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

Pull perf fixes from Ingo Molnar: "Mostly tooling fixes, plus a static key fix fixing /sys/devices/cpu/rdpmc" * 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: perf tools: Really allow to specify custom CC, AR or LD perf auxtrace: Fix misplaced check for HAVE_SYNC_COMPARE_AND_SWAP_SUPPORT perf hists browser: Take the --comm, --dsos, etc filters into account perf symbols: Store if there is a filter in place x86, perf: Fix static_key bug in load_mm_cr4() tools: Copy lib/hweight.c from the kernel sources perf tools: Fix the detached tarball wrt rbtree copy perf thread_map: Fix the sizeof() calculation for map entries tools lib: Improve clean target perf stat: Fix shadow declaration of close perf tools: Fix lockup using 32-bit compat vdso

Linus Torvalds committed 11 years ago

Commit f79a17b2 parents 59ee762 a6acd6a15 files · +101 −26Browse files

Unified · Split

15 files

arch/x86/include/asm/mmu_context.h +1 −1
tools/lib/api/Makefile +1 −1
tools/lib/hweight.c +62 −0
tools/lib/traceevent/Makefile +1 −1
tools/perf/MANIFEST +1 −1
tools/perf/Makefile.perf +16 −3
tools/perf/builtin-stat.c +2 −2
tools/perf/ui/browsers/hists.c +1 −1
tools/perf/util/Build +1 −1
tools/perf/util/auxtrace.c +5 −5
tools/perf/util/python-ext-sources +2 −2
tools/perf/util/symbol.c +2 −0
tools/perf/util/symbol.h +2 −1
tools/perf/util/thread_map.c +1 −2
tools/perf/util/vdso.c +3 −5

arch/x86/include/asm/mmu_context.h+1−1

@@ -23,7 +23,7 @@ extern struct static_key rdpmc_always_available; · 22 unmodified lines
23	23
24	24	static inline void load_mm_cr4(struct mm_struct *mm)
25	25	{
26		− if (static_key_true(&rdpmc_always_available) \|\|
	26	+ if (static_key_false(&rdpmc_always_available) \|\|
27	27	atomic_read(&mm->context.perf_rdpmc_allowed))
28	28	cr4_set_bits(X86_CR4_PCE);
29	29	else

tools/lib/api/Makefile+1−1

@@ -36,7 +36,7 @@ $(LIBFILE): $(API_IN) · 35 unmodified lines
36	36
37	37	clean:
38	38	$(call QUIET_CLEAN, libapi) $(RM) $(LIBFILE); \
39		− find $(if $(OUTPUT),$(OUTPUT),.) -name \*.o \| xargs $(RM)
	39	+ find $(if $(OUTPUT),$(OUTPUT),.) -name \.o -or -name \.o.cmd -or -name \*.o.d \| xargs $(RM)
40	40
41	41	FORCE:
42	42

tools/lib/hweight.c+62−0

@@ -0,0 +1,62 @@
	1	+#include <linux/bitops.h>
	2	+#include <asm/types.h>
	3	+
	4	+/**
	5	+ * hweightN - returns the hamming weight of a N-bit word
	6	+ * @x: the word to weigh
	7	+ *
	8	+ * The Hamming Weight of a number is the total number of bits set in it.
	9	+ */
	10	+
	11	+unsigned int __sw_hweight32(unsigned int w)
	12	+{
	13	+#ifdef CONFIG_ARCH_HAS_FAST_MULTIPLIER
	14	+ w -= (w >> 1) & 0x55555555;
	15	+ w = (w & 0x33333333) + ((w >> 2) & 0x33333333);
	16	+ w = (w + (w >> 4)) & 0x0f0f0f0f;
	17	+ return (w * 0x01010101) >> 24;
	18	+#else
	19	+ unsigned int res = w - ((w >> 1) & 0x55555555);
	20	+ res = (res & 0x33333333) + ((res >> 2) & 0x33333333);
	21	+ res = (res + (res >> 4)) & 0x0F0F0F0F;
	22	+ res = res + (res >> 8);
	23	+ return (res + (res >> 16)) & 0x000000FF;
	24	+#endif
	25	+}
	26	+
	27	+unsigned int __sw_hweight16(unsigned int w)
	28	+{
	29	+ unsigned int res = w - ((w >> 1) & 0x5555);
	30	+ res = (res & 0x3333) + ((res >> 2) & 0x3333);
	31	+ res = (res + (res >> 4)) & 0x0F0F;
	32	+ return (res + (res >> 8)) & 0x00FF;
	33	+}
	34	+
	35	+unsigned int __sw_hweight8(unsigned int w)
	36	+{
	37	+ unsigned int res = w - ((w >> 1) & 0x55);
	38	+ res = (res & 0x33) + ((res >> 2) & 0x33);
	39	+ return (res + (res >> 4)) & 0x0F;
	40	+}
	41	+
	42	+unsigned long __sw_hweight64(__u64 w)
	43	+{
	44	+#if BITS_PER_LONG == 32
	45	+ return __sw_hweight32((unsigned int)(w >> 32)) +
	46	+ __sw_hweight32((unsigned int)w);
	47	+#elif BITS_PER_LONG == 64
	48	+#ifdef CONFIG_ARCH_HAS_FAST_MULTIPLIER
	49	+ w -= (w >> 1) & 0x5555555555555555ul;
	50	+ w = (w & 0x3333333333333333ul) + ((w >> 2) & 0x3333333333333333ul);
	51	+ w = (w + (w >> 4)) & 0x0f0f0f0f0f0f0f0ful;
	52	+ return (w * 0x0101010101010101ul) >> 56;
	53	+#else
	54	+ __u64 res = w - ((w >> 1) & 0x5555555555555555ul);
	55	+ res = (res & 0x3333333333333333ul) + ((res >> 2) & 0x3333333333333333ul);
	56	+ res = (res + (res >> 4)) & 0x0F0F0F0F0F0F0F0Ful;
	57	+ res = res + (res >> 8);
	58	+ res = res + (res >> 16);
	59	+ return (res + (res >> 32)) & 0x00000000000000FFul;
	60	+#endif
	61	+#endif
	62	+}

