Skip to contentSkip to conversation

perf tools: Add traceevent plugins support

PATCH In review Jiri Olsa 5 participants 70 messages Dec 3, 2013 – Dec 10, 2013

Subsystems (2)
lore
Share & tools
lore

Jiri Olsa

author
#0 lore raw

v3 cover perf tools: Add traceevent plugins support

hi,
backporting traceevent plugin support from trace-cmd.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

Initial RFC is here:
http://marc.info/?l=linux-kernel&m=137803257921043&w=2

Having plugins allow traceevent to properly parse 'print fmt' line of tracepoint format and display proper/additional data in perf script command (check examples below).

Plugins backported from trace-cmd:
plugin_function.c
plugin_hrtimer.c
plugin_jbd2.c
plugin_kmem.c
plugin_kvm.c
plugin_mac80211.c
plugin_sched_switch.c

Plugins newly added:
plugin_cfg80211.c
plugin_scsi.c
plugin_xen.c

v3 changes:
- rebased to current Arnaldo's perf/core
- patches order changed so the build framework gets in before actual plugins
- the perf/core_plugins branch includes also change from Steven, which got acked but is not in yet (it's not included in the patchset, just in branch)
- following patch got in:
perf tools: Include test-stackprotector-all.c in test-all
- factored plugin_function.c stack helpers re/allocation
- indented 'diff' like output in several changelogs

v2 changes:
- removing usage of malloc_or_die from event-pluginc.c
- using static methods in plugins where possible
- added Steven's Signed-off-by tags
- changed traceevent_host_bigendian with code suggested by Steven
- using pevent_print_func_field in hrtimer_start handler
- plugin_function.c changes:
* switching malloc_or_die into malloc
* removing index field from struct func_stack as it's not needed
* renaming get_index into add_and_get_index
* using '%*X' format string capability instead of the loop
- changing pevent_parse_format interface instead
of the pevent_parse_event as requested by Steven
- breakout LIBTRACEEVENT_FLAGS in Makefile.perf as suggested by Namhyung
- added better error handling for filename__read_str suggested by Nmahyung
- using _eprintf code in pr_stat function as suggested by Namhyung
- removing stackprotector feature check as it's not needed and is causing warnings - build failures on old gcc

Changes from RFC:
- added comment about repo/HEAD of the backport source
- omitted options plugin support
- change plugins installation path to:

    $(perfexecdir)/traceevent/plugins

The code is available in here:
git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
perf/core_plugins

thanks,
jirka

Examples (diffs of 'perf script' output):

* data was generated by 'perf record -e 'cfg80211:*' -a' --- script.cfg80211.old
+++ script.cfg80211.new
- ifconfig 2705 [003] 662.896560: cfg80211:cfg80211_tx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[c0, 00,
+ ifconfig 2705 [003] 662.896560: cfg80211:cfg80211_tx_mlme_mgmt: netdev:wlan0(3), ftype:0xc0
- kworker/u16:0 1697 [002] 664.808210: cfg80211:cfg80211_rx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[b0, 00,
+ kworker/u16:0 1697 [002] 664.808210: cfg80211:cfg80211_rx_mlme_mgmt: netdev:wlan0(3), ftype:0xb0

* data was generated by 'perf record -e sched:sched_switch -a' --- script.sched_switch.old
+++ script.sched_switch.new
- perf 577 [002] 30965.311852: sched:sched_switch: prev_comm=perf prev_pid=577 prev_prio=120 prev_state=S ==> next_comm=sw
+ perf 577 [002] 30965.311852: sched:sched_switch: perf:577 [120] S ==> swapper/2:0 [120]

* data was generated by 'perf record -e ftrace:function ls' --- script.function.old
+++ script.function.new
- ls 10781 [001] 32667.291379: ftrace:function: ffffffff811adb80 <-- ffffffff811afc48
- ls 10781 [001] 32667.291379: ftrace:function: ffffffff811b35d0 <-- ffffffff811adb9b
- ls 10781 [001] 32667.291380: ftrace:function: ffffffff811b3520 <-- ffffffff811b35e8
- ls 10781 [001] 32667.291380: ftrace:function: ffffffff811b2720 <-- ffffffff811b3549
- ls 10781 [001] 32667.291381: ftrace:function: ffffffff81297e10 <-- ffffffff811b356c
- ls 10781 [001] 32667.291381: ftrace:function: ffffffff81298f40 <-- ffffffff81297e2c
- ls 10781 [001] 32667.291382: ftrace:function: ffffffff81076160 <-- ffffffff811afbf0
- ls 10781 [001] 32667.291383: ftrace:function: ffffffff811c3eb0 <-- ffffffff811afbfc
- ls 10781 [001] 32667.291383: ftrace:function: ffffffff8164e100 <-- ffffffff811c3ed8
- ls 10781 [001] 32667.291384: ftrace:function: ffffffff811a5d10 <-- ffffffff811c3f53
- ls 10781 [001] 32667.291384: ftrace:function: ffffffff811e8e70 <-- ffffffff811a5d58
- ls 10781 [001] 32667.291385: ftrace:function: ffffffff811f38e0 <-- ffffffff811a5d63
- ls 10781 [001] 32667.291385: ftrace:function: ffffffff811a9ff0 <-- ffffffff811a5d6b
- ls 10781 [001] 32667.291386: ftrace:function: ffffffff811a9fa0 <-- ffffffff811aa015
- ls 10781 [001] 32667.291387: ftrace:function: ffffffff810851c0 <-- ffffffff811aa053
- ls 10781 [001] 32667.291387: ftrace:function: ffffffff81090e00 <-- ffffffff81085211
+ ls 10781 [001] 32667.291379: ftrace:function: would_dump <-- setup_new_exec
+ ls 10781 [001] 32667.291379: ftrace:function: inode_permission <-- would_dump
+ ls 10781 [001] 32667.291380: ftrace:function: __inode_permission <-- inode_permission
+ ls 10781 [001] 32667.291380: ftrace:function: generic_permission <-- __inode_permission
+ ls 10781 [001] 32667.291381: ftrace:function: security_inode_permission <-- __inode_permission
+ ls 10781 [001] 32667.291381: ftrace:function: cap_inode_permission <-- security_inode_permission
+ ls 10781 [001] 32667.291382: ftrace:function: flush_signal_handlers <-- setup_new_exec
+ ls 10781 [001] 32667.291383: ftrace:function: do_close_on_exec <-- setup_new_exec
+ ls 10781 [001] 32667.291383: ftrace:function: _raw_spin_lock <-- do_close_on_exec
+ ls 10781 [001] 32667.291384: ftrace:function: filp_close <-- do_close_on_exec
+ ls 10781 [001] 32667.291384: ftrace:function: dnotify_flush <-- filp_close
+ ls 10781 [001] 32667.291385: ftrace:function: locks_remove_posix <-- filp_close
+ ls 10781 [001] 32667.291385: ftrace:function: fput <-- filp_close
+ ls 10781 [001] 32667.291386: ftrace:function: file_sb_list_del <-- fput
+ ls 10781 [001] 32667.291387: ftrace:function: task_work_add <-- fput
+ ls 10781 [001] 32667.291387: ftrace:function: kick_process <-- task_work_add

* data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a'
--- script.kvm.old
+++ script.kvm.new
qemu-system-x86 17414 [000] 6868.995053: kvm:kvm_exit: reason EPT_VIOLATION rip 0xfff0 info 184 0 qemu-system-x86 17414 [000] 6868.995109: kvm:kvm_emulate_insn: f0000:c46b:e4 71 (real)
- qemu-system-x86 3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=0
+ qemu-system-x86 3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: new sp gfn 0 0/4 q0 direct --- !pge !nxe root 0 sync
- qemu-system-x86 3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root
+ qemu-system-x86 3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: 0/4 q0 direct --- !pge !nxe root 1 sync qemu-system-x86 3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xfff
- insmod 2576 [001] 781.731666: kvmmmu:kvm_mmu_sync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=1
+ insmod 2576 [001] 781.731666: kvmmmu:kvm_mmu_sync_page: 3/0 q0 --- !pge !nxe root 10 unsync
- insmod 2576 [001] 781.731668: kvmmmu:kvm_mmu_unsync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count
+ insmod 2576 [001] 781.731668: kvmmmu:kvm_mmu_unsync_page: 3/0 q0 --- !pge !nxe root 10 unsync

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>

---
Jiri Olsa (28):
perf tools: Remove stackprotector feature check
tools lib traceevent: Add plugin support
tools lib traceevent: Add plugin build support
tools lib traceevent: Add traceevent_host_bigendian function tools lib traceevent: Change pevent_parse_format to include pevent handle tools lib traceevent: Harmonize the install messages in lib-traceevent perf tools: Add build and install plugins targets
perf tools: Add filename__read_str util function
perf tools: Add trace-event object
perf tools: Add trace-event global object for tracepoint interface perf tools: Overload pr_stat traceevent print function
tools lib traceevent: Add jbd2 plugin
tools lib traceevent: Add hrtimer plugin
tools lib traceevent: Add kmem plugin
tools lib traceevent: Add kvm plugin
tools lib traceevent: Add mac80211 plugin
tools lib traceevent: Add sched_switch plugin
tools lib traceevent: Add function plugin
tools lib traceevent: Add xen plugin
tools lib traceevent: Add scsi plugin
tools lib traceevent: Add cfg80211 plugin
tools lib traceevent: Remove malloc_or_die from event-plugin.c tools lib traceevent: Use static functions in jbd2 plugin tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler tools lib traceevent: Several cleanups for function plugin tools lib traceevent: Remove malloc_or_die from plugin_function.c tools lib traceevent: Update kvm plugin with is_writable_pte helper perf tools: Add udis86 disassembler feature check

 tools/lib/traceevent/Makefile                          |  71 +++++++++++--
 tools/lib/traceevent/event-parse.c                     |  56 ++++++----
 tools/lib/traceevent/event-parse.h                     |  19 +++-
 tools/lib/traceevent/event-plugin.c                    | 215 +++++++++++++++++++++++++++++++++++++
 tools/lib/traceevent/plugin_cfg80211.c                 |  24 +++++
 tools/lib/traceevent/plugin_function.c                 | 160 ++++++++++++++++++++++++++++
 tools/lib/traceevent/plugin_hrtimer.c                  |  78 ++++++++++++++
 tools/lib/traceevent/plugin_jbd2.c                     |  68 ++++++++++++
 tools/lib/traceevent/plugin_kmem.c                     |  72 +++++++++++++
 tools/lib/traceevent/plugin_kvm.c                      | 436 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tools/lib/traceevent/plugin_mac80211.c                 |  95 +++++++++++++++++
 tools/lib/traceevent/plugin_sched_switch.c             | 148 ++++++++++++++++++++++++++
 tools/lib/traceevent/plugin_scsi.c                     | 423 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tools/lib/traceevent/plugin_xen.c                      | 130 +++++++++++++++++++++++
 tools/perf/Makefile.perf                               |  18 +++-
 tools/perf/builtin-script.c                            |   2 +-
 tools/perf/builtin-trace.c                             |   5 +-
 tools/perf/config/Makefile                             |  21 +++-
 tools/perf/config/feature-checks/Makefile              |  12 +--
 tools/perf/config/feature-checks/test-all.c            |   5 +
 tools/perf/config/feature-checks/test-stackprotector.c |   6 --
 tools/perf/config/feature-checks/test-udis86.c         |  11 ++
 tools/perf/util/debug.c                                |  30 +++++-
 tools/perf/util/debug.h                                |   2 +
 tools/perf/util/evsel.c                                |  44 +-------
 tools/perf/util/header.c                               |   8 +-
 tools/perf/util/python-ext-sources                     |   1 +
 tools/perf/util/session.h                              |   3 +-
 tools/perf/util/trace-event-parse.c                    |  13 ---
 tools/perf/util/trace-event-read.c                     |  20 ++--
 tools/perf/util/trace-event.c                          |  82 +++++++++++++++
 tools/perf/util/trace-event.h                          |  15 ++-
 tools/perf/util/util.c                                 |  49 +++++++++
 tools/perf/util/util.h                                 |   1 +
 34 files changed, 2214 insertions(+), 129 deletions(-)
 create mode 100644 tools/lib/traceevent/event-plugin.c
 create mode 100644 tools/lib/traceevent/plugin_cfg80211.c
 create mode 100644 tools/lib/traceevent/plugin_function.c
 create mode 100644 tools/lib/traceevent/plugin_hrtimer.c
 create mode 100644 tools/lib/traceevent/plugin_jbd2.c
 create mode 100644 tools/lib/traceevent/plugin_kmem.c
 create mode 100644 tools/lib/traceevent/plugin_kvm.c
 create mode 100644 tools/lib/traceevent/plugin_mac80211.c
 create mode 100644 tools/lib/traceevent/plugin_sched_switch.c
 create mode 100644 tools/lib/traceevent/plugin_scsi.c
 create mode 100644 tools/lib/traceevent/plugin_xen.c
 delete mode 100644 tools/perf/config/feature-checks/test-stackprotector.c
 create mode 100644 tools/perf/config/feature-checks/test-udis86.c
 create mode 100644 tools/perf/util/trace-event.c
author
#1 lore raw

PATCH v3 1/28 perf tools: Remove stackprotector feature check

We use -fstack-protector-all option to enable stack protecting for all available functions. There's no reason for enabling -Wstack-protector to get warning for unprotected functions.

