On Sat, Jul 31, 2021 at 10:11 PM Rafael David Tinoco
<rafaeldtinoco@xxxxxxxxx> wrote:
>
> Allow kprobe tracepoint events creation through legacy interface, as the
> kprobe dynamic PMUs support, used by default, was only created in v4.17.
>
> After commit "bpf: implement minimal BPF perf link", it was allowed that
> some extra - to the link - information is accessed through container_of
> struct bpf_link. This allows the tracing perf event legacy name, and
> information whether it is a retprobe, to be saved outside bpf_link
> structure, which would not be optimal.
>
> This enables CO.RE support for older kernels.
nit: it's CO-RE (or Compile Once - Run Everywhere), let's keep
consistent spelling.
>
> Cc: Andrii Nakryiko <andrii.nakryiko@xxxxxxxxx>
> Signed-off-by: Rafael David Tinoco <rafaeldtinoco@xxxxxxxxx>
> ---
Looks good overall, modulo various nits and skipped error handling.
Please wait for bpf_perf_link changes to get applied and submit the
next revision based on it. Thanks!
> tools/lib/bpf/libbpf.c | 127 ++++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 125 insertions(+), 2 deletions(-)
>
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index e1b7b2b6618c..40037340a3e7 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -8985,9 +8985,55 @@ int bpf_link__unpin(struct bpf_link *link)
> return 0;
> }
>
> +static int poke_kprobe_events(bool add, const char *name, bool retprobe, uint64_t offset) {
'{' should be on separate line
Please use scripts/checkpatch.pl on your changes, there are a bunch of
stylistic problems in your code, which you would have caught yourself
if you ran this script.
> + int fd, ret = 0;
> + pid_t p = getpid();
> + char cmd[192] = {}, probename[128] = {}, probefunc[128] = {};
> + const char *file = "/sys/kernel/debug/tracing/kprobe_events";
> +
> + if (retprobe)
> + snprintf(probename, sizeof(probename), "kretprobes/%s_libbpf_%u", name, p);
> + else
> + snprintf(probename, sizeof(probename), "kprobes/%s_libbpf_%u", name, p);
> +
> + if (offset)
> + snprintf(probefunc, sizeof(probefunc), "%s+%lu", name, offset);
> +
> + if (add) {
> + snprintf(cmd, sizeof(cmd), "%c:%s %s",
> + retprobe ? 'r' : 'p',
> + probename,
> + offset ? probefunc : name);
> + } else {
> + snprintf(cmd, sizeof(cmd), "-:%s", probename);
> + }
> +
> + fd = open(file, O_WRONLY | O_APPEND, 0);
> + if (!fd)
> + return -errno;
> + ret = write(fd, cmd, strlen(cmd));
> + if (ret < 0)
> + ret = -errno;
> + close(fd);
> +
> + return ret;
> +}
> +
> +static inline int add_kprobe_event_legacy(const char *name, bool retprobe, uint64_t offset)
> +{
> + return poke_kprobe_events(true, name, retprobe, offset);
> +}
> +
> +static inline int remove_kprobe_event_legacy(const char *name, bool retprobe)
> +{
> + return poke_kprobe_events(false, name, retprobe, 0);
> +}
> +
> struct bpf_link_perf {
> struct bpf_link link;
> int perf_event_fd;
> + char *legacy_name;
My initial reaction was to have a still different struct specifically
for legacy kprobe (e.g., struct bpf_link_kprobe_legacy or something
like that). But we'll need to do similar changes to uprobes as well,
so now I think it's ok to have this as an optional extra field on
perf_event-based link. If that causes problems we can always change
that later.
To make it clearer, can you name it "legacy_probe_name" and leave a
short comment mentioning that this is used to remember original
identifier we used to create legacy kprobe?
> + bool is_retprobe;
> };
>
> static int bpf_link_perf_detach(struct bpf_link *link)
> @@ -9002,6 +9048,10 @@ static int bpf_link_perf_detach(struct bpf_link *link)
> close(perf_link->perf_event_fd);
> close(link->fd);
>
> + /* legacy kprobe needs to be removed after perf event fd closure */
> + if (perf_link->legacy_name)
> + remove_kprobe_event_legacy(perf_link->legacy_name, perf_link->is_retprobe);
check and propagate error?