tools/lib/traceevent/Makefile+1−1

@@ -268,7 +268,7 @@ install: install_lib · 267 unmodified lines
268	268
269	269	clean:
270	270	$(call QUIET_CLEAN, libtraceevent) \
271		− $(RM) .o ~ $(TARGETS) .a .so $(VERSION_FILES) .*.d \
	271	+ $(RM) .o ~ $(TARGETS) .a .so $(VERSION_FILES) ..d ..cmd \
272	272	$(RM) TRACEEVENT-CFLAGS tags TAGS
273	273
274	274	PHONY += force plugins

tools/perf/MANIFEST+1−1

@@ -18,6 +18,7 @@ tools/arch/x86/include/asm/atomic.h · 17 unmodified lines
18	18	tools/arch/x86/include/asm/rmwcc.h
19	19	tools/lib/traceevent
20	20	tools/lib/api
	21	+tools/lib/hweight.c
21	22	tools/lib/rbtree.c
22	23	tools/lib/symbol/kallsyms.c
23	24	tools/lib/symbol/kallsyms.h
@@ -57,7 +58,6 @@ include/linux/perf_event.h · 33 unmodified lines
57	58	include/linux/list.h
58	59	include/linux/hash.h
59	60	include/linux/stringify.h
60		−lib/hweight.c
61	61	include/linux/swab.h
62	62	arch//include/asm/unistd.h
63	63	arch//include/uapi/asm/unistd.h

tools/perf/Makefile.perf+16−3

@@ -109,9 +109,22 @@ $(OUTPUT)PERF-VERSION-FILE: ../../.git/HEAD · 108 unmodified lines
109	109	$(Q)$(SHELL_PATH) util/PERF-VERSION-GEN $(OUTPUT)
110	110	$(Q)touch $(OUTPUT)PERF-VERSION-FILE
111	111
112		−CC = $(CROSS_COMPILE)gcc
113		−LD ?= $(CROSS_COMPILE)ld
114		−AR = $(CROSS_COMPILE)ar
	112	+# Makefiles suck: This macro sets a default value of $(2) for the
	113	+# variable named by $(1), unless the variable has been set by
	114	+# environment or command line. This is necessary for CC and AR
	115	+# because make sets default values, so the simpler ?= approach
	116	+# won't work as expected.
	117	+define allow-override
	118	+ $(if $(or $(findstring environment,$(origin $(1))),\
	119	+ $(findstring command line,$(origin $(1)))),,\
	120	+ $(eval $(1) = $(2)))
	121	+endef
	122	+
	123	+# Allow setting CC and AR and LD, or setting CROSS_COMPILE as a prefix.
	124	+$(call allow-override,CC,$(CROSS_COMPILE)gcc)
	125	+$(call allow-override,AR,$(CROSS_COMPILE)ar)
	126	+$(call allow-override,LD,$(CROSS_COMPILE)ld)
	127	+
115	128	PKG_CONFIG = $(CROSS_COMPILE)pkg-config
116	129
117	130	RM = rm -f

tools/perf/builtin-stat.c+2−2

@@ -343,7 +343,7 @@ static int read_counter(struct perf_evsel *counter) · 342 unmodified lines
343	343	return 0;
344	344	}
345	345
346		−static void read_counters(bool close)
	346	+static void read_counters(bool close_counters)
347	347	{
348	348	struct perf_evsel *counter;
349	349
@@ -354,7 +354,7 @@ static void read_counters(bool close) · 4 unmodified lines
354	354	if (process_counter(counter))
355	355	pr_warning("failed to process counter %s\n", counter->name);
356	356
357		− if (close) {
	357	+ if (close_counters) {
358	358	perf_evsel__close_fd(counter, perf_evsel__nr_cpus(counter),
359	359	thread_map__nr(evsel_list->threads));
360	360	}

