* [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links
@ 2024-01-18 9:54 Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records Jiri Olsa
` (8 more replies)
0 siblings, 9 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
hi,
this patchset adds support to retrieve cookies from existing tracing
links that still did not support it plus changes to bpftool to display
them. It's leftover we discussed some time ago [1].
thanks,
jirka
[1] https://lore.kernel.org/bpf/CALOAHbAZ6=A9j3VFCLoAC_WhgQKU7injMf06=cM2sU4Hi4Sx+Q@mail.gmail.com/
---
Jiri Olsa (8):
bpf: Add cookie to perf_event bpf_link_info records
bpf: Store cookies in kprobe_multi bpf_link_info data
bpftool: Fix wrong free call in do_show_link
selftests/bpf: Add cookies check for kprobe_multi fill_link_info test
selftests/bpf: Add cookies check for perf_event fill_link_info test
selftests/bpf: Add fill_link_info test for perf event
bpftool: Display cookie for perf event link probes
bpftool: Display cookie for kprobe multi link
include/uapi/linux/bpf.h | 5 +++++
kernel/bpf/syscall.c | 4 ++++
kernel/trace/bpf_trace.c | 15 +++++++++++++
tools/bpf/bpftool/link.c | 87 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------
tools/include/uapi/linux/bpf.h | 5 +++++
tools/testing/selftests/bpf/prog_tests/fill_link_info.c | 114 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------
tools/testing/selftests/bpf/progs/test_fill_link_info.c | 6 +++++
7 files changed, 204 insertions(+), 32 deletions(-)
^ permalink raw reply [flat|nested] 19+ messages in thread
* [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 12:24 ` Yafang Shao
2024-01-18 9:54 ` [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data Jiri Olsa
` (7 subsequent siblings)
8 siblings, 1 reply; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
At the moment we don't store cookie for perf_event probes,
while we do that for the rest of the probes.
Adding cookie fields to struct bpf_link_info perf event
probe records:
perf_event.uprobe
perf_event.kprobe
perf_event.tracepoint
perf_event.perf_event
And the code to store that in bpf_link_info struct.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/uapi/linux/bpf.h | 4 ++++
kernel/bpf/syscall.c | 4 ++++
tools/include/uapi/linux/bpf.h | 4 ++++
3 files changed, 12 insertions(+)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index a00f8a5623e1..b823d367a83c 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -6582,6 +6582,7 @@ struct bpf_link_info {
__aligned_u64 file_name; /* in/out */
__u32 name_len;
__u32 offset; /* offset from file_name */
+ __u64 cookie;
} uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
struct {
__aligned_u64 func_name; /* in/out */
@@ -6589,14 +6590,17 @@ struct bpf_link_info {
__u32 offset; /* offset from func_name */
__u64 addr;
__u64 missed;
+ __u64 cookie;
} kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
struct {
__aligned_u64 tp_name; /* in/out */
__u32 name_len;
+ __u64 cookie;
} tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
struct {
__u64 config;
__u32 type;
+ __u64 cookie;
} event; /* BPF_PERF_EVENT_EVENT */
};
} perf_event;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index a1f18681721c..13193aaafb64 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3501,6 +3501,7 @@ static int bpf_perf_link_fill_kprobe(const struct perf_event *event,
if (!kallsyms_show_value(current_cred()))
addr = 0;
info->perf_event.kprobe.addr = addr;
+ info->perf_event.kprobe.cookie = event->bpf_cookie;
return 0;
}
#endif
@@ -3526,6 +3527,7 @@ static int bpf_perf_link_fill_uprobe(const struct perf_event *event,
else
info->perf_event.type = BPF_PERF_EVENT_UPROBE;
info->perf_event.uprobe.offset = offset;
+ info->perf_event.uprobe.cookie = event->bpf_cookie;
return 0;
}
#endif
@@ -3553,6 +3555,7 @@ static int bpf_perf_link_fill_tracepoint(const struct perf_event *event,
uname = u64_to_user_ptr(info->perf_event.tracepoint.tp_name);
ulen = info->perf_event.tracepoint.name_len;
info->perf_event.type = BPF_PERF_EVENT_TRACEPOINT;
+ info->perf_event.tracepoint.cookie = event->bpf_cookie;
return bpf_perf_link_fill_common(event, uname, ulen, NULL, NULL, NULL, NULL);
}
@@ -3561,6 +3564,7 @@ static int bpf_perf_link_fill_perf_event(const struct perf_event *event,
{
info->perf_event.event.type = event->attr.type;
info->perf_event.event.config = event->attr.config;
+ info->perf_event.event.cookie = event->bpf_cookie;
info->perf_event.type = BPF_PERF_EVENT_EVENT;
return 0;
}
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index a00f8a5623e1..b823d367a83c 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -6582,6 +6582,7 @@ struct bpf_link_info {
__aligned_u64 file_name; /* in/out */
__u32 name_len;
__u32 offset; /* offset from file_name */
+ __u64 cookie;
} uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
struct {
__aligned_u64 func_name; /* in/out */
@@ -6589,14 +6590,17 @@ struct bpf_link_info {
__u32 offset; /* offset from func_name */
__u64 addr;
__u64 missed;
+ __u64 cookie;
} kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
struct {
__aligned_u64 tp_name; /* in/out */
__u32 name_len;
+ __u64 cookie;
} tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
struct {
__u64 config;
__u32 type;
+ __u64 cookie;
} event; /* BPF_PERF_EVENT_EVENT */
};
} perf_event;
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 13:21 ` Yafang Shao
2024-01-18 9:54 ` [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link Jiri Olsa
` (6 subsequent siblings)
8 siblings, 1 reply; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Storing cookies in kprobe_multi bpf_link_info data. The cookies
field is optional and if provided it needs to be an array of
__u64 with kprobe_multi.count length.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/uapi/linux/bpf.h | 1 +
kernel/trace/bpf_trace.c | 15 +++++++++++++++
tools/include/uapi/linux/bpf.h | 1 +
3 files changed, 17 insertions(+)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index b823d367a83c..199cb93dca7f 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -6563,6 +6563,7 @@ struct bpf_link_info {
__u32 count; /* in/out: kprobe_multi function count */
__u32 flags;
__u64 missed;
+ __aligned_u64 cookies;
} kprobe_multi;
struct {
__aligned_u64 path;
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 7ac6c52b25eb..c98c20abaf99 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -2679,6 +2679,7 @@ static void bpf_kprobe_multi_link_dealloc(struct bpf_link *link)
static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
struct bpf_link_info *info)
{
+ u64 __user *ucookies = u64_to_user_ptr(info->kprobe_multi.cookies);
u64 __user *uaddrs = u64_to_user_ptr(info->kprobe_multi.addrs);
struct bpf_kprobe_multi_link *kmulti_link;
u32 ucount = info->kprobe_multi.count;
@@ -2686,6 +2687,8 @@ static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
if (!uaddrs ^ !ucount)
return -EINVAL;
+ if (ucookies && !ucount)
+ return -EINVAL;
kmulti_link = container_of(link, struct bpf_kprobe_multi_link, link);
info->kprobe_multi.count = kmulti_link->cnt;
@@ -2699,6 +2702,18 @@ static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
else
ucount = kmulti_link->cnt;
+ if (ucookies) {
+ if (kmulti_link->cookies) {
+ if (copy_to_user(ucookies, kmulti_link->cookies, ucount * sizeof(u64)))
+ return -EFAULT;
+ } else {
+ for (i = 0; i < ucount; i++) {
+ if (put_user(0, ucookies + i))
+ return -EFAULT;
+ }
+ }
+ }
+
if (kallsyms_show_value(current_cred())) {
if (copy_to_user(uaddrs, kmulti_link->addrs, ucount * sizeof(u64)))
return -EFAULT;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index b823d367a83c..199cb93dca7f 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -6563,6 +6563,7 @@ struct bpf_link_info {
__u32 count; /* in/out: kprobe_multi function count */
__u32 flags;
__u64 missed;
+ __aligned_u64 cookies;
} kprobe_multi;
struct {
__aligned_u64 path;
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 12:29 ` Yafang Shao
2024-01-18 9:54 ` [PATCH bpf-next 4/8] selftests/bpf: Add cookies check for kprobe_multi fill_link_info test Jiri Olsa
` (5 subsequent siblings)
8 siblings, 1 reply; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
The error path frees wrong array, it should be ref_ctr_offsets.