Removing stackprotector feature check which was used to enable the -Wstack-protector option.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/config/Makefile                             | 5 -----
 tools/perf/config/feature-checks/Makefile              | 6 +-----
 tools/perf/config/feature-checks/test-stackprotector.c | 6 ------
 3 files changed, 1 insertion(+), 16 deletions(-)
 delete mode 100644 tools/perf/config/feature-checks/test-stackprotector.c
tools/perf/config/Makefile +0 -5
index 36e66ac..0761d57 100644
@@ -141,7 +141,6 @@ CORE_FEATURE_TESTS =			\
 	libslang			\
 	libunwind			\
 	on-exit				\
-	stackprotector			\
 	stackprotector-all		\
 	timerfd
 
@@ -209,10 +208,6 @@ ifeq ($(feature-stackprotector-all), 1)
   CFLAGS += -fstack-protector-all
 endif
 
-ifeq ($(feature-stackprotector), 1)
-  CFLAGS += -Wstack-protector
-endif
-
 ifeq ($(DEBUG),0)
   ifeq ($(feature-fortify-source), 1)
     CFLAGS += -D_FORTIFY_SOURCE=2
tools/perf/config/feature-checks/Makefile +1 -5
index 87e7900..b8bb749 100644
@@ -26,7 +26,6 @@ FILES=					\
 	test-libunwind-debug-frame	\
 	test-on-exit			\
 	test-stackprotector-all		\
-	test-stackprotector		\
 	test-timerfd
 
 CC := $(CC) -MD
@@ -38,7 +37,7 @@ BUILD = $(CC) $(CFLAGS) $(LDFLAGS) -o $(OUTPUT)$@ [email protected]
 ###############################
 
 test-all:
-	$(BUILD) -Werror -fstack-protector -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl
+	$(BUILD) -Werror -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl
 
 test-hello:
 	$(BUILD)
@@ -46,9 +45,6 @@ test-hello:
 test-stackprotector-all:
 	$(BUILD) -Werror -fstack-protector-all
 
-test-stackprotector:
-	$(BUILD) -Werror -fstack-protector -Wstack-protector
-
 test-fortify-source:
 	$(BUILD) -O2 -Werror -D_FORTIFY_SOURCE=2
 
tools/perf/config/feature-checks/test-stackprotector.c deleted +0 -6
deleted file mode 100644
index c9f398d..0000000
@@ -1,6 +0,0 @@
-#include <stdio.h>
-
-int main(void)
-{
-	return puts("hi");
-}
Signature · 1 line
1.8.3.1
#2 lore raw

[tip:perf/core] perf tools: Remove stackprotector feature check

perf tools: Remove stackprotector feature check
Commit-ID:  e6d7cee1ecac099786d3ba7a71b1e69cf0ae9a7b
Gitweb:     http://git.kernel.org/tip/e6d7cee1ecac099786d3ba7a71b1e69cf0ae9a7b
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:15 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 13:46:36 -0300

perf tools: Remove stackprotector feature check

We use -fstack-protector-all option to enable stack protecting for all available functions. There's no reason for enabling -Wstack-protector to get warning for unprotected functions.

Removing stackprotector feature check which was used to enable the -Wstack-protector option.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/config/Makefile                             | 5 -----
 tools/perf/config/feature-checks/Makefile              | 6 +-----
 tools/perf/config/feature-checks/test-stackprotector.c | 6 ------
 3 files changed, 1 insertion(+), 16 deletions(-)
tools/perf/config/Makefile +0 -5
index 36e66ac..0761d57 100644
@@ -141,7 +141,6 @@ CORE_FEATURE_TESTS =			\
 	libslang			\
 	libunwind			\
 	on-exit				\
-	stackprotector			\
 	stackprotector-all		\
 	timerfd
 
@@ -209,10 +208,6 @@ ifeq ($(feature-stackprotector-all), 1)
   CFLAGS += -fstack-protector-all
 endif
 
-ifeq ($(feature-stackprotector), 1)
-  CFLAGS += -Wstack-protector
-endif
-
 ifeq ($(DEBUG),0)
   ifeq ($(feature-fortify-source), 1)
     CFLAGS += -D_FORTIFY_SOURCE=2
tools/perf/config/feature-checks/Makefile +1 -5
index 87e7900..b8bb749 100644
@@ -26,7 +26,6 @@ FILES=					\
 	test-libunwind-debug-frame	\
 	test-on-exit			\
 	test-stackprotector-all		\
-	test-stackprotector		\
 	test-timerfd
 
 CC := $(CC) -MD
@@ -38,7 +37,7 @@ BUILD = $(CC) $(CFLAGS) $(LDFLAGS) -o $(OUTPUT)$@ [email protected]
 ###############################
 
 test-all:
-	$(BUILD) -Werror -fstack-protector -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl
+	$(BUILD) -Werror -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl
 
 test-hello:
 	$(BUILD)
@@ -46,9 +45,6 @@ test-hello:
 test-stackprotector-all:
 	$(BUILD) -Werror -fstack-protector-all
 
-test-stackprotector:
-	$(BUILD) -Werror -fstack-protector -Wstack-protector
-
 test-fortify-source:
 	$(BUILD) -O2 -Werror -D_FORTIFY_SOURCE=2
 
tools/perf/config/feature-checks/test-stackprotector.c deleted +0 -6
deleted file mode 100644
index c9f398d..0000000
@@ -1,6 +0,0 @@
-#include <stdio.h>
-
-int main(void)
-{
-	return puts("hi");
-}
author
#3 lore raw

PATCH v3 2/28 tools lib traceevent: Add plugin support

In reply to Jiri Olsa

Backporting plugin support for traceevent lib.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

It's now possible to use following interface to load plugins (shared objects) to enhance pevent object functionality.

The plugin interface/hooks are as follows:
(taken from event-parse.h comments)

- 'pevent_plugin_loader' (required)
The function name to initialized the plugin.

int pevent_plugin_loader(struct pevent *pevent)

- 'pevent_plugin_unloader' (optional)
The function called just before unloading

int pevent_plugin_unloader(void)

- 'pevent_plugin_options' (optional)
Plugin options that can be set before loading

struct plugin_option pevent_plugin_options[] = {
{
.name = "option-name",
.plugin_alias = "overide-file-name", (optional) .description = "description of option to show users",
},
{
.name = NULL,
},
};

Array must end with .name = NULL;

The plugin_alias (below) can be used to give a shorter name to access the variable. Useful if a plugin handles more than one event.

NOTE options support is not backported yet.

- 'pevent_plugin_alias' (optional)
The name to use for finding options (uses filename if not defined)

New traceevent functions are added to search and load available plugins:

struct plugin_list* traceevent_load_plugins(struct pevent *pevent)
- loads plusing for 'struct pevent' object and returns loaded plugins list

void traceevent_unload_plugins(struct plugin_list *plugin_list);
- unload plugin list

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile       |   6 +-
 tools/lib/traceevent/event-parse.h  |   5 +
 tools/lib/traceevent/event-plugin.c | 202 ++++++++++++++++++++++++++++++++++++
 3 files changed, 212 insertions(+), 1 deletion(-)
 create mode 100644 tools/lib/traceevent/event-plugin.c
tools/lib/traceevent/Makefile +5 -1
index fc15020..2ccb5bc 100644
@@ -180,7 +180,11 @@ $(obj)/%.o: $(src)/%.c
 %.o: $(src)/%.c
 	$(Q)$(call do_compile)
 
-PEVENT_LIB_OBJS = event-parse.o trace-seq.o parse-filter.o parse-utils.o
+PEVENT_LIB_OBJS  = event-parse.o
+PEVENT_LIB_OBJS += event-plugin.o
+PEVENT_LIB_OBJS += trace-seq.o
+PEVENT_LIB_OBJS += parse-filter.o
+PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
 ALL_OBJS = $(PEVENT_LIB_OBJS)
tools/lib/traceevent/event-parse.h +5 -0
index 8d73d25..a288860 100644
@@ -377,6 +377,11 @@ enum pevent_errno {
 };
 #undef _PE
 
+struct plugin_list;
+
+struct plugin_list *traceevent_load_plugins(struct pevent *pevent);
+void traceevent_unload_plugins(struct plugin_list *plugin_list);
+
 struct cmdline;
 struct cmdline_list;
 struct func_map;
tools/lib/traceevent/event-plugin.c new file +202 -0

202 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#4 lore raw

[tip:perf/core] tools lib traceevent: Add plugin support

tools lib traceevent: Add plugin support
Commit-ID:  c877bbd8eceb14c5eac6779cc804fa8b34044736
Gitweb:     http://git.kernel.org/tip/c877bbd8eceb14c5eac6779cc804fa8b34044736
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:16 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:16:44 -0300

tools lib traceevent: Add plugin support

Backporting plugin support for traceevent lib.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

It's now possible to use following interface to load plugins (shared objects) to enhance pevent object functionality.

The plugin interface/hooks are as follows:
(taken from event-parse.h comments)

- 'pevent_plugin_loader' (required)
The function name to initialized the plugin.

int pevent_plugin_loader(struct pevent *pevent)

- 'pevent_plugin_unloader' (optional)
The function called just before unloading

int pevent_plugin_unloader(void)

- 'pevent_plugin_options' (optional)
Plugin options that can be set before loading

struct plugin_option pevent_plugin_options[] = {
{
.name = "option-name",
.plugin_alias = "overide-file-name", (optional) .description = "description of option to show users",
},
{
.name = NULL,
},
};

Array must end with .name = NULL;

The plugin_alias (below) can be used to give a shorter name to access the variable. Useful if a plugin handles more than one event.

NOTE options support is not backported yet.

- 'pevent_plugin_alias' (optional)
The name to use for finding options (uses filename if not defined)

New traceevent functions are added to search and load available plugins:

struct plugin_list* traceevent_load_plugins(struct pevent *pevent)
- loads plusing for 'struct pevent' object and returns loaded plugins list

void traceevent_unload_plugins(struct plugin_list *plugin_list);
- unload plugin list

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile       |   6 +-
 tools/lib/traceevent/event-parse.h  |   5 +
 tools/lib/traceevent/event-plugin.c | 202 ++++++++++++++++++++++++++++++++++++
 3 files changed, 212 insertions(+), 1 deletion(-)
tools/lib/traceevent/Makefile +5 -1
index fc15020..2ccb5bc 100644
@@ -180,7 +180,11 @@ $(obj)/%.o: $(src)/%.c
 %.o: $(src)/%.c
 	$(Q)$(call do_compile)
 
-PEVENT_LIB_OBJS = event-parse.o trace-seq.o parse-filter.o parse-utils.o
+PEVENT_LIB_OBJS  = event-parse.o
+PEVENT_LIB_OBJS += event-plugin.o
+PEVENT_LIB_OBJS += trace-seq.o
+PEVENT_LIB_OBJS += parse-filter.o
+PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
 ALL_OBJS = $(PEVENT_LIB_OBJS)
tools/lib/traceevent/event-parse.h +5 -0
index 8d73d25..a288860 100644
@@ -377,6 +377,11 @@ enum pevent_errno {
 };
 #undef _PE
 
+struct plugin_list;
+
+struct plugin_list *traceevent_load_plugins(struct pevent *pevent);
+void traceevent_unload_plugins(struct plugin_list *plugin_list);
+
 struct cmdline;
 struct cmdline_list;
 struct func_map;
tools/lib/traceevent/event-plugin.c new file +202 -0

202 lines not on this page · raw message

author
#5 lore raw

PATCH v3 3/28 tools lib traceevent: Add plugin build support

In reply to Jiri Olsa

Backporting missing pieces of plugin building infrastructure:
- adding Makefile 'plugins' target to build all defined plugins
- adding Makefile 'install_plugins' target as 'install_lib' target dependency
- link plugin objects with shared object building

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

Plugins are by default installed into following locations:
'$(HOME)/.traceevent/plugins'
- if we are installing under $(HOME)
'$(prefix)/lib/traceevent/plugins'
- otherwise

This path is propagated to the plugin object as a plugins search path.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile | 52 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 47 insertions(+), 5 deletions(-)
tools/lib/traceevent/Makefile +47 -5
index 2ccb5bc..1526798 100644
@@ -43,6 +43,30 @@ man_dir_SQ = '$(subst ','\'',$(man_dir))'
 export man_dir man_dir_SQ INSTALL
 export DESTDIR DESTDIR_SQ
 
+set_plugin_dir := 1
+
+# Set plugin_dir to preffered global plugin location
+# If we install under $HOME directory we go under
+# $(HOME)/.traceevent/plugins
+#
+# We dont set PLUGIN_DIR in case we install under $HOME
+# directory, because by default the code looks under:
+# $(HOME)/.traceevent/plugins by default.
+#
+ifeq ($(plugin_dir),)
+ifeq ($(prefix),$(HOME))
+override plugin_dir = $(HOME)/.traceevent/plugins
+set_plugin_dir := 0
+else
+override plugin_dir = $(prefix)/lib/traceevent/plugins
+endif
+endif
+
+ifeq ($(set_plugin_dir),1)
+PLUGIN_DIR = -DPLUGIN_DIR="$(DESTDIR)/$(plugin_dir)"
+PLUGIN_DIR_SQ = '$(subst ','\'',$(PLUGIN_DIR))'
+endif
+
 # copy a bit from Linux kbuild
 
 ifeq ("$(origin V)", "command line")
@@ -96,6 +120,7 @@ export prefix bindir src obj
 # Shell quotes
 bindir_SQ = $(subst ','\'',$(bindir))
 bindir_relative_SQ = $(subst ','\'',$(bindir_relative))
+plugin_dir_SQ = $(subst ','\'',$(plugin_dir))
 
 LIB_FILE = libtraceevent.a libtraceevent.so
 
@@ -138,8 +163,8 @@ else
   print_app_build =		echo '  BUILD    '$(OBJ);
   print_fpic_compile =		echo '  CC FPIC  '$(OBJ);
   print_shared_lib_compile =	echo '  BUILD    SHARED LIB '$(OBJ);
-  print_plugin_obj_compile =	echo '  BUILD    PLUGIN OBJ '$(OBJ);
-  print_plugin_build =		echo '  BUILD    PLUGIN     '$(OBJ);
+  print_plugin_obj_compile =	echo '  CC FPIC  '$(OBJ);
+  print_plugin_build =		echo '  BUILD    PLUGIN '$(OBJ);
   print_static_lib_build =	echo '  BUILD    STATIC LIB '$(OBJ);
   print_install =		echo '  INSTALL  '$1'	to	$(DESTDIR_SQ)$2';
 endif
@@ -187,9 +212,11 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
-ALL_OBJS = $(PEVENT_LIB_OBJS)
+PLUGINS := $(PLUGIN_OBJS:.o=.so)
+
+ALL_OBJS = $(PEVENT_LIB_OBJS) $(PLUGIN_OBJS)
 
-CMD_TARGETS = $(LIB_FILE)
+CMD_TARGETS = $(LIB_FILE) $(PLUGINS)
 
 TARGETS = $(CMD_TARGETS)
 
@@ -204,9 +231,17 @@ libtraceevent.so: $(PEVENT_LIB_OBJS)
 libtraceevent.a: $(PEVENT_LIB_OBJS)
 	$(Q)$(do_build_static_lib)
 
