Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf tooling updates form Ingo Molnar:
 "A final batch of perf tooling changes: mostly fixes and small
  improvements"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (29 commits)
  perf session: Add comment for perf_session__register_idle_thread()
  perf thread-stack: Fix thread stack processing for the idle task
  perf thread-stack: Allocate an array of thread stacks
  perf thread-stack: Factor out thread_stack__init()
  perf thread-stack: Allow for a thread stack array
  perf thread-stack: Avoid direct reference to the thread's stack
  perf thread-stack: Tidy thread_stack__bottom() usage
  perf thread-stack: Simplify some code in thread_stack__process()
  tools gpio: Allow overriding CFLAGS
  tools power turbostat: Override CFLAGS assignments and add LDFLAGS to build command
  tools thermal tmon: Allow overriding CFLAGS assignments
  tools power x86_energy_perf_policy: Override CFLAGS assignments and add LDFLAGS to build command
  perf c2c: Increase the HITM ratio limit for displayed cachelines
  perf c2c: Change the default coalesce setup
  perf trace beauty ioctl: Beautify USBDEVFS_ commands
  perf trace beauty: Export function to get the files for a thread
  perf trace: Wire up ioctl's USBDEBFS_ cmd table generator
  perf beauty ioctl: Add generator for USBDEVFS_ ioctl commands
  tools headers uapi: Grab a copy of usbdevice_fs.h
  perf trace: Store the major number for a file when storing its pathname
  ...

tools/perf/util/dump-insn.c

@@ -13,3 +13,11 @@ const char *dump_insn(struct perf_insn *x __maybe_unused,
 		*lenp = 0;
 	return "?";
 }
+
+__weak
+int arch_is_branch(const unsigned char *buf __maybe_unused,
+		   size_t len __maybe_unused,
+		   int x86_64 __maybe_unused)
+{
+	return 0;
+}

tools/perf/util/dump-insn.h

@@ -20,4 +20,6 @@ struct perf_insn {
 
 const char *dump_insn(struct perf_insn *x, u64 ip,
 		      u8 *inbuf, int inlen, int *lenp);
+int arch_is_branch(const unsigned char *buf, size_t len, int x86_64);
+
 #endif

tools/perf/util/intel-bts.c

@@ -451,7 +451,7 @@ static int intel_bts_process_buffer(struct intel_bts_queue *btsq,
 			continue;
 		intel_bts_get_branch_type(btsq, branch);
 		if (btsq->bts->synth_opts.thread_stack)
-			thread_stack__event(thread, btsq->sample_flags,
+			thread_stack__event(thread, btsq->cpu, btsq->sample_flags,
 					    le64_to_cpu(branch->from),
 					    le64_to_cpu(branch->to),
 					    btsq->intel_pt_insn.length,
@@ -523,7 +523,7 @@ static int intel_bts_process_queue(struct intel_bts_queue *btsq, u64 *timestamp)
 	    !btsq->bts->synth_opts.thread_stack && thread &&
 	    (!old_buffer || btsq->bts->sampling_mode ||
 	     (btsq->bts->snapshot_mode && !buffer->consecutive)))
-		thread_stack__set_trace_nr(thread, buffer->buffer_nr + 1);
+		thread_stack__set_trace_nr(thread, btsq->cpu, buffer->buffer_nr + 1);
 
 	err = intel_bts_process_buffer(btsq, buffer, thread);
 

tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c

@@ -180,6 +180,14 @@ int intel_pt_get_insn(const unsigned char *buf, size_t len, int x86_64,
 	return 0;
 }
 
+int arch_is_branch(const unsigned char *buf, size_t len, int x86_64)
+{
+	struct intel_pt_insn in;
+	if (intel_pt_get_insn(buf, len, x86_64, &in) < 0)
+		return -1;
+	return in.branch != INTEL_PT_BR_NO_BRANCH;
+}
+
 const char *dump_insn(struct perf_insn *x, uint64_t ip __maybe_unused,
 		      u8 *inbuf, int inlen, int *lenp)
 {

tools/perf/util/intel-pt.c

@@ -1174,7 +1174,7 @@ static void intel_pt_prep_sample(struct intel_pt *pt,
 	intel_pt_prep_b_sample(pt, ptq, event, sample);
 
