility to re-attach the loaded tracing and
lsm programs.

Fixing missing unlock with proper cleanup goto jump reported
by Julia.

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 kernel/bpf/syscall.c    | 23 +++++++++++++++++------
 kernel/bpf/trampoline.c |  2 +-
 2 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 6428634da57e..f02c6a871b4f 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -2645,14 +2645,25 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 	 *   target_btf_id using the link_create API.
 	 *
 	 * - if tgt_prog == NULL when this function was called using the old
-         *   raw_tracepoint_open API, and we need a target from prog->aux
-         *
-         * The combination of no saved target in prog->aux, and no target
-         * specified on load is illegal, and we reject that here.
+	 *   raw_tracepoint_open API, and we need a target from prog->aux
+	 *
+	 * - 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 && !tgt_prog) {
-		err = -ENOENT;
-		goto out_unlock;
+		/*
+		 * Allow re-attach for TRACING and LSM programs. If it's
+		 * currently linked, bpf_trampoline_link_prog will fail.
+		 * EXT programs need to specify tgt_prog_fd, so they
+		 * re-attach in separate code path.
+		 */
+		if (prog->type != BPF_PROG_TYPE_TRACING &&
+		    prog->type != BPF_PROG_TYPE_LSM) {
+			err = -EINVAL;
+			goto out_unlock;
+		}
+		btf_id = prog->aux->attach_btf_id;
+		key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf, btf_id);
 	}
 
 	if (!prog->aux->dst_trampoline ||
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 1f3a4be4b175..48b8b9916aa2 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -437,7 +437,7 @@ int bpf_trampoline_unlink_prog(struct bpf_prog *prog, struct bpf_trampoline *tr)
 		tr->extension_prog = NULL;
 		goto out;
 	}
-	hlist_del(&prog->aux->tramp_hlist);
+	hlist_del_init(&prog->aux->tramp_hlist);
 	tr->progs_cnt[kind]--;
 	err = bpf_trampoline_update(tr);
 out:
-- 
2.30.2

[PATCHv4 bpf-next 2/5] selftests/bpf: Add re-attach test to fentry_test

[Jiri Olsa]

Adding the test to re-attach (detach/attach again) tracing
fentry programs, plus check that already linked program can't
be attached again.

Also switching to ASSERT* macros and adding missing ';' in
ASSERT_ERR_PTR macro.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 .../selftests/bpf/prog_tests/fentry_test.c    | 51 +++++++++++++------
 tools/testing/selftests/bpf/test_progs.h      |  2 +-
 2 files changed, 37 insertions(+), 16 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/fentry_test.c b/tools/testing/selftests/bpf/prog_tests/fentry_test.c
index 04ebbf1cb390..f440c74f5367 100644
--- a/tools/testing/selftests/bpf/prog_tests/fentry_test.c
+++ b/tools/testing/selftests/bpf/prog_tests/fentry_test.c
@@ -3,35 +3,56 @@
 #include <test_progs.h>
 #include "fentry_test.skel.h"
 
-void test_fentry_test(void)
+static int fentry_test(struct fentry_test *fentry_skel)
 {
-	struct fentry_test *fentry_skel = NULL;
 	int err, prog_fd, i;
 	__u32 duration = 0, retval;
+	struct bpf_link *link;
 	__u64 *result;
 
-	fentry_skel = fentry_test__open_and_load();
-	if (CHECK(!fentry_skel, "fentry_skel_load", "fentry skeleton failed\n"))
-		goto cleanup;
-
 	err = fentry_test__attach(fentry_skel);
-	if (CHECK(err, "fentry_attach", "fentry attach failed: %d\n", err))
-		goto cleanup;
+	if (!ASSERT_OK(err, "fentry_attach"))
+		return err;
+
+	/* Check that already linked program can't be attached again. */
+	link = bpf_program__attach(fentry_skel->progs.test1);
+	if (!ASSERT_ERR_PTR(link, "fentry_attach_link"))
+		return -1;
 
 	prog_fd = bpf_program__fd(fentry_skel->progs.test1);
 	err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
 				NULL, NULL, &retval, &duration);
-	CHECK(err || retval, "test_run",
-	      "err %d errno %d retval %d duration %d\n",
-	      err, errno, retval, duration);
+	ASSERT_OK(err || retval, "test_run");
 
 	result = (__u64 *)fentry_skel->bss;
-	for (i = 0; i < 6; i++) {
-		if (CHECK(result[i] != 1, "result",
-			  "fentry_test%d failed err %lld\n", i + 1, result[i]))
-			goto cleanup;
+	for (i = 0; i < sizeof(*fentry_skel->bss) / sizeof(__u64); i++) {
+		if (!ASSERT_EQ(result[i], 1, "fentry_result"))
+			return -1;
 	}
 