+plugins: $(PLUGINS)
+
 $(PEVENT_LIB_OBJS): %.o: $(src)/%.c TRACEEVENT-CFLAGS
 	$(Q)$(do_fpic_compile)
 
+$(PLUGIN_OBJS): %.o : $(src)/%.c
+	$(Q)$(do_compile_plugin_obj)
+
+$(PLUGINS): %.so: %.o
+	$(Q)$(do_plugin_build)
+
 define make_version.h
 	(echo '/* This file is automatically generated. Do not modify. */';		\
 	echo \#define VERSION_CODE $(shell						\
@@ -294,9 +329,16 @@ define do_install
 	$(INSTALL) $1 '$(DESTDIR_SQ)$2'
 endef
 
-install_lib: all_cmd
+install_lib: all_cmd install_plugins
 	$(Q)$(call do_install,$(LIB_FILE),$(bindir_SQ))
 
+PLUGINS_INSTALL = $(subst .so,.install,$(PLUGINS))
+
+$(PLUGINS_INSTALL): %.install : %.so force
+	$(Q)$(call do_install,$<,$(plugin_dir_SQ))
+
+install_plugins: $(PLUGINS_INSTALL)
+
 install: install_lib
 
 clean:
Signature · 1 line
1.8.3.1
#6 lore raw

[tip:perf/core] tools lib traceevent: Add plugin build support

tools lib traceevent: Add plugin build support
Commit-ID:  e0e96d03f004e8953a731053b61e275f276fff01
Gitweb:     http://git.kernel.org/tip/e0e96d03f004e8953a731053b61e275f276fff01
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:17 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:17:16 -0300

tools lib traceevent: Add plugin build support

Backporting missing pieces of plugin building infrastructure:

- Adding Makefile 'plugins' target to build all defined plugins

- Adding Makefile 'install_plugins' target as 'install_lib' target dependency

- Link plugin objects with shared object building

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

Plugins are by default installed into following locations:

'$(HOME)/.traceevent/plugins'
- If we are installing under $(HOME)

'$(prefix)/lib/traceevent/plugins'
- Otherwise

This path is propagated to the plugin object as a plugins search path.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile | 52 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 47 insertions(+), 5 deletions(-)
tools/lib/traceevent/Makefile +47 -5
index 2ccb5bc..1526798 100644
@@ -43,6 +43,30 @@ man_dir_SQ = '$(subst ','\'',$(man_dir))'
 export man_dir man_dir_SQ INSTALL
 export DESTDIR DESTDIR_SQ
 
+set_plugin_dir := 1
+
+# Set plugin_dir to preffered global plugin location
+# If we install under $HOME directory we go under
+# $(HOME)/.traceevent/plugins
+#
+# We dont set PLUGIN_DIR in case we install under $HOME
+# directory, because by default the code looks under:
+# $(HOME)/.traceevent/plugins by default.
+#
+ifeq ($(plugin_dir),)
+ifeq ($(prefix),$(HOME))
+override plugin_dir = $(HOME)/.traceevent/plugins
+set_plugin_dir := 0
+else
+override plugin_dir = $(prefix)/lib/traceevent/plugins
+endif
+endif
+
+ifeq ($(set_plugin_dir),1)
+PLUGIN_DIR = -DPLUGIN_DIR="$(DESTDIR)/$(plugin_dir)"
+PLUGIN_DIR_SQ = '$(subst ','\'',$(PLUGIN_DIR))'
+endif
+
 # copy a bit from Linux kbuild
 
 ifeq ("$(origin V)", "command line")
@@ -96,6 +120,7 @@ export prefix bindir src obj
 # Shell quotes
 bindir_SQ = $(subst ','\'',$(bindir))
 bindir_relative_SQ = $(subst ','\'',$(bindir_relative))
+plugin_dir_SQ = $(subst ','\'',$(plugin_dir))
 
 LIB_FILE = libtraceevent.a libtraceevent.so
 
@@ -138,8 +163,8 @@ else
   print_app_build =		echo '  BUILD    '$(OBJ);
   print_fpic_compile =		echo '  CC FPIC  '$(OBJ);
   print_shared_lib_compile =	echo '  BUILD    SHARED LIB '$(OBJ);
-  print_plugin_obj_compile =	echo '  BUILD    PLUGIN OBJ '$(OBJ);
-  print_plugin_build =		echo '  BUILD    PLUGIN     '$(OBJ);
+  print_plugin_obj_compile =	echo '  CC FPIC  '$(OBJ);
+  print_plugin_build =		echo '  BUILD    PLUGIN '$(OBJ);
   print_static_lib_build =	echo '  BUILD    STATIC LIB '$(OBJ);
   print_install =		echo '  INSTALL  '$1'	to	$(DESTDIR_SQ)$2';
 endif
@@ -187,9 +212,11 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
-ALL_OBJS = $(PEVENT_LIB_OBJS)
+PLUGINS := $(PLUGIN_OBJS:.o=.so)
+
+ALL_OBJS = $(PEVENT_LIB_OBJS) $(PLUGIN_OBJS)
 
-CMD_TARGETS = $(LIB_FILE)
+CMD_TARGETS = $(LIB_FILE) $(PLUGINS)
 
 TARGETS = $(CMD_TARGETS)
 
@@ -204,9 +231,17 @@ libtraceevent.so: $(PEVENT_LIB_OBJS)
 libtraceevent.a: $(PEVENT_LIB_OBJS)
 	$(Q)$(do_build_static_lib)
 
+plugins: $(PLUGINS)
+
 $(PEVENT_LIB_OBJS): %.o: $(src)/%.c TRACEEVENT-CFLAGS
 	$(Q)$(do_fpic_compile)
 
+$(PLUGIN_OBJS): %.o : $(src)/%.c
+	$(Q)$(do_compile_plugin_obj)
+
+$(PLUGINS): %.so: %.o
+	$(Q)$(do_plugin_build)
+
 define make_version.h
 	(echo '/* This file is automatically generated. Do not modify. */';		\
 	echo \#define VERSION_CODE $(shell						\
@@ -294,9 +329,16 @@ define do_install
 	$(INSTALL) $1 '$(DESTDIR_SQ)$2'
 endef
 
-install_lib: all_cmd
+install_lib: all_cmd install_plugins
 	$(Q)$(call do_install,$(LIB_FILE),$(bindir_SQ))
 
+PLUGINS_INSTALL = $(subst .so,.install,$(PLUGINS))
+
+$(PLUGINS_INSTALL): %.install : %.so force
+	$(Q)$(call do_install,$<,$(plugin_dir_SQ))
+
+install_plugins: $(PLUGINS_INSTALL)
+
 install: install_lib
 
 clean:
author
#7 lore raw

PATCH v3 4/28 tools lib traceevent: Add traceevent_host_bigendian function

In reply to Jiri Olsa

Adding traceevent_host_bigendian function to get host endianity. It's used in following patches.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/event-parse.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
tools/lib/traceevent/event-parse.h +10 -0
index a288860..54273c0 100644
@@ -23,6 +23,7 @@
 #include <stdbool.h>
 #include <stdarg.h>
 #include <regex.h>
+#include <string.h>
 
 #ifndef __maybe_unused
 #define __maybe_unused __attribute__((unused))
@@ -527,6 +528,15 @@ __data2host8(struct pevent *pevent, unsigned long long data)
 	__data2host8(pevent, __val);				\
 })
 
+static inline int traceevent_host_bigendian(void)
+{
+	unsigned char str[] = { 0x1, 0x2, 0x3, 0x4 };
+	unsigned int val;
+
+	memcpy(&val, str, 4);
+	return val == 0x01020304;
+}
+
 /* taken from kernel/trace/trace.h */
 enum trace_flag_type {
 	TRACE_FLAG_IRQS_OFF		= 0x01,
Signature · 1 line
1.8.3.1
#8 lore raw

[tip:perf/core] tools lib traceevent: Add traceevent_host_bigendian function

tools lib traceevent: Add traceevent_host_bigendian function
Commit-ID:  91a058ad3856b0a37cb21250eb4c478d0297a1a3
Gitweb:     http://git.kernel.org/tip/91a058ad3856b0a37cb21250eb4c478d0297a1a3
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:18 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:17:58 -0300

tools lib traceevent: Add traceevent_host_bigendian function

Adding traceevent_host_bigendian function to get host endianity. It's used in following patches.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/event-parse.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
tools/lib/traceevent/event-parse.h +10 -0
index a288860..54273c0 100644
@@ -23,6 +23,7 @@
 #include <stdbool.h>
 #include <stdarg.h>
 #include <regex.h>
+#include <string.h>
 
 #ifndef __maybe_unused
 #define __maybe_unused __attribute__((unused))
@@ -527,6 +528,15 @@ __data2host8(struct pevent *pevent, unsigned long long data)
 	__data2host8(pevent, __val);				\
 })
 
+static inline int traceevent_host_bigendian(void)
+{
+	unsigned char str[] = { 0x1, 0x2, 0x3, 0x4 };
+	unsigned int val;
+
+	memcpy(&val, str, 4);
+	return val == 0x01020304;
+}
+
 /* taken from kernel/trace/trace.h */
 enum trace_flag_type {
 	TRACE_FLAG_IRQS_OFF		= 0x01,
author
#9 lore raw

PATCH v3 5/28 tools lib traceevent: Change pevent_parse_format to include pevent handle

In reply to Jiri Olsa

Changing the pevent_parse_format interface to include the pevent handle.

The goal is to always use pevent object when dealing with traceevent library. The reason is that we might need additional processing (like plugins), which is not possible otherwise.

Patches follow to make this happen completely.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/event-parse.c | 56 ++++++++++++++++++++++++--------------
 tools/lib/traceevent/event-parse.h |  4 ++-
 tools/perf/util/evsel.c            |  2 +-
 3 files changed, 39 insertions(+), 23 deletions(-)
tools/lib/traceevent/event-parse.c +35 -21
index 74007ab..22566c2 100644
@@ -5129,8 +5129,38 @@ enum pevent_errno __pevent_parse_format(struct event_format **eventp,
 	return ret;
 }
 
+static enum pevent_errno
+__pevent_parse_event(struct pevent *pevent,
+		     struct event_format **eventp,
+		     const char *buf, unsigned long size,
+		     const char *sys)
+{
+	int ret = __pevent_parse_format(eventp, pevent, buf, size, sys);
+	struct event_format *event = *eventp;
+
+	if (event == NULL)
+		return ret;
+
+	if (pevent && add_event(pevent, event)) {
+		ret = PEVENT_ERRNO__MEM_ALLOC_FAILED;
+		goto event_add_failed;
+	}
+
+#define PRINT_ARGS 0
+	if (PRINT_ARGS && event->print_fmt.args)
+		print_args(event->print_fmt.args);
+
+	return 0;
+
+event_add_failed:
+	pevent_free_format(event);
+	return ret;
+}
+
 /**
  * pevent_parse_format - parse the event format
+ * @pevent: the handle to the pevent
+ * @eventp: returned format
  * @buf: the buffer storing the event format string
  * @size: the size of @buf
  * @sys: the system the event belongs to
@@ -5142,10 +5172,12 @@ enum pevent_errno __pevent_parse_format(struct event_format **eventp,
  *
  * /sys/kernel/debug/tracing/events/.../.../format
  */
-enum pevent_errno pevent_parse_format(struct event_format **eventp, const char *buf,
+enum pevent_errno pevent_parse_format(struct pevent *pevent,
+				      struct event_format **eventp,
+				      const char *buf,
 				      unsigned long size, const char *sys)
 {
-	return __pevent_parse_format(eventp, NULL, buf, size, sys);
+	return __pevent_parse_event(pevent, eventp, buf, size, sys);
 }
 
 /**
@@ -5166,25 +5198,7 @@ enum pevent_errno pevent_parse_event(struct pevent *pevent, const char *buf,
 				     unsigned long size, const char *sys)
 {
 	struct event_format *event = NULL;
-	int ret = __pevent_parse_format(&event, pevent, buf, size, sys);
-
-	if (event == NULL)
-		return ret;
-
-	if (add_event(pevent, event)) {
-		ret = PEVENT_ERRNO__MEM_ALLOC_FAILED;
-		goto event_add_failed;
-	}
-
-#define PRINT_ARGS 0
-	if (PRINT_ARGS && event->print_fmt.args)
-		print_args(event->print_fmt.args);
-
-	return 0;
-
-event_add_failed:
-	pevent_free_format(event);
-	return ret;
+	return __pevent_parse_event(pevent, &event, buf, size, sys);
 }
 
 #undef _PE
tools/lib/traceevent/event-parse.h +3 -1
index 54273c0..620c27a 100644
@@ -562,7 +562,9 @@ int pevent_parse_header_page(struct pevent *pevent, char *buf, unsigned long siz
 
 enum pevent_errno pevent_parse_event(struct pevent *pevent, const char *buf,
 				     unsigned long size, const char *sys);
-enum pevent_errno pevent_parse_format(struct event_format **eventp, const char *buf,
+enum pevent_errno pevent_parse_format(struct pevent *pevent,
+				      struct event_format **eventp,
+				      const char *buf,
 				      unsigned long size, const char *sys);
 void pevent_free_format(struct event_format *event);
 
tools/perf/util/evsel.c +1 -1
index b5fe7f9..6a046ed 100644
@@ -210,7 +210,7 @@ struct event_format *event_format__new(const char *sys, const char *name)
 		size += n;
 	} while (n > 0);
 
-	pevent_parse_format(&format, bf, size, sys);
+	pevent_parse_format(NULL, &format, bf, size, sys);
 
 out_free_bf:
 	free(bf);
Signature · 1 line
1.8.3.1
#10 lore raw

[tip:perf/core] tools lib traceevent: Change pevent_parse_format to include pevent handle

tools lib traceevent: Change pevent_parse_format to include pevent handle
Commit-ID:  71ad9583ff15d2e95cb8c67a5c963321d9d1877b
Gitweb:     http://git.kernel.org/tip/71ad9583ff15d2e95cb8c67a5c963321d9d1877b
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:19 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:18:42 -0300

tools lib traceevent: Change pevent_parse_format to include pevent handle

Changing the pevent_parse_format interface to include the pevent handle.

The goal is to always use pevent object when dealing with traceevent library. The reason is that we might need additional processing (like plugins), which is not possible otherwise.

