diff options
| author | Masami Hiramatsu <mhiramat@kernel.org> | 2019-05-15 14:39:05 +0900 |
|---|---|---|
| committer | Steven Rostedt (VMware) <rostedt@goodmis.org> | 2019-05-25 23:04:42 -0400 |
| commit | 1e032f7cfa141b4424827b0ecb0ea899f84e182e (patch) | |
| tree | 929053b3dc3a806529788e54d9700f556b48911b /tools/perf/util/probe-file.h | |
| parent | bdf2b8cbf076568f7d17dc62467348d409142298 (diff) | |
perf-probe: Add user memory access attribute support
Add user memory access attribute for kprobe event arguments.
If a given 'local variable' is in user-space, User can
specify memory access method by '@user' suffix. This is
not only for string but also for data structure.
If we access a field of data structure in user memory from
kernel on some arch, it will fail. e.g.
perf probe -a "sched_setscheduler param->sched_priority"
This will fail to access the "param->sched_priority" because
the param is __user pointer. Instead, we can now specify
@user suffix for such argument.
perf probe -a "sched_setscheduler param->sched_priority@user"
Note that kernel memory access with "@user" must always fail
on any arch.
Link: http://lkml.kernel.org/r/155789874562.26965.10836126971405890891.stgit@devnote2
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Diffstat (limited to 'tools/perf/util/probe-file.h')
| -rw-r--r-- | tools/perf/util/probe-file.h | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/tools/perf/util/probe-file.h b/tools/perf/util/probe-file.h index 2a249182f2a6..986c1c94f64f 100644 --- a/tools/perf/util/probe-file.h +++ b/tools/perf/util/probe-file.h @@ -70,6 +70,7 @@ int probe_cache__show_all_caches(struct strfilter *filter); bool probe_type_is_available(enum probe_type type); bool kretprobe_offset_is_supported(void); bool uprobe_ref_ctr_is_supported(void); +bool user_access_is_supported(void); #else /* ! HAVE_LIBELF_SUPPORT */ static inline struct probe_cache *probe_cache__new(const char *tgt __maybe_unused, struct nsinfo *nsi __maybe_unused) { |