+	fentry_test__detach(fentry_skel);
+
+	/* zero results for re-attach test */
+	memset(fentry_skel->bss, 0, sizeof(*fentry_skel->bss));
+	return 0;
+}
+
+void test_fentry_test(void)
+{
+	struct fentry_test *fentry_skel = NULL;
+	int err;
+
+	fentry_skel = fentry_test__open_and_load();
+	if (!ASSERT_OK_PTR(fentry_skel, "fentry_skel_load"))
+		goto cleanup;
+
+	err = fentry_test(fentry_skel);
+	if (!ASSERT_OK(err, "fentry_first_attach"))
+		goto cleanup;
+
+	err = fentry_test(fentry_skel);
+	ASSERT_OK(err, "fentry_second_attach");
+
 cleanup:
 	fentry_test__destroy(fentry_skel);
 }
diff --git a/tools/testing/selftests/bpf/test_progs.h b/tools/testing/selftests/bpf/test_progs.h
index e87c8546230e..ee7e3b45182a 100644
--- a/tools/testing/selftests/bpf/test_progs.h
+++ b/tools/testing/selftests/bpf/test_progs.h
@@ -210,7 +210,7 @@ extern int test__join_cgroup(const char *path);
 #define ASSERT_ERR_PTR(ptr, name) ({					\
 	static int duration = 0;					\
 	const void *___res = (ptr);					\
-	bool ___ok = IS_ERR(___res)					\
+	bool ___ok = IS_ERR(___res);					\
 	CHECK(!___ok, (name), "unexpected pointer: %p\n", ___res);	\
 	___ok;								\
 })
-- 
2.30.2

[PATCHv4 bpf-next 3/5] selftests/bpf: Add re-attach test to fexit_test

[Jiri Olsa]

Adding the test to re-attach (detach/attach again) tracing
fexit programs, plus check that already linked program can't
be attached again.

Also switching to ASSERT* macros.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 .../selftests/bpf/prog_tests/fexit_test.c     | 51 +++++++++++++------
 1 file changed, 36 insertions(+), 15 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/fexit_test.c b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
index 78d7a2765c27..c48e10c138bc 100644
--- a/tools/testing/selftests/bpf/prog_tests/fexit_test.c
+++ b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
@@ -3,35 +3,56 @@
 #include <test_progs.h>
 #include "fexit_test.skel.h"
 
-void test_fexit_test(void)
+static int fexit_test(struct fexit_test *fexit_skel)
 {
-	struct fexit_test *fexit_skel = NULL;
 	int err, prog_fd, i;
 	__u32 duration = 0, retval;
+	struct bpf_link *link;
 	__u64 *result;
 
-	fexit_skel = fexit_test__open_and_load();
-	if (CHECK(!fexit_skel, "fexit_skel_load", "fexit skeleton failed\n"))
-		goto cleanup;
-
 	err = fexit_test__attach(fexit_skel);
-	if (CHECK(err, "fexit_attach", "fexit attach failed: %d\n", err))
-		goto cleanup;
+	if (!ASSERT_OK(err, "fexit_attach"))
+		return err;
+
+	/* Check that already linked program can't be attached again. */
+	link = bpf_program__attach(fexit_skel->progs.test1);
+	if (!ASSERT_ERR_PTR(link, "fexit_attach_link"))
+		return -1;
 
 	prog_fd = bpf_program__fd(fexit_skel->progs.test1);
 	err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
 				NULL, NULL, &retval, &duration);
-	CHECK(err || retval, "test_run",
-	      "err %d errno %d retval %d duration %d\n",
-	      err, errno, retval, duration);
+	ASSERT_OK(err || retval, "test_run");
 
 	result = (__u64 *)fexit_skel->bss;
-	for (i = 0; i < 6; i++) {
-		if (CHECK(result[i] != 1, "result",
-			  "fexit_test%d failed err %lld\n", i + 1, result[i]))
-			goto cleanup;
+	for (i = 0; i < sizeof(*fexit_skel->bss) / sizeof(__u64); i++) {
+		if (!ASSERT_EQ(result[i], 1, "fexit_result"))
+			return -1;
 	}
 
+	fexit_test__detach(fexit_skel);
+
+	/* zero results for re-attach test */
+	memset(fexit_skel->bss, 0, sizeof(*fexit_skel->bss));
+	return 0;
+}
+
+void test_fexit_test(void)
+{
+	struct fexit_test *fexit_skel = NULL;
+	int err;
+
+	fexit_skel = fexit_test__open_and_load();
+	if (!ASSERT_OK_PTR(fexit_skel, "fexit_skel_load"))
+		goto cleanup;
+
+	err = fexit_test(fexit_skel);
+	if (!ASSERT_OK(err, "fexit_first_attach"))
+		goto cleanup;
+
+	err = fexit_test(fexit_skel);
+	ASSERT_OK(err, "fexit_second_attach");
+
 cleanup:
 	fexit_test__destroy(fexit_skel);
 }
-- 
2.30.2

[PATCHv4 bpf-next 4/5] selftests/bpf: Add re-attach test to lsm test

[Jiri Olsa]

Adding the test to re-attach (detach/attach again) lsm programs,
plus check that already linked program can't be attached again.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 .../selftests/bpf/prog_tests/test_lsm.c       | 48 +++++++++++++++----
 1 file changed, 38 insertions(+), 10 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/test_lsm.c b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
index 2755e4f81499..d492e76e01cf 100644
--- a/tools/testing/selftests/bpf/prog_tests/test_lsm.c
+++ b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
@@ -18,6 +18,8 @@ char *CMD_ARGS[] = {"true", NULL};
 #define GET_PAGE_ADDR(ADDR, PAGE_SIZE)					\
 	(char *)(((unsigned long) (ADDR + PAGE_SIZE)) & ~(PAGE_SIZE-1))
 
+static int duration = 0;
+
 int stack_mprotect(void)
 {
 	void *buf;
@@ -51,23 +53,25 @@ int exec_cmd(int *monitored_pid)
 	return -EINVAL;
 }
 