Patches follow to make this happen completely.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/event-parse.c | 56 ++++++++++++++++++++++++--------------
 tools/lib/traceevent/event-parse.h |  4 ++-
 tools/perf/util/evsel.c            |  2 +-
 3 files changed, 39 insertions(+), 23 deletions(-)
tools/lib/traceevent/event-parse.c +35 -21
index 900fca0..9849873 100644
@@ -5121,8 +5121,38 @@ enum pevent_errno __pevent_parse_format(struct event_format **eventp,
 	return ret;
 }
 
+static enum pevent_errno
+__pevent_parse_event(struct pevent *pevent,
+		     struct event_format **eventp,
+		     const char *buf, unsigned long size,
+		     const char *sys)
+{
+	int ret = __pevent_parse_format(eventp, pevent, buf, size, sys);
+	struct event_format *event = *eventp;
+
+	if (event == NULL)
+		return ret;
+
+	if (pevent && add_event(pevent, event)) {
+		ret = PEVENT_ERRNO__MEM_ALLOC_FAILED;
+		goto event_add_failed;
+	}
+
+#define PRINT_ARGS 0
+	if (PRINT_ARGS && event->print_fmt.args)
+		print_args(event->print_fmt.args);
+
+	return 0;
+
+event_add_failed:
+	pevent_free_format(event);
+	return ret;
+}
+
 /**
  * pevent_parse_format - parse the event format
+ * @pevent: the handle to the pevent
+ * @eventp: returned format
  * @buf: the buffer storing the event format string
  * @size: the size of @buf
  * @sys: the system the event belongs to
@@ -5134,10 +5164,12 @@ enum pevent_errno __pevent_parse_format(struct event_format **eventp,
  *
  * /sys/kernel/debug/tracing/events/.../.../format
  */
-enum pevent_errno pevent_parse_format(struct event_format **eventp, const char *buf,
+enum pevent_errno pevent_parse_format(struct pevent *pevent,
+				      struct event_format **eventp,
+				      const char *buf,
 				      unsigned long size, const char *sys)
 {
-	return __pevent_parse_format(eventp, NULL, buf, size, sys);
+	return __pevent_parse_event(pevent, eventp, buf, size, sys);
 }
 
 /**
@@ -5158,25 +5190,7 @@ enum pevent_errno pevent_parse_event(struct pevent *pevent, const char *buf,
 				     unsigned long size, const char *sys)
 {
 	struct event_format *event = NULL;
-	int ret = __pevent_parse_format(&event, pevent, buf, size, sys);
-
-	if (event == NULL)
-		return ret;
-
-	if (add_event(pevent, event)) {
-		ret = PEVENT_ERRNO__MEM_ALLOC_FAILED;
-		goto event_add_failed;
-	}
-
-#define PRINT_ARGS 0
-	if (PRINT_ARGS && event->print_fmt.args)
-		print_args(event->print_fmt.args);
-
-	return 0;
-
-event_add_failed:
-	pevent_free_format(event);
-	return ret;
+	return __pevent_parse_event(pevent, &event, buf, size, sys);
 }
 
 #undef _PE
tools/lib/traceevent/event-parse.h +3 -1
index 54273c0..620c27a 100644
@@ -562,7 +562,9 @@ int pevent_parse_header_page(struct pevent *pevent, char *buf, unsigned long siz
 
 enum pevent_errno pevent_parse_event(struct pevent *pevent, const char *buf,
 				     unsigned long size, const char *sys);
-enum pevent_errno pevent_parse_format(struct event_format **eventp, const char *buf,
+enum pevent_errno pevent_parse_format(struct pevent *pevent,
+				      struct event_format **eventp,
+				      const char *buf,
 				      unsigned long size, const char *sys);
 void pevent_free_format(struct event_format *event);
 
tools/perf/util/evsel.c +1 -1
index b5fe7f9..6a046ed 100644
@@ -210,7 +210,7 @@ struct event_format *event_format__new(const char *sys, const char *name)
 		size += n;
 	} while (n > 0);
 
-	pevent_parse_format(&format, bf, size, sys);
+	pevent_parse_format(NULL, &format, bf, size, sys);
 
 out_free_bf:
 	free(bf);
author
#11 lore raw

PATCH v3 6/28 tools lib traceevent: Harmonize the install messages in lib-traceevent

In reply to Jiri Olsa

Removing the 'to ...' part out of the install message, because it does not fit to the rest of the build messages we use.

Before:

  INSTALL  plugin_hrtimer.so    to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_jbd2.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kmem.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kvm.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_mac80211.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_sched_switch.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_function.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_xen.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_scsi.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins

Now:

  INSTALL  plugin_jbd2.so
  INSTALL  plugin_hrtimer.so
  INSTALL  plugin_kmem.so
  INSTALL  plugin_kvm.so
  INSTALL  plugin_mac80211.so
  INSTALL  plugin_sched_switch.so
  INSTALL  plugin_function.so
  INSTALL  plugin_xen.so
  INSTALL  plugin_scsi.so
Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/lib/traceevent/Makefile +1 -1
index 1526798..f946851 100644
@@ -166,7 +166,7 @@ else
   print_plugin_obj_compile =	echo '  CC FPIC  '$(OBJ);
   print_plugin_build =		echo '  BUILD    PLUGIN '$(OBJ);
   print_static_lib_build =	echo '  BUILD    STATIC LIB '$(OBJ);
-  print_install =		echo '  INSTALL  '$1'	to	$(DESTDIR_SQ)$2';
+  print_install =		echo '  INSTALL  '$1;
 endif
 
 do_fpic_compile =					\
Signature · 1 line
1.8.3.1
#12 lore raw

[tip:perf/core] tools lib traceevent: Harmonize the install messages in lib-traceevent

tools lib traceevent: Harmonize the install messages in lib-traceevent
Commit-ID:  f33c5cd73172c4fb81daa83a4f6d322210172fa0
Gitweb:     http://git.kernel.org/tip/f33c5cd73172c4fb81daa83a4f6d322210172fa0
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:20 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:19:09 -0300

tools lib traceevent: Harmonize the install messages in lib-traceevent

Removing the 'to ...' part out of the install message, because it does not fit to the rest of the build messages we use.

Before:

  INSTALL  plugin_hrtimer.so    to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_jbd2.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kmem.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kvm.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_mac80211.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_sched_switch.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_function.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_xen.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_scsi.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins

Now:

  INSTALL  plugin_jbd2.so
  INSTALL  plugin_hrtimer.so
  INSTALL  plugin_kmem.so
  INSTALL  plugin_kvm.so
  INSTALL  plugin_mac80211.so
  INSTALL  plugin_sched_switch.so
  INSTALL  plugin_function.so
  INSTALL  plugin_xen.so
  INSTALL  plugin_scsi.so
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/lib/traceevent/Makefile +1 -1
index 1526798..f946851 100644
@@ -166,7 +166,7 @@ else
   print_plugin_obj_compile =	echo '  CC FPIC  '$(OBJ);
   print_plugin_build =		echo '  BUILD    PLUGIN '$(OBJ);
   print_static_lib_build =	echo '  BUILD    STATIC LIB '$(OBJ);
-  print_install =		echo '  INSTALL  '$1'	to	$(DESTDIR_SQ)$2';
+  print_install =		echo '  INSTALL  '$1;
 endif
 
 do_fpic_compile =					\
author
#13 lore raw

PATCH v3 7/28 perf tools: Add build and install plugins targets

In reply to Jiri Olsa

Adding 'plugins' target along with the libtraceevent.a, so plugins are built together with traceevent library.

Adding 'install-traceevent-plugins' Makefile install target, instructing perf to install plugins into:
$(HOME)/.traceevent/plugins
- if installed localy under $HOME
$(DESTDIR)/$(prefix)/$(libdir)/traceevent/plugins
- if installed globally

Examples:

  $ make install
  ...
  $ find ~/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/plugin_mac80211.so
  /home/jolsa/.traceevent/plugins/plugin_kvm.so
  /home/jolsa/.traceevent/plugins/plugin_scsi.so
  /home/jolsa/.traceevent/plugins/plugin_sched_switch.so
  /home/jolsa/.traceevent/plugins/plugin_xen.so
  /home/jolsa/.traceevent/plugins/plugin_cfg80211.so
  /home/jolsa/.traceevent/plugins/plugin_function.so
  /home/jolsa/.traceevent/plugins/plugin_kmem.so
  /home/jolsa/.traceevent/plugins/plugin_hrtimer.so
  /home/jolsa/.traceevent/plugins/plugin_jbd2.so
  $ sudo make install DESTDIR=/opt/perf/
  ...
  $ find /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/plugin_kvm.so
  /opt/perf/lib64/traceevent/plugins/plugin_scsi.so
  /opt/perf/lib64/traceevent/plugins/plugin_mac80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_hrtimer.so
  /opt/perf/lib64/traceevent/plugins/plugin_kmem.so
  /opt/perf/lib64/traceevent/plugins/plugin_jbd2.so
  /opt/perf/lib64/traceevent/plugins/plugin_sched_switch.so
  /opt/perf/lib64/traceevent/plugins/plugin_function.so
  /opt/perf/lib64/traceevent/plugins/plugin_cfg80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_xen.so
Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/Makefile.perf   | 17 ++++++++++++-----
 tools/perf/config/Makefile |  8 ++++++++
 2 files changed, 20 insertions(+), 5 deletions(-)
tools/perf/Makefile.perf +12 -5
index e416ccc..abc7ac3 100644
@@ -105,7 +105,7 @@ ifeq ($(config),1)
 include config/Makefile
 endif
 
-export prefix bindir sharedir sysconfdir
+export prefix bindir sharedir sysconfdir DESTDIR
 
 # sparse is architecture-neutral, which means that we need to tell it
 # explicitly what architecture to check for. Fix this up for yours..
@@ -710,13 +710,20 @@ $(LIB_FILE): $(LIB_OBJS)
 # libtraceevent.a
 TE_SOURCES = $(wildcard $(TRACE_EVENT_DIR)*.[ch])
 
-$(LIBTRACEEVENT): $(TE_SOURCES)
-	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(QUIET_SUBDIR1) O=$(OUTPUT) CFLAGS="-g -Wall $(EXTRA_CFLAGS)" libtraceevent.a
+LIBTRACEEVENT_FLAGS  = $(QUIET_SUBDIR1) O=$(OUTPUT)
+LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS)"
+LIBTRACEEVENT_FLAGS += plugin_dir=$(plugindir_SQ)
+
+$(LIBTRACEEVENT): $(TE_SOURCES) $(OUTPUT)PERF-CFLAGS
+	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(LIBTRACEEVENT_FLAGS) libtraceevent.a plugins
 
 $(LIBTRACEEVENT)-clean:
 	$(call QUIET_CLEAN, libtraceevent)
 	@$(MAKE) -C $(TRACE_EVENT_DIR) O=$(OUTPUT) clean >/dev/null
 
+install-traceevent-plugins:
+	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(LIBTRACEEVENT_FLAGS) install_plugins
+
 LIBLK_SOURCES = $(wildcard $(LK_PATH)*.[ch])
 
 # if subdir is set, we've been called from above so target has been built
@@ -785,7 +792,7 @@ cscope:
 
 ### Detect prefix changes
 TRACK_CFLAGS = $(subst ','\'',$(CFLAGS)):\
-             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ)
+             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ):$(plugindir_SQ)
 
 $(OUTPUT)PERF-CFLAGS: .FORCE-PERF-CFLAGS
 	@FLAGS='$(TRACK_CFLAGS)'; \
@@ -849,7 +856,7 @@ endif
 		$(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/tests/attr'; \
 		$(INSTALL) tests/attr/* '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/tests/attr'
 
-install: install-bin try-install-man
+install: install-bin try-install-man install-traceevent-plugins
 
 install-python_ext:
 	$(PYTHON_WORD) util/setup.py --quiet install --root='/$(DESTDIR_SQ)'
tools/perf/config/Makefile +8 -0
index 0761d57..bae1072 100644
@@ -593,3 +593,11 @@ else
 perfexec_instdir = $(prefix)/$(perfexecdir)
 endif
 perfexec_instdir_SQ = $(subst ','\'',$(perfexec_instdir))
+
+# If we install to $(HOME) we keep the traceevent default:
+# $(HOME)/.traceevent/plugins
+# Otherwise we install plugins into the global $(libdir).
+ifdef DESTDIR
+plugindir=$(libdir)/traceevent/plugins
+plugindir_SQ= $(subst ','\'',$(prefix)/$(plugindir))
+endif
Signature · 1 line
1.8.3.1
#14 lore raw

[tip:perf/core] perf tools: Add build and install plugins targets

perf tools: Add build and install plugins targets
Commit-ID:  3d7c0144491bd8c21d53b43032274a85efdfe434
Gitweb:     http://git.kernel.org/tip/3d7c0144491bd8c21d53b43032274a85efdfe434
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:21 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:19:35 -0300

perf tools: Add build and install plugins targets

Adding 'plugins' target along with the libtraceevent.a, so plugins are built together with traceevent library.

