Skip to contentSkip to conversation

perf/urgent fixes

GIT PULL New Arnaldo Carvalho de Melo 2 participants 5 messages Feb 3, 2017 perf events

lore
Share & tools
lore

Arnaldo Carvalho de Melo

author maintainer
#0 lore raw

Hi Ingo,

Please consider pulling,

- Arnaldo

Test results at the end of this message, as usual.

The following changes since commit fff4b87e594ad3d2e4f51e8d3d86a6f9d3d8b654:

perf/x86/intel/uncore: Make package handling more robust (2017-02-01 08:37:27 +0100)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-urgent-for-mingo-4.10-20170203

for you to fetch changes up to aa33b9b9a2ebb00d33c83a5312d4fbf2d5aeba36:

perf callchain: Reference count maps (2017-02-02 11:39:09 -0300)

----------------------------------------------------------------
perf/urgent fixes:

. Reference count maps in callchains, fixing a SEGFAULT when referencing a
map after it is freed (Krister Johansen)

- Fix segfault on 'perf diff -o N' option (Namhyung Kim)

- Fix 'perf diff -o/--order' option behavior (Namhyung Kim)

----------------------------------------------------------------
Krister Johansen (1):
perf callchain: Reference count maps

Namhyung Kim (2):
perf diff: Fix segfault on 'perf diff -o N' option perf diff: Fix -o/--order option behavior (again)

 tools/perf/builtin-diff.c   |  2 +-
 tools/perf/ui/hist.c        | 10 ++++++++++
 tools/perf/util/callchain.c | 11 +++++++++--
 tools/perf/util/callchain.h |  6 ++++++
 tools/perf/util/hist.c      |  7 +++++++
 tools/perf/util/hist.h      |  7 +++++++
 6 files changed, 40 insertions(+), 3 deletions(-)

Test results:

The first ones are container (docker) based builds of tools/perf with and without libelf support, objtool where it is supported and samples/bpf/, ditto.

Several are cross builds, the ones with -x-ARCH, and the android one, and those may not have all the features built, due to lack of multi-arch devel packages, available and being used so far on just a few, like debian:experimental-x-{arm64,mipsel}.

The 'perf test' one will perform a variety of tests exercising tools/perf/util/, tools/lib/{bpf,traceevent,etc}, as well as run perf commands with a variety of command line event specifications to then intercept the sys_perf_event syscall to check that the perf_event_attr fields are set up as expected, among a variety of other unit tests.

Then there is the 'make -C tools/perf build-test' ones, that build tools/perf/ with a variety of feature sets, exercising the build with an incomplete set of features as well as with a complete one. It is planned to have it run on each of the containers mentioned above, using some container orchestration infrastructure. Get in contact if interested in helping having this in place.