 	if (pt->synth_opts.callchain) {
-		thread_stack__sample(ptq->thread, ptq->chain,
+		thread_stack__sample(ptq->thread, ptq->cpu, ptq->chain,
 				     pt->synth_opts.callchain_sz + 1,
 				     sample->ip, pt->kernel_start);
 		sample->callchain = ptq->chain;
@@ -1526,11 +1526,11 @@ static int intel_pt_sample(struct intel_pt_queue *ptq)
 		return 0;
 
 	if (pt->synth_opts.callchain || pt->synth_opts.thread_stack)
-		thread_stack__event(ptq->thread, ptq->flags, state->from_ip,
+		thread_stack__event(ptq->thread, ptq->cpu, ptq->flags, state->from_ip,
 				    state->to_ip, ptq->insn_len,
 				    state->trace_nr);
 	else
-		thread_stack__set_trace_nr(ptq->thread, state->trace_nr);
+		thread_stack__set_trace_nr(ptq->thread, ptq->cpu, state->trace_nr);
 
 	if (pt->sample_branches) {
 		err = intel_pt_synth_branch_sample(ptq);

tools/perf/util/python.c

@@ -939,7 +939,8 @@ static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist,
 
 		file = PyFile_FromFile(fp, "perf", "r", NULL);
 #else
-		file = PyFile_FromFd(evlist->pollfd.entries[i].fd, "perf", "r", -1, NULL, NULL, NULL, 1);
+		file = PyFile_FromFd(evlist->pollfd.entries[i].fd, "perf", "r", -1,
+				     NULL, NULL, NULL, 0);
 #endif
 		if (file == NULL)
 			goto free_list;

tools/perf/util/session.c

@@ -1527,6 +1527,13 @@ struct thread *perf_session__findnew(struct perf_session *session, pid_t pid)
 	return machine__findnew_thread(&session->machines.host, -1, pid);
 }
 
+/*
+ * Threads are identified by pid and tid, and the idle task has pid == tid == 0.
+ * So here a single thread is created for that, but actually there is a separate
+ * idle task per cpu, so there should be one 'struct thread' per cpu, but there
+ * is only 1. That causes problems for some tools, requiring workarounds. For
+ * example get_idle_thread() in builtin-sched.c, or thread_stack__per_cpu().
+ */
 int perf_session__register_idle_thread(struct perf_session *session)
 {
 	struct thread *thread;

tools/perf/util/thread-stack.c

@@ -15,6 +15,7 @@
 
 #include <linux/rbtree.h>
 #include <linux/list.h>
+#include <linux/log2.h>
 #include <errno.h>
 #include "thread.h"
 #include "event.h"
@@ -60,6 +61,7 @@ struct thread_stack_entry {
  * @last_time: last timestamp
  * @crp: call/return processor
  * @comm: current comm
+ * @arr_sz: size of array if this is the first element of an array
  */
 struct thread_stack {
 	struct thread_stack_entry *stack;
@@ -71,8 +73,19 @@ struct thread_stack {
 	u64 last_time;
 	struct call_return_processor *crp;
 	struct comm *comm;
+	unsigned int arr_sz;
 };
 
+/*
+ * Assume pid == tid == 0 identifies the idle task as defined by
+ * perf_session__register_idle_thread(). The idle task is really 1 task per cpu,
+ * and therefore requires a stack for each cpu.
+ */
+static inline bool thread_stack__per_cpu(struct thread *thread)
+{
+	return !(thread->tid || thread->pid_);
+}
+
 static int thread_stack__grow(struct thread_stack *ts)
 {
 	struct thread_stack_entry *new_stack;
@@ -91,19 +104,14 @@ static int thread_stack__grow(struct thread_stack *ts)
 	return 0;
 }
 
-static struct thread_stack *thread_stack__new(struct thread *thread,
-					      struct call_return_processor *crp)
+static int thread_stack__init(struct thread_stack *ts, struct thread *thread,
+			      struct call_return_processor *crp)
 {
-	struct thread_stack *ts;
+	int err;
 