Adding 'install-traceevent-plugins' Makefile install target, instructing perf to install plugins into:

$(HOME)/.traceevent/plugins
- If installed localy under $HOME

$(DESTDIR)/$(prefix)/$(libdir)/traceevent/plugins
- If installed globally

Examples:

  $ make install
  ...
  $ find ~/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/plugin_mac80211.so
  /home/jolsa/.traceevent/plugins/plugin_kvm.so
  /home/jolsa/.traceevent/plugins/plugin_scsi.so
  /home/jolsa/.traceevent/plugins/plugin_sched_switch.so
  /home/jolsa/.traceevent/plugins/plugin_xen.so
  /home/jolsa/.traceevent/plugins/plugin_cfg80211.so
  /home/jolsa/.traceevent/plugins/plugin_function.so
  /home/jolsa/.traceevent/plugins/plugin_kmem.so
  /home/jolsa/.traceevent/plugins/plugin_hrtimer.so
  /home/jolsa/.traceevent/plugins/plugin_jbd2.so
  $ sudo make install DESTDIR=/opt/perf/
  ...
  $ find /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/plugin_kvm.so
  /opt/perf/lib64/traceevent/plugins/plugin_scsi.so
  /opt/perf/lib64/traceevent/plugins/plugin_mac80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_hrtimer.so
  /opt/perf/lib64/traceevent/plugins/plugin_kmem.so
  /opt/perf/lib64/traceevent/plugins/plugin_jbd2.so
  /opt/perf/lib64/traceevent/plugins/plugin_sched_switch.so
  /opt/perf/lib64/traceevent/plugins/plugin_function.so
  /opt/perf/lib64/traceevent/plugins/plugin_cfg80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_xen.so
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/Makefile.perf   | 17 ++++++++++++-----
 tools/perf/config/Makefile |  8 ++++++++
 2 files changed, 20 insertions(+), 5 deletions(-)
tools/perf/Makefile.perf +12 -5
index e416ccc..abc7ac3 100644
@@ -105,7 +105,7 @@ ifeq ($(config),1)
 include config/Makefile
 endif
 
-export prefix bindir sharedir sysconfdir
+export prefix bindir sharedir sysconfdir DESTDIR
 
 # sparse is architecture-neutral, which means that we need to tell it
 # explicitly what architecture to check for. Fix this up for yours..
@@ -710,13 +710,20 @@ $(LIB_FILE): $(LIB_OBJS)
 # libtraceevent.a
 TE_SOURCES = $(wildcard $(TRACE_EVENT_DIR)*.[ch])
 
-$(LIBTRACEEVENT): $(TE_SOURCES)
-	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(QUIET_SUBDIR1) O=$(OUTPUT) CFLAGS="-g -Wall $(EXTRA_CFLAGS)" libtraceevent.a
+LIBTRACEEVENT_FLAGS  = $(QUIET_SUBDIR1) O=$(OUTPUT)
+LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS)"
+LIBTRACEEVENT_FLAGS += plugin_dir=$(plugindir_SQ)
+
+$(LIBTRACEEVENT): $(TE_SOURCES) $(OUTPUT)PERF-CFLAGS
+	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(LIBTRACEEVENT_FLAGS) libtraceevent.a plugins
 
 $(LIBTRACEEVENT)-clean:
 	$(call QUIET_CLEAN, libtraceevent)
 	@$(MAKE) -C $(TRACE_EVENT_DIR) O=$(OUTPUT) clean >/dev/null
 
+install-traceevent-plugins:
+	$(QUIET_SUBDIR0)$(TRACE_EVENT_DIR) $(LIBTRACEEVENT_FLAGS) install_plugins
+
 LIBLK_SOURCES = $(wildcard $(LK_PATH)*.[ch])
 
 # if subdir is set, we've been called from above so target has been built
@@ -785,7 +792,7 @@ cscope:
 
 ### Detect prefix changes
 TRACK_CFLAGS = $(subst ','\'',$(CFLAGS)):\
-             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ)
+             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ):$(plugindir_SQ)
 
 $(OUTPUT)PERF-CFLAGS: .FORCE-PERF-CFLAGS
 	@FLAGS='$(TRACK_CFLAGS)'; \
@@ -849,7 +856,7 @@ endif
 		$(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/tests/attr'; \
 		$(INSTALL) tests/attr/* '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/tests/attr'
 
-install: install-bin try-install-man
+install: install-bin try-install-man install-traceevent-plugins
 
 install-python_ext:
 	$(PYTHON_WORD) util/setup.py --quiet install --root='/$(DESTDIR_SQ)'
tools/perf/config/Makefile +8 -0
index 0761d57..bae1072 100644
@@ -593,3 +593,11 @@ else
 perfexec_instdir = $(prefix)/$(perfexecdir)
 endif
 perfexec_instdir_SQ = $(subst ','\'',$(perfexec_instdir))
+
+# If we install to $(HOME) we keep the traceevent default:
+# $(HOME)/.traceevent/plugins
+# Otherwise we install plugins into the global $(libdir).
+ifdef DESTDIR
+plugindir=$(libdir)/traceevent/plugins
+plugindir_SQ= $(subst ','\'',$(prefix)/$(plugindir))
+endif
author
#15 lore raw

PATCH v3 8/28 perf tools: Add filename__read_str util function

In reply to Jiri Olsa

Adding filename__read_str util function to read text file and return it in the char array.

The interface is:
int filename__read_str(const char *filename, char **buf, size_t *sizep)

Returns 0/-1 if the read suceeded/fail respectively.

buf - place to store the data pointer size - place to store data size

v2 change:
- better error handling suggested by Namhyung Kim.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/util/util.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/util.h |  1 +
 2 files changed, 50 insertions(+)
tools/perf/util/util.c +49 -0
index b1d5376..bae8756 100644
@@ -6,6 +6,8 @@
 #endif
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
+#include <errno.h>
 #include <linux/kernel.h>
 
 /*
@@ -433,3 +435,50 @@ int filename__read_int(const char *filename, int *value)
 	close(fd);
 	return err;
 }
+
+int filename__read_str(const char *filename, char **buf, size_t *sizep)
+{
+	size_t size = 0, alloc_size = 0;
+	void *bf = NULL, *nbf;
+	int fd, n, err = 0;
+
+	fd = open(filename, O_RDONLY);
+	if (fd < 0)
+		return -errno;
+
+	do {
+		if (size == alloc_size) {
+			alloc_size += BUFSIZ;
+			nbf = realloc(bf, alloc_size);
+			if (!nbf) {
+				err = -ENOMEM;
+				break;
+			}
+
+			bf = nbf;
+		}
+
+		n = read(fd, bf + size, alloc_size - size);
+		if (n < 0) {
+			if (size) {
+				pr_warning("read failed %d: %s\n",
+					   errno, strerror(errno));
+				err = 0;
+			} else
+				err = -errno;
+
+			break;
+		}
+
+		size += n;
+	} while (n > 0);
+
+	if (!err) {
+		*sizep = size;
+		*buf   = bf;
+	} else
+		free(bf);
+
+	close(fd);
+	return err;
+}
tools/perf/util/util.h +1 -0
index ce0f73d..adb39f2 100644
@@ -308,4 +308,5 @@ char *get_srcline(struct dso *dso, unsigned long addr);
 void free_srcline(char *srcline);
 
 int filename__read_int(const char *filename, int *value);
+int filename__read_str(const char *filename, char **buf, size_t *sizep);
 #endif /* GIT_COMPAT_UTIL_H */
Signature · 1 line
1.8.3.1
#16 lore raw

[tip:perf/core] perf tools: Add filename__read_str util function

perf tools: Add filename__read_str util function
Commit-ID:  cef82c9f5a3deab94c1f1f02cfc25213844852ed
Gitweb:     http://git.kernel.org/tip/cef82c9f5a3deab94c1f1f02cfc25213844852ed
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:22 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:20:12 -0300

perf tools: Add filename__read_str util function

Adding filename__read_str util function to read text file and return it in the char array.

The interface is:
int filename__read_str(const char *filename, char **buf, size_t *sizep)

Returns 0/-1 if the read suceeded/fail respectively.

buf - place to store the data pointer size - place to store data size

v2 change:
- better error handling suggested by Namhyung Kim.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/util/util.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/util.h |  1 +
 2 files changed, 50 insertions(+)
tools/perf/util/util.c +49 -0
index b1d5376..bae8756 100644
@@ -6,6 +6,8 @@
 #endif
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
+#include <errno.h>
 #include <linux/kernel.h>
 
 /*
@@ -433,3 +435,50 @@ int filename__read_int(const char *filename, int *value)
 	close(fd);
 	return err;
 }
+
+int filename__read_str(const char *filename, char **buf, size_t *sizep)
+{
+	size_t size = 0, alloc_size = 0;
+	void *bf = NULL, *nbf;
+	int fd, n, err = 0;
+
+	fd = open(filename, O_RDONLY);
+	if (fd < 0)
+		return -errno;
+
+	do {
+		if (size == alloc_size) {
+			alloc_size += BUFSIZ;
+			nbf = realloc(bf, alloc_size);
+			if (!nbf) {
+				err = -ENOMEM;
+				break;
+			}
+
+			bf = nbf;
+		}
+
+		n = read(fd, bf + size, alloc_size - size);
+		if (n < 0) {
+			if (size) {
+				pr_warning("read failed %d: %s\n",
+					   errno, strerror(errno));
+				err = 0;
+			} else
+				err = -errno;
+
+			break;
+		}
+
+		size += n;
+	} while (n > 0);
+
+	if (!err) {
+		*sizep = size;
+		*buf   = bf;
+	} else
+		free(bf);
+
+	close(fd);
+	return err;
+}
tools/perf/util/util.h +1 -0
index ce0f73d..adb39f2 100644
@@ -308,4 +308,5 @@ char *get_srcline(struct dso *dso, unsigned long addr);
 void free_srcline(char *srcline);
 
 int filename__read_int(const char *filename, int *value);
+int filename__read_str(const char *filename, char **buf, size_t *sizep);
 #endif /* GIT_COMPAT_UTIL_H */
author
#17 lore raw

PATCH v3 9/28 perf tools: Add trace-event object

In reply to Jiri Olsa

Add trace-event object to keep together 'struct pevent' object with its loaded plugins with following interface:

int trace_event__init(struct trace_event *t);
- initalizes 'struct pevent' object and loads plugins for it

void trace_event__cleanup(struct trace_event *t);
- cleanups both 'struct pevent' and plugins

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/Makefile.perf            |  1 +
 tools/perf/builtin-script.c         |  2 +-
 tools/perf/util/header.c            |  8 ++++----
 tools/perf/util/python-ext-sources  |  1 +
 tools/perf/util/session.h           |  3 ++-
 tools/perf/util/trace-event-parse.c | 13 -------------
 tools/perf/util/trace-event-read.c  | 20 +++++++++++---------
 tools/perf/util/trace-event.c       | 21 +++++++++++++++++++++
 tools/perf/util/trace-event.h       | 13 ++++++++++---
 9 files changed, 51 insertions(+), 31 deletions(-)
 create mode 100644 tools/perf/util/trace-event.c
tools/perf/Makefile.perf +1 -0
index abc7ac3..ca3b87d 100644
@@ -353,6 +353,7 @@ LIB_OBJS += $(OUTPUT)util/pmu-bison.o
 LIB_OBJS += $(OUTPUT)util/trace-event-read.o
 LIB_OBJS += $(OUTPUT)util/trace-event-info.o
 LIB_OBJS += $(OUTPUT)util/trace-event-scripting.o
+LIB_OBJS += $(OUTPUT)util/trace-event.o
 LIB_OBJS += $(OUTPUT)util/svghelper.o
 LIB_OBJS += $(OUTPUT)util/sort.o
 LIB_OBJS += $(OUTPUT)util/hist.o
tools/perf/builtin-script.c +1 -1
index 952dce9..715766d 100644
@@ -1785,7 +1785,7 @@ int cmd_script(int argc, const char **argv, const char *prefix __maybe_unused)
 			return -1;
 		}
 
-		err = scripting_ops->generate_script(session->pevent,
+		err = scripting_ops->generate_script(session->tevent.pevent,
 						     "perf-script");
 		goto out;
 	}
tools/perf/util/header.c +4 -4
index 3e755f2..125cdc9 100644
@@ -2834,11 +2834,11 @@ int perf_session__read_header(struct perf_session *session)
 
 	symbol_conf.nr_events = nr_attrs;
 
-	perf_header__process_sections(header, fd, &session->pevent,
+	perf_header__process_sections(header, fd, &session->tevent,
 				      perf_file_section__process);
 
 	if (perf_evlist__prepare_tracepoint_events(session->evlist,
-						   session->pevent))
+						   session->tevent.pevent))
 		goto out_delete_evlist;
 
 	return 0;
@@ -3003,7 +3003,7 @@ int perf_event__process_tracing_data(struct perf_tool *tool __maybe_unused,
 	lseek(fd, offset + sizeof(struct tracing_data_event),
 	      SEEK_SET);
 
-	size_read = trace_report(fd, &session->pevent,
+	size_read = trace_report(fd, &session->tevent,
 				 session->repipe);
 	padding = PERF_ALIGN(size_read, sizeof(u64)) - size_read;
 
@@ -3025,7 +3025,7 @@ int perf_event__process_tracing_data(struct perf_tool *tool __maybe_unused,
 	}
 
 	perf_evlist__prepare_tracepoint_events(session->evlist,
-					       session->pevent);
+					       session->tevent.pevent);
 
 	return size_read + padding;
 }
tools/perf/util/python-ext-sources +1 -0
index 239036f..595bfc7 100644
@@ -18,4 +18,5 @@ util/cgroup.c
 util/rblist.c
 util/strlist.c
 util/fs.c
+util/trace-event.c
 ../../lib/rbtree.c
tools/perf/util/session.h +2 -1
index 50f6409..83cba9b 100644
@@ -1,6 +1,7 @@
 #ifndef __PERF_SESSION_H
 #define __PERF_SESSION_H
 
+#include "trace-event.h"
 #include "hist.h"
 #include "event.h"
 #include "header.h"
@@ -32,7 +33,7 @@ struct perf_session {
 	struct perf_header	header;
 	struct machines		machines;
 	struct perf_evlist	*evlist;
-	struct pevent		*pevent;
+	struct trace_event	tevent;
 	struct events_stats	stats;
 	bool			repipe;
 	struct ordered_samples	ordered_samples;
tools/perf/util/trace-event-parse.c +0 -13
index 6681f71..e0d6d07f 100644
@@ -28,19 +28,6 @@
 #include "util.h"
 #include "trace-event.h"
 
-struct pevent *read_trace_init(int file_bigendian, int host_bigendian)
-{
-	struct pevent *pevent = pevent_alloc();
-
-	if (pevent != NULL) {
-		pevent_set_flag(pevent, PEVENT_NSEC_OUTPUT);
-		pevent_set_file_bigendian(pevent, file_bigendian);
-		pevent_set_host_bigendian(pevent, host_bigendian);
-	}
-
-	return pevent;
-}
-
 static int get_common_field(struct scripting_context *context,
 			    int *offset, int *size, const char *type)
 {
tools/perf/util/trace-event-read.c +11 -9
index f211227..e113e18 100644
@@ -343,7 +343,7 @@ static int read_event_files(struct pevent *pevent)
 	return 0;
 }
 
-ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
+ssize_t trace_report(int fd, struct trace_event *tevent, bool __repipe)
 {
 	char buf[BUFSIZ];
 	char test[] = { 23, 8, 68 };
@@ -356,11 +356,9 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 	int host_bigendian;
 	int file_long_size;
 	int file_page_size;
-	struct pevent *pevent;
+	struct pevent *pevent = NULL;
 	int err;
 