tools/perf/ui/browsers/hists.c+1−1

@@ -48,7 +48,7 @@ static struct rb_node hists__filter_entries(struct rb_node nd, · 47 unmodified lines
48	48
49	49	static bool hist_browser__has_filter(struct hist_browser *hb)
50	50	{
51		− return hists__has_filter(hb->hists) \|\| hb->min_pcnt;
	51	+ return hists__has_filter(hb->hists) \|\| hb->min_pcnt \|\| symbol_conf.has_filter;
52	52	}
53	53
54	54	static int hist_browser__get_folding(struct hist_browser *browser)

tools/perf/util/Build+1−1

@@ -143,6 +143,6 @@ $(OUTPUT)util/rbtree.o: ../lib/rbtree.c FORCE · 142 unmodified lines
143	143	$(call rule_mkdir)
144	144	$(call if_changed_dep,cc_o_c)
145	145
146		−$(OUTPUT)util/hweight.o: ../../lib/hweight.c FORCE
	146	+$(OUTPUT)util/hweight.o: ../lib/hweight.c FORCE
147	147	$(call rule_mkdir)
148	148	$(call if_changed_dep,cc_o_c)

tools/perf/util/auxtrace.c+5−5

@@ -53,11 +53,6 @@ int auxtrace_mmap__mmap(struct auxtrace_mmap *mm, · 52 unmodified lines
53	53	{
54	54	struct perf_event_mmap_page *pc = userpg;
55	55
56		−#if BITS_PER_LONG != 64 && !defined(HAVE_SYNC_COMPARE_AND_SWAP_SUPPORT)
57		− pr_err("Cannot use AUX area tracing mmaps\n");
58		− return -1;
59		−#endif
60		−
61	56	WARN_ONCE(mm->base, "Uninitialized auxtrace_mmap\n");
62	57
63	58	mm->userpg = userpg;
@@ -73,6 +68,11 @@ int auxtrace_mmap__mmap(struct auxtrace_mmap *mm, · 9 unmodified lines
73	68	return 0;
74	69	}
75	70
	71	+#if BITS_PER_LONG != 64 && !defined(HAVE_SYNC_COMPARE_AND_SWAP_SUPPORT)
	72	+ pr_err("Cannot use AUX area tracing mmaps\n");
	73	+ return -1;
	74	+#endif
	75	+
76	76	pc->aux_offset = mp->offset;
77	77	pc->aux_size = mp->len;
78	78

tools/perf/util/python-ext-sources+2−2

@@ -10,7 +10,7 @@ util/ctype.c · 9 unmodified lines
10	10	util/evlist.c
11	11	util/evsel.c
12	12	util/cpumap.c
13		−../../lib/hweight.c
	13	+../lib/hweight.c
14	14	util/thread_map.c
15	15	util/util.c
16	16	util/xyarray.c
@@ -19,5 +19,5 @@ util/rblist.c · 2 unmodified lines
19	19	util/stat.c
20	20	util/strlist.c
21	21	util/trace-event.c
22		−../../lib/rbtree.c
	22	+../lib/rbtree.c
23	23	util/string.c

tools/perf/util/symbol.c+2−0

@@ -1911,6 +1911,8 @@ int setup_list(struct strlist *list, const char list_str, · 1910 unmodified lines
1911	1911	pr_err("problems parsing %s list\n", list_name);
1912	1912	return -1;
1913	1913	}
	1914	+
	1915	+ symbol_conf.has_filter = true;
1914	1916	return 0;
1915	1917	}
1916	1918

tools/perf/util/symbol.h+2−1

@@ -105,7 +105,8 @@ struct symbol_conf { · 104 unmodified lines
105	105	demangle_kernel,
106	106	filter_relative,
107	107	show_hist_headers,
108		− branch_callstack;
	108	+ branch_callstack,
	109	+ has_filter;
109	110	const char *vmlinux_name,
110	111	*kallsyms_name,
111	112	*source_prefix,

tools/perf/util/thread_map.c+1−2

@@ -136,8 +136,7 @@ struct thread_map *thread_map__new_by_uid(uid_t uid) · 135 unmodified lines
136	136	if (grow) {
137	137	struct thread_map *tmp;
138	138
139		− tmp = realloc(threads, (sizeof(*threads) +
140		− max_threads * sizeof(pid_t)));
	139	+ tmp = thread_map__realloc(threads, max_threads);
141	140	if (tmp == NULL)
142	141	goto out_free_namelist;
143	142

tools/perf/util/vdso.c+3−5

@@ -236,18 +236,16 @@ static struct dso __machine__findnew_compat(struct machine machine, · 235 unmodified lines
236	236	const char *file_name;
237	237	struct dso *dso;
238	238
239		− pthread_rwlock_wrlock(&machine->dsos.lock);
240	239	dso = __dsos__find(&machine->dsos, vdso_file->dso_name, true);
241	240	if (dso)
242		− goto out_unlock;
	241	+ goto out;
243	242
244	243	file_name = vdso__get_compat_file(vdso_file);
245	244	if (!file_name)
246		− goto out_unlock;
	245	+ goto out;
247	246
248	247	dso = __machine__addnew_vdso(machine, vdso_file->dso_name, file_name);
249		−out_unlock:
250		− pthread_rwlock_unlock(&machine->dsos.lock);
	248	+out:
251	249	return dso;
252	250	}
253	251