  # dm
   1 alpine:3.4: Ok
   2 android-ndk:r12b-arm: Ok
   3 archlinux:latest: Ok
   4 centos:5: Ok
   5 centos:6: Ok
   6 centos:7: Ok
   7 debian:7: Ok
   8 debian:8: Ok
   9 debian:experimental: Ok
  10 debian:experimental-x-arm64: Ok
  11 debian:experimental-x-mips: Ok
  12 debian:experimental-x-mips64: Ok
  13 debian:experimental-x-mipsel: Ok
  14 fedora:20: Ok
  15 fedora:21: Ok
  16 fedora:22: Ok
  17 fedora:23: Ok
  18 fedora:24: Ok
  19 fedora:24-x-ARC-uClibc: Ok
  20 fedora:25: Ok
  21 fedora:rawhide: Ok
  22 mageia:5: Ok
  23 opensuse:13.2: Ok
  24 opensuse:42.1: Ok
  25 opensuse:tumbleweed: Ok
  26 ubuntu:12.04.5: Ok
  27 ubuntu:14.04.4-x-linaro-arm64: Ok
  28 ubuntu:15.10: Ok
  29 ubuntu:16.04: Ok
  30 ubuntu:16.04-x-arm: Ok
  31 ubuntu:16.04-x-arm64: Ok
  32 ubuntu:16.04-x-powerpc: Ok
  33 ubuntu:16.04-x-powerpc64: Ok
  34 ubuntu:16.04-x-powerpc64el: Ok
  35 ubuntu:16.04-x-s390: Ok
  36 ubuntu:16.10: Ok
  # uname -a
  Linux jouet 4.9.6-200.fc25.x86_64 #1 SMP Thu Jan 26 10:17:45 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
  # perf test
   1: vmlinux symtab matches kallsyms            : Ok
   2: Detect openat syscall event                : Ok
   3: Detect openat syscall event on all cpus    : Ok
   4: Read samples using the mmap interface      : Ok
   5: Parse event definition strings             : Ok
   6: PERF_RECORD_* events & perf_sample fields  : Ok
   7: Parse perf pmu format                      : Ok
   8: DSO data read                              : Ok
   9: DSO data cache                             : Ok
  10: DSO data reopen                            : Ok
  11: Roundtrip evsel->name                      : Ok
  12: Parse sched tracepoints fields             : Ok
  13: syscalls:sys_enter_openat event fields     : Ok
  14: Setup struct perf_event_attr               : Ok
  15: Match and link multiple hists              : Ok
  16: 'import perf' in python                    : Ok
  17: Breakpoint overflow signal handler         : Ok
  18: Breakpoint overflow sampling               : Ok
  19: Number of exit events of a simple workload : Ok
  20: Software clock events period values        : Ok
  21: Object code reading                        : Ok
  22: Sample parsing                             : Ok
  23: Use a dummy software event to keep tracking: Ok
  24: Parse with no sample_id_all bit set        : Ok
  25: Filter hist entries                        : Ok
  26: Lookup mmap thread                         : Ok
  27: Share thread mg                            : Ok
  28: Sort output of hist entries                : Ok
  29: Cumulate child hist entries                : Ok
  30: Track with sched_switch                    : Ok
  31: Filter fds with revents mask in a fdarray  : Ok
  32: Add fd to a fdarray, making it autogrow    : Ok
  33: kmod_path__parse                           : Ok
  34: Thread map                                 : Ok
  35: LLVM search and compile                    :
  35.1: Basic BPF llvm compile                    : Ok
  35.2: kbuild searching                          : Ok
  35.3: Compile source for BPF prologue generation: Ok
  35.4: Compile source for BPF relocation         : Ok
  36: Session topology                           : Ok
  37: BPF filter                                 :
  37.1: Basic BPF filtering                      : Ok
  37.2: BPF prologue generation                  : Ok
  37.3: BPF relocation checker                   : Ok
  38: Synthesize thread map                      : Ok
  39: Remove thread map                          : Ok
  40: Synthesize cpu map                         : Ok
  41: Synthesize stat config                     : Ok
  42: Synthesize stat                            : Ok
  43: Synthesize stat round                      : Ok
  44: Synthesize attr update                     : Ok
  45: Event times                                : Ok
  46: Read backward ring buffer                  : Ok
  47: Print cpu map                              : Ok
  48: Probe SDT events                           : Ok
  49: is_printable_array                         : Ok
  50: Print bitmap                               : Ok
  51: perf hooks                                 : Ok
  52: builtin clang support                      : Skip (not compiled in)
  53: x86 rdpmc                                  : Ok
  54: Convert perf time to TSC                   : Ok
  55: DWARF unwind                               : Ok
  56: x86 instruction decoder - new instructions : Ok
  57: Intel cqm nmi context read                 : Skip
  $ make -C tools/perf build-test
  make: Entering directory '/home/acme/git/linux/tools/perf'
  - tarpkg: ./tests/perf-targz-src-pkg .
  make FEATURE_DUMP_COPY=/home/acme/git/linux/tools/perf/BUILD_TEST_FEATURE_DUMP feature-dump
           make_no_libpython_O: make NO_LIBPYTHON=1
            make_no_demangle_O: make NO_DEMANGLE=1
  make_no_libdw_dwarf_unwind_O: make NO_LIBDW_DWARF_UNWIND=1
           make_no_backtrace_O: make NO_BACKTRACE=1
                   make_pure_O: make
             make_no_scripts_O: make NO_LIBPYTHON=1 NO_LIBPERL=1
   make_install_prefix_slash_O: make install prefix=/tmp/krava/
             make_no_libperl_O: make NO_LIBPERL=1
             make_no_libnuma_O: make NO_LIBNUMA=1
                  make_debug_O: make DEBUG=1
       make_util_pmu_bison_o_O: make util/pmu-bison.o
            make_no_auxtrace_O: make NO_AUXTRACE=1
         make_with_clangllvm_O: make LIBCLANGLLVM=1
                make_no_gtk2_O: make NO_GTK2=1
                  make_no_ui_O: make NO_NEWT=1 NO_SLANG=1 NO_GTK2=1
                make_install_O: make install
               make_no_slang_O: make NO_SLANG=1
                 make_static_O: make LDFLAGS=-static
            make_no_libaudit_O: make NO_LIBAUDIT=1
              make_no_libelf_O: make NO_LIBELF=1
                   make_help_O: make help
        make_with_babeltrace_O: make LIBBABELTRACE=1
              make_no_libbpf_O: make NO_LIBBPF=1
         make_install_prefix_O: make install prefix=/tmp/krava
           make_no_libunwind_O: make NO_LIBUNWIND=1
                    make_doc_O: make doc
           make_no_libbionic_O: make NO_LIBBIONIC=1
            make_install_bin_O: make install-bin
             make_util_map_o_O: make util/map.o
                make_minimal_O: make NO_LIBPERL=1 NO_LIBPYTHON=1 NO_NEWT=1 NO_GTK2=1 NO_DEMANGLE=1 NO_LIBELF=1 NO_LIBUNWIND=1 NO_BACKTRACE=1 NO_LIBNUMA=1 NO_LIBAUDIT=1 NO_LIBBIONIC=1 NO_LIBDW_DWARF_UNWIND=1 NO_AUXTRACE=1 NO_LIBBPF=1 NO_LIBCRYPTO=1 NO_SDT=1 NO_JVMTI=1
                   make_tags_O: make tags
              make_clean_all_O: make clean all
                 make_perf_o_O: make perf.o
                make_no_newt_O: make NO_NEWT=1
  OK
  make: Leaving directory '/home/acme/git/linux/tools/perf'
  $
author maintainer
#1 lore raw