-	ts = zalloc(sizeof(struct thread_stack));
-	if (!ts)
-		return NULL;
-
-	if (thread_stack__grow(ts)) {
-		free(ts);
-		return NULL;
-	}
+	err = thread_stack__grow(ts);
+	if (err)
+		return err;
 
 	if (thread->mg && thread->mg->machine)
 		ts->kernel_start = machine__kernel_start(thread->mg->machine);
@@ -111,9 +119,72 @@ static struct thread_stack *thread_stack__new(struct thread *thread,
 		ts->kernel_start = 1ULL << 63;
 	ts->crp = crp;
 
+	return 0;
+}
+
+static struct thread_stack *thread_stack__new(struct thread *thread, int cpu,
+					      struct call_return_processor *crp)
+{
+	struct thread_stack *ts = thread->ts, *new_ts;
+	unsigned int old_sz = ts ? ts->arr_sz : 0;
+	unsigned int new_sz = 1;
+
+	if (thread_stack__per_cpu(thread) && cpu > 0)
+		new_sz = roundup_pow_of_two(cpu + 1);
+
+	if (!ts || new_sz > old_sz) {
+		new_ts = calloc(new_sz, sizeof(*ts));
+		if (!new_ts)
+			return NULL;
+		if (ts)
+			memcpy(new_ts, ts, old_sz * sizeof(*ts));
+		new_ts->arr_sz = new_sz;
+		zfree(&thread->ts);
+		thread->ts = new_ts;
+		ts = new_ts;
+	}
+
+	if (thread_stack__per_cpu(thread) && cpu > 0 &&
+	    (unsigned int)cpu < ts->arr_sz)
+		ts += cpu;
+
+	if (!ts->stack &&
+	    thread_stack__init(ts, thread, crp))
+		return NULL;
+
 	return ts;
 }
 
+static struct thread_stack *thread__cpu_stack(struct thread *thread, int cpu)
+{
+	struct thread_stack *ts = thread->ts;
+
+	if (cpu < 0)
+		cpu = 0;
+
+	if (!ts || (unsigned int)cpu >= ts->arr_sz)
+		return NULL;
+
+	ts += cpu;
+
+	if (!ts->stack)
+		return NULL;
+
+	return ts;
+}
+
+static inline struct thread_stack *thread__stack(struct thread *thread,
+						    int cpu)
+{
+	if (!thread)
+		return NULL;
+
+	if (thread_stack__per_cpu(thread))
+		return thread__cpu_stack(thread, cpu);
+
+	return thread->ts;
+}
+
 static int thread_stack__push(struct thread_stack *ts, u64 ret_addr,
 			      bool trace_end)
 {
@@ -226,25 +297,37 @@ static int __thread_stack__flush(struct thread *thread, struct thread_stack *ts)
 
 int thread_stack__flush(struct thread *thread)
 {
-	if (thread->ts)
-		return __thread_stack__flush(thread, thread->ts);
+	struct thread_stack *ts = thread->ts;
+	unsigned int pos;
+	int err = 0;
 
-	return 0;
+	if (ts) {
+		for (pos = 0; pos < ts->arr_sz; pos++) {
+			int ret = __thread_stack__flush(thread, ts + pos);
+
+			if (ret)
+				err = ret;
+		}
+	}
+
+	return err;
 }
 
-int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
+int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip,
 			u64 to_ip, u16 insn_len, u64 trace_nr)
 {
+	struct thread_stack *ts = thread__stack(thread, cpu);
+
 	if (!thread)
 		return -EINVAL;
 
-	if (!thread->ts) {
-		thread->ts = thread_stack__new(thread, NULL);
-		if (!thread->ts) {
+	if (!ts) {
+		ts = thread_stack__new(thread, cpu, NULL);
+		if (!ts) {
 			pr_warning("Out of memory: no thread stack\n");
 			return -ENOMEM;
 		}
-		thread->ts->trace_nr = trace_nr;
+		ts->trace_nr = trace_nr;
 	}
 
 	/*
@@ -252,14 +335,14 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
 	 * the stack might be completely invalid.  Better to report nothing than
 	 * to report something misleading, so flush the stack.
 	 */
-	if (trace_nr != thread->ts->trace_nr) {
-		if (thread->ts->trace_nr)
-			__thread_stack__flush(thread, thread->ts);
-		thread->ts->trace_nr = trace_nr;
+	if (trace_nr != ts->trace_nr) {
+		if (ts->trace_nr)
+			__thread_stack__flush(thread, ts);
+		ts->trace_nr = trace_nr;
 	}
 