> +
> return libbpf_err(err);
> }
>
> @@ -9009,6 +9059,9 @@ static void bpf_link_perf_dealloc(struct bpf_link *link)
> {
> struct bpf_link_perf *perf_link = container_of(link, struct bpf_link_perf, link);
>
> + if (perf_link->legacy_name)
> + free(perf_link->legacy_name);
free() handles NULL properly, no need for if
> +
> free(perf_link);
> }
>
> @@ -9122,6 +9175,26 @@ static int parse_uint_from_file(const char *file, const char *fmt)
> return ret;
> }
>
> +static bool determine_kprobe_legacy(void)
> +{
> + const char *file = "/sys/bus/event_source/devices/kprobe/type";
> +
> + return access(file, 0) == 0 ? false : true;
> +}
> +
> +static int determine_kprobe_perf_type_legacy(const char *func_name, bool is_retprobe)
> +{
> + char file[192];
> +
extra empty line
> + const char *fname = "/sys/kernel/debug/tracing/events/%s/%s_libbpf_%d/id";
> +
> + snprintf(file, sizeof(file), fname,
> + is_retprobe ? "kretprobes" : "kprobes",
> + func_name, getpid());
> +
> + return parse_uint_from_file(file, "%d\n");
> +}
> +
> static int determine_kprobe_perf_type(void)
> {
> const char *file = "/sys/bus/event_source/devices/kprobe/type";
> @@ -9197,6 +9270,41 @@ static int perf_event_open_probe(bool uprobe, bool retprobe, const char *name,
> return pfd;
> }
>
> +static int perf_event_kprobe_open_legacy(bool retprobe, const char *name, uint64_t offset, int pid)
> +{
> + struct perf_event_attr attr = {};
> + char errmsg[STRERR_BUFSIZE];
> + int type, pfd, err;
> +
> + err = add_kprobe_event_legacy(name, retprobe, offset);
> + if (err < 0) {
> + pr_warn("failed to add legacy kprobe event: %s\n",
> + libbpf_strerror_r(err, errmsg, sizeof(errmsg)));
> + return err;
> + }
> + type = determine_kprobe_perf_type_legacy(name, retprobe);
> + if (type < 0) {
> + pr_warn("failed to determine legacy kprobe event id: %s\n",
> + libbpf_strerror_r(type, errmsg, sizeof(errmsg)));
> + return type;
> + }
> + attr.size = sizeof(attr);
> + attr.config = type;
> + attr.type = PERF_TYPE_TRACEPOINT;
> +
> + pfd = syscall(__NR_perf_event_open, &attr,
> + pid < 0 ? -1 : pid, /* pid */
> + pid == -1 ? 0 : -1, /* cpu */
> + -1 /* group_fd */, PERF_FLAG_FD_CLOEXEC);
> + if (pfd < 0) {
> + err = -errno;
> + pr_warn("legacy kprobe perf_event_open() failed: %s\n",
> + libbpf_strerror_r(err, errmsg, sizeof(errmsg)));
> + return err;
> + }
> + return pfd;
> +}
> +
> struct bpf_link *
> bpf_program__attach_kprobe_opts(struct bpf_program *prog,
> const char *func_name,
> @@ -9208,6 +9316,7 @@ bpf_program__attach_kprobe_opts(struct bpf_program *prog,
> unsigned long offset;
> bool retprobe;
> int pfd, err;
> + bool legacy;
>
> if (!OPTS_VALID(opts, bpf_kprobe_opts))
> return libbpf_err_ptr(-EINVAL);
> @@ -9216,8 +9325,16 @@ bpf_program__attach_kprobe_opts(struct bpf_program *prog,
> offset = OPTS_GET(opts, offset, 0);
> pe_opts.user_ctx = OPTS_GET(opts, user_ctx, 0);
>
> - pfd = perf_event_open_probe(false /* uprobe */, retprobe, func_name,
> - offset, -1 /* pid */);
> + legacy = determine_kprobe_legacy();
> + if (!legacy) {
> + pfd = perf_event_open_probe(false /* uprobe */,
> + retprobe, func_name,
> + offset, -1 /* pid */);
> + } else {
> + pfd = perf_event_kprobe_open_legacy(retprobe, func_name,
> + 0 /* offset */,
> + -1 /* pid */);
> + }
> if (pfd < 0) {
> pr_warn("prog '%s': failed to create %s '%s' perf event: %s\n",
> prog->name, retprobe ? "kretprobe" : "kprobe", func_name,
> @@ -9233,6 +9350,12 @@ bpf_program__attach_kprobe_opts(struct bpf_program *prog,
> libbpf_strerror_r(err, errmsg, sizeof(errmsg)));
> return libbpf_err_ptr(err);
> }
> + if (legacy) {
> + struct bpf_link_perf *perf_link = container_of(link, struct bpf_link_perf, link);
empty line between variable declaration and statements
> + perf_link->legacy_name = strdup(func_name);
check NULL, could be ENOMEM. Probably easier to allocate this at the
beginning, before we do heavy-lifting of perf_event_open(), clean up
will be simpler.
> + perf_link->is_retprobe = retprobe;
> + }
> +
> return link;
> }
>
> --
> 2.30.2
>