Fixes: a7795698f8b6 ("bpftool: Add support to display uprobe_multi links")
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/bpf/bpftool/link.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
index cb46667a6b2e..35b6859dd7c3 100644
--- a/tools/bpf/bpftool/link.c
+++ b/tools/bpf/bpftool/link.c
@@ -977,7 +977,7 @@ static int do_show_link(int fd)
cookies = calloc(count, sizeof(__u64));
if (!cookies) {
p_err("mem alloc failed");
- free(cookies);
+ free(ref_ctr_offsets);
free(offsets);
close(fd);
return -ENOMEM;
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 4/8] selftests/bpf: Add cookies check for kprobe_multi fill_link_info test
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (2 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 5/8] selftests/bpf: Add cookies check for perf_event " Jiri Olsa
` (4 subsequent siblings)
8 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Adding cookies check for kprobe_multi fill_link_info test,
plus tests for invalid values related to cookies.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/fill_link_info.c | 48 ++++++++++++++-----
1 file changed, 36 insertions(+), 12 deletions(-)
diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
index d4b1901f7879..2c2a02010c0b 100644
--- a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
+++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
@@ -19,6 +19,7 @@ static const char *kmulti_syms[] = {
};
#define KMULTI_CNT ARRAY_SIZE(kmulti_syms)
static __u64 kmulti_addrs[KMULTI_CNT];
+static __u64 kmulti_cookies[] = { 3, 1, 2 };
#define KPROBE_FUNC "bpf_fentry_test1"
static __u64 kprobe_addr;
@@ -195,11 +196,11 @@ static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
bpf_link__destroy(link);
}
-static int verify_kmulti_link_info(int fd, bool retprobe)
+static int verify_kmulti_link_info(int fd, bool retprobe, bool has_cookies)
{
+ __u64 addrs[KMULTI_CNT], cookies[KMULTI_CNT];
struct bpf_link_info info;
__u32 len = sizeof(info);
- __u64 addrs[KMULTI_CNT];
int flags, i, err;
memset(&info, 0, sizeof(info));
@@ -221,18 +222,22 @@ static int verify_kmulti_link_info(int fd, bool retprobe)
if (!info.kprobe_multi.addrs) {
info.kprobe_multi.addrs = ptr_to_u64(addrs);
+ info.kprobe_multi.cookies = ptr_to_u64(cookies);
goto again;
}
- for (i = 0; i < KMULTI_CNT; i++)
+ for (i = 0; i < KMULTI_CNT; i++) {
ASSERT_EQ(addrs[i], kmulti_addrs[i], "kmulti_addrs");
+ ASSERT_EQ(cookies[i], has_cookies ? kmulti_cookies[i] : 0,
+ "kmulti_cookies_value");
+ }
return 0;
}
static void verify_kmulti_invalid_user_buffer(int fd)
{
+ __u64 addrs[KMULTI_CNT], cookies[KMULTI_CNT];
struct bpf_link_info info;
__u32 len = sizeof(info);
- __u64 addrs[KMULTI_CNT];
int err, i;
memset(&info, 0, sizeof(info));
@@ -266,7 +271,20 @@ static void verify_kmulti_invalid_user_buffer(int fd)
info.kprobe_multi.count = KMULTI_CNT;
info.kprobe_multi.addrs = 0x1; /* invalid addr */
err = bpf_link_get_info_by_fd(fd, &info, &len);
- ASSERT_EQ(err, -EFAULT, "invalid_buff");
+ ASSERT_EQ(err, -EFAULT, "invalid_buff_addrs");
+
+ info.kprobe_multi.count = KMULTI_CNT;
+ info.kprobe_multi.addrs = ptr_to_u64(addrs);
+ info.kprobe_multi.cookies = 0x1; /* invalid addr */
+ err = bpf_link_get_info_by_fd(fd, &info, &len);
+ ASSERT_EQ(err, -EFAULT, "invalid_buff_cookies");
+
+ /* cookies && !count */
+ info.kprobe_multi.count = 0;
+ info.kprobe_multi.addrs = ptr_to_u64(NULL);
+ info.kprobe_multi.cookies = ptr_to_u64(cookies);
+ err = bpf_link_get_info_by_fd(fd, &info, &len);
+ ASSERT_EQ(err, -EINVAL, "invalid_cookies_count");
}
static int symbols_cmp_r(const void *a, const void *b)
@@ -278,13 +296,15 @@ static int symbols_cmp_r(const void *a, const void *b)
}
static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
- bool retprobe, bool invalid)
+ bool retprobe, bool cookies,
+ bool invalid)
{
LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
struct bpf_link *link;
int link_fd, err;
opts.syms = kmulti_syms;
+ opts.cookies = cookies ? kmulti_cookies : NULL;
opts.cnt = KMULTI_CNT;
opts.retprobe = retprobe;
link = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run, NULL, &opts);
@@ -293,7 +313,7 @@ static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
link_fd = bpf_link__fd(link);
if (!invalid) {
- err = verify_kmulti_link_info(link_fd, retprobe);
+ err = verify_kmulti_link_info(link_fd, retprobe, cookies);
ASSERT_OK(err, "verify_kmulti_link_info");
} else {
verify_kmulti_invalid_user_buffer(link_fd);
@@ -523,12 +543,16 @@ void test_fill_link_info(void)
qsort(kmulti_syms, KMULTI_CNT, sizeof(kmulti_syms[0]), symbols_cmp_r);
for (i = 0; i < KMULTI_CNT; i++)
kmulti_addrs[i] = ksym_get_addr(kmulti_syms[i]);
- if (test__start_subtest("kprobe_multi_link_info"))
- test_kprobe_multi_fill_link_info(skel, false, false);
- if (test__start_subtest("kretprobe_multi_link_info"))
- test_kprobe_multi_fill_link_info(skel, true, false);
+ if (test__start_subtest("kprobe_multi_link_info")) {
+ test_kprobe_multi_fill_link_info(skel, false, false, false);
+ test_kprobe_multi_fill_link_info(skel, false, true, false);
+ }
+ if (test__start_subtest("kretprobe_multi_link_info")) {
+ test_kprobe_multi_fill_link_info(skel, true, false, false);
+ test_kprobe_multi_fill_link_info(skel, true, true, false);
+ }
if (test__start_subtest("kprobe_multi_invalid_ubuff"))
- test_kprobe_multi_fill_link_info(skel, true, true);
+ test_kprobe_multi_fill_link_info(skel, true, true, true);
if (test__start_subtest("uprobe_multi_link_info"))
test_uprobe_multi_fill_link_info(skel, false, false);
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 5/8] selftests/bpf: Add cookies check for perf_event fill_link_info test
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (3 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 4/8] selftests/bpf: Add cookies check for kprobe_multi fill_link_info test Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 6/8] selftests/bpf: Add fill_link_info test for perf event Jiri Olsa
` (3 subsequent siblings)
8 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Now that we get cookies for perf_event probes, adding tests
for cookie for kprobe/uprobe/tracepoint.