 	/* Stop here if thread_stack__process() is in use */
-	if (thread->ts->crp)
+	if (ts->crp)
 		return 0;
 
 	if (flags & PERF_IP_FLAG_CALL) {
@@ -270,7 +353,7 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
 		ret_addr = from_ip + insn_len;
 		if (ret_addr == to_ip)
 			return 0; /* Zero-length calls are excluded */
-		return thread_stack__push(thread->ts, ret_addr,
+		return thread_stack__push(ts, ret_addr,
 					  flags & PERF_IP_FLAG_TRACE_END);
 	} else if (flags & PERF_IP_FLAG_TRACE_BEGIN) {
 		/*
@@ -280,32 +363,52 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
 		 * address, so try to pop that. Also, do not expect a call made
 		 * when the trace ended, to return, so pop that.
 		 */
-		thread_stack__pop(thread->ts, to_ip);
-		thread_stack__pop_trace_end(thread->ts);
+		thread_stack__pop(ts, to_ip);
+		thread_stack__pop_trace_end(ts);
 	} else if ((flags & PERF_IP_FLAG_RETURN) && from_ip) {
-		thread_stack__pop(thread->ts, to_ip);
+		thread_stack__pop(ts, to_ip);
 	}
 
 	return 0;
 }
 
-void thread_stack__set_trace_nr(struct thread *thread, u64 trace_nr)
+void thread_stack__set_trace_nr(struct thread *thread, int cpu, u64 trace_nr)
 {
-	if (!thread || !thread->ts)
+	struct thread_stack *ts = thread__stack(thread, cpu);
+
+	if (!ts)
 		return;
 
-	if (trace_nr != thread->ts->trace_nr) {
-		if (thread->ts->trace_nr)
-			__thread_stack__flush(thread, thread->ts);
-		thread->ts->trace_nr = trace_nr;
+	if (trace_nr != ts->trace_nr) {
+		if (ts->trace_nr)
+			__thread_stack__flush(thread, ts);
+		ts->trace_nr = trace_nr;
 	}
 }
 
+static void __thread_stack__free(struct thread *thread, struct thread_stack *ts)
+{
+	__thread_stack__flush(thread, ts);
+	zfree(&ts->stack);
+}
+
+static void thread_stack__reset(struct thread *thread, struct thread_stack *ts)
+{
+	unsigned int arr_sz = ts->arr_sz;
+
+	__thread_stack__free(thread, ts);
+	memset(ts, 0, sizeof(*ts));
+	ts->arr_sz = arr_sz;
+}
+
 void thread_stack__free(struct thread *thread)
 {
-	if (thread->ts) {
-		__thread_stack__flush(thread, thread->ts);
-		zfree(&thread->ts->stack);
+	struct thread_stack *ts = thread->ts;
+	unsigned int pos;
+
+	if (ts) {
+		for (pos = 0; pos < ts->arr_sz; pos++)
+			__thread_stack__free(thread, ts + pos);
 		zfree(&thread->ts);
 	}
 }
@@ -315,9 +418,11 @@ static inline u64 callchain_context(u64 ip, u64 kernel_start)
 	return ip < kernel_start ? PERF_CONTEXT_USER : PERF_CONTEXT_KERNEL;
 }
 
-void thread_stack__sample(struct thread *thread, struct ip_callchain *chain,
+void thread_stack__sample(struct thread *thread, int cpu,
+			  struct ip_callchain *chain,
 			  size_t sz, u64 ip, u64 kernel_start)
 {
+	struct thread_stack *ts = thread__stack(thread, cpu);
 	u64 context = callchain_context(ip, kernel_start);
 	u64 last_context;
 	size_t i, j;
@@ -330,15 +435,15 @@ void thread_stack__sample(struct thread *thread, struct ip_callchain *chain,
 	chain->ips[0] = context;
 	chain->ips[1] = ip;
 
-	if (!thread || !thread->ts) {
+	if (!ts) {
 		chain->nr = 2;
 		return;
 	}
 
 	last_context = context;
 
