Merge tag 'perf-urgent-for-mingo-5.7-20200403' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent

Pull perf/urgent fixes and improvements from Arnaldo Carvalho de Melo:

perf python:

  Arnaldo Carvalho de Melo:

  - Fix clang detection to strip out options passed in $CC.

build:

  He Zhe:

  - Normalize gcc parameter when generating arch errno table, fixing
    the build by removing options from $(CC).

  Sam Lunt:

  - Support Python 3.8+ in Makefile.

perf report/top:

  Arnaldo Carvalho de Melo:

  - Fix title line formatting.

perf script:

  Andreas Gerstmayr:

  - Fix SEGFAULT when using DWARF mode.

  - Fix invalid read of directory entry after closedir(), found with valgrind.

  Hagen Paul Pfeifer:

  - Introduce --deltatime option.

  Stephane Eranian:

  - Allow --symbol to accept hexadecimal addresses.

  Ian Rogers:

  - Add -S/--symbols documentation

  Namhyung Kim:

  - Add --show-cgroup-events option.

perf python:

  Arnaldo Carvalho de Melo:

  - Include rwsem.c in the python binding, needed by the cgroups improvements.

build-test:

  Arnaldo Carvalho de Melo:

  - Honour JOBS to override detection of number of cores

perf top:

  Jin Yao:

  - Support --group-sort-idx to change the sort order

  - perf top: Support hotkey to change sort order

perf pmu-events x86:

  Jin Yao:

  - Use CPU_CLK_UNHALTED.THREAD in Kernel_Utilization metric

perf symbols arm64:

  Kemeng Shi:

  - Fix arm64 gap between kernel start and module end

kernel perf subsystem:

  Namhyung Kim:

  - Add PERF_RECORD_CGROUP event and Add PERF_SAMPLE_CGROUP feature,
    to allow cgroup tracking, saving a link between cgroup path and
    its id number.

perf cgroup:

  Namhyung Kim:

  - Maintain cgroup hierarchy.

perf report:

  Namhyung Kim:

  - Add 'cgroup' sort key.

perf record:

  Namhyung Kim:

  - Support synthesizing cgroup events for pre-existing cgroups.

  - Add --all-cgroups option

Documentation:

  Tony Jones:

  - Update docs regarding kernel/user space unwinding.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

100 files changed, 3182 insertions, 622 deletions

diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 8768a39b5258..9c3e7619c929 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1020,6 +1020,7 @@ struct perf_sample_data {
 	u64				stack_user_size;
 
 	u64				phys_addr;
+	u64				cgroup;
 } ____cacheline_aligned;
 
 /* default value for data source */
diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index 397cfd65b3fe..7b2d6fc9e6ed 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -142,8 +142,9 @@ enum perf_event_sample_format {
 	PERF_SAMPLE_REGS_INTR			= 1U << 18,
 	PERF_SAMPLE_PHYS_ADDR			= 1U << 19,
 	PERF_SAMPLE_AUX				= 1U << 20,
+	PERF_SAMPLE_CGROUP			= 1U << 21,
 
-	PERF_SAMPLE_MAX = 1U << 21,		/* non-ABI */
+	PERF_SAMPLE_MAX = 1U << 22,		/* non-ABI */
 
 	__PERF_SAMPLE_CALLCHAIN_EARLY		= 1ULL << 63, /* non-ABI; internal use */
 };
@@ -381,7 +382,8 @@ struct perf_event_attr {
 				ksymbol        :  1, /* include ksymbol events */
 				bpf_event      :  1, /* include bpf events */
 				aux_output     :  1, /* generate AUX records instead of events */
-				__reserved_1   : 32;
+				cgroup         :  1, /* include cgroup events */
+				__reserved_1   : 31;
 
 	union {
 		__u32		wakeup_events;	  /* wakeup every n events */
@@ -1012,6 +1014,16 @@ enum perf_event_type {
 	 */
 	PERF_RECORD_BPF_EVENT			= 18,
 
+	/*
+	 * struct {
+	 *	struct perf_event_header	header;
+	 *	u64				id;
+	 *	char				path[];
+	 *	struct sample_id		sample_id;
+	 * };
+	 */
+	PERF_RECORD_CGROUP			= 19,
+
 	PERF_RECORD_MAX,			/* non-ABI */
 };
 
diff --git a/init/Kconfig b/init/Kconfig
index 697d109387ed..93e1fafa88ce 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1030,7 +1030,8 @@ config CGROUP_PERF
 	help
 	  This option extends the perf per-cpu mode to restrict monitoring
 	  to threads which belong to the cgroup specified and run on the
-	  designated cpu.
+	  designated cpu.  Or this can be used to have cgroup ID in samples
+	  so that it can monitor performance events among cgroups.
 