PATCH 1/3 perf diff: Fix segfault on 'perf diff -o N' option

From: Namhyung Kim <[email protected]>

The -o/--order option is to select column number to sort a diff result.

It does the job by adding a hpp field at the beginning of the sort list. But it should not be added to the output field list as it has no callbacks required by a output field.

During the setup_sorting(), the perf_hpp__setup_output_field() appends the given sort keys to the output field if it's not there already.

Originally it was checked by fmt->list being non-empty. But commit 3f931f2c4274 ("perf hists: Make hpp setup function generic") changed it to check the ->equal callback.

Anyways, we don't need to add the pseudo hpp field to the output field list since it won't be used for output. So just skip fields if they have no ->color or ->entry callbacks.

Cc Peter Zijlstra <[email protected]>
Fixes 3f931f2c4274 ("perf hists: Make hpp setup function generic")
---
 tools/perf/ui/hist.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/perf/ui/hist.c +4 -0
index 37388397b5bc..4ec79b2f9416 100644
@@ -560,6 +560,10 @@ void perf_hpp__setup_output_field(struct perf_hpp_list *list)
 	perf_hpp_list__for_each_sort_list(list, fmt) {
 		struct perf_hpp_fmt *pos;
 
+		/* skip sort-only fields ("sort_compute" in perf diff) */
+		if (!fmt->entry && !fmt->color)
+			continue;
+
 		perf_hpp_list__for_each_format(list, pos) {
 			if (fmt_equal(fmt, pos))
 				goto next;
Signature · 1 line
2.9.3
author maintainer
#2 lore raw

PATCH 2/3 perf diff: Fix -o/--order option behavior (again)

In reply to Arnaldo Carvalho de Melo

From: Namhyung Kim <[email protected]>

Commit 21e6d8428664 ("perf diff: Use perf_hpp__register_sort_field interface") changed list_add() to perf_hpp__register_sort_field().

This resulted in a behavior change since the field was added to the tail instead of the head. So the -o option is mostly ignored due to its order in the list.

This patch fixes it by adding perf_hpp__prepend_sort_field().