The perf_event test needs to be added completely and is coming
in following change.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/fill_link_info.c | 26 +++++++++++++++----
1 file changed, 21 insertions(+), 5 deletions(-)
diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
index 2c2a02010c0b..20e105001bc3 100644
--- a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
+++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
@@ -32,6 +32,8 @@ static noinline void uprobe_func(void)
asm volatile ("");
}
+#define PERF_EVENT_COOKIE 0xdeadbeef
+
static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long addr,
ssize_t offset, ssize_t entry_offset)
{
@@ -63,6 +65,8 @@ static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long add
ASSERT_EQ(info.perf_event.kprobe.addr, addr + entry_offset,
"kprobe_addr");
+ ASSERT_EQ(info.perf_event.kprobe.cookie, PERF_EVENT_COOKIE, "kprobe_cookie");
+
if (!info.perf_event.kprobe.func_name) {
ASSERT_EQ(info.perf_event.kprobe.name_len, 0, "name_len");
info.perf_event.kprobe.func_name = ptr_to_u64(&buf);
@@ -82,6 +86,8 @@ static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long add
goto again;
}
+ ASSERT_EQ(info.perf_event.tracepoint.cookie, PERF_EVENT_COOKIE, "tracepoint_cookie");
+
err = strncmp(u64_to_ptr(info.perf_event.tracepoint.tp_name), TP_NAME,
strlen(TP_NAME));
ASSERT_EQ(err, 0, "cmp_tp_name");
@@ -97,6 +103,8 @@ static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long add
goto again;
}
+ ASSERT_EQ(info.perf_event.uprobe.cookie, PERF_EVENT_COOKIE, "uprobe_cookie");
+
err = strncmp(u64_to_ptr(info.perf_event.uprobe.file_name), UPROBE_FILE,
strlen(UPROBE_FILE));
ASSERT_EQ(err, 0, "cmp_file_name");
@@ -140,6 +148,7 @@ static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
.attach_mode = PROBE_ATTACH_MODE_LINK,
.retprobe = type == BPF_PERF_EVENT_KRETPROBE,
+ .bpf_cookie = PERF_EVENT_COOKIE,
);
ssize_t entry_offset = 0;
struct bpf_link *link;
@@ -164,10 +173,13 @@ static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
static void test_tp_fill_link_info(struct test_fill_link_info *skel)
{
+ DECLARE_LIBBPF_OPTS(bpf_tracepoint_opts, opts,
+ .bpf_cookie = PERF_EVENT_COOKIE,
+ );
struct bpf_link *link;
int link_fd, err;
- link = bpf_program__attach_tracepoint(skel->progs.tp_run, TP_CAT, TP_NAME);
+ link = bpf_program__attach_tracepoint_opts(skel->progs.tp_run, TP_CAT, TP_NAME, &opts);
if (!ASSERT_OK_PTR(link, "attach_tp"))
return;
@@ -180,13 +192,17 @@ static void test_tp_fill_link_info(struct test_fill_link_info *skel)
static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
enum bpf_perf_event_type type)
{
+ DECLARE_LIBBPF_OPTS(bpf_uprobe_opts, opts,
+ .retprobe = type == BPF_PERF_EVENT_URETPROBE,
+ .bpf_cookie = PERF_EVENT_COOKIE,
+ );
struct bpf_link *link;
int link_fd, err;
- link = bpf_program__attach_uprobe(skel->progs.uprobe_run,
- type == BPF_PERF_EVENT_URETPROBE,
- 0, /* self pid */
- UPROBE_FILE, uprobe_offset);
+ link = bpf_program__attach_uprobe_opts(skel->progs.uprobe_run,
+ 0, /* self pid */
+ UPROBE_FILE, uprobe_offset,
+ &opts);
if (!ASSERT_OK_PTR(link, "attach_uprobe"))
return;
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 6/8] selftests/bpf: Add fill_link_info test for perf event
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (4 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 5/8] selftests/bpf: Add cookies check for perf_event " Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes Jiri Olsa
` (2 subsequent siblings)
8 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Adding fill_link_info test for perf event and testing we
get its values back through the bpf_link_info interface.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/fill_link_info.c | 40 +++++++++++++++++++
.../selftests/bpf/progs/test_fill_link_info.c | 6 +++
2 files changed, 46 insertions(+)
diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
index 20e105001bc3..f3932941bbaa 100644
--- a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
+++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
@@ -109,6 +109,11 @@ static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long add
strlen(UPROBE_FILE));
ASSERT_EQ(err, 0, "cmp_file_name");
break;
+ case BPF_PERF_EVENT_EVENT:
+ ASSERT_EQ(info.perf_event.event.type, PERF_TYPE_SOFTWARE, "event_type");
+ ASSERT_EQ(info.perf_event.event.config, PERF_COUNT_SW_PAGE_FAULTS, "event_config");
+ ASSERT_EQ(info.perf_event.event.cookie, PERF_EVENT_COOKIE, "event_cookie");
+ break;
default:
err = -1;
break;
@@ -189,6 +194,39 @@ static void test_tp_fill_link_info(struct test_fill_link_info *skel)
bpf_link__destroy(link);
}
+static void test_event_fill_link_info(struct test_fill_link_info *skel)
+{
+ DECLARE_LIBBPF_OPTS(bpf_perf_event_opts, opts,
+ .bpf_cookie = PERF_EVENT_COOKIE,
+ );
+ struct bpf_link *link;
+ int link_fd, err, pfd;
+ struct perf_event_attr attr = {
+ .type = PERF_TYPE_SOFTWARE,
+ .config = PERF_COUNT_SW_PAGE_FAULTS,
+ .freq = 1,
+ .sample_freq = 1,
+ .size = sizeof(struct perf_event_attr),
+ };
+
+ pfd = syscall(__NR_perf_event_open, &attr, -1 /* pid */, 0 /* cpu 0 */,
+ -1 /* group id */, 0 /* flags */);
+ if (!ASSERT_GE(pfd, 0, "perf_event_open"))
+ return;
+
+ link = bpf_program__attach_perf_event_opts(skel->progs.event_run, pfd, &opts);
+ if (!ASSERT_OK_PTR(link, "attach_event"))
+ goto error;
+
+ link_fd = bpf_link__fd(link);
+ err = verify_perf_link_info(link_fd, BPF_PERF_EVENT_EVENT, 0, 0, 0);
+ ASSERT_OK(err, "verify_perf_link_info");
+ bpf_link__destroy(link);
+
+error:
+ close(pfd);
+}
+
static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
enum bpf_perf_event_type type)
{
@@ -549,6 +587,8 @@ void test_fill_link_info(void)
test_kprobe_fill_link_info(skel, BPF_PERF_EVENT_KPROBE, true);
if (test__start_subtest("tracepoint_link_info"))
test_tp_fill_link_info(skel);
+ if (test__start_subtest("event_link_info"))
+ test_event_fill_link_info(skel);
uprobe_offset = get_uprobe_offset(&uprobe_func);
if (test__start_subtest("uprobe_link_info"))
diff --git a/tools/testing/selftests/bpf/progs/test_fill_link_info.c b/tools/testing/selftests/bpf/progs/test_fill_link_info.c
index 69509f8bb680..6afa834756e9 100644
--- a/tools/testing/selftests/bpf/progs/test_fill_link_info.c
+++ b/tools/testing/selftests/bpf/progs/test_fill_link_info.c
@@ -33,6 +33,12 @@ int BPF_PROG(tp_run)
return 0;
}
+SEC("perf_event")
+int event_run(void *ctx)
+{
+ return 0;
+}
+
SEC("kprobe.multi")
int BPF_PROG(kmulti_run)
{
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (5 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 6/8] selftests/bpf: Add fill_link_info test for perf event Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 17:51 ` Quentin Monnet
2024-01-18 9:54 ` [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link Jiri Olsa
2024-01-18 12:42 ` [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Yafang Shao
8 siblings, 1 reply; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Displaying cookie for perf event link probes, in plain mode:
# bpftool link
17: perf_event prog 90
kprobe ffffffff82b1c2b0 bpf_fentry_test1 cookie 3735928559
18: perf_event prog 90
kretprobe ffffffff82b1c2b0 bpf_fentry_test1 cookie 3735928559
20: perf_event prog 92
tracepoint sched_switch cookie 3735928559
21: perf_event prog 93
event software:page-faults cookie 3735928559
22: perf_event prog 91
uprobe /proc/self/exe+0xd703c cookie 3735928559
And in json mode:
# bpftool link -j | jq
{
"id": 30,
"type": "perf_event",
"prog_id": 160,
"retprobe": false,
"addr": 18446744071607272112,
"func": "bpf_fentry_test1",
"offset": 0,
"missed": 0,
"cookie": 3735928559
}
{
"id": 33,
"type": "perf_event",
"prog_id": 162,
"tracepoint": "sched_switch",
"cookie": 3735928559
}
{
"id": 34,
"type": "perf_event",
"prog_id": 163,
"event_type": "software",
"event_config": "page-faults",
"cookie": 3735928559
}
{
"id": 35,
"type": "perf_event",
"prog_id": 161,
"retprobe": false,
"file": "/proc/self/exe",
"offset": 880700,
"cookie": 3735928559
}
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/bpf/bpftool/link.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
index 35b6859dd7c3..b66a1598b87c 100644
--- a/tools/bpf/bpftool/link.c
+++ b/tools/bpf/bpftool/link.c
@@ -334,6 +334,7 @@ show_perf_event_kprobe_json(struct bpf_link_info *info, json_writer_t *wtr)
u64_to_ptr(info->perf_event.kprobe.func_name));
jsonw_uint_field(wtr, "offset", info->perf_event.kprobe.offset);
jsonw_uint_field(wtr, "missed", info->perf_event.kprobe.missed);
+ jsonw_uint_field(wtr, "cookie", info->perf_event.kprobe.cookie);
}
static void
@@ -343,6 +344,7 @@ show_perf_event_uprobe_json(struct bpf_link_info *info, json_writer_t *wtr)
jsonw_string_field(wtr, "file",
u64_to_ptr(info->perf_event.uprobe.file_name));
jsonw_uint_field(wtr, "offset", info->perf_event.uprobe.offset);
+ jsonw_uint_field(wtr, "cookie", info->perf_event.uprobe.cookie);
}
static void
@@ -350,6 +352,7 @@ show_perf_event_tracepoint_json(struct bpf_link_info *info, json_writer_t *wtr)
{
jsonw_string_field(wtr, "tracepoint",
u64_to_ptr(info->perf_event.tracepoint.tp_name));
+ jsonw_uint_field(wtr, "cookie", info->perf_event.tracepoint.cookie);
}
static char *perf_config_hw_cache_str(__u64 config)
@@ -426,6 +429,8 @@ show_perf_event_event_json(struct bpf_link_info *info, json_writer_t *wtr)
else
jsonw_uint_field(wtr, "event_config", config);
+ jsonw_uint_field(wtr, "cookie", info->perf_event.event.cookie);
+
if (type == PERF_TYPE_HW_CACHE && perf_config)
free((void *)perf_config);
}
@@ -754,6 +759,8 @@ static void show_perf_event_kprobe_plain(struct bpf_link_info *info)
printf("+%#x", info->perf_event.kprobe.offset);
if (info->perf_event.kprobe.missed)
printf(" missed %llu", info->perf_event.kprobe.missed);
+ if (info->perf_event.kprobe.cookie)
+ printf(" cookie %llu", info->perf_event.kprobe.cookie);
printf(" ");
}
@@ -770,6 +777,8 @@ static void show_perf_event_uprobe_plain(struct bpf_link_info *info)
else
printf("\n\tuprobe ");
printf("%s+%#x ", buf, info->perf_event.uprobe.offset);
+ if (info->perf_event.uprobe.cookie)
+ printf("cookie %llu ", info->perf_event.uprobe.cookie);
}
static void show_perf_event_tracepoint_plain(struct bpf_link_info *info)
@@ -781,6 +790,8 @@ static void show_perf_event_tracepoint_plain(struct bpf_link_info *info)
return;
printf("\n\ttracepoint %s ", buf);
+ if (info->perf_event.tracepoint.cookie)
+ printf("cookie %llu ", info->perf_event.tracepoint.cookie);
}
static void show_perf_event_event_plain(struct bpf_link_info *info)
@@ -802,6 +813,9 @@ static void show_perf_event_event_plain(struct bpf_link_info *info)
else
printf("%llu ", config);
+ if (info->perf_event.event.cookie)
+ printf("cookie %llu ", info->perf_event.event.cookie);
+
if (type == PERF_TYPE_HW_CACHE && perf_config)
free((void *)perf_config);
}
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (6 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes Jiri Olsa
@ 2024-01-18 9:54 ` Jiri Olsa
2024-01-18 17:51 ` Quentin Monnet
2024-01-18 12:42 ` [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Yafang Shao
8 siblings, 1 reply; 19+ messages in thread
From: Jiri Olsa @ 2024-01-18 9:54 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao,
Quentin Monnet
Displaying cookies for kprobe multi link, in plain mode:
# bpftool link
...