-	*ppevent = NULL;
-
 	repipe = __repipe;
 	input_fd = fd;
 
@@ -390,12 +388,17 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 	file_bigendian = buf[0];
 	host_bigendian = bigendian();
 
-	pevent = read_trace_init(file_bigendian, host_bigendian);
-	if (pevent == NULL) {
-		pr_debug("read_trace_init failed");
+	if (trace_event__init(tevent)) {
+		pr_debug("trace_event__init failed");
 		goto out;
 	}
 
+	pevent = tevent->pevent;
+
+	pevent_set_flag(pevent, PEVENT_NSEC_OUTPUT);
+	pevent_set_file_bigendian(pevent, file_bigendian);
+	pevent_set_host_bigendian(pevent, host_bigendian);
+
 	if (do_read(buf, 1) < 0)
 		goto out;
 	file_long_size = buf[0];
@@ -432,11 +435,10 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 		pevent_print_printk(pevent);
 	}
 
-	*ppevent = pevent;
 	pevent = NULL;
 
 out:
 	if (pevent)
-		pevent_free(pevent);
+		trace_event__cleanup(tevent);
 	return size;
 }
tools/perf/util/trace-event.c new file +21 -0
new file mode 100644
index 0000000..a155a77
@@ -0,0 +1,21 @@
+
+#include <traceevent/event-parse.h>
+#include "trace-event.h"
+
+int trace_event__init(struct trace_event *t)
+{
+	struct pevent *pevent = pevent_alloc();
+
+	if (pevent) {
+		t->plugin_list = traceevent_load_plugins(pevent);
+		t->pevent  = pevent;
+	}
+
+	return pevent ? 0 : -1;
+}
+
+void trace_event__cleanup(struct trace_event *t)
+{
+	pevent_free(t->pevent);
+	traceevent_unload_plugins(t->plugin_list);
+}
tools/perf/util/trace-event.h +10 -3
index 04df631..75dded3 100644
@@ -3,17 +3,24 @@
 
 #include <traceevent/event-parse.h>
 #include "parse-events.h"
-#include "session.h"
 
 struct machine;
 struct perf_sample;
 union perf_event;
 struct perf_tool;
 struct thread;
+struct plugin_list;
+
+struct trace_event {
+	struct pevent		*pevent;
+	struct plugin_list	*plugin_list;
+};
+
+int trace_event__init(struct trace_event *t);
+void trace_event__cleanup(struct trace_event *t);
 
 int bigendian(void);
 
-struct pevent *read_trace_init(int file_bigendian, int host_bigendian);
 void event_format__print(struct event_format *event,
 			 int cpu, void *data, int size);
 
@@ -27,7 +34,7 @@ raw_field_value(struct event_format *event, const char *name, void *data);
 void parse_proc_kallsyms(struct pevent *pevent, char *file, unsigned int size);
 void parse_ftrace_printk(struct pevent *pevent, char *file, unsigned int size);
 
-ssize_t trace_report(int fd, struct pevent **pevent, bool repipe);
+ssize_t trace_report(int fd, struct trace_event *tevent, bool repipe);
 
 struct event_format *trace_find_next_event(struct pevent *pevent,
 					   struct event_format *event);
Signature · 1 line
1.8.3.1
#18 lore raw

[tip:perf/core] perf tools: Add trace-event object

perf tools: Add trace-event object
Commit-ID:  29f5ffd3d3ff9abdfc98928f7c4cb4097cefe434
Gitweb:     http://git.kernel.org/tip/29f5ffd3d3ff9abdfc98928f7c4cb4097cefe434
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:23 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:20:52 -0300

perf tools: Add trace-event object

Add trace-event object to keep together 'struct pevent' object with its loaded plugins with following interface:

int trace_event__init(struct trace_event *t);

- Initalizes 'struct pevent' object and loads plugins for it

void trace_event__cleanup(struct trace_event *t);

- Cleanups both 'struct pevent' and plugins

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/Makefile.perf            |  1 +
 tools/perf/builtin-script.c         |  2 +-
 tools/perf/util/header.c            |  8 ++++----
 tools/perf/util/python-ext-sources  |  1 +
 tools/perf/util/session.h           |  3 ++-
 tools/perf/util/trace-event-parse.c | 13 -------------
 tools/perf/util/trace-event-read.c  | 20 +++++++++++---------
 tools/perf/util/trace-event.c       | 21 +++++++++++++++++++++
 tools/perf/util/trace-event.h       | 13 ++++++++++---
 9 files changed, 51 insertions(+), 31 deletions(-)
tools/perf/Makefile.perf +1 -0
index abc7ac3..ca3b87d 100644
@@ -353,6 +353,7 @@ LIB_OBJS += $(OUTPUT)util/pmu-bison.o
 LIB_OBJS += $(OUTPUT)util/trace-event-read.o
 LIB_OBJS += $(OUTPUT)util/trace-event-info.o
 LIB_OBJS += $(OUTPUT)util/trace-event-scripting.o
+LIB_OBJS += $(OUTPUT)util/trace-event.o
 LIB_OBJS += $(OUTPUT)util/svghelper.o
 LIB_OBJS += $(OUTPUT)util/sort.o
 LIB_OBJS += $(OUTPUT)util/hist.o
tools/perf/builtin-script.c +1 -1
index c555bdd..4484886 100644
@@ -1786,7 +1786,7 @@ int cmd_script(int argc, const char **argv, const char *prefix __maybe_unused)
 			return -1;
 		}
 
-		err = scripting_ops->generate_script(session->pevent,
+		err = scripting_ops->generate_script(session->tevent.pevent,
 						     "perf-script");
 		goto out;
 	}
tools/perf/util/header.c +4 -4
index 3e755f2..125cdc9 100644
@@ -2834,11 +2834,11 @@ int perf_session__read_header(struct perf_session *session)
 
 	symbol_conf.nr_events = nr_attrs;
 
-	perf_header__process_sections(header, fd, &session->pevent,
+	perf_header__process_sections(header, fd, &session->tevent,
 				      perf_file_section__process);
 
 	if (perf_evlist__prepare_tracepoint_events(session->evlist,
-						   session->pevent))
+						   session->tevent.pevent))
 		goto out_delete_evlist;
 
 	return 0;
@@ -3003,7 +3003,7 @@ int perf_event__process_tracing_data(struct perf_tool *tool __maybe_unused,
 	lseek(fd, offset + sizeof(struct tracing_data_event),
 	      SEEK_SET);
 
-	size_read = trace_report(fd, &session->pevent,
+	size_read = trace_report(fd, &session->tevent,
 				 session->repipe);
 	padding = PERF_ALIGN(size_read, sizeof(u64)) - size_read;
 
@@ -3025,7 +3025,7 @@ int perf_event__process_tracing_data(struct perf_tool *tool __maybe_unused,
 	}
 
 	perf_evlist__prepare_tracepoint_events(session->evlist,
-					       session->pevent);
+					       session->tevent.pevent);
 
 	return size_read + padding;
 }
tools/perf/util/python-ext-sources +1 -0
index 239036f..595bfc7 100644
@@ -18,4 +18,5 @@ util/cgroup.c
 util/rblist.c
 util/strlist.c
 util/fs.c
+util/trace-event.c
 ../../lib/rbtree.c
tools/perf/util/session.h +2 -1
index 7e5d4398..004d3e8 100644
@@ -1,6 +1,7 @@
 #ifndef __PERF_SESSION_H
 #define __PERF_SESSION_H
 
+#include "trace-event.h"
 #include "hist.h"
 #include "event.h"
 #include "header.h"
@@ -32,7 +33,7 @@ struct perf_session {
 	struct perf_header	header;
 	struct machines		machines;
 	struct perf_evlist	*evlist;
-	struct pevent		*pevent;
+	struct trace_event	tevent;
 	struct events_stats	stats;
 	bool			repipe;
 	struct ordered_samples	ordered_samples;
tools/perf/util/trace-event-parse.c +0 -13
index 6681f71..e0d6d07f 100644
@@ -28,19 +28,6 @@
 #include "util.h"
 #include "trace-event.h"
 
-struct pevent *read_trace_init(int file_bigendian, int host_bigendian)
-{
-	struct pevent *pevent = pevent_alloc();
-
-	if (pevent != NULL) {
-		pevent_set_flag(pevent, PEVENT_NSEC_OUTPUT);
-		pevent_set_file_bigendian(pevent, file_bigendian);
-		pevent_set_host_bigendian(pevent, host_bigendian);
-	}
-
-	return pevent;
-}
-
 static int get_common_field(struct scripting_context *context,
 			    int *offset, int *size, const char *type)
 {
tools/perf/util/trace-event-read.c +11 -9
index f211227..e113e18 100644
@@ -343,7 +343,7 @@ static int read_event_files(struct pevent *pevent)
 	return 0;
 }
 
-ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
+ssize_t trace_report(int fd, struct trace_event *tevent, bool __repipe)
 {
 	char buf[BUFSIZ];
 	char test[] = { 23, 8, 68 };
@@ -356,11 +356,9 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 	int host_bigendian;
 	int file_long_size;
 	int file_page_size;
-	struct pevent *pevent;
+	struct pevent *pevent = NULL;
 	int err;
 
-	*ppevent = NULL;
-
 	repipe = __repipe;
 	input_fd = fd;
 
@@ -390,12 +388,17 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 	file_bigendian = buf[0];
 	host_bigendian = bigendian();
 
-	pevent = read_trace_init(file_bigendian, host_bigendian);
-	if (pevent == NULL) {
-		pr_debug("read_trace_init failed");
+	if (trace_event__init(tevent)) {
+		pr_debug("trace_event__init failed");
 		goto out;
 	}
 
+	pevent = tevent->pevent;
+
+	pevent_set_flag(pevent, PEVENT_NSEC_OUTPUT);
+	pevent_set_file_bigendian(pevent, file_bigendian);
+	pevent_set_host_bigendian(pevent, host_bigendian);
+
 	if (do_read(buf, 1) < 0)
 		goto out;
 	file_long_size = buf[0];
@@ -432,11 +435,10 @@ ssize_t trace_report(int fd, struct pevent **ppevent, bool __repipe)
 		pevent_print_printk(pevent);
 	}
 
-	*ppevent = pevent;
 	pevent = NULL;
 
 out:
 	if (pevent)
-		pevent_free(pevent);
+		trace_event__cleanup(tevent);
 	return size;
 }
tools/perf/util/trace-event.c new file +21 -0
new file mode 100644
index 0000000..a155a77
@@ -0,0 +1,21 @@
+
+#include <traceevent/event-parse.h>
+#include "trace-event.h"
+
+int trace_event__init(struct trace_event *t)
+{
+	struct pevent *pevent = pevent_alloc();
+
+	if (pevent) {
+		t->plugin_list = traceevent_load_plugins(pevent);
+		t->pevent  = pevent;
+	}
+
+	return pevent ? 0 : -1;
+}
+
+void trace_event__cleanup(struct trace_event *t)
+{
+	pevent_free(t->pevent);
+	traceevent_unload_plugins(t->plugin_list);
+}
tools/perf/util/trace-event.h +10 -3
index 04df631..75dded3 100644
@@ -3,17 +3,24 @@
 
 #include <traceevent/event-parse.h>
 #include "parse-events.h"
-#include "session.h"
 
 struct machine;
 struct perf_sample;
 union perf_event;
 struct perf_tool;
 struct thread;
+struct plugin_list;
+
+struct trace_event {
+	struct pevent		*pevent;
+	struct plugin_list	*plugin_list;
+};
+
+int trace_event__init(struct trace_event *t);
+void trace_event__cleanup(struct trace_event *t);
 
 int bigendian(void);
 
-struct pevent *read_trace_init(int file_bigendian, int host_bigendian);
 void event_format__print(struct event_format *event,
 			 int cpu, void *data, int size);
 
@@ -27,7 +34,7 @@ raw_field_value(struct event_format *event, const char *name, void *data);
 void parse_proc_kallsyms(struct pevent *pevent, char *file, unsigned int size);
 void parse_ftrace_printk(struct pevent *pevent, char *file, unsigned int size);
 
-ssize_t trace_report(int fd, struct pevent **pevent, bool repipe);
+ssize_t trace_report(int fd, struct trace_event *tevent, bool repipe);
 
 struct event_format *trace_find_next_event(struct pevent *pevent,
 					   struct event_format *event);
author
#19 lore raw

PATCH v3 10/28 perf tools: Add trace-event global object for tracepoint interface

In reply to Jiri Olsa

In order to get the proper plugins processing we need to use full trace-event interface when creating tracepoint events. So far we were using shortcut to get the parsed format.

Moving current 'event_format__new' function into trace-event object as 'trace_event__tp_format'.

This function uses properly initialized global trace-event object, ensuring proper plugins processing.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/builtin-trace.c    |  5 ++--
 tools/perf/util/evsel.c       | 44 ++-----------------------------
 tools/perf/util/trace-event.c | 61 +++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/trace-event.h |  2 ++
 4 files changed, 68 insertions(+), 44 deletions(-)
tools/perf/builtin-trace.c +3 -2
index 9f2a242..56afe33 100644
@@ -11,6 +11,7 @@
 #include "util/intlist.h"
 #include "util/thread_map.h"
 #include "util/stat.h"
+#include "trace-event.h"
 
 #include <libaudit.h>
 #include <stdlib.h>
@@ -1430,11 +1431,11 @@ static int trace__read_syscall_info(struct trace *trace, int id)
 	sc->fmt  = syscall_fmt__find(sc->name);
 
 	snprintf(tp_name, sizeof(tp_name), "sys_enter_%s", sc->name);
-	sc->tp_format = event_format__new("syscalls", tp_name);
+	sc->tp_format = trace_event__tp_format("syscalls", tp_name);
 
 	if (sc->tp_format == NULL && sc->fmt && sc->fmt->alias) {
 		snprintf(tp_name, sizeof(tp_name), "sys_enter_%s", sc->fmt->alias);
-		sc->tp_format = event_format__new("syscalls", tp_name);
+		sc->tp_format = trace_event__tp_format("syscalls", tp_name);
 	}
 
 	if (sc->tp_format == NULL)
tools/perf/util/evsel.c +2 -42

62 lines not on this page · raw message

tools/perf/util/trace-event.c +61 -0

70 lines not on this page · raw message

tools/perf/util/trace-event.h +2 -0
index 75dded3..3a01618 100644
@@ -18,6 +18,8 @@ struct trace_event {
 
 int trace_event__init(struct trace_event *t);
 void trace_event__cleanup(struct trace_event *t);
+struct event_format*
+trace_event__tp_format(const char *sys, const char *name);
 
 int bigendian(void);
 
Signature · 1 line
1.8.3.1
#20 lore raw

[tip:perf/core] perf tools: Add trace-event global object for tracepoint interface

perf tools: Add trace-event global object for tracepoint interface
Commit-ID:  97978b3e30a512beba43720365f53fea8d432c97
Gitweb:     http://git.kernel.org/tip/97978b3e30a512beba43720365f53fea8d432c97
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:24 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:21:25 -0300

perf tools: Add trace-event global object for tracepoint interface

In order to get the proper plugins processing we need to use full trace-event interface when creating tracepoint events. So far we were using shortcut to get the parsed format.