 	  Say N if unsure.
 
diff --git a/kernel/events/core.c b/kernel/events/core.c
index d22e4ba59dfa..1569979c8912 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -387,6 +387,7 @@ static atomic_t nr_freq_events __read_mostly;
 static atomic_t nr_switch_events __read_mostly;
 static atomic_t nr_ksymbol_events __read_mostly;
 static atomic_t nr_bpf_events __read_mostly;
+static atomic_t nr_cgroup_events __read_mostly;
 
 static LIST_HEAD(pmus);
 static DEFINE_MUTEX(pmus_lock);
@@ -1861,6 +1862,9 @@ static void __perf_event_header_size(struct perf_event *event, u64 sample_type)
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		size += sizeof(data->phys_addr);
 
+	if (sample_type & PERF_SAMPLE_CGROUP)
+		size += sizeof(data->cgroup);
+
 	event->header_size = size;
 }
 
@@ -4608,6 +4612,8 @@ static void unaccount_event(struct perf_event *event)
 		atomic_dec(&nr_comm_events);
 	if (event->attr.namespaces)
 		atomic_dec(&nr_namespaces_events);
+	if (event->attr.cgroup)
+		atomic_dec(&nr_cgroup_events);
 	if (event->attr.task)
 		atomic_dec(&nr_task_events);
 	if (event->attr.freq)
@@ -6864,6 +6870,9 @@ void perf_output_sample(struct perf_output_handle *handle,
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		perf_output_put(handle, data->phys_addr);
 
+	if (sample_type & PERF_SAMPLE_CGROUP)
+		perf_output_put(handle, data->cgroup);
+
 	if (sample_type & PERF_SAMPLE_AUX) {
 		perf_output_put(handle, data->aux_size);
 
@@ -7063,6 +7072,16 @@ void perf_prepare_sample(struct perf_event_header *header,
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		data->phys_addr = perf_virt_to_phys(data->addr);
 
+#ifdef CONFIG_CGROUP_PERF
+	if (sample_type & PERF_SAMPLE_CGROUP) {
+		struct cgroup *cgrp;
+
+		/* protected by RCU */
+		cgrp = task_css_check(current, perf_event_cgrp_id, 1)->cgroup;
+		data->cgroup = cgroup_id(cgrp);
+	}
+#endif
+
 	if (sample_type & PERF_SAMPLE_AUX) {
 		u64 size;
 
@@ -7736,6 +7755,105 @@ void perf_event_namespaces(struct task_struct *task)
 }
 
 /*
+ * cgroup tracking
+ */
+#ifdef CONFIG_CGROUP_PERF
+
+struct perf_cgroup_event {
+	char				*path;
+	int				path_size;
+	struct {
+		struct perf_event_header	header;
+		u64				id;
+		char				path[];
+	} event_id;
+};
+
+static int perf_event_cgroup_match(struct perf_event *event)
+{
+	return event->attr.cgroup;
+}
+
+static void perf_event_cgroup_output(struct perf_event *event, void *data)
+{
+	struct perf_cgroup_event *cgroup_event = data;
+	struct perf_output_handle handle;
+	struct perf_sample_data sample;
+	u16 header_size = cgroup_event->event_id.header.size;
+	int ret;
+
+	if (!perf_event_cgroup_match(event))
+		return;
+
+	perf_event_header__init_id(&cgroup_event->event_id.header,
+				   &sample, event);
+	ret = perf_output_begin(&handle, event,
+				cgroup_event->event_id.header.size);
+	if (ret)
+		goto out;
+
+	perf_output_put(&handle, cgroup_event->event_id);
+	__output_copy(&handle, cgroup_event->path, cgroup_event->path_size);
+
+	perf_event__output_id_sample(event, &handle, &sample);
+
+	perf_output_end(&handle);
+out:
+	cgroup_event->event_id.header.size = header_size;
+}
+
+static void perf_event_cgroup(struct cgroup *cgrp)
+{
+	struct perf_cgroup_event cgroup_event;
+	char path_enomem[16] = "//enomem";
+	char *pathname;
+	size_t size;
+
+	if (!atomic_read(&nr_cgroup_events))
+		return;
+
+	cgroup_event = (struct perf_cgroup_event){
+		.event_id  = {
+			.header = {
+				.type = PERF_RECORD_CGROUP,
+				.misc = 0,
+				.size = sizeof(cgroup_event.event_id),
+			},
+			.id = cgroup_id(cgrp),
+		},
+	};
+
+	pathname = kmalloc(PATH_MAX, GFP_KERNEL);
+	if (pathname == NULL) {
+		cgroup_event.path = path_enomem;
+	} else {
+		/* just to be sure to have enough space for alignment */
+		cgroup_path(cgrp, pathname, PATH_MAX - sizeof(u64));
+		cgroup_event.path = pathname;
+	}
+
+	/*
+	 * Since our buffer works in 8 byte units we need to align our string
+	 * size to a multiple of 8. However, we must guarantee the tail end is
+	 * zero'd out to avoid leaking random bits to userspace.
+	 */
+	size = strlen(cgroup_event.path) + 1;
+	while (!IS_ALIGNED(size, sizeof(u64)))
+		cgroup_event.path[size++] = '\0';
+
+	cgroup_event.event_id.header.size += size;
+	cgroup_event.path_size = size;
+
+	perf_iterate_sb(perf_event_cgroup_output,
+			&cgroup_event,
+			NULL);
+
+	kfree(pathname);
+}
+
+#endif