1397: kprobe_multi prog 47532
kretprobe.multi func_cnt 3
addr cookie func [module]
ffffffff82b370c0 3 bpf_fentry_test1
ffffffff82b39780 1 bpf_fentry_test2
ffffffff82b397a0 2 bpf_fentry_test3
And in json mode:
# bpftool link -j | jq
...
{
"id": 1397,
"type": "kprobe_multi",
"prog_id": 47532,
"retprobe": true,
"func_cnt": 3,
"missed": 0,
"funcs": [
{
"addr": 18446744071607382208,
"func": "bpf_fentry_test1",
"module": null,
"cookie": 3
},
{
"addr": 18446744071607392128,
"func": "bpf_fentry_test2",
"module": null,
"cookie": 1
},
{
"addr": 18446744071607392160,
"func": "bpf_fentry_test3",
"module": null,
"cookie": 2
}
]
}
Cookie is attached to specific address, and because we sort addresses
before printing, we need to sort cookies the same way, hence adding
the struct addr_cookie to keep and sort them together.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/bpf/bpftool/link.c | 71 ++++++++++++++++++++++++++++++++--------
1 file changed, 57 insertions(+), 14 deletions(-)
diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
index b66a1598b87c..fd862afe6c6f 100644
--- a/tools/bpf/bpftool/link.c
+++ b/tools/bpf/bpftool/link.c
@@ -249,18 +249,44 @@ static int get_prog_info(int prog_id, struct bpf_prog_info *info)
return err;
}
-static int cmp_u64(const void *A, const void *B)
+struct addr_cookie {
+ __u64 addr;
+ __u64 cookie;
+};
+
+static int cmp_addr_cookie(const void *A, const void *B)
+{
+ const struct addr_cookie *a = A, *b = B;
+
+ if (a->addr == b->addr)
+ return 0;
+ return a->addr < b->addr ? -1 : 1;
+}
+
+static struct addr_cookie *
+get_addr_cookie_array(__u64 *addrs, __u64 *cookies, __u32 count)
{
- const __u64 *a = A, *b = B;
+ struct addr_cookie *data;
+ __u32 i;
- return *a - *b;
+ data = calloc(count, sizeof(data[0]));
+ if (!data) {
+ p_err("mem alloc failed");
+ return NULL;
+ }
+ for (i = 0; i < count; i++) {
+ data[i].addr = addrs[i];
+ data[i].cookie = cookies[i];
+ }
+ qsort(data, count, sizeof(data[0]), cmp_addr_cookie);
+ return data;
}
static void
show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
{
+ struct addr_cookie *data;
__u32 i, j = 0;
- __u64 *addrs;
jsonw_bool_field(json_wtr, "retprobe",
info->kprobe_multi.flags & BPF_F_KPROBE_MULTI_RETURN);
@@ -268,14 +294,17 @@ show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
jsonw_uint_field(json_wtr, "missed", info->kprobe_multi.missed);
jsonw_name(json_wtr, "funcs");
jsonw_start_array(json_wtr);
- addrs = u64_to_ptr(info->kprobe_multi.addrs);
- qsort(addrs, info->kprobe_multi.count, sizeof(addrs[0]), cmp_u64);
+ data = get_addr_cookie_array(u64_to_ptr(info->kprobe_multi.addrs),
+ u64_to_ptr(info->kprobe_multi.cookies),
+ info->kprobe_multi.count);
+ if (!data)
+ return;
/* Load it once for all. */
if (!dd.sym_count)
kernel_syms_load(&dd);
for (i = 0; i < dd.sym_count; i++) {
- if (dd.sym_mapping[i].address != addrs[j])
+ if (dd.sym_mapping[i].address != data[j].addr)
continue;
jsonw_start_object(json_wtr);
jsonw_uint_field(json_wtr, "addr", dd.sym_mapping[i].address);
@@ -287,11 +316,13 @@ show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
} else {
jsonw_string_field(json_wtr, "module", dd.sym_mapping[i].module);
}
+ jsonw_uint_field(json_wtr, "cookie", data[j].cookie);
jsonw_end_object(json_wtr);
if (j++ == info->kprobe_multi.count)
break;
}
jsonw_end_array(json_wtr);
+ free(data);
}
static __u64 *u64_to_arr(__u64 val)
@@ -675,8 +706,8 @@ void netfilter_dump_plain(const struct bpf_link_info *info)
static void show_kprobe_multi_plain(struct bpf_link_info *info)
{
+ struct addr_cookie *data;
__u32 i, j = 0;
- __u64 *addrs;
if (!info->kprobe_multi.count)
return;
@@ -688,8 +719,11 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
printf("func_cnt %u ", info->kprobe_multi.count);
if (info->kprobe_multi.missed)
printf("missed %llu ", info->kprobe_multi.missed);
- addrs = (__u64 *)u64_to_ptr(info->kprobe_multi.addrs);
- qsort(addrs, info->kprobe_multi.count, sizeof(__u64), cmp_u64);
+ data = get_addr_cookie_array(u64_to_ptr(info->kprobe_multi.addrs),
+ u64_to_ptr(info->kprobe_multi.cookies),
+ info->kprobe_multi.count);
+ if (!data)
+ return;
/* Load it once for all. */
if (!dd.sym_count)
@@ -697,12 +731,12 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
if (!dd.sym_count)
return;
- printf("\n\t%-16s %s", "addr", "func [module]");
+ printf("\n\t%-16s %-16s %s", "addr", "cookie", "func [module]");
for (i = 0; i < dd.sym_count; i++) {
- if (dd.sym_mapping[i].address != addrs[j])
+ if (dd.sym_mapping[i].address != data[j].addr)
continue;
- printf("\n\t%016lx %s",
- dd.sym_mapping[i].address, dd.sym_mapping[i].name);
+ printf("\n\t%016lx %-16llx %s",
+ dd.sym_mapping[i].address, data[j].cookie, dd.sym_mapping[i].name);
if (dd.sym_mapping[i].module[0] != '\0')
printf(" [%s] ", dd.sym_mapping[i].module);
else
@@ -711,6 +745,7 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
if (j++ == info->kprobe_multi.count)
break;
}
+ free(data);
}
static void show_uprobe_multi_plain(struct bpf_link_info *info)
@@ -966,6 +1001,14 @@ static int do_show_link(int fd)
return -ENOMEM;
}
info.kprobe_multi.addrs = ptr_to_u64(addrs);
+ cookies = calloc(count, sizeof(__u64));
+ if (!cookies) {
+ p_err("mem alloc failed");
+ free(addrs);
+ close(fd);
+ return -ENOMEM;
+ }
+ info.kprobe_multi.cookies = ptr_to_u64(cookies);
goto again;
}
}
--
2.43.0
^ permalink raw reply related [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records
2024-01-18 9:54 ` [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records Jiri Olsa
@ 2024-01-18 12:24 ` Yafang Shao
2024-01-19 8:03 ` Jiri Olsa
0 siblings, 1 reply; 19+ messages in thread
From: Yafang Shao @ 2024-01-18 12:24 UTC (permalink / raw)
To: Jiri Olsa
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 5:54 PM Jiri Olsa <jolsa@kernel.org> wrote:
>
> At the moment we don't store cookie for perf_event probes,
> while we do that for the rest of the probes.