Moving current 'event_format__new' function into trace-event object as 'trace_event__tp_format'.

This function uses properly initialized global trace-event object, ensuring proper plugins processing.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/builtin-trace.c    |  5 ++--
 tools/perf/util/evsel.c       | 44 ++-----------------------------
 tools/perf/util/trace-event.c | 61 +++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/trace-event.h |  2 ++
 4 files changed, 68 insertions(+), 44 deletions(-)
tools/perf/builtin-trace.c +3 -2
index 9f2a242..56afe33 100644
@@ -11,6 +11,7 @@
 #include "util/intlist.h"
 #include "util/thread_map.h"
 #include "util/stat.h"
+#include "trace-event.h"
 
 #include <libaudit.h>
 #include <stdlib.h>
@@ -1430,11 +1431,11 @@ static int trace__read_syscall_info(struct trace *trace, int id)
 	sc->fmt  = syscall_fmt__find(sc->name);
 
 	snprintf(tp_name, sizeof(tp_name), "sys_enter_%s", sc->name);
-	sc->tp_format = event_format__new("syscalls", tp_name);
+	sc->tp_format = trace_event__tp_format("syscalls", tp_name);
 
 	if (sc->tp_format == NULL && sc->fmt && sc->fmt->alias) {
 		snprintf(tp_name, sizeof(tp_name), "sys_enter_%s", sc->fmt->alias);
-		sc->tp_format = event_format__new("syscalls", tp_name);
+		sc->tp_format = trace_event__tp_format("syscalls", tp_name);
 	}
 
 	if (sc->tp_format == NULL)
tools/perf/util/evsel.c +2 -42

62 lines not on this page · raw message

tools/perf/util/trace-event.c +61 -0

70 lines not on this page · raw message

tools/perf/util/trace-event.h +2 -0
index 75dded3..3a01618 100644
@@ -18,6 +18,8 @@ struct trace_event {
 
 int trace_event__init(struct trace_event *t);
 void trace_event__cleanup(struct trace_event *t);
+struct event_format*
+trace_event__tp_format(const char *sys, const char *name);
 
 int bigendian(void);
 
author
#21 lore raw

PATCH v3 11/28 perf tools: Overload pr_stat traceevent print function

In reply to Jiri Olsa

The traceevent lib uses pr_stat to display all standard info. It's defined as __weak. Overloading it with perf version plugged into perf output system logic.

Displaying the pr_stat stuff under '-v' option.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/util/debug.c | 30 +++++++++++++++++++++++++++---
 tools/perf/util/debug.h |  2 ++
 2 files changed, 29 insertions(+), 3 deletions(-)
tools/perf/util/debug.c +27 -3
index 399e74c..8640a91 100644
@@ -16,13 +16,11 @@
 int verbose;
 bool dump_trace = false, quiet = false;
 
-int eprintf(int level, const char *fmt, ...)
+static int _eprintf(int level, const char *fmt, va_list args)
 {
-	va_list args;
 	int ret = 0;
 
 	if (verbose >= level) {
-		va_start(args, fmt);
 		if (use_browser >= 1)
 			ui_helpline__vshow(fmt, args);
 		else
@@ -33,6 +31,32 @@ int eprintf(int level, const char *fmt, ...)
 	return ret;
 }
 
+int eprintf(int level, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = _eprintf(level, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+/*
+ * Overloading libtraceevent standard info print
+ * function, display with -v in perf.
+ */
+void pr_stat(const char *fmt, ...)
+{
+	va_list args;
+
+	va_start(args, fmt);
+	_eprintf(1, fmt, args);
+	va_end(args);
+	eprintf(1, "\n");
+}
+
 int dump_printf(const char *fmt, ...)
 {
 	va_list args;
tools/perf/util/debug.h +2 -0
index efbd988..443694c 100644
@@ -17,4 +17,6 @@ void trace_event(union perf_event *event);
 int ui__error(const char *format, ...) __attribute__((format(printf, 1, 2)));
 int ui__warning(const char *format, ...) __attribute__((format(printf, 1, 2)));
 
+void pr_stat(const char *fmt, ...);
+
 #endif	/* __PERF_DEBUG_H */
Signature · 1 line
1.8.3.1
#22 lore raw

[tip:perf/core] perf tools: Overload pr_stat traceevent print function

perf tools: Overload pr_stat traceevent print function
Commit-ID:  f772abc63462a48df605259fbe1af3fe9b223926
Gitweb:     http://git.kernel.org/tip/f772abc63462a48df605259fbe1af3fe9b223926
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:25 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:23:55 -0300

perf tools: Overload pr_stat traceevent print function

The traceevent lib uses pr_stat to display all standard info. It's defined as __weak. Overloading it with perf version plugged into perf output system logic.

Displaying the pr_stat stuff under '-v' option.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/perf/util/debug.c | 30 +++++++++++++++++++++++++++---
 tools/perf/util/debug.h |  2 ++
 2 files changed, 29 insertions(+), 3 deletions(-)
tools/perf/util/debug.c +27 -3
index 399e74c..8640a91 100644
@@ -16,13 +16,11 @@
 int verbose;
 bool dump_trace = false, quiet = false;
 
-int eprintf(int level, const char *fmt, ...)
+static int _eprintf(int level, const char *fmt, va_list args)
 {
-	va_list args;
 	int ret = 0;
 
 	if (verbose >= level) {
-		va_start(args, fmt);
 		if (use_browser >= 1)
 			ui_helpline__vshow(fmt, args);
 		else
@@ -33,6 +31,32 @@ int eprintf(int level, const char *fmt, ...)
 	return ret;
 }
 
+int eprintf(int level, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = _eprintf(level, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+/*
+ * Overloading libtraceevent standard info print
+ * function, display with -v in perf.
+ */
+void pr_stat(const char *fmt, ...)
+{
+	va_list args;
+
+	va_start(args, fmt);
+	_eprintf(1, fmt, args);
+	va_end(args);
+	eprintf(1, "\n");
+}
+
 int dump_printf(const char *fmt, ...)
 {
 	va_list args;
tools/perf/util/debug.h +2 -0
index efbd988..443694c 100644
@@ -17,4 +17,6 @@ void trace_event(union perf_event *event);
 int ui__error(const char *format, ...) __attribute__((format(printf, 1, 2)));
 int ui__warning(const char *format, ...) __attribute__((format(printf, 1, 2)));
 
+void pr_stat(const char *fmt, ...);
+
 #endif	/* __PERF_DEBUG_H */
author
#23 lore raw

PATCH v3 12/28 tools lib traceevent: Add jbd2 plugin

In reply to Jiri Olsa

Backporting jbd2 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following tracepoint events:
jbd2:jbd2_checkpoint_stats jbd2:jbd2_run_stats

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'jbd2:jbd2_run_stats,jbd2:jbd2_checkpoint_stats' -a')

--- script.jbd2.old
+++ script.jbd2.new

  -     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bbb chp_time=0x0 forced_to_close=0 written=0 dropped=2
  +     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: dev 253,3 tid 4139963 chp_time 0 forced_to_close 0 written 0 dropped 2
  -     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bc0 wait=0x0 request_delay=0x0 running=0x138d locked=0x0 flushing=0x0 logging=0x68 handle_count=2995 blocks=17 blocks_logged=18
  +     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: dev 253,3 tid 4139968 wait 0 request_delay 0 running 5005 locked 0 flushing 0 logging 104 handle_count 2995 blocks 17 blocks_logged 18

This plugin also adds jbd2_dev_to_name function, which was removed from jdb2 tracepoints recently. Keeping it in for backward compatibility reasons.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile      |  2 ++
 tools/lib/traceevent/plugin_jbd2.c | 66 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 68 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_jbd2.c
tools/lib/traceevent/Makefile +2 -0
index f946851..6c65400 100644
@@ -212,6 +212,8 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
+PLUGIN_OBJS = plugin_jbd2.o
+
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
 ALL_OBJS = $(PEVENT_LIB_OBJS) $(PLUGIN_OBJS)
tools/lib/traceevent/plugin_jbd2.c new file +66 -0

66 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#24 lore raw

[tip:perf/core] tools lib traceevent: Add jbd2 plugin

tools lib traceevent: Add jbd2 plugin
Commit-ID:  e0549f11e8d57c1a606773342b3b71c74f9986eb
Gitweb:     http://git.kernel.org/tip/e0549f11e8d57c1a606773342b3b71c74f9986eb
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:26 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:24:47 -0300

tools lib traceevent: Add jbd2 plugin

Backporting jbd2 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following tracepoint events:

jbd2:jbd2_checkpoint_stats
jbd2:jbd2_run_stats

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'jbd2:jbd2_run_stats,jbd2:jbd2_checkpoint_stats' -a')

--- script.jbd2.old
+++ script.jbd2.new

  -     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bbb chp_time=0x0 forced_to_close=0 written=0 dropped=2
  +     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: dev 253,3 tid 4139963 chp_time 0 forced_to_close 0 written 0 dropped 2
  -     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bc0 wait=0x0 request_delay=0x0 running=0x138d locked=0x0 flushing=0x0 logging=0x68 handle_count=2995 blocks=17 blocks_logged=18
  +     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: dev 253,3 tid 4139968 wait 0 request_delay 0 running 5005 locked 0 flushing 0 logging 104 handle_count 2995 blocks 17 blocks_logged 18

This plugin also adds jbd2_dev_to_name function, which was removed from jdb2 tracepoints recently. Keeping it in for backward compatibility reasons.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile      |  2 ++
 tools/lib/traceevent/plugin_jbd2.c | 66 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 68 insertions(+)
tools/lib/traceevent/Makefile +2 -0
index f946851..6c65400 100644
@@ -212,6 +212,8 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
+PLUGIN_OBJS = plugin_jbd2.o
+
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
 ALL_OBJS = $(PEVENT_LIB_OBJS) $(PLUGIN_OBJS)
tools/lib/traceevent/plugin_jbd2.c new file +66 -0

66 lines not on this page · raw message

author
#25 lore raw

PATCH v3 13/28 tools lib traceevent: Add hrtimer plugin

In reply to Jiri Olsa

Backporting hrtimer plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function field resolving for following tracepoint events:
timer:hrtimer_expire_entry timer:hrtimer_start

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'timer:hrtimer*' -a')

--- script.hrtimer.old
+++ script.hrtimer.new

  -         swapper     0 [000] 27405.519092: timer:hrtimer_start: [FAILED TO PARSE] hrtimer=0xffff88021e20e800 function=0xffffffff810c0e10 expires=27398383000000 softexpires=27398383000000
  +         swapper     0 [000] 27405.519103: timer:hrtimer_start: hrtimer=0xffff88021e20e800 function=tick_sched_timer expires=27398383000000 softexpires=27398383000000
  -         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: [FAILED TO PARSE] hrtimer=0xffff880211334058 now=27398294182491 function=0xffffffff81086f20
  +         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: hrtimer=0xffff880211334058 now=27398294182491 function=posix_timer_fn/0x0

Check the 'function' field is translated into the function name.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile         |  3 +-
 tools/lib/traceevent/plugin_hrtimer.c | 93 +++++++++++++++++++++++++++++++++++
 2 files changed, 95 insertions(+), 1 deletion(-)
 create mode 100644 tools/lib/traceevent/plugin_hrtimer.c
tools/lib/traceevent/Makefile +2 -1
index 6c65400..9ff2e25 100644
@@ -212,7 +212,8 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
-PLUGIN_OBJS = plugin_jbd2.o
+PLUGIN_OBJS  = plugin_jbd2.o
+PLUGIN_OBJS += plugin_hrtimer.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_hrtimer.c new file +93 -0

93 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#26 lore raw

[tip:perf/core] tools lib traceevent: Add hrtimer plugin

tools lib traceevent: Add hrtimer plugin
Commit-ID:  d9d13f8be6237950753e093683e9f415f31eccbc
Gitweb:     http://git.kernel.org/tip/d9d13f8be6237950753e093683e9f415f31eccbc
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:27 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:25:41 -0300

tools lib traceevent: Add hrtimer plugin

Backporting hrtimer plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function field resolving for following tracepoint events:

timer:hrtimer_expire_entry
timer:hrtimer_start

The diff of 'perf script' output generated by old and new code: (data was generated by 'perf record -e 'timer:hrtimer*' -a')

--- script.hrtimer.old
+++ script.hrtimer.new

  -         swapper     0 [000] 27405.519092: timer:hrtimer_start: [FAILED TO PARSE] hrtimer=0xffff88021e20e800 function=0xffffffff810c0e10 expires=27398383000000 softexpires=27398383000000
  +         swapper     0 [000] 27405.519103: timer:hrtimer_start: hrtimer=0xffff88021e20e800 function=tick_sched_timer expires=27398383000000 softexpires=27398383000000
  -         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: [FAILED TO PARSE] hrtimer=0xffff880211334058 now=27398294182491 function=0xffffffff81086f20
  +         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: hrtimer=0xffff880211334058 now=27398294182491 function=posix_timer_fn/0x0

Check the 'function' field is translated into the function name.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile         |  3 +-
 tools/lib/traceevent/plugin_hrtimer.c | 93 +++++++++++++++++++++++++++++++++++
 2 files changed, 95 insertions(+), 1 deletion(-)
tools/lib/traceevent/Makefile +2 -1
index 6c65400..9ff2e25 100644
@@ -212,7 +212,8 @@ PEVENT_LIB_OBJS += parse-filter.o
 PEVENT_LIB_OBJS += parse-utils.o
 PEVENT_LIB_OBJS += kbuffer-parse.o
 