-	for (i = 2, j = 1; i < sz && j <= thread->ts->cnt; i++, j++) {
-		ip = thread->ts->stack[thread->ts->cnt - j].ret_addr;
+	for (i = 2, j = 1; i < sz && j <= ts->cnt; i++, j++) {
+		ip = ts->stack[ts->cnt - j].ret_addr;
 		context = callchain_context(ip, kernel_start);
 		if (context != last_context) {
 			if (i >= sz - 1)
@@ -449,7 +554,7 @@ static int thread_stack__pop_cp(struct thread *thread, struct thread_stack *ts,
 	return 1;
 }
 
-static int thread_stack__bottom(struct thread *thread, struct thread_stack *ts,
+static int thread_stack__bottom(struct thread_stack *ts,
 				struct perf_sample *sample,
 				struct addr_location *from_al,
 				struct addr_location *to_al, u64 ref)
@@ -474,7 +579,7 @@ static int thread_stack__bottom(struct thread *thread, struct thread_stack *ts,
 	if (!cp)
 		return -ENOMEM;
 
-	return thread_stack__push_cp(thread->ts, ip, sample->time, ref, cp,
+	return thread_stack__push_cp(ts, ip, sample->time, ref, cp,
 				     true, false);
 }
 
@@ -590,24 +695,19 @@ int thread_stack__process(struct thread *thread, struct comm *comm,
 			  struct addr_location *to_al, u64 ref,
 			  struct call_return_processor *crp)
 {
-	struct thread_stack *ts = thread->ts;
+	struct thread_stack *ts = thread__stack(thread, sample->cpu);
 	int err = 0;
 
-	if (ts) {
-		if (!ts->crp) {
-			/* Supersede thread_stack__event() */
-			thread_stack__free(thread);
-			thread->ts = thread_stack__new(thread, crp);
-			if (!thread->ts)
-				return -ENOMEM;
-			ts = thread->ts;
-			ts->comm = comm;
-		}
-	} else {
-		thread->ts = thread_stack__new(thread, crp);
-		if (!thread->ts)
+	if (ts && !ts->crp) {
+		/* Supersede thread_stack__event() */
+		thread_stack__reset(thread, ts);
+		ts = NULL;
+	}
+
+	if (!ts) {
+		ts = thread_stack__new(thread, sample->cpu, crp);
+		if (!ts)
 			return -ENOMEM;
-		ts = thread->ts;
 		ts->comm = comm;
 	}
 
@@ -621,8 +721,7 @@ int thread_stack__process(struct thread *thread, struct comm *comm,
 
 	/* If the stack is empty, put the current symbol on the stack */
 	if (!ts->cnt) {
-		err = thread_stack__bottom(thread, ts, sample, from_al, to_al,
-					   ref);
+		err = thread_stack__bottom(ts, sample, from_al, to_al, ref);
 		if (err)
 			return err;
 	}
@@ -671,9 +770,11 @@ int thread_stack__process(struct thread *thread, struct comm *comm,
 	return err;
 }
 
-size_t thread_stack__depth(struct thread *thread)
+size_t thread_stack__depth(struct thread *thread, int cpu)
 {
-	if (!thread->ts)
+	struct thread_stack *ts = thread__stack(thread, cpu);
+
+	if (!ts)
 		return 0;
-	return thread->ts->cnt;
+	return ts->cnt;
 }

tools/perf/util/thread-stack.h

@@ -80,14 +80,14 @@ struct call_return_processor {
 	void *data;
 };
 
-int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
+int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip,
 			u64 to_ip, u16 insn_len, u64 trace_nr);
-void thread_stack__set_trace_nr(struct thread *thread, u64 trace_nr);
-void thread_stack__sample(struct thread *thread, struct ip_callchain *chain,
+void thread_stack__set_trace_nr(struct thread *thread, int cpu, u64 trace_nr);
+void thread_stack__sample(struct thread *thread, int cpu, struct ip_callchain *chain,
 			  size_t sz, u64 ip, u64 kernel_start);
 int thread_stack__flush(struct thread *thread);
 void thread_stack__free(struct thread *thread);
-size_t thread_stack__depth(struct thread *thread);
+size_t thread_stack__depth(struct thread *thread, int cpu);
 
 struct call_return_processor *
 call_return_processor__new(int (*process)(struct call_return *cr, void *data),