>
> Adding cookie fields to struct bpf_link_info perf event
> probe records:
>
> perf_event.uprobe
> perf_event.kprobe
> perf_event.tracepoint
> perf_event.perf_event
>
> And the code to store that in bpf_link_info struct.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
> include/uapi/linux/bpf.h | 4 ++++
> kernel/bpf/syscall.c | 4 ++++
> tools/include/uapi/linux/bpf.h | 4 ++++
> 3 files changed, 12 insertions(+)
>
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index a00f8a5623e1..b823d367a83c 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -6582,6 +6582,7 @@ struct bpf_link_info {
> __aligned_u64 file_name; /* in/out */
> __u32 name_len;
> __u32 offset; /* offset from file_name */
> + __u64 cookie;
> } uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
> struct {
> __aligned_u64 func_name; /* in/out */
> @@ -6589,14 +6590,17 @@ struct bpf_link_info {
> __u32 offset; /* offset from func_name */
> __u64 addr;
> __u64 missed;
> + __u64 cookie;
> } kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
> struct {
> __aligned_u64 tp_name; /* in/out */
> __u32 name_len;
It might be beneficial to include an alignment pad '__u32 :32;' here,
following the pattern used in other instances within this file.
> + __u64 cookie;
> } tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
> struct {
> __u64 config;
> __u32 type;
Same here.
> + __u64 cookie;
> } event; /* BPF_PERF_EVENT_EVENT */
> };
> } perf_event;
> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> index a1f18681721c..13193aaafb64 100644
> --- a/kernel/bpf/syscall.c
> +++ b/kernel/bpf/syscall.c
> @@ -3501,6 +3501,7 @@ static int bpf_perf_link_fill_kprobe(const struct perf_event *event,
> if (!kallsyms_show_value(current_cred()))
> addr = 0;
> info->perf_event.kprobe.addr = addr;
> + info->perf_event.kprobe.cookie = event->bpf_cookie;
> return 0;
> }
> #endif
> @@ -3526,6 +3527,7 @@ static int bpf_perf_link_fill_uprobe(const struct perf_event *event,
> else
> info->perf_event.type = BPF_PERF_EVENT_UPROBE;
> info->perf_event.uprobe.offset = offset;
> + info->perf_event.uprobe.cookie = event->bpf_cookie;
> return 0;
> }
> #endif
> @@ -3553,6 +3555,7 @@ static int bpf_perf_link_fill_tracepoint(const struct perf_event *event,
> uname = u64_to_user_ptr(info->perf_event.tracepoint.tp_name);
> ulen = info->perf_event.tracepoint.name_len;
> info->perf_event.type = BPF_PERF_EVENT_TRACEPOINT;
> + info->perf_event.tracepoint.cookie = event->bpf_cookie;
> return bpf_perf_link_fill_common(event, uname, ulen, NULL, NULL, NULL, NULL);
> }
>
> @@ -3561,6 +3564,7 @@ static int bpf_perf_link_fill_perf_event(const struct perf_event *event,
> {
> info->perf_event.event.type = event->attr.type;
> info->perf_event.event.config = event->attr.config;
> + info->perf_event.event.cookie = event->bpf_cookie;
> info->perf_event.type = BPF_PERF_EVENT_EVENT;
> return 0;
> }
> diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
> index a00f8a5623e1..b823d367a83c 100644
> --- a/tools/include/uapi/linux/bpf.h
> +++ b/tools/include/uapi/linux/bpf.h
> @@ -6582,6 +6582,7 @@ struct bpf_link_info {
> __aligned_u64 file_name; /* in/out */
> __u32 name_len;
> __u32 offset; /* offset from file_name */
> + __u64 cookie;
> } uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
> struct {
> __aligned_u64 func_name; /* in/out */
> @@ -6589,14 +6590,17 @@ struct bpf_link_info {
> __u32 offset; /* offset from func_name */
> __u64 addr;
> __u64 missed;
> + __u64 cookie;
> } kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
> struct {
> __aligned_u64 tp_name; /* in/out */
> __u32 name_len;
> + __u64 cookie;
> } tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
> struct {
> __u64 config;
> __u32 type;
> + __u64 cookie;
> } event; /* BPF_PERF_EVENT_EVENT */
> };
> } perf_event;
> --
> 2.43.0
>
--
Regards
Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link
2024-01-18 9:54 ` [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link Jiri Olsa
@ 2024-01-18 12:29 ` Yafang Shao
2024-01-18 17:51 ` Quentin Monnet
0 siblings, 1 reply; 19+ messages in thread
From: Yafang Shao @ 2024-01-18 12:29 UTC (permalink / raw)
To: Jiri Olsa
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 5:55 PM Jiri Olsa <jolsa@kernel.org> wrote:
>
> The error path frees wrong array, it should be ref_ctr_offsets.
>
> Fixes: a7795698f8b6 ("bpftool: Add support to display uprobe_multi links")
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
> ---
> tools/bpf/bpftool/link.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
> index cb46667a6b2e..35b6859dd7c3 100644
> --- a/tools/bpf/bpftool/link.c
> +++ b/tools/bpf/bpftool/link.c
> @@ -977,7 +977,7 @@ static int do_show_link(int fd)
> cookies = calloc(count, sizeof(__u64));
> if (!cookies) {
> p_err("mem alloc failed");
> - free(cookies);
> + free(ref_ctr_offsets);
> free(offsets);
> close(fd);
> return -ENOMEM;
> --
> 2.43.0
>
--
Regards
Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
` (7 preceding siblings ...)
2024-01-18 9:54 ` [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link Jiri Olsa
@ 2024-01-18 12:42 ` Yafang Shao
2024-01-19 8:03 ` Jiri Olsa
8 siblings, 1 reply; 19+ messages in thread
From: Yafang Shao @ 2024-01-18 12:42 UTC (permalink / raw)
To: Jiri Olsa
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 5:54 PM Jiri Olsa <jolsa@kernel.org> wrote:
>
> hi,
> this patchset adds support to retrieve cookies from existing tracing
> links that still did not support it plus changes to bpftool to display
> them. It's leftover we discussed some time ago [1].
Thanks for your work.
bpf cookie is also displayed in the pid_iter [0]. After we add support
for the cookie for other progs like kprobe_multi and uprobe_multi, I
think we should update this file as well.
[0]. tools/bpf/bpftool/skeleton/pid_iter.bpf.c
>
> thanks,
> jirka
>
>
> [1] https://lore.kernel.org/bpf/CALOAHbAZ6=A9j3VFCLoAC_WhgQKU7injMf06=cM2sU4Hi4Sx+Q@mail.gmail.com/
> ---
> Jiri Olsa (8):
> bpf: Add cookie to perf_event bpf_link_info records
> bpf: Store cookies in kprobe_multi bpf_link_info data
> bpftool: Fix wrong free call in do_show_link
> selftests/bpf: Add cookies check for kprobe_multi fill_link_info test
> selftests/bpf: Add cookies check for perf_event fill_link_info test
> selftests/bpf: Add fill_link_info test for perf event
> bpftool: Display cookie for perf event link probes
> bpftool: Display cookie for kprobe multi link
>
> include/uapi/linux/bpf.h | 5 +++++
> kernel/bpf/syscall.c | 4 ++++
> kernel/trace/bpf_trace.c | 15 +++++++++++++
> tools/bpf/bpftool/link.c | 87 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------
> tools/include/uapi/linux/bpf.h | 5 +++++
> tools/testing/selftests/bpf/prog_tests/fill_link_info.c | 114 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------
> tools/testing/selftests/bpf/progs/test_fill_link_info.c | 6 +++++
> 7 files changed, 204 insertions(+), 32 deletions(-)
--
Regards
Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data
2024-01-18 9:54 ` [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data Jiri Olsa
@ 2024-01-18 13:21 ` Yafang Shao
0 siblings, 0 replies; 19+ messages in thread
From: Yafang Shao @ 2024-01-18 13:21 UTC (permalink / raw)
To: Jiri Olsa
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 5:54 PM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Storing cookies in kprobe_multi bpf_link_info data. The cookies
> field is optional and if provided it needs to be an array of
> __u64 with kprobe_multi.count length.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
LGTM.
