Merge tag 'perf-core-2020-06-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf updates from Ingo Molnar:
 "Kernel side changes:

   - Add AMD Fam17h RAPL support

   - Introduce CAP_PERFMON to kernel and user space

   - Add Zhaoxin CPU support

   - Misc fixes and cleanups

  Tooling changes:

   - perf record:

     Introduce '--switch-output-event' to use arbitrary events to be
     setup and read from a side band thread and, when they take place a
     signal be sent to the main 'perf record' thread, reusing the core
     for '--switch-output' to take perf.data snapshots from the ring
     buffer used for '--overwrite', e.g.:

	# perf record --overwrite -e sched:* \
		      --switch-output-event syscalls:*connect* \
		      workload

     will take perf.data.YYYYMMDDHHMMSS snapshots up to around the
     connect syscalls.

     Add '--num-synthesize-threads' option to control degree of
     parallelism of the synthesize_mmap() code which is scanning
     /proc/PID/task/PID/maps and can be time consuming. This mimics
     pre-existing behaviour in 'perf top'.

   - perf bench:

     Add a multi-threaded synthesize benchmark and kallsyms parsing
     benchmark.

   - Intel PT support:

     Stitch LBR records from multiple samples to get deeper backtraces,
     there are caveats, see the csets for details.

     Allow using Intel PT to synthesize callchains for regular events.

     Add support for synthesizing branch stacks for regular events
     (cycles, instructions, etc) from Intel PT data.

  Misc changes:

   - Updated perf vendor events for power9 and Coresight.

   - Add flamegraph.py script via 'perf flamegraph'

   - Misc other changes, fixes and cleanups - see the Git log for details

  Also, since over the last couple of years perf tooling has matured and
  decoupled from the kernel perf changes to a large degree, going
  forward Arnaldo is going to send perf tooling changes via direct pull
  requests"

* tag 'perf-core-2020-06-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (163 commits)
  perf/x86/rapl: Add AMD Fam17h RAPL support
  perf/x86/rapl: Make perf_probe_msr() more robust and flexible
  perf/x86/rapl: Flip logic on default events visibility
  perf/x86/rapl: Refactor to share the RAPL code between Intel and AMD CPUs
  perf/x86/rapl: Move RAPL support to common x86 code
  perf/core: Replace zero-length array with flexible-array
  perf/x86: Replace zero-length array with flexible-array
  perf/x86/intel: Add more available bits for OFFCORE_RESPONSE of Intel Tremont
  perf/x86/rapl: Add Ice Lake RAPL support
  perf flamegraph: Use /bin/bash for report and record scripts
  perf cs-etm: Move definition of 'traceid_list' global variable from header file
  libsymbols kallsyms: Move hex2u64 out of header
  libsymbols kallsyms: Parse using io api
  perf bench: Add kallsyms parsing
  perf: cs-etm: Update to build with latest opencsd version.
  perf symbol: Fix kernel symbol address display
  perf inject: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
  perf annotate: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
  perf trace: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
  perf script: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
  ...

4 files changed, 20 insertions, 15 deletions

diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c
index c2b41a263166..b1991043b7d8 100644
--- a/kernel/events/callchain.c
+++ b/kernel/events/callchain.c
@@ -16,7 +16,7 @@
 
 struct callchain_cpus_entries {
 	struct rcu_head			rcu_head;
-	struct perf_callchain_entry	*cpu_entries[0];
+	struct perf_callchain_entry	*cpu_entries[];
 };
 
 int sysctl_perf_event_max_stack __read_mostly = PERF_MAX_STACK_DEPTH;
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 633b4ae72ed5..e296c5c59c6f 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -95,11 +95,11 @@ static void remote_function(void *data)
  * @info:	the function call argument
  *
  * Calls the function @func when the task is currently running. This might
- * be on the current CPU, which just calls the function directly
+ * be on the current CPU, which just calls the function directly.  This will
+ * retry due to any failures in smp_call_function_single(), such as if the
+ * task_cpu() goes offline concurrently.
  *
- * returns: @func return value, or
- *	    -ESRCH  - when the process isn't running
- *	    -EAGAIN - when the process moved away
+ * returns @func return value or -ESRCH when the process isn't running
  */
 static int
 task_function_call(struct task_struct *p, remote_function_f func, void *info)
@@ -112,11 +112,16 @@ task_function_call(struct task_struct *p, remote_function_f func, void *info)
 	};
 	int ret;
 
-	do {
-		ret = smp_call_function_single(task_cpu(p), remote_function, &data, 1);
-		if (!ret)
-			ret = data.ret;
-	} while (ret == -EAGAIN);
+	for (;;) {
+		ret = smp_call_function_single(task_cpu(p), remote_function,
+					       &data, 1);
+		ret = !ret ? data.ret : -EAGAIN;
+
+		if (ret != -EAGAIN)
+			break;
+
+		cond_resched();
+	}
 
 	return ret;
 }
@@ -9404,7 +9409,7 @@ static int perf_kprobe_event_init(struct perf_event *event)
 	if (event->attr.type != perf_kprobe.type)
 		return -ENOENT;
 
-	if (!capable(CAP_SYS_ADMIN))
+	if (!perfmon_capable())
 		return -EACCES;
 
 	/*
@@ -9464,7 +9469,7 @@ static int perf_uprobe_event_init(struct perf_event *event)
 	if (event->attr.type != perf_uprobe.type)
 		return -ENOENT;
 
-	if (!capable(CAP_SYS_ADMIN))
+	if (!perfmon_capable())
 		return -EACCES;
 
 	/*
@@ -11511,7 +11516,7 @@ SYSCALL_DEFINE5(perf_event_open,
 	}
 
 	if (attr.namespaces) {
-		if (!capable(CAP_SYS_ADMIN))
+		if (!perfmon_capable())
 			return -EACCES;
 	}
 
diff --git a/kernel/events/internal.h b/kernel/events/internal.h
index f16f66b6b655..fcbf5616a441 100644
--- a/kernel/events/internal.h
+++ b/kernel/events/internal.h
@@ -55,7 +55,7 @@ struct perf_buffer {
 	void				*aux_priv;
 
 	struct perf_event_mmap_page	*user_page;
-	void				*data_pages[0];
+	void				*data_pages[];
 };
 
 extern void rb_free(struct perf_buffer *rb);
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index a010edc37ee0..92ba69b716dc 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -1500,7 +1500,7 @@ int perf_event_query_prog_array(struct perf_event *event, void __user *info)
 	u32 *ids, prog_cnt, ids_len;
 	int ret;
 
-	if (!capable(CAP_SYS_ADMIN))
+	if (!perfmon_capable())
 		return -EPERM;
 	if (event->attr.type != PERF_TYPE_TRACEPOINT)
 		return -EINVAL;