-void test_test_lsm(void)
+static int test_lsm(struct lsm *skel)
 {
-	struct lsm *skel = NULL;
-	int err, duration = 0;
+	struct bpf_link *link;
 	int buf = 1234;
-
-	skel = lsm__open_and_load();
-	if (CHECK(!skel, "skel_load", "lsm skeleton failed\n"))
-		goto close_prog;
+	int err;
 
 	err = lsm__attach(skel);
 	if (CHECK(err, "attach", "lsm attach failed: %d\n", err))
-		goto close_prog;
+		return err;
+
+	/* Check that already linked program can't be attached again. */
+	link = bpf_program__attach(skel->progs.test_int_hook);
+	if (CHECK(!IS_ERR(link), "attach_link",
+		  "re-attach without detach should not succeed"))
+		return -1;
 
 	err = exec_cmd(&skel->bss->monitored_pid);
 	if (CHECK(err < 0, "exec_cmd", "err %d errno %d\n", err, errno))
-		goto close_prog;
+		return err;
 
 	CHECK(skel->bss->bprm_count != 1, "bprm_count", "bprm_count = %d\n",
 	      skel->bss->bprm_count);
@@ -77,7 +81,7 @@ void test_test_lsm(void)
 	err = stack_mprotect();
 	if (CHECK(errno != EPERM, "stack_mprotect", "want err=EPERM, got %d\n",
 		  errno))
-		goto close_prog;
+		return err;
 
 	CHECK(skel->bss->mprotect_count != 1, "mprotect_count",
 	      "mprotect_count = %d\n", skel->bss->mprotect_count);
@@ -89,6 +93,30 @@ void test_test_lsm(void)
 	CHECK(skel->bss->copy_test != 3, "copy_test",
 	      "copy_test = %d\n", skel->bss->copy_test);
 
+	lsm__detach(skel);
+
+	skel->bss->copy_test = 0;
+	skel->bss->bprm_count = 0;
+	skel->bss->mprotect_count = 0;
+	return 0;
+}
+
+void test_test_lsm(void)
+{
+	struct lsm *skel = NULL;
+	int err;
+
+	skel = lsm__open_and_load();
+	if (CHECK(!skel, "lsm_skel_load", "lsm skeleton failed\n"))
+		goto close_prog;
+
+	err = test_lsm(skel);
+	if (CHECK(err, "test_lsm", "first attach failed\n"))
+		goto close_prog;
+
+	err = test_lsm(skel);
+	CHECK(err, "test_lsm", "second attach failed\n");
+
 close_prog:
 	lsm__destroy(skel);
 }
-- 
2.30.2

[PATCHv4 bpf-next 5/5] selftests/bpf: Test that module can't be unloaded with attached trampoline

[Jiri Olsa]

Adding test to verify that once we attach module's trampoline,
the module can't be unloaded.

Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 .../selftests/bpf/prog_tests/module_attach.c  | 23 +++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/tools/testing/selftests/bpf/prog_tests/module_attach.c b/tools/testing/selftests/bpf/prog_tests/module_attach.c
index 5bc53d53d86e..d85a69b7ce44 100644
--- a/tools/testing/selftests/bpf/prog_tests/module_attach.c
+++ b/tools/testing/selftests/bpf/prog_tests/module_attach.c
@@ -45,12 +45,18 @@ static int trigger_module_test_write(int write_sz)
 	return 0;
 }
 
+static int delete_module(const char *name, int flags)
+{
+	return syscall(__NR_delete_module, name, flags);
+}
+
 void test_module_attach(void)
 {
 	const int READ_SZ = 456;
 	const int WRITE_SZ = 457;
 	struct test_module_attach* skel;
 	struct test_module_attach__bss *bss;
+	struct bpf_link *link;
 	int err;
 
 	skel = test_module_attach__open();
@@ -84,6 +90,23 @@ void test_module_attach(void)
 	ASSERT_EQ(bss->fexit_ret, -EIO, "fexit_tet");
 	ASSERT_EQ(bss->fmod_ret_read_sz, READ_SZ, "fmod_ret");
 
+	test_module_attach__detach(skel);
+
+	/* attach fentry/fexit and make sure it get's module reference */
+	link = bpf_program__attach(skel->progs.handle_fentry);
+	if (!ASSERT_OK_PTR(link, "attach_fentry"))
+		goto cleanup;
+
+	ASSERT_ERR(delete_module("bpf_testmod", 0), "delete_module");
+	bpf_link__destroy(link);
+
+	link = bpf_program__attach(skel->progs.handle_fexit);
+	if (!ASSERT_OK_PTR(link, "attach_fexit"))
+		goto cleanup;
+
+	ASSERT_ERR(delete_module("bpf_testmod", 0), "delete_module");
+	bpf_link__destroy(link);
+
 cleanup:
 	test_module_attach__destroy(skel);
 }
-- 
2.30.2

Re: [PATCHv4 bpf-next 2/5] selftests/bpf: Add re-attach test to fentry_test

[Andrii Nakryiko]