Acked-by: Yafang Shao <laoar.shao@gmail.com>
> ---
> include/uapi/linux/bpf.h | 1 +
> kernel/trace/bpf_trace.c | 15 +++++++++++++++
> tools/include/uapi/linux/bpf.h | 1 +
> 3 files changed, 17 insertions(+)
>
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index b823d367a83c..199cb93dca7f 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -6563,6 +6563,7 @@ struct bpf_link_info {
> __u32 count; /* in/out: kprobe_multi function count */
> __u32 flags;
> __u64 missed;
> + __aligned_u64 cookies;
> } kprobe_multi;
> struct {
> __aligned_u64 path;
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 7ac6c52b25eb..c98c20abaf99 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -2679,6 +2679,7 @@ static void bpf_kprobe_multi_link_dealloc(struct bpf_link *link)
> static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
> struct bpf_link_info *info)
> {
> + u64 __user *ucookies = u64_to_user_ptr(info->kprobe_multi.cookies);
> u64 __user *uaddrs = u64_to_user_ptr(info->kprobe_multi.addrs);
> struct bpf_kprobe_multi_link *kmulti_link;
> u32 ucount = info->kprobe_multi.count;
> @@ -2686,6 +2687,8 @@ static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
>
> if (!uaddrs ^ !ucount)
> return -EINVAL;
> + if (ucookies && !ucount)
> + return -EINVAL;
>
> kmulti_link = container_of(link, struct bpf_kprobe_multi_link, link);
> info->kprobe_multi.count = kmulti_link->cnt;
> @@ -2699,6 +2702,18 @@ static int bpf_kprobe_multi_link_fill_link_info(const struct bpf_link *link,
> else
> ucount = kmulti_link->cnt;
>
> + if (ucookies) {
> + if (kmulti_link->cookies) {
> + if (copy_to_user(ucookies, kmulti_link->cookies, ucount * sizeof(u64)))
> + return -EFAULT;
> + } else {
> + for (i = 0; i < ucount; i++) {
> + if (put_user(0, ucookies + i))
> + return -EFAULT;
> + }
> + }
> + }
> +
> if (kallsyms_show_value(current_cred())) {
> if (copy_to_user(uaddrs, kmulti_link->addrs, ucount * sizeof(u64)))
> return -EFAULT;
> diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
> index b823d367a83c..199cb93dca7f 100644
> --- a/tools/include/uapi/linux/bpf.h
> +++ b/tools/include/uapi/linux/bpf.h
> @@ -6563,6 +6563,7 @@ struct bpf_link_info {
> __u32 count; /* in/out: kprobe_multi function count */
> __u32 flags;
> __u64 missed;
> + __aligned_u64 cookies;
> } kprobe_multi;
> struct {
> __aligned_u64 path;
> --
> 2.43.0
>
--
Regards
Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link
2024-01-18 12:29 ` Yafang Shao
@ 2024-01-18 17:51 ` Quentin Monnet
0 siblings, 0 replies; 19+ messages in thread
From: Quentin Monnet @ 2024-01-18 17:51 UTC (permalink / raw)
To: Yafang Shao, Jiri Olsa
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo
2024-01-18 12:30 UTC+0000 ~ Yafang Shao <laoar.shao@gmail.com>
> On Thu, Jan 18, 2024 at 5:55 PM Jiri Olsa <jolsa@kernel.org> wrote:
>>
>> The error path frees wrong array, it should be ref_ctr_offsets.
>>
>> Fixes: a7795698f8b6 ("bpftool: Add support to display uprobe_multi links")
>> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
>
> Acked-by: Yafang Shao <laoar.shao@gmail.com>
Reviewed-by: Quentin Monnet <quentin@isovalent.com>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes
2024-01-18 9:54 ` [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes Jiri Olsa
@ 2024-01-18 17:51 ` Quentin Monnet
0 siblings, 0 replies; 19+ messages in thread
From: Quentin Monnet @ 2024-01-18 17:51 UTC (permalink / raw)
To: Jiri Olsa, Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao
2024-01-18 09:55 UTC+0000 ~ Jiri Olsa <jolsa@kernel.org>
> Displaying cookie for perf event link probes, in plain mode:
>
> # bpftool link
> 17: perf_event prog 90
> kprobe ffffffff82b1c2b0 bpf_fentry_test1 cookie 3735928559
> 18: perf_event prog 90
> kretprobe ffffffff82b1c2b0 bpf_fentry_test1 cookie 3735928559
> 20: perf_event prog 92
> tracepoint sched_switch cookie 3735928559
> 21: perf_event prog 93
> event software:page-faults cookie 3735928559
> 22: perf_event prog 91
> uprobe /proc/self/exe+0xd703c cookie 3735928559
>
> And in json mode:
>
> # bpftool link -j | jq
>
> {
> "id": 30,
> "type": "perf_event",
> "prog_id": 160,
> "retprobe": false,
> "addr": 18446744071607272112,
> "func": "bpf_fentry_test1",
> "offset": 0,
> "missed": 0,
> "cookie": 3735928559
> }
>
> {
> "id": 33,
> "type": "perf_event",
> "prog_id": 162,
> "tracepoint": "sched_switch",
> "cookie": 3735928559
> }
>
> {
> "id": 34,
> "type": "perf_event",
> "prog_id": 163,
> "event_type": "software",
> "event_config": "page-faults",
> "cookie": 3735928559
> }
>
> {
> "id": 35,
> "type": "perf_event",
> "prog_id": 161,
> "retprobe": false,
> "file": "/proc/self/exe",
> "offset": 880700,
> "cookie": 3735928559
> }
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Reviewed-by: Quentin Monnet <quentin@isovalent.com>
Thanks!
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link
2024-01-18 9:54 ` [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link Jiri Olsa
@ 2024-01-18 17:51 ` Quentin Monnet
2024-01-19 8:03 ` Jiri Olsa
0 siblings, 1 reply; 19+ messages in thread
From: Quentin Monnet @ 2024-01-18 17:51 UTC (permalink / raw)
To: Jiri Olsa, Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao
2024-01-18 09:55 UTC+0000 ~ Jiri Olsa <jolsa@kernel.org>
> Displaying cookies for kprobe multi link, in plain mode:
>
> # bpftool link
> ...
> 1397: kprobe_multi prog 47532
> kretprobe.multi func_cnt 3
> addr cookie func [module]
> ffffffff82b370c0 3 bpf_fentry_test1
> ffffffff82b39780 1 bpf_fentry_test2
> ffffffff82b397a0 2 bpf_fentry_test3
>
> And in json mode:
>
> # bpftool link -j | jq
> ...