-PLUGIN_OBJS = plugin_jbd2.o
+PLUGIN_OBJS  = plugin_jbd2.o
+PLUGIN_OBJS += plugin_hrtimer.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_hrtimer.c new file +93 -0

93 lines not on this page · raw message

author
#27 lore raw

PATCH v3 14/28 tools lib traceevent: Add kmem plugin

In reply to Jiri Olsa

Backporting kmem plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds call_site field resolving for following tracepoint events:
kmem:kfree
kmem:kmalloc
kmem:kmalloc_node
kmem:kmem_cache_alloc kmem:kmem_cache_alloc_node kmem:kmem_cache_free

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kmem:*' -a')

--- script.kmem.old
+++ script.kmem.new

  -            perf 27846 [001] 29643.403319: kmem:kfree: call_site=ffffffff810e64f6 ptr=(nil)
  +            perf 27846 [001] 29643.403238: kmem:kfree: (__audit_syscall_exit+0x1f6) call_site=ffffffff810e64f6 ptr=(nil)
  -            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  +            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: (radix_tree_preload+0x35) call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  -            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  +            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: (jbd2_journal_stop+0x221) call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  -         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  +         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: (__alloc_skb+0x4e) call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  -            perf 27846 [001] 29643.445510: kmem:kmalloc: call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  +            perf 27846 [001] 29643.445510: kmem:kmalloc: (ext4_ext_find_extent+0x362) call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  -               X   888 [002] 29643.445546: kmem:kmalloc_node: call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1
  +               X   888 [002] 29643.445546: kmem:kmalloc_node: (__alloc_skb+0x7e) call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1
Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile      |  1 +
 tools/lib/traceevent/plugin_kmem.c | 72 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 73 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_kmem.c
tools/lib/traceevent/Makefile +1 -0
index 9ff2e25..245d6b4 100644
@@ -214,6 +214,7 @@ PEVENT_LIB_OBJS += kbuffer-parse.o
 
 PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
+PLUGIN_OBJS += plugin_kmem.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_kmem.c new file +72 -0

72 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#28 lore raw

[tip:perf/core] tools lib traceevent: Add kmem plugin

tools lib traceevent: Add kmem plugin
Commit-ID:  f825628840e581f92f1c2e79dfa943f29c054e67
Gitweb:     http://git.kernel.org/tip/f825628840e581f92f1c2e79dfa943f29c054e67
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:28 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:26:05 -0300

tools lib traceevent: Add kmem plugin

Backporting kmem plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds call_site field resolving for following tracepoint events:

kmem:kfree
kmem:kmalloc
kmem:kmalloc_node
kmem:kmem_cache_alloc
kmem:kmem_cache_alloc_node
kmem:kmem_cache_free

The diff of 'perf script' output generated by old and new code: (data was generated by 'perf record -e 'kmem:*' -a')

--- script.kmem.old
+++ script.kmem.new

  -            perf 27846 [001] 29643.403319: kmem:kfree: call_site=ffffffff810e64f6 ptr=(nil)
  +            perf 27846 [001] 29643.403238: kmem:kfree: (__audit_syscall_exit+0x1f6) call_site=ffffffff810e64f6 ptr=(nil)
  -            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  +            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: (radix_tree_preload+0x35) call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  -            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  +            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: (jbd2_journal_stop+0x221) call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  -         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  +         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: (__alloc_skb+0x4e) call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  -            perf 27846 [001] 29643.445510: kmem:kmalloc: call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  +            perf 27846 [001] 29643.445510: kmem:kmalloc: (ext4_ext_find_extent+0x362) call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  -               X   888 [002] 29643.445546: kmem:kmalloc_node: call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1
  +               X   888 [002] 29643.445546: kmem:kmalloc_node: (__alloc_skb+0x7e) call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile      |  1 +
 tools/lib/traceevent/plugin_kmem.c | 72 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 73 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 9ff2e25..245d6b4 100644
@@ -214,6 +214,7 @@ PEVENT_LIB_OBJS += kbuffer-parse.o
 
 PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
+PLUGIN_OBJS += plugin_kmem.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_kmem.c new file +72 -0

72 lines not on this page · raw message

author
#29 lore raw

PATCH v3 15/28 tools lib traceevent: Add kvm plugin

In reply to Jiri Olsa

Backporting kvm plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following tracepoint events:
kvm:kvm_exit
kvm:kvm_emulate_insn
kvm:kvm_nested_vmexit kvm:kvm_nested_vmexit_inject kvmmmu:kvm_mmu_get_page kvmmmu:kvm_mmu_sync_page kvmmmu:kvm_mmu_unsync_page kvmmmu:kvm_mmu_zap_page kvmmmu:kvm_mmu_prepare_zap_page

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

--- script.kvm.old
+++ script.kvm.new

    qemu-system-x86 17414 [000]  6868.995053: kvm:kvm_exit: reason EPT_VIOLATION rip 0xfff0 info 184 0
    qemu-system-x86 17414 [000]  6868.995109: kvm:kvm_emulate_insn: f0000:c46b:e4 71 (real)
  - qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=0 unsync=0 created=1
  + qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: new sp gfn 0 0/4 q0 direct --- !pge !nxe root 0 sync
  - qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=1 unsync=0
  + qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: 0/4 q0 direct --- !pge !nxe root 1 sync
    qemu-system-x86  3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
  -          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: 3/0 q0 --- !pge !nxe root 10 unsync
  -          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: 3/0 q0 --- !pge !nxe root 10 unsync

Note:
- kvm_mmu_zap_page is replaced by kvm_mmu_prepare_zap_page in current kernel, keeping it for backward compatibility
- some of the tracepoints keep the same output even with the plugin handling: kvm:kvm_exit, kvm:kvm_emulate_insn
- the 'kvmmmu:fast_page_fault' is still broken because of missing is_writable_pte function and is fixed in another patch
- ommited following tracepoints from backport because the output was buggy
kvm:kvm_nested_vmexit
kvm:kvm_nested_vmexit_inject

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile     |   1 +
 tools/lib/traceevent/plugin_kvm.c | 419 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 420 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_kvm.c
tools/lib/traceevent/Makefile +1 -0
index 245d6b4..226a8f9 100644
@@ -215,6 +215,7 @@ PEVENT_LIB_OBJS += kbuffer-parse.o
 PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
+PLUGIN_OBJS += plugin_kvm.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_kvm.c new file +419 -0

419 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#30 lore raw

[tip:perf/core] tools lib traceevent: Add kvm plugin

tools lib traceevent: Add kvm plugin
Commit-ID:  35d79f9ff88765af3fba6981db8d071872dcc5a8
Gitweb:     http://git.kernel.org/tip/35d79f9ff88765af3fba6981db8d071872dcc5a8
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:29 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:29:25 -0300

tools lib traceevent: Add kvm plugin

Backporting kvm plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following tracepoint events:
kvm:kvm_exit
kvm:kvm_emulate_insn
kvm:kvm_nested_vmexit kvm:kvm_nested_vmexit_inject kvmmmu:kvm_mmu_get_page kvmmmu:kvm_mmu_sync_page kvmmmu:kvm_mmu_unsync_page kvmmmu:kvm_mmu_zap_page kvmmmu:kvm_mmu_prepare_zap_page

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

--- script.kvm.old
+++ script.kvm.new

    qemu-system-x86 17414 [000]  6868.995053: kvm:kvm_exit: reason EPT_VIOLATION rip 0xfff0 info 184 0
    qemu-system-x86 17414 [000]  6868.995109: kvm:kvm_emulate_insn: f0000:c46b:e4 71 (real)
  - qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=0 unsync=0 created=1
  + qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: new sp gfn 0 0/4 q0 direct --- !pge !nxe root 0 sync
  - qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=1 unsync=0
  + qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: 0/4 q0 direct --- !pge !nxe root 1 sync
    qemu-system-x86  3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
  -          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: 3/0 q0 --- !pge !nxe root 10 unsync
  -          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: 3/0 q0 --- !pge !nxe root 10 unsync

Note:
- kvm_mmu_zap_page is replaced by kvm_mmu_prepare_zap_page in current kernel, keeping it for backward compatibility
- some of the tracepoints keep the same output even with the plugin handling: kvm:kvm_exit, kvm:kvm_emulate_insn
- the 'kvmmmu:fast_page_fault' is still broken because of missing is_writable_pte function and is fixed in another patch
- ommited following tracepoints from backport because the output was buggy
kvm:kvm_nested_vmexit
kvm:kvm_nested_vmexit_inject

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile     |   1 +
 tools/lib/traceevent/plugin_kvm.c | 419 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 420 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 245d6b4..226a8f9 100644
@@ -215,6 +215,7 @@ PEVENT_LIB_OBJS += kbuffer-parse.o
 PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
+PLUGIN_OBJS += plugin_kvm.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_kvm.c new file +419 -0

419 lines not on this page · raw message

author
#31 lore raw

PATCH v3 16/28 tools lib traceevent: Add mac80211 plugin

In reply to Jiri Olsa

Backporting mac80211 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds changed field resolving for
mac80211:drv_bss_info_changed tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'mac80211:drv_bss_info_changed' -a')

--- script.mac80211.old
+++ script.mac80211.new

  -        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2) changed:0x309f
  +        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2)
  +                                                                 assoc:0 aid:2 cts:0 shortpre:0 shortslot:0 dtimper:1
  +                                                                 bcnint:102 assoc_cap:0x431 basic_rates:0xf enable_beacon:0
  +                                                                 ht_operation_mode:0

Omitting the mac80211:drv_config tracepoint handling because the kernel tracepoint changed its prototype and the plugin handler is no longer working.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile          |  1 +
 tools/lib/traceevent/plugin_mac80211.c | 95 ++++++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_mac80211.c
tools/lib/traceevent/Makefile +1 -0
index 226a8f9..d0e0ca1 100644
@@ -216,6 +216,7 @@ PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
+PLUGIN_OBJS += plugin_mac80211.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_mac80211.c new file +95 -0

95 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#32 lore raw

[tip:perf/core] tools lib traceevent: Add mac80211 plugin

tools lib traceevent: Add mac80211 plugin
Commit-ID:  de705e2e88442c0c5dc6ec8c1f3fb1931def102e
Gitweb:     http://git.kernel.org/tip/de705e2e88442c0c5dc6ec8c1f3fb1931def102e
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:30 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:30:36 -0300

tools lib traceevent: Add mac80211 plugin

Backporting mac80211 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds changed field resolving for
mac80211:drv_bss_info_changed tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'mac80211:drv_bss_info_changed' -a')

--- script.mac80211.old
+++ script.mac80211.new

  -        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2) changed:0x309f
  +        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2)
  +                                                                 assoc:0 aid:2 cts:0 shortpre:0 shortslot:0 dtimper:1
  +                                                                 bcnint:102 assoc_cap:0x431 basic_rates:0xf enable_beacon:0
  +                                                                 ht_operation_mode:0

Omitting the mac80211:drv_config tracepoint handling because the kernel tracepoint changed its prototype and the plugin handler is no longer working.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile          |  1 +
 tools/lib/traceevent/plugin_mac80211.c | 95 ++++++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 226a8f9..d0e0ca1 100644
@@ -216,6 +216,7 @@ PLUGIN_OBJS  = plugin_jbd2.o
 PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
+PLUGIN_OBJS += plugin_mac80211.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_mac80211.c new file +95 -0

95 lines not on this page · raw message

author
#33 lore raw

PATCH v3 17/28 tools lib traceevent: Add sched_switch plugin

In reply to Jiri Olsa

Backporting sched_switch plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds fields resolving for
sched:sched_switch tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e sched:sched_switch -a')

--- script.sched_switch.old
+++ script.sched_switch.new

  -            perf   577 [002] 30965.311852: sched:sched_switch: prev_comm=perf prev_pid=577 prev_prio=120 prev_state=S ==> next_comm=swapper/2 next_pid=0 next_prio=120
  +            perf   577 [002] 30965.311852: sched:sched_switch: perf:577 [120] S ==> swapper/2:0 [120]
Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile              |   1 +
 tools/lib/traceevent/plugin_sched_switch.c | 148 +++++++++++++++++++++++++++++
 2 files changed, 149 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_sched_switch.c
tools/lib/traceevent/Makefile +1 -0
index d0e0ca1..1c319a8 100644
@@ -217,6 +217,7 @@ PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
+PLUGIN_OBJS += plugin_sched_switch.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_sched_switch.c new file +148 -0

148 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#34 lore raw

[tip:perf/core] tools lib traceevent: Add sched_switch plugin

tools lib traceevent: Add sched_switch plugin
Commit-ID:  83e815ee1c6b5fdb570374404ab44b24c1325ccb
Gitweb:     http://git.kernel.org/tip/83e815ee1c6b5fdb570374404ab44b24c1325ccb
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:31 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:31:11 -0300

tools lib traceevent: Add sched_switch plugin

Backporting sched_switch plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds fields resolving for sched:sched_switch tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e sched:sched_switch -a')

--- script.sched_switch.old
+++ script.sched_switch.new

  -            perf   577 [002] 30965.311852: sched:sched_switch: prev_comm=perf prev_pid=577 prev_prio=120 prev_state=S ==> next_comm=swapper/2 next_pid=0 next_prio=120
  +            perf   577 [002] 30965.311852: sched:sched_switch: perf:577 [120] S ==> swapper/2:0 [120]
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile              |   1 +
 tools/lib/traceevent/plugin_sched_switch.c | 148 +++++++++++++++++++++++++++++
 2 files changed, 149 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index d0e0ca1..1c319a8 100644
@@ -217,6 +217,7 @@ PLUGIN_OBJS += plugin_hrtimer.o
 PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
+PLUGIN_OBJS += plugin_sched_switch.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_sched_switch.c new file +148 -0

148 lines not on this page · raw message

author
#35 lore raw

PATCH v3 18/28 tools lib traceevent: Add function plugin

In reply to Jiri Olsa

Backporting function plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function and parent function fields resolving for ftrace:function tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e ftrace:function ls')

--- script.function.old
+++ script.function.new

  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811adb80 <-- ffffffff811afc48
  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811