Cc Peter Zijlstra <[email protected]>
Fixes 21e6d8428664 ("perf diff: Use perf_hpp__register_sort_field interface")
---
 tools/perf/builtin-diff.c | 2 +-
 tools/perf/ui/hist.c      | 6 ++++++
 tools/perf/util/hist.h    | 7 +++++++
 3 files changed, 14 insertions(+), 1 deletion(-)
tools/perf/builtin-diff.c +1 -1
index 9ff0db4e2d0c..933aeec46f4a 100644
@@ -1199,7 +1199,7 @@ static int ui_init(void)
 		BUG_ON(1);
 	}
 
-	perf_hpp__register_sort_field(fmt);
+	perf_hpp__prepend_sort_field(fmt);
 	return 0;
 }
 
tools/perf/ui/hist.c +6 -0
index 4ec79b2f9416..18cfcdc90356 100644
@@ -521,6 +521,12 @@ void perf_hpp_list__register_sort_field(struct perf_hpp_list *list,
 	list_add_tail(&format->sort_list, &list->sorts);
 }
 
+void perf_hpp_list__prepend_sort_field(struct perf_hpp_list *list,
+				       struct perf_hpp_fmt *format)
+{
+	list_add(&format->sort_list, &list->sorts);
+}
+
 void perf_hpp__column_unregister(struct perf_hpp_fmt *format)
 {
 	list_del(&format->list);
tools/perf/util/hist.h +7 -0
index d4b6514eeef5..28c216e3d5b7 100644
@@ -283,6 +283,8 @@ void perf_hpp_list__column_register(struct perf_hpp_list *list,
 				    struct perf_hpp_fmt *format);
 void perf_hpp_list__register_sort_field(struct perf_hpp_list *list,
 					struct perf_hpp_fmt *format);
+void perf_hpp_list__prepend_sort_field(struct perf_hpp_list *list,
+				       struct perf_hpp_fmt *format);
 
 static inline void perf_hpp__column_register(struct perf_hpp_fmt *format)
 {
@@ -294,6 +296,11 @@ static inline void perf_hpp__register_sort_field(struct perf_hpp_fmt *format)
 	perf_hpp_list__register_sort_field(&perf_hpp_list, format);
 }
 
+static inline void perf_hpp__prepend_sort_field(struct perf_hpp_fmt *format)
+{
+	perf_hpp_list__prepend_sort_field(&perf_hpp_list, format);
+}
+
 #define perf_hpp_list__for_each_format(_list, format) \
 	list_for_each_entry(format, &(_list)->fields, list)
 
Signature · 1 line
2.9.3
author maintainer
#3 lore raw

PATCH 3/3 perf callchain: Reference count maps

In reply to Arnaldo Carvalho de Melo

From: Krister Johansen <[email protected]>

If dso__load_kcore frees all of the existing maps, but one has already been attached to a callchain cursor node, then we can get a SIGSEGV in any function that happens to try to use this invalid cursor. Use the existing map refcount mechanism to forestall cleanup of a map until the cursor iterates past the node.

Cc Frederic Weisbecker <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Fixes 84c2cafa2889 ("perf tools: Reference count struct map")
---
 tools/perf/util/callchain.c | 11 +++++++++--
 tools/perf/util/callchain.h |  6 ++++++
 tools/perf/util/hist.c      |  7 +++++++
 3 files changed, 22 insertions(+), 2 deletions(-)
tools/perf/util/callchain.c +9 -2
index 42922512c1c6..8b610dd9e2f6 100644
@@ -437,7 +437,7 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor)
 		}
 		call->ip = cursor_node->ip;
 		call->ms.sym = cursor_node->sym;
-		call->ms.map = cursor_node->map;
+		call->ms.map = map__get(cursor_node->map);
 
 		if (cursor_node->branch) {
 			call->branch_count = 1;
@@ -477,6 +477,7 @@ add_child(struct callchain_node *parent,
 
 		list_for_each_entry_safe(call, tmp, &new->val, list) {
 			list_del(&call->list);
+			map__zput(call->ms.map);
 			free(call);
 		}
 		free(new);
@@ -761,6 +762,7 @@ merge_chain_branch(struct callchain_cursor *cursor,
 					list->ms.map, list->ms.sym,
 					false, NULL, 0, 0);
 		list_del(&list->list);
+		map__zput(list->ms.map);
 		free(list);
 	}
 
@@ -811,7 +813,8 @@ int callchain_cursor_append(struct callchain_cursor *cursor,
 	}
 
 	node->ip = ip;
-	node->map = map;
+	map__zput(node->map);
+	node->map = map__get(map);
 	node->sym = sym;
 	node->branch = branch;
 	node->nr_loop_iter = nr_loop_iter;
@@ -1142,11 +1145,13 @@ static void free_callchain_node(struct callchain_node *node)
 
 	list_for_each_entry_safe(list, tmp, &node->parent_val, list) {
 		list_del(&list->list);
+		map__zput(list->ms.map);
 		free(list);
 	}
 
 	list_for_each_entry_safe(list, tmp, &node->val, list) {
 		list_del(&list->list);
+		map__zput(list->ms.map);
 		free(list);
 	}
 
