cleanup the legacy probe_event on failed scenario
Subsystems (4)
v4 cover cleanup the legacy probe_event on failed scenario
A potential scenario, when an error is returned after add_uprobe_event_legacy() in perf_event_uprobe_open_legacy(), or bpf_program__attach_perf_event_opts() in bpf_program__attach_uprobe_opts() returns an error, the uprobe_event that was previously created is not cleaned.
At the same time, the legacy kprobe_event also have similar problems.
With these patches, whenever an error is returned, it ensures that the created kprobe_event/uprobe_event is cleaned.
V1 -> v3:
- add detail commits
- call remove_kprobe_event_legacy() on failed bpf_program__attach_perf_event_opts()
v3 -> v4:
- cleanup the legacy kprobe_event on failed add/attach_event
Chuang Wang (3):
libbpf: cleanup the legacy kprobe_event on failed add/attach_event()
libbpf: fix wrong variable used in perf_event_uprobe_open_legacy()
libbpf: cleanup the legacy uprobe_event on failed add/attach_event()
tools/lib/bpf/libbpf.c | 37 +++++++++++++++++++++++++++++-------- 1 file changed, 29 insertions(+), 8 deletions(-)
Signature · 1 line
2.34.1
PATCH v4 1/3 libbpf: cleanup the legacy kprobe_event on failed add/attach_event()
Before the 0bc11ed5ab60 commit ("kprobes: Allow kprobes coexist with livepatch"), in a scenario where livepatch and kprobe coexist on the same function entry, the creation of kprobe_event using add_kprobe_event_legacy() will be successful, at the same time as a trace event (e.g. /debugfs/tracing/events/kprobe/XXX) will exist, but perf_event_open() will return an error because both livepatch and kprobe use FTRACE_OPS_FL_IPMODIFY. As follows:
1) add a livepatch
$ insmod livepatch-XXX.ko
2) add a kprobe using tracefs API (i.e. add_kprobe_event_legacy)
$ echo 'p:mykprobe XXX' > /sys/kernel/debug/tracing/kprobe_events
3) enable this kprobe (i.e. sys_perf_event_open)
This will return an error, -EBUSY.
On Andrii Nakryiko's comment, few error paths in bpf_program__attach_kprobe_opts() that should need to call remove_kprobe_event_legacy().
With this patch, whenever an error is returned after add_kprobe_event_legacy() or bpf_program__attach_perf_event_opts(), this ensures that the created kprobe_event is cleaned.
--- tools/lib/bpf/libbpf.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-)
tools/lib/bpf/libbpf.c +15 -4
@@ -9868,10 +9868,11 @@ static int perf_event_kprobe_open_legacy(const char *probe_name, bool retprobe, } type =determine_kprobe_perf_type_legacy (probe_name, retprobe);if (type <0 ) { + err = type;pr_warn ("failed to determine legacy kprobe event id for '%s+0x%zx': %s\n" , kfunc_name, offset, -libbpf_strerror_r (type, errmsg,sizeof (errmsg))); -return type; +libbpf_strerror_r (err, errmsg,sizeof (errmsg))); + goto err_clean_legacy; } attr.size =sizeof (attr); attr.config = type;
@@ -9885,9 +9886,14 @@ static int perf_event_kprobe_open_legacy(const char *probe_name, bool retprobe, err = -errno;pr_warn ("legacy kprobe perf_event_open() failed: %s\n" ,libbpf_strerror_r (err, errmsg,sizeof (errmsg))); -return err; + goto err_clean_legacy; }return pfd; + +err_clean_legacy: +/* Clear the newly added legacy kprobe_event */ +remove_kprobe_event_legacy (probe_name, retprobe); +return err; } struct bpf_link *
@@ -9944,7 +9950,7 @@ bpf_program__attach_kprobe_opts(const struct bpf_program *prog, prog->name, retprobe ?"kretprobe" :"kprobe" , func_name, offset, libbpf_strerror_r(err, errmsg, sizeof(errmsg))); - goto err_out; + goto err_clean_legacy; }if (legacy) { struct bpf_link_perf *perf_link = container_of(link, struct bpf_link_perf, link);
@@ -9955,6 +9961,11 @@ bpf_program__attach_kprobe_opts(const struct bpf_program *prog, }return link; + +err_clean_legacy: +if (legacy) +remove_kprobe_event_legacy (legacy_probe, retprobe); + err_out:free (legacy_probe);return libbpf_err_ptr (err);
Signature · 1 line
2.34.1
PATCH v4 2/3 libbpf: fix wrong variable used in perf_event_uprobe_open_legacy()
In reply to Chuang WangUse "type" as opposed to "err" in pr_warn() after determine_uprobe_perf_type_legacy() returns an error.