On Mon, Apr 12, 2021 at 9:29 AM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Adding the test to re-attach (detach/attach again) tracing
> fentry programs, plus check that already linked program can't
> be attached again.
>
> Also switching to ASSERT* macros and adding missing ';' in
> ASSERT_ERR_PTR macro.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
>  .../selftests/bpf/prog_tests/fentry_test.c    | 51 +++++++++++++------
>  tools/testing/selftests/bpf/test_progs.h      |  2 +-
>  2 files changed, 37 insertions(+), 16 deletions(-)
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/fentry_test.c b/tools/testing/selftests/bpf/prog_tests/fentry_test.c
> index 04ebbf1cb390..f440c74f5367 100644
> --- a/tools/testing/selftests/bpf/prog_tests/fentry_test.c
> +++ b/tools/testing/selftests/bpf/prog_tests/fentry_test.c
> @@ -3,35 +3,56 @@
>  #include <test_progs.h>
>  #include "fentry_test.skel.h"
>
> -void test_fentry_test(void)
> +static int fentry_test(struct fentry_test *fentry_skel)
>  {
> -       struct fentry_test *fentry_skel = NULL;
>         int err, prog_fd, i;
>         __u32 duration = 0, retval;
> +       struct bpf_link *link;
>         __u64 *result;
>
> -       fentry_skel = fentry_test__open_and_load();
> -       if (CHECK(!fentry_skel, "fentry_skel_load", "fentry skeleton failed\n"))
> -               goto cleanup;
> -
>         err = fentry_test__attach(fentry_skel);
> -       if (CHECK(err, "fentry_attach", "fentry attach failed: %d\n", err))
> -               goto cleanup;
> +       if (!ASSERT_OK(err, "fentry_attach"))
> +               return err;
> +
> +       /* Check that already linked program can't be attached again. */
> +       link = bpf_program__attach(fentry_skel->progs.test1);
> +       if (!ASSERT_ERR_PTR(link, "fentry_attach_link"))
> +               return -1;
>
>         prog_fd = bpf_program__fd(fentry_skel->progs.test1);
>         err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
>                                 NULL, NULL, &retval, &duration);
> -       CHECK(err || retval, "test_run",
> -             "err %d errno %d retval %d duration %d\n",
> -             err, errno, retval, duration);
> +       ASSERT_OK(err || retval, "test_run");

this is quite misleading, even if will result in a correct check. Toke
did this in his patch set:

ASSERT_OK(err, ...);
ASSERT_EQ(retval, 0, ...);

It is a better and more straightforward way to validate the checks
instead of relying on (err || retval) -> bool (true) -> int (1) -> !=
0 chain.


>
>         result = (__u64 *)fentry_skel->bss;
> -       for (i = 0; i < 6; i++) {
> -               if (CHECK(result[i] != 1, "result",
> -                         "fentry_test%d failed err %lld\n", i + 1, result[i]))
> -                       goto cleanup;
> +       for (i = 0; i < sizeof(*fentry_skel->bss) / sizeof(__u64); i++) {
> +               if (!ASSERT_EQ(result[i], 1, "fentry_result"))
> +                       return -1;
>         }
>
> +       fentry_test__detach(fentry_skel);
> +
> +       /* zero results for re-attach test */
> +       memset(fentry_skel->bss, 0, sizeof(*fentry_skel->bss));
> +       return 0;
> +}
> +
> +void test_fentry_test(void)
> +{
> +       struct fentry_test *fentry_skel = NULL;
> +       int err;
> +
> +       fentry_skel = fentry_test__open_and_load();
> +       if (!ASSERT_OK_PTR(fentry_skel, "fentry_skel_load"))
> +               goto cleanup;
> +
> +       err = fentry_test(fentry_skel);
> +       if (!ASSERT_OK(err, "fentry_first_attach"))
> +               goto cleanup;
> +
> +       err = fentry_test(fentry_skel);
> +       ASSERT_OK(err, "fentry_second_attach");
> +
>  cleanup:
>         fentry_test__destroy(fentry_skel);
>  }
> diff --git a/tools/testing/selftests/bpf/test_progs.h b/tools/testing/selftests/bpf/test_progs.h
> index e87c8546230e..ee7e3b45182a 100644
> --- a/tools/testing/selftests/bpf/test_progs.h
> +++ b/tools/testing/selftests/bpf/test_progs.h
> @@ -210,7 +210,7 @@ extern int test__join_cgroup(const char *path);
>  #define ASSERT_ERR_PTR(ptr, name) ({                                   \
>         static int duration = 0;                                        \
>         const void *___res = (ptr);                                     \
> -       bool ___ok = IS_ERR(___res)                                     \
> +       bool ___ok = IS_ERR(___res);                                    \

heh, it probably deserves a separate patch with Fixes tag...

>         CHECK(!___ok, (name), "unexpected pointer: %p\n", ___res);      \
>         ___ok;                                                          \
>  })
> --
> 2.30.2
>

Re: [PATCHv4 bpf-next 3/5] selftests/bpf: Add re-attach test to fexit_test

[Andrii Nakryiko]