> {
> "id": 1397,
> "type": "kprobe_multi",
> "prog_id": 47532,
> "retprobe": true,
> "func_cnt": 3,
> "missed": 0,
> "funcs": [
> {
> "addr": 18446744071607382208,
> "func": "bpf_fentry_test1",
> "module": null,
> "cookie": 3
> },
> {
> "addr": 18446744071607392128,
> "func": "bpf_fentry_test2",
> "module": null,
> "cookie": 1
> },
> {
> "addr": 18446744071607392160,
> "func": "bpf_fentry_test3",
> "module": null,
> "cookie": 2
> }
> ]
> }
>
> Cookie is attached to specific address, and because we sort addresses
> before printing, we need to sort cookies the same way, hence adding
> the struct addr_cookie to keep and sort them together.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
> tools/bpf/bpftool/link.c | 71 ++++++++++++++++++++++++++++++++--------
> 1 file changed, 57 insertions(+), 14 deletions(-)
>
> diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
> index b66a1598b87c..fd862afe6c6f 100644
> --- a/tools/bpf/bpftool/link.c
> +++ b/tools/bpf/bpftool/link.c
> @@ -249,18 +249,44 @@ static int get_prog_info(int prog_id, struct bpf_prog_info *info)
> return err;
> }
>
> -static int cmp_u64(const void *A, const void *B)
> +struct addr_cookie {
> + __u64 addr;
> + __u64 cookie;
> +};
> +
> +static int cmp_addr_cookie(const void *A, const void *B)
> +{
> + const struct addr_cookie *a = A, *b = B;
> +
> + if (a->addr == b->addr)
> + return 0;
> + return a->addr < b->addr ? -1 : 1;
> +}
> +
> +static struct addr_cookie *
> +get_addr_cookie_array(__u64 *addrs, __u64 *cookies, __u32 count)
> {
> - const __u64 *a = A, *b = B;
> + struct addr_cookie *data;
> + __u32 i;
>
> - return *a - *b;
> + data = calloc(count, sizeof(data[0]));
> + if (!data) {
> + p_err("mem alloc failed");
> + return NULL;
> + }
> + for (i = 0; i < count; i++) {
> + data[i].addr = addrs[i];
> + data[i].cookie = cookies[i];
> + }
> + qsort(data, count, sizeof(data[0]), cmp_addr_cookie);
> + return data;
> }
>
> static void
> show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
> {
> + struct addr_cookie *data;
> __u32 i, j = 0;
> - __u64 *addrs;
>
> jsonw_bool_field(json_wtr, "retprobe",
> info->kprobe_multi.flags & BPF_F_KPROBE_MULTI_RETURN);
> @@ -268,14 +294,17 @@ show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
> jsonw_uint_field(json_wtr, "missed", info->kprobe_multi.missed);
> jsonw_name(json_wtr, "funcs");
> jsonw_start_array(json_wtr);
> - addrs = u64_to_ptr(info->kprobe_multi.addrs);
> - qsort(addrs, info->kprobe_multi.count, sizeof(addrs[0]), cmp_u64);
> + data = get_addr_cookie_array(u64_to_ptr(info->kprobe_multi.addrs),
> + u64_to_ptr(info->kprobe_multi.cookies),
> + info->kprobe_multi.count);
> + if (!data)
> + return;
>
> /* Load it once for all. */
> if (!dd.sym_count)
> kernel_syms_load(&dd);
> for (i = 0; i < dd.sym_count; i++) {
> - if (dd.sym_mapping[i].address != addrs[j])
> + if (dd.sym_mapping[i].address != data[j].addr)
> continue;
> jsonw_start_object(json_wtr);
> jsonw_uint_field(json_wtr, "addr", dd.sym_mapping[i].address);
> @@ -287,11 +316,13 @@ show_kprobe_multi_json(struct bpf_link_info *info, json_writer_t *wtr)
> } else {
> jsonw_string_field(json_wtr, "module", dd.sym_mapping[i].module);
> }
> + jsonw_uint_field(json_wtr, "cookie", data[j].cookie);
> jsonw_end_object(json_wtr);
> if (j++ == info->kprobe_multi.count)
> break;
> }
> jsonw_end_array(json_wtr);
> + free(data);
> }
>
> static __u64 *u64_to_arr(__u64 val)
> @@ -675,8 +706,8 @@ void netfilter_dump_plain(const struct bpf_link_info *info)
>
> static void show_kprobe_multi_plain(struct bpf_link_info *info)
> {
> + struct addr_cookie *data;
> __u32 i, j = 0;
> - __u64 *addrs;
>
> if (!info->kprobe_multi.count)
> return;
> @@ -688,8 +719,11 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> printf("func_cnt %u ", info->kprobe_multi.count);
> if (info->kprobe_multi.missed)
> printf("missed %llu ", info->kprobe_multi.missed);
> - addrs = (__u64 *)u64_to_ptr(info->kprobe_multi.addrs);
> - qsort(addrs, info->kprobe_multi.count, sizeof(__u64), cmp_u64);
> + data = get_addr_cookie_array(u64_to_ptr(info->kprobe_multi.addrs),
> + u64_to_ptr(info->kprobe_multi.cookies),
> + info->kprobe_multi.count);
> + if (!data)
> + return;
>
> /* Load it once for all. */
> if (!dd.sym_count)
> @@ -697,12 +731,12 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> if (!dd.sym_count)
> return;
Don't we need to free(data) if we return here?
>
> - printf("\n\t%-16s %s", "addr", "func [module]");
> + printf("\n\t%-16s %-16s %s", "addr", "cookie", "func [module]");
> for (i = 0; i < dd.sym_count; i++) {
> - if (dd.sym_mapping[i].address != addrs[j])
> + if (dd.sym_mapping[i].address != data[j].addr)
> continue;
> - printf("\n\t%016lx %s",
> - dd.sym_mapping[i].address, dd.sym_mapping[i].name);
> + printf("\n\t%016lx %-16llx %s",
> + dd.sym_mapping[i].address, data[j].cookie, dd.sym_mapping[i].name);
> if (dd.sym_mapping[i].module[0] != '\0')
> printf(" [%s] ", dd.sym_mapping[i].module);
> else
> @@ -711,6 +745,7 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> if (j++ == info->kprobe_multi.count)
> break;
> }
> + free(data);
> }
>
> static void show_uprobe_multi_plain(struct bpf_link_info *info)
> @@ -966,6 +1001,14 @@ static int do_show_link(int fd)
> return -ENOMEM;
> }
> info.kprobe_multi.addrs = ptr_to_u64(addrs);
> + cookies = calloc(count, sizeof(__u64));
> + if (!cookies) {
> + p_err("mem alloc failed");
> + free(addrs);
> + close(fd);
> + return -ENOMEM;
> + }
> + info.kprobe_multi.cookies = ptr_to_u64(cookies);
> goto again;
> }
> }
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link
2024-01-18 17:51 ` Quentin Monnet
@ 2024-01-19 8:03 ` Jiri Olsa
0 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-19 8:03 UTC (permalink / raw)
To: Quentin Monnet
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Yafang Shao
On Thu, Jan 18, 2024 at 05:51:17PM +0000, Quentin Monnet wrote:
SNIP
> > static __u64 *u64_to_arr(__u64 val)
> > @@ -675,8 +706,8 @@ void netfilter_dump_plain(const struct bpf_link_info *info)
> >
> > static void show_kprobe_multi_plain(struct bpf_link_info *info)
> > {
> > + struct addr_cookie *data;
> > __u32 i, j = 0;
> > - __u64 *addrs;
> >
> > if (!info->kprobe_multi.count)
> > return;
> > @@ -688,8 +719,11 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> > printf("func_cnt %u ", info->kprobe_multi.count);
> > if (info->kprobe_multi.missed)
> > printf("missed %llu ", info->kprobe_multi.missed);
> > - addrs = (__u64 *)u64_to_ptr(info->kprobe_multi.addrs);
> > - qsort(addrs, info->kprobe_multi.count, sizeof(__u64), cmp_u64);
> > + data = get_addr_cookie_array(u64_to_ptr(info->kprobe_multi.addrs),
> > + u64_to_ptr(info->kprobe_multi.cookies),
> > + info->kprobe_multi.count);
> > + if (!data)
> > + return;
> >
> > /* Load it once for all. */
> > if (!dd.sym_count)
> > @@ -697,12 +731,12 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> > if (!dd.sym_count)
> > return;
>
> Don't we need to free(data) if we return here?
good catch! I guess I got distracted by show_kprobe_multi_plain being
similar to show_kprobe_multi_json, which does not check dd.sym_count
and does not return, which it should :-\ I'll include that fix as well
thanks,
jirka
>
> >
> > - printf("\n\t%-16s %s", "addr", "func [module]");
> > + printf("\n\t%-16s %-16s %s", "addr", "cookie", "func [module]");
> > for (i = 0; i < dd.sym_count; i++) {
> > - if (dd.sym_mapping[i].address != addrs[j])
> > + if (dd.sym_mapping[i].address != data[j].addr)
> > continue;
> > - printf("\n\t%016lx %s",
> > - dd.sym_mapping[i].address, dd.sym_mapping[i].name);
> > + printf("\n\t%016lx %-16llx %s",
> > + dd.sym_mapping[i].address, data[j].cookie, dd.sym_mapping[i].name);
> > if (dd.sym_mapping[i].module[0] != '\0')
> > printf(" [%s] ", dd.sym_mapping[i].module);
> > else
> > @@ -711,6 +745,7 @@ static void show_kprobe_multi_plain(struct bpf_link_info *info)
> > if (j++ == info->kprobe_multi.count)
> > break;
> > }
> > + free(data);
> > }
> >
> > static void show_uprobe_multi_plain(struct bpf_link_info *info)
> > @@ -966,6 +1001,14 @@ static int do_show_link(int fd)
> > return -ENOMEM;
> > }
> > info.kprobe_multi.addrs = ptr_to_u64(addrs);
> > + cookies = calloc(count, sizeof(__u64));
> > + if (!cookies) {
> > + p_err("mem alloc failed");
> > + free(addrs);
> > + close(fd);
> > + return -ENOMEM;
> > + }
> > + info.kprobe_multi.cookies = ptr_to_u64(cookies);
> > goto again;
> > }
> > }
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records
2024-01-18 12:24 ` Yafang Shao
@ 2024-01-19 8:03 ` Jiri Olsa
0 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-19 8:03 UTC (permalink / raw)
To: Yafang Shao
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 08:24:58PM +0800, Yafang Shao wrote:
> On Thu, Jan 18, 2024 at 5:54 PM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > At the moment we don't store cookie for perf_event probes,
> > while we do that for the rest of the probes.
> >
> > Adding cookie fields to struct bpf_link_info perf event
> > probe records:
> >
> > perf_event.uprobe
> > perf_event.kprobe
> > perf_event.tracepoint
> > perf_event.perf_event
> >
> > And the code to store that in bpf_link_info struct.
