perf/core: Take over CAP_SYS_PTRACE creds to CAP_PERFMON capability

Open access to per-process monitoring for CAP_PERFMON only
privileged processes [1]. Extend ptrace_may_access() check
in perf_events subsystem with perfmon_capable() to simplify
user experience and make monitoring more secure by reducing
attack surface.

[1] https://lore.kernel.org/lkml/7776fa40-6c65-2aa6-1322-eb3a01201000@linux.intel.com/

Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/6e8392ff-4732-0012-2949-e1587709f0f6@linux.intel.com

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 78e69e1..41e0cef 100644 (file)
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -11689,7 +11689,7 @@ SYSCALL_DEFINE5(perf_event_open,
                        goto err_task;
 
                /*
-                * Reuse ptrace permission checks for now.
+                * Preserve ptrace permission check for backwards compatibility.
                 *
                 * We must hold exec_update_mutex across this and any potential
                 * perf_install_in_context() call for this new event to
@@ -11697,7 +11697,7 @@ SYSCALL_DEFINE5(perf_event_open,
                 * perf_event_exit_task() that could imply).
                 */
                err = -EACCES;
-               if (!ptrace_may_access(task, PTRACE_MODE_READ_REALCREDS))
+               if (!perfmon_capable() && !ptrace_may_access(task, PTRACE_MODE_READ_REALCREDS))
                        goto err_cred;
        }