On Mon, Apr 12, 2021 at 9:30 AM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Adding the test to re-attach (detach/attach again) tracing
> fexit programs, plus check that already linked program can't
> be attached again.
>
> Also switching to ASSERT* macros.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
>  .../selftests/bpf/prog_tests/fexit_test.c     | 51 +++++++++++++------
>  1 file changed, 36 insertions(+), 15 deletions(-)
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/fexit_test.c b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> index 78d7a2765c27..c48e10c138bc 100644
> --- a/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> +++ b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> @@ -3,35 +3,56 @@
>  #include <test_progs.h>
>  #include "fexit_test.skel.h"
>
> -void test_fexit_test(void)
> +static int fexit_test(struct fexit_test *fexit_skel)
>  {
> -       struct fexit_test *fexit_skel = NULL;
>         int err, prog_fd, i;
>         __u32 duration = 0, retval;
> +       struct bpf_link *link;
>         __u64 *result;
>
> -       fexit_skel = fexit_test__open_and_load();
> -       if (CHECK(!fexit_skel, "fexit_skel_load", "fexit skeleton failed\n"))
> -               goto cleanup;
> -
>         err = fexit_test__attach(fexit_skel);
> -       if (CHECK(err, "fexit_attach", "fexit attach failed: %d\n", err))
> -               goto cleanup;
> +       if (!ASSERT_OK(err, "fexit_attach"))
> +               return err;
> +
> +       /* Check that already linked program can't be attached again. */
> +       link = bpf_program__attach(fexit_skel->progs.test1);
> +       if (!ASSERT_ERR_PTR(link, "fexit_attach_link"))
> +               return -1;
>
>         prog_fd = bpf_program__fd(fexit_skel->progs.test1);
>         err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
>                                 NULL, NULL, &retval, &duration);
> -       CHECK(err || retval, "test_run",
> -             "err %d errno %d retval %d duration %d\n",
> -             err, errno, retval, duration);
> +       ASSERT_OK(err || retval, "test_run");

same as in previous patch

With this fixed, feel free to add my ack to this and previous patch. Thanks!

>
>         result = (__u64 *)fexit_skel->bss;
> -       for (i = 0; i < 6; i++) {
> -               if (CHECK(result[i] != 1, "result",
> -                         "fexit_test%d failed err %lld\n", i + 1, result[i]))
> -                       goto cleanup;
> +       for (i = 0; i < sizeof(*fexit_skel->bss) / sizeof(__u64); i++) {
> +               if (!ASSERT_EQ(result[i], 1, "fexit_result"))
> +                       return -1;
>         }
>
> +       fexit_test__detach(fexit_skel);
> +
> +       /* zero results for re-attach test */
> +       memset(fexit_skel->bss, 0, sizeof(*fexit_skel->bss));
> +       return 0;
> +}
> +
> +void test_fexit_test(void)
> +{
> +       struct fexit_test *fexit_skel = NULL;
> +       int err;
> +
> +       fexit_skel = fexit_test__open_and_load();
> +       if (!ASSERT_OK_PTR(fexit_skel, "fexit_skel_load"))
> +               goto cleanup;
> +
> +       err = fexit_test(fexit_skel);
> +       if (!ASSERT_OK(err, "fexit_first_attach"))
> +               goto cleanup;
> +
> +       err = fexit_test(fexit_skel);
> +       ASSERT_OK(err, "fexit_second_attach");
> +
>  cleanup:
>         fexit_test__destroy(fexit_skel);
>  }
> --
> 2.30.2
>

Re: [PATCHv4 bpf-next 4/5] selftests/bpf: Add re-attach test to lsm test

[Andrii Nakryiko]

On Mon, Apr 12, 2021 at 9:31 AM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Adding the test to re-attach (detach/attach again) lsm programs,
> plus check that already linked program can't be attached again.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
>  .../selftests/bpf/prog_tests/test_lsm.c       | 48 +++++++++++++++----
>  1 file changed, 38 insertions(+), 10 deletions(-)
>

Surprised you didn't switch this one to ASSERT, but ok, we can do it
some other time ;)

Acked-by: Andrii Nakryiko <andrii@kernel.org>


> diff --git a/tools/testing/selftests/bpf/prog_tests/test_lsm.c b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> index 2755e4f81499..d492e76e01cf 100644
> --- a/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> +++ b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> @@ -18,6 +18,8 @@ char *CMD_ARGS[] = {"true", NULL};
>  #define GET_PAGE_ADDR(ADDR, PAGE_SIZE)                                 \
>         (char *)(((unsigned long) (ADDR + PAGE_SIZE)) & ~(PAGE_SIZE-1))
>

[...]

Re: [PATCHv4 bpf-next 1/5] bpf: Allow trampoline re-attach for tracing and lsm programs

[Andrii Nakryiko]

On Mon, Apr 12, 2021 at 9:28 AM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Currently we don't allow re-attaching of trampolines. Once
> it's detached, it can't be re-attach even when the program
> is still loaded.
>
> Adding the possibility to re-attach the loaded tracing and
> lsm programs.
>
> Fixing missing unlock with proper cleanup goto jump reported
> by Julia.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Reported-by: Julia Lawall <julia.lawall@lip6.fr>
> Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
>  kernel/bpf/syscall.c    | 23 +++++++++++++++++------
>  kernel/bpf/trampoline.c |  2 +-
>  2 files changed, 18 insertions(+), 7 deletions(-)
>
> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> index 6428634da57e..f02c6a871b4f 100644
> --- a/kernel/bpf/syscall.c
> +++ b/kernel/bpf/syscall.c
> @@ -2645,14 +2645,25 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
>          *   target_btf_id using the link_create API.
>          *
>          * - if tgt_prog == NULL when this function was called using the old
> -         *   raw_tracepoint_open API, and we need a target from prog->aux
> -         *
> -         * The combination of no saved target in prog->aux, and no target
> -         * specified on load is illegal, and we reject that here.
> +        *   raw_tracepoint_open API, and we need a target from prog->aux
> +        *
> +        * - 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 && !tgt_prog) {
> -               err = -ENOENT;
> -               goto out_unlock;
> +               /*
> +                * Allow re-attach for TRACING and LSM programs. If it's
> +                * currently linked, bpf_trampoline_link_prog will fail.
> +                * EXT programs need to specify tgt_prog_fd, so they
> +                * re-attach in separate code path.
> +                */
> +               if (prog->type != BPF_PROG_TYPE_TRACING &&
> +                   prog->type != BPF_PROG_TYPE_LSM) {
> +                       err = -EINVAL;
> +                       goto out_unlock;
> +               }
> +               btf_id = prog->aux->attach_btf_id;
> +               key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf, btf_id);
>         }
>
>         if (!prog->aux->dst_trampoline ||
> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> index 1f3a4be4b175..48b8b9916aa2 100644
> --- a/kernel/bpf/trampoline.c
> +++ b/kernel/bpf/trampoline.c
> @@ -437,7 +437,7 @@ int bpf_trampoline_unlink_prog(struct bpf_prog *prog, struct bpf_trampoline *tr)
>                 tr->extension_prog = NULL;
>                 goto out;
>         }
> -       hlist_del(&prog->aux->tramp_hlist);
> +       hlist_del_init(&prog->aux->tramp_hlist);