@@ -1210,6 +1215,7 @@ int callchain_node__make_parent_list(struct callchain_node *node)
 				goto out;
 			*new = *chain;
 			new->has_children = false;
+			map__get(new->ms.map);
 			list_add_tail(&new->list, &head);
 		}
 		parent = parent->parent;
@@ -1230,6 +1236,7 @@ int callchain_node__make_parent_list(struct callchain_node *node)
 out:
 	list_for_each_entry_safe(chain, new, &head, list) {
 		list_del(&chain->list);
+		map__zput(chain->ms.map);
 		free(chain);
 	}
 	return -ENOMEM;
tools/perf/util/callchain.h +6 -0
index 35c8e379530f..4f4b60f1558a 100644
@@ -5,6 +5,7 @@
 #include <linux/list.h>
 #include <linux/rbtree.h>
 #include "event.h"
+#include "map.h"
 #include "symbol.h"
 
 #define HELP_PAD "\t\t\t\t"
@@ -184,8 +185,13 @@ int callchain_merge(struct callchain_cursor *cursor,
  */
 static inline void callchain_cursor_reset(struct callchain_cursor *cursor)
 {
+	struct callchain_cursor_node *node;
+
 	cursor->nr = 0;
 	cursor->last = &cursor->first;
+
+	for (node = cursor->first; node != NULL; node = node->next)
+		map__zput(node->map);
 }
 
 int callchain_cursor_append(struct callchain_cursor *cursor, u64 ip,
tools/perf/util/hist.c +7 -0
index 6770a9645609..7d1b7d33e644 100644
@@ -1,6 +1,7 @@
 #include "util.h"
 #include "build-id.h"
 #include "hist.h"
+#include "map.h"
 #include "session.h"
 #include "sort.h"
 #include "evlist.h"
@@ -1019,6 +1020,10 @@ int hist_entry_iter__add(struct hist_entry_iter *iter, struct addr_location *al,
 			 int max_stack_depth, void *arg)
 {
 	int err, err2;
+	struct map *alm = NULL;
+
+	if (al && al->map)
+		alm = map__get(al->map);
 
 	err = sample__resolve_callchain(iter->sample, &callchain_cursor, &iter->parent,
 					iter->evsel, al, max_stack_depth);
@@ -1058,6 +1063,8 @@ int hist_entry_iter__add(struct hist_entry_iter *iter, struct addr_location *al,
 	if (!err)
 		err = err2;
 
+	map__put(alm);
+
 	return err;
 }
 
Signature · 1 line
2.9.3
#4 lore raw
In reply to Arnaldo Carvalho de Melo
Quoted 40 lines · Arnaldo Carvalho de Melo

Hi Ingo,

Please consider pulling,

- Arnaldo

Test results at the end of this message, as usual.

The following changes since commit fff4b87e594ad3d2e4f51e8d3d86a6f9d3d8b654:

perf/x86/intel/uncore: Make package handling more robust (2017-02-01 08:37:27 +0100)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-urgent-for-mingo-4.10-20170203

for you to fetch changes up to aa33b9b9a2ebb00d33c83a5312d4fbf2d5aeba36:

perf callchain: Reference count maps (2017-02-02 11:39:09 -0300)

----------------------------------------------------------------
perf/urgent fixes:

. Reference count maps in callchains, fixing a SEGFAULT when referencing a
map after it is freed (Krister Johansen)

- Fix segfault on 'perf diff -o N' option (Namhyung Kim)

- Fix 'perf diff -o/--order' option behavior (Namhyung Kim)

----------------------------------------------------------------
Krister Johansen (1):
perf callchain: Reference count maps

Namhyung Kim (2):
perf diff: Fix segfault on 'perf diff -o N' option perf diff: Fix -o/--order option behavior (again)

Quoted 7 lines
 tools/perf/builtin-diff.c   |  2 +-
 tools/perf/ui/hist.c        | 10 ++++++++++
 tools/perf/util/callchain.c | 11 +++++++++--
 tools/perf/util/callchain.h |  6 ++++++
 tools/perf/util/hist.c      |  7 +++++++
 tools/perf/util/hist.h      |  7 +++++++
 6 files changed, 40 insertions(+), 3 deletions(-)

Pulled, thanks a lot Arnaldo!

Ingo