> >
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> > include/uapi/linux/bpf.h | 4 ++++
> > kernel/bpf/syscall.c | 4 ++++
> > tools/include/uapi/linux/bpf.h | 4 ++++
> > 3 files changed, 12 insertions(+)
> >
> > diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> > index a00f8a5623e1..b823d367a83c 100644
> > --- a/include/uapi/linux/bpf.h
> > +++ b/include/uapi/linux/bpf.h
> > @@ -6582,6 +6582,7 @@ struct bpf_link_info {
> > __aligned_u64 file_name; /* in/out */
> > __u32 name_len;
> > __u32 offset; /* offset from file_name */
> > + __u64 cookie;
> > } uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
> > struct {
> > __aligned_u64 func_name; /* in/out */
> > @@ -6589,14 +6590,17 @@ struct bpf_link_info {
> > __u32 offset; /* offset from func_name */
> > __u64 addr;
> > __u64 missed;
> > + __u64 cookie;
> > } kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
> > struct {
> > __aligned_u64 tp_name; /* in/out */
> > __u32 name_len;
>
> It might be beneficial to include an alignment pad '__u32 :32;' here,
> following the pattern used in other instances within this file.
good cactch, thanks
jirka
>
> > + __u64 cookie;
> > } tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
> > struct {
> > __u64 config;
> > __u32 type;
>
> Same here.
>
> > + __u64 cookie;
> > } event; /* BPF_PERF_EVENT_EVENT */
> > };
> > } perf_event;
> > diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> > index a1f18681721c..13193aaafb64 100644
> > --- a/kernel/bpf/syscall.c
> > +++ b/kernel/bpf/syscall.c
> > @@ -3501,6 +3501,7 @@ static int bpf_perf_link_fill_kprobe(const struct perf_event *event,
> > if (!kallsyms_show_value(current_cred()))
> > addr = 0;
> > info->perf_event.kprobe.addr = addr;
> > + info->perf_event.kprobe.cookie = event->bpf_cookie;
> > return 0;
> > }
> > #endif
> > @@ -3526,6 +3527,7 @@ static int bpf_perf_link_fill_uprobe(const struct perf_event *event,
> > else
> > info->perf_event.type = BPF_PERF_EVENT_UPROBE;
> > info->perf_event.uprobe.offset = offset;
> > + info->perf_event.uprobe.cookie = event->bpf_cookie;
> > return 0;
> > }
> > #endif
> > @@ -3553,6 +3555,7 @@ static int bpf_perf_link_fill_tracepoint(const struct perf_event *event,
> > uname = u64_to_user_ptr(info->perf_event.tracepoint.tp_name);
> > ulen = info->perf_event.tracepoint.name_len;
> > info->perf_event.type = BPF_PERF_EVENT_TRACEPOINT;
> > + info->perf_event.tracepoint.cookie = event->bpf_cookie;
> > return bpf_perf_link_fill_common(event, uname, ulen, NULL, NULL, NULL, NULL);
> > }
> >
> > @@ -3561,6 +3564,7 @@ static int bpf_perf_link_fill_perf_event(const struct perf_event *event,
> > {
> > info->perf_event.event.type = event->attr.type;
> > info->perf_event.event.config = event->attr.config;
> > + info->perf_event.event.cookie = event->bpf_cookie;
> > info->perf_event.type = BPF_PERF_EVENT_EVENT;
> > return 0;
> > }
> > diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
> > index a00f8a5623e1..b823d367a83c 100644
> > --- a/tools/include/uapi/linux/bpf.h
> > +++ b/tools/include/uapi/linux/bpf.h
> > @@ -6582,6 +6582,7 @@ struct bpf_link_info {
> > __aligned_u64 file_name; /* in/out */
> > __u32 name_len;
> > __u32 offset; /* offset from file_name */
> > + __u64 cookie;
> > } uprobe; /* BPF_PERF_EVENT_UPROBE, BPF_PERF_EVENT_URETPROBE */
> > struct {
> > __aligned_u64 func_name; /* in/out */
> > @@ -6589,14 +6590,17 @@ struct bpf_link_info {
> > __u32 offset; /* offset from func_name */
> > __u64 addr;
> > __u64 missed;
> > + __u64 cookie;
> > } kprobe; /* BPF_PERF_EVENT_KPROBE, BPF_PERF_EVENT_KRETPROBE */
> > struct {
> > __aligned_u64 tp_name; /* in/out */
> > __u32 name_len;
> > + __u64 cookie;
> > } tracepoint; /* BPF_PERF_EVENT_TRACEPOINT */
> > struct {
> > __u64 config;
> > __u32 type;
> > + __u64 cookie;
> > } event; /* BPF_PERF_EVENT_EVENT */
> > };
> > } perf_event;
> > --
> > 2.43.0
> >
>
>
> --
> Regards
> Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links
2024-01-18 12:42 ` [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Yafang Shao
@ 2024-01-19 8:03 ` Jiri Olsa
0 siblings, 0 replies; 19+ messages in thread
From: Jiri Olsa @ 2024-01-19 8:03 UTC (permalink / raw)
To: Yafang Shao
Cc: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko, bpf,
Martin KaFai Lau, Song Liu, Yonghong Song, John Fastabend,
KP Singh, Stanislav Fomichev, Hao Luo, Quentin Monnet
On Thu, Jan 18, 2024 at 08:42:29PM +0800, Yafang Shao wrote:
> On Thu, Jan 18, 2024 at 5:54 PM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > hi,
> > this patchset adds support to retrieve cookies from existing tracing
> > links that still did not support it plus changes to bpftool to display
> > them. It's leftover we discussed some time ago [1].
>
> Thanks for your work.
>
> bpf cookie is also displayed in the pid_iter [0]. After we add support
> for the cookie for other progs like kprobe_multi and uprobe_multi, I
> think we should update this file as well.
ok, I'll check
thanks,
jirka
>
> [0]. tools/bpf/bpftool/skeleton/pid_iter.bpf.c
>
> >
> > thanks,
> > jirka
> >
> >
> > [1] https://lore.kernel.org/bpf/CALOAHbAZ6=A9j3VFCLoAC_WhgQKU7injMf06=cM2sU4Hi4Sx+Q@mail.gmail.com/
> > ---
> > Jiri Olsa (8):
> > bpf: Add cookie to perf_event bpf_link_info records
> > bpf: Store cookies in kprobe_multi bpf_link_info data
> > bpftool: Fix wrong free call in do_show_link
> > selftests/bpf: Add cookies check for kprobe_multi fill_link_info test
> > selftests/bpf: Add cookies check for perf_event fill_link_info test
> > selftests/bpf: Add fill_link_info test for perf event
> > bpftool: Display cookie for perf event link probes
> > bpftool: Display cookie for kprobe multi link
> >
> > include/uapi/linux/bpf.h | 5 +++++
> > kernel/bpf/syscall.c | 4 ++++
> > kernel/trace/bpf_trace.c | 15 +++++++++++++
> > tools/bpf/bpftool/link.c | 87 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------
> > tools/include/uapi/linux/bpf.h | 5 +++++
> > tools/testing/selftests/bpf/prog_tests/fill_link_info.c | 114 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------
> > tools/testing/selftests/bpf/progs/test_fill_link_info.c | 6 +++++
> > 7 files changed, 204 insertions(+), 32 deletions(-)
>
>
>
> --
> Regards
> Yafang
^ permalink raw reply [flat|nested] 19+ messages in thread
end of thread, other threads:[~2024-01-19 8:03 UTC | newest]
Thread overview: 19+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-01-18 9:54 [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 1/8] bpf: Add cookie to perf_event bpf_link_info records Jiri Olsa
2024-01-18 12:24 ` Yafang Shao
2024-01-19 8:03 ` Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 2/8] bpf: Store cookies in kprobe_multi bpf_link_info data Jiri Olsa
2024-01-18 13:21 ` Yafang Shao
2024-01-18 9:54 ` [PATCH bpf-next 3/8] bpftool: Fix wrong free call in do_show_link Jiri Olsa
2024-01-18 12:29 ` Yafang Shao
2024-01-18 17:51 ` Quentin Monnet
2024-01-18 9:54 ` [PATCH bpf-next 4/8] selftests/bpf: Add cookies check for kprobe_multi fill_link_info test Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 5/8] selftests/bpf: Add cookies check for perf_event " Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 6/8] selftests/bpf: Add fill_link_info test for perf event Jiri Olsa
2024-01-18 9:54 ` [PATCH bpf-next 7/8] bpftool: Display cookie for perf event link probes Jiri Olsa
2024-01-18 17:51 ` Quentin Monnet
2024-01-18 9:54 ` [PATCH bpf-next 8/8] bpftool: Display cookie for kprobe multi link Jiri Olsa
2024-01-18 17:51 ` Quentin Monnet
2024-01-19 8:03 ` Jiri Olsa
2024-01-18 12:42 ` [PATCH bpf-next 0/8] bpf: Add cookies retrieval for perf/kprobe multi links Yafang Shao
2024-01-19 8:03 ` Jiri Olsa
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).