there is another hlist_del few lines above in error handling path of
bpf_trampoline_link_prog(), it should probably be also updated to
hlist_del_init(), no?

>         tr->progs_cnt[kind]--;
>         err = bpf_trampoline_update(tr);
>  out:
> --
> 2.30.2
>

Re: [PATCHv4 bpf-next 4/5] selftests/bpf: Add re-attach test to lsm test

[Jiri Olsa]

On Tue, Apr 13, 2021 at 02:57:26PM -0700, Andrii Nakryiko wrote:
> On Mon, Apr 12, 2021 at 9:31 AM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > Adding the test to re-attach (detach/attach again) lsm programs,
> > plus check that already linked program can't be attached again.
> >
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> >  .../selftests/bpf/prog_tests/test_lsm.c       | 48 +++++++++++++++----
> >  1 file changed, 38 insertions(+), 10 deletions(-)
> >
> 
> Surprised you didn't switch this one to ASSERT, but ok, we can do it
> some other time ;)

yep, I commented on that in the previous version ;-)

  - used ASSERT* macros apart from lsm test, which is using
    CHECKs all over the place [Andrii]

I think it should go to separate patch, so it won't shade
the actual change

jirka

> 
> Acked-by: Andrii Nakryiko <andrii@kernel.org>
> 
> 
> > diff --git a/tools/testing/selftests/bpf/prog_tests/test_lsm.c b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> > index 2755e4f81499..d492e76e01cf 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/test_lsm.c
> > @@ -18,6 +18,8 @@ char *CMD_ARGS[] = {"true", NULL};
> >  #define GET_PAGE_ADDR(ADDR, PAGE_SIZE)                                 \
> >         (char *)(((unsigned long) (ADDR + PAGE_SIZE)) & ~(PAGE_SIZE-1))
> >
> 
> [...]
> 

Re: [PATCHv4 bpf-next 2/5] selftests/bpf: Add re-attach test to fentry_test

[Jiri Olsa]

On Tue, Apr 13, 2021 at 02:54:10PM -0700, Andrii Nakryiko wrote:

SNIP

> >         __u32 duration = 0, retval;
> > +       struct bpf_link *link;
> >         __u64 *result;
> >
> > -       fentry_skel = fentry_test__open_and_load();
> > -       if (CHECK(!fentry_skel, "fentry_skel_load", "fentry skeleton failed\n"))
> > -               goto cleanup;
> > -
> >         err = fentry_test__attach(fentry_skel);
> > -       if (CHECK(err, "fentry_attach", "fentry attach failed: %d\n", err))
> > -               goto cleanup;
> > +       if (!ASSERT_OK(err, "fentry_attach"))
> > +               return err;
> > +
> > +       /* Check that already linked program can't be attached again. */
> > +       link = bpf_program__attach(fentry_skel->progs.test1);
> > +       if (!ASSERT_ERR_PTR(link, "fentry_attach_link"))
> > +               return -1;
> >
> >         prog_fd = bpf_program__fd(fentry_skel->progs.test1);
> >         err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
> >                                 NULL, NULL, &retval, &duration);
> > -       CHECK(err || retval, "test_run",
> > -             "err %d errno %d retval %d duration %d\n",
> > -             err, errno, retval, duration);
> > +       ASSERT_OK(err || retval, "test_run");
> 
> this is quite misleading, even if will result in a correct check. Toke
> did this in his patch set:
> 
> ASSERT_OK(err, ...);
> ASSERT_EQ(retval, 0, ...);
> 
> It is a better and more straightforward way to validate the checks
> instead of relying on (err || retval) -> bool (true) -> int (1) -> !=
> 0 chain.