--- tools/lib/bpf/libbpf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/lib/bpf/libbpf.c +1 -1
@@ -10241,7 +10241,7 @@ static int perf_event_uprobe_open_legacy(const char *probe_name, bool retprobe, type =determine_uprobe_perf_type_legacy (probe_name, retprobe);if (type <0 ) {pr_warn ("failed to determine legacy uprobe event id for %s:0x%zx: %d\n" , - binary_path, offset, err); + binary_path, offset, type);return type; }
Signature · 1 line
2.34.1
PATCH v4 3/3 libbpf: cleanup the legacy uprobe_event on failed add/attach_event()
In reply to Chuang WangA potential scenario, when an error is returned after add_uprobe_event_legacy() in perf_event_uprobe_open_legacy(), or bpf_program__attach_perf_event_opts() in bpf_program__attach_uprobe_opts() returns an error, the uprobe_event that was previously created is not cleaned.
So, with this patch, when an error is returned, fix this by adding remove_uprobe_event_legacy()
--- tools/lib/bpf/libbpf.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-)
tools/lib/bpf/libbpf.c +15 -5
@@ -10240,9 +10240,10 @@ static int perf_event_uprobe_open_legacy(const char *probe_name, bool retprobe, } type =determine_uprobe_perf_type_legacy (probe_name, retprobe);if (type <0 ) { + err = type;pr_warn ("failed to determine legacy uprobe event id for %s:0x%zx: %d\n" , - binary_path, offset, type); -return type; + binary_path, offset, err); + goto err_clean_legacy; }memset (&attr,0 ,sizeof (attr));
@@ -10257,9 +10258,14 @@ static int perf_event_uprobe_open_legacy(const char *probe_name, bool retprobe,if (pfd <0 ) { err = -errno;pr_warn ("legacy uprobe perf_event_open() failed: %d\n" , err); -return err; + goto err_clean_legacy; }return pfd; + +err_clean_legacy: +/* Clear the newly added legacy uprobe_event */ +remove_uprobe_event_legacy (probe_name, retprobe); +return err; }/* Return next ELF section of sh_type after scn, or first of that type if scn is NULL. */
@@ -10593,7 +10599,7 @@ bpf_program__attach_uprobe_opts(const struct bpf_program *prog, pid_t pid, prog->name, retprobe ?"uretprobe" :"uprobe" , binary_path, func_offset, libbpf_strerror_r(err, errmsg, sizeof(errmsg))); - goto err_out; + goto err_clean_legacy; }if (legacy) { struct bpf_link_perf *perf_link = container_of(link, struct bpf_link_perf, link);
@@ -10603,10 +10609,14 @@ bpf_program__attach_uprobe_opts(const struct bpf_program *prog, pid_t pid, perf_link->legacy_is_retprobe = retprobe; }return link; + +err_clean_legacy: +if (legacy) +remove_uprobe_event_legacy (legacy_probe, retprobe); + err_out:free (legacy_probe);return libbpf_err_ptr (err); - }/* Format of u[ret]probe section definition supporting auto-attach:
Signature · 1 line
2.34.1
This series was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:
Hello:
This series was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:
Quoted 9 lines · On Wed, 29 Jun 2022 23:18:44 +0800 you wrote:
A potential scenario, when an error is returned after
add_uprobe_event_legacy()inperf_event_uprobe_open_legacy(), orbpf_program__attach_perf_event_opts()inbpf_program__attach_uprobe_opts()returns an error, theuprobe_eventthat was previously created is not cleaned.At the same time, the legacy
kprobe_eventalso have similar problems.[...]
Here is the summary with links:
- [v4,1/3] libbpf: cleanup the legacy kprobe_event on failed add/attach_event()
https://git.kernel.org/bpf/bpf-next/c/8094029330a2 - [v4,2/3] libbpf: fix wrong variable used in perf_event_uprobe_open_legacy() https://git.kernel.org/bpf/bpf-next/c/5666fc997ccb - [v4,3/3] libbpf: cleanup the legacy uprobe_event on failed add/attach_event() https://git.kernel.org/bpf/bpf-next/c/2655144fb49b
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html