ok, makes sense

SNIP

> > +void test_fentry_test(void)
> > +{
> > +       struct fentry_test *fentry_skel = NULL;
> > +       int err;
> > +
> > +       fentry_skel = fentry_test__open_and_load();
> > +       if (!ASSERT_OK_PTR(fentry_skel, "fentry_skel_load"))
> > +               goto cleanup;
> > +
> > +       err = fentry_test(fentry_skel);
> > +       if (!ASSERT_OK(err, "fentry_first_attach"))
> > +               goto cleanup;
> > +
> > +       err = fentry_test(fentry_skel);
> > +       ASSERT_OK(err, "fentry_second_attach");
> > +
> >  cleanup:
> >         fentry_test__destroy(fentry_skel);
> >  }
> > diff --git a/tools/testing/selftests/bpf/test_progs.h b/tools/testing/selftests/bpf/test_progs.h
> > index e87c8546230e..ee7e3b45182a 100644
> > --- a/tools/testing/selftests/bpf/test_progs.h
> > +++ b/tools/testing/selftests/bpf/test_progs.h
> > @@ -210,7 +210,7 @@ extern int test__join_cgroup(const char *path);
> >  #define ASSERT_ERR_PTR(ptr, name) ({                                   \
> >         static int duration = 0;                                        \
> >         const void *___res = (ptr);                                     \
> > -       bool ___ok = IS_ERR(___res)                                     \
> > +       bool ___ok = IS_ERR(___res);                                    \
> 
> heh, it probably deserves a separate patch with Fixes tag...

va bene

jirka

Re: [PATCHv4 bpf-next 1/5] bpf: Allow trampoline re-attach for tracing and lsm programs

[Jiri Olsa]

On Tue, Apr 13, 2021 at 03:03:27PM -0700, Andrii Nakryiko wrote:
> On Mon, Apr 12, 2021 at 9:28 AM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > Currently we don't allow re-attaching of trampolines. Once
> > it's detached, it can't be re-attach even when the program
> > is still loaded.
> >
> > Adding the possibility to re-attach the loaded tracing and
> > lsm programs.
> >
> > Fixing missing unlock with proper cleanup goto jump reported
> > by Julia.
> >
> > Reported-by: kernel test robot <lkp@intel.com>
> > Reported-by: Julia Lawall <julia.lawall@lip6.fr>
> > Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> >  kernel/bpf/syscall.c    | 23 +++++++++++++++++------
> >  kernel/bpf/trampoline.c |  2 +-
> >  2 files changed, 18 insertions(+), 7 deletions(-)
> >
> > diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> > index 6428634da57e..f02c6a871b4f 100644
> > --- a/kernel/bpf/syscall.c
> > +++ b/kernel/bpf/syscall.c
> > @@ -2645,14 +2645,25 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
> >          *   target_btf_id using the link_create API.
> >          *
> >          * - if tgt_prog == NULL when this function was called using the old
> > -         *   raw_tracepoint_open API, and we need a target from prog->aux
> > -         *
> > -         * The combination of no saved target in prog->aux, and no target
> > -         * specified on load is illegal, and we reject that here.
> > +        *   raw_tracepoint_open API, and we need a target from prog->aux
> > +        *
> > +        * - 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 && !tgt_prog) {
> > -               err = -ENOENT;
> > -               goto out_unlock;
> > +               /*
> > +                * Allow re-attach for TRACING and LSM programs. If it's
> > +                * currently linked, bpf_trampoline_link_prog will fail.
> > +                * EXT programs need to specify tgt_prog_fd, so they
> > +                * re-attach in separate code path.
> > +                */
> > +               if (prog->type != BPF_PROG_TYPE_TRACING &&
> > +                   prog->type != BPF_PROG_TYPE_LSM) {
> > +                       err = -EINVAL;
> > +                       goto out_unlock;
> > +               }
> > +               btf_id = prog->aux->attach_btf_id;
> > +               key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf, btf_id);
> >         }
> >
> >         if (!prog->aux->dst_trampoline ||
> > diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> > index 1f3a4be4b175..48b8b9916aa2 100644
> > --- a/kernel/bpf/trampoline.c
> > +++ b/kernel/bpf/trampoline.c
> > @@ -437,7 +437,7 @@ int bpf_trampoline_unlink_prog(struct bpf_prog *prog, struct bpf_trampoline *tr)
> >                 tr->extension_prog = NULL;
> >                 goto out;
> >         }
> > -       hlist_del(&prog->aux->tramp_hlist);
> > +       hlist_del_init(&prog->aux->tramp_hlist);
> 
> there is another hlist_del few lines above in error handling path of
> bpf_trampoline_link_prog(), it should probably be also updated to
> hlist_del_init(), no?

ugh, that one is missing.. will fix

thanks,
jirka

> 
> >         tr->progs_cnt[kind]--;
> >         err = bpf_trampoline_update(tr);
> >  out:
> > --
> > 2.30.2
> >
> 

Re: [PATCHv4 bpf-next 3/5] selftests/bpf: Add re-attach test to fexit_test

[Jiri Olsa]

On Tue, Apr 13, 2021 at 02:55:32PM -0700, Andrii Nakryiko wrote:
> On Mon, Apr 12, 2021 at 9:30 AM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > Adding the test to re-attach (detach/attach again) tracing
> > fexit programs, plus check that already linked program can't
> > be attached again.
> >
> > Also switching to ASSERT* macros.
> >
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> >  .../selftests/bpf/prog_tests/fexit_test.c     | 51 +++++++++++++------
> >  1 file changed, 36 insertions(+), 15 deletions(-)
> >
> > diff --git a/tools/testing/selftests/bpf/prog_tests/fexit_test.c b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> > index 78d7a2765c27..c48e10c138bc 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/fexit_test.c
> > @@ -3,35 +3,56 @@
> >  #include <test_progs.h>
> >  #include "fexit_test.skel.h"
> >
> > -void test_fexit_test(void)
> > +static int fexit_test(struct fexit_test *fexit_skel)
> >  {
> > -       struct fexit_test *fexit_skel = NULL;
> >         int err, prog_fd, i;
> >         __u32 duration = 0, retval;
> > +       struct bpf_link *link;
> >         __u64 *result;
> >
> > -       fexit_skel = fexit_test__open_and_load();
> > -       if (CHECK(!fexit_skel, "fexit_skel_load", "fexit skeleton failed\n"))
> > -               goto cleanup;
> > -
> >         err = fexit_test__attach(fexit_skel);
> > -       if (CHECK(err, "fexit_attach", "fexit attach failed: %d\n", err))
> > -               goto cleanup;
> > +       if (!ASSERT_OK(err, "fexit_attach"))
> > +               return err;
> > +
> > +       /* Check that already linked program can't be attached again. */
> > +       link = bpf_program__attach(fexit_skel->progs.test1);
> > +       if (!ASSERT_ERR_PTR(link, "fexit_attach_link"))
> > +               return -1;
> >
> >         prog_fd = bpf_program__fd(fexit_skel->progs.test1);
> >         err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
> >                                 NULL, NULL, &retval, &duration);
> > -       CHECK(err || retval, "test_run",
> > -             "err %d errno %d retval %d duration %d\n",
> > -             err, errno, retval, duration);
> > +       ASSERT_OK(err || retval, "test_run");
> 
> same as in previous patch
> 
> With this fixed, feel free to add my ack to this and previous patch. Thanks!

ok, thanks,
jirka

Re: [PATCHv4 bpf-next 2/5] selftests/bpf: Add re-attach test to fentry_test

[Andrii Nakryiko]

On Wed, Apr 14, 2021 at 3:57 AM Jiri Olsa <jolsa@redhat.com> wrote:
>
> On Tue, Apr 13, 2021 at 02:54:10PM -0700, Andrii Nakryiko wrote:
>
> SNIP
>
> > >         __u32 duration = 0, retval;
> > > +       struct bpf_link *link;
> > >         __u64 *result;
> > >
> > > -       fentry_skel = fentry_test__open_and_load();
> > > -       if (CHECK(!fentry_skel, "fentry_skel_load", "fentry skeleton failed\n"))
> > > -               goto cleanup;
> > > -
> > >         err = fentry_test__attach(fentry_skel);
> > > -       if (CHECK(err, "fentry_attach", "fentry attach failed: %d\n", err))
> > > -               goto cleanup;
> > > +       if (!ASSERT_OK(err, "fentry_attach"))
> > > +               return err;
> > > +
> > > +       /* Check that already linked program can't be attached again. */
> > > +       link = bpf_program__attach(fentry_skel->progs.test1);
> > > +       if (!ASSERT_ERR_PTR(link, "fentry_attach_link"))
> > > +               return -1;
> > >
> > >         prog_fd = bpf_program__fd(fentry_skel->progs.test1);
> > >         err = bpf_prog_test_run(prog_fd, 1, NULL, 0,
> > >                                 NULL, NULL, &retval, &duration);
> > > -       CHECK(err || retval, "test_run",
> > > -             "err %d errno %d retval %d duration %d\n",
> > > -             err, errno, retval, duration);
> > > +       ASSERT_OK(err || retval, "test_run");
> >
> > this is quite misleading, even if will result in a correct check. Toke
> > did this in his patch set:
> >
> > ASSERT_OK(err, ...);
> > ASSERT_EQ(retval, 0, ...);
> >
> > It is a better and more straightforward way to validate the checks
> > instead of relying on (err || retval) -> bool (true) -> int (1) -> !=
> > 0 chain.
>
> ok, makes sense
>
> SNIP
>
> > > +void test_fentry_test(void)
> > > +{
> > > +       struct fentry_test *fentry_skel = NULL;
> > > +       int err;
> > > +
> > > +       fentry_skel = fentry_test__open_and_load();
> > > +       if (!ASSERT_OK_PTR(fentry_skel, "fentry_skel_load"))
> > > +               goto cleanup;
> > > +
> > > +       err = fentry_test(fentry_skel);
> > > +       if (!ASSERT_OK(err, "fentry_first_attach"))
> > > +               goto cleanup;
> > > +
> > > +       err = fentry_test(fentry_skel);
> > > +       ASSERT_OK(err, "fentry_second_attach");
> > > +
> > >  cleanup:
> > >         fentry_test__destroy(fentry_skel);
> > >  }
> > > diff --git a/tools/testing/selftests/bpf/test_progs.h b/tools/testing/selftests/bpf/test_progs.h
> > > index e87c8546230e..ee7e3b45182a 100644
> > > --- a/tools/testing/selftests/bpf/test_progs.h
> > > +++ b/tools/testing/selftests/bpf/test_progs.h
> > > @@ -210,7 +210,7 @@ extern int test__join_cgroup(const char *path);
> > >  #define ASSERT_ERR_PTR(ptr, name) ({                                   \
> > >         static int duration = 0;                                        \
> > >         const void *___res = (ptr);                                     \
> > > -       bool ___ok = IS_ERR(___res)                                     \
> > > +       bool ___ok = IS_ERR(___res);                                    \
> >
> > heh, it probably deserves a separate patch with Fixes tag...
>
> va bene

Where would I learn some Italian if not on bpf@vger :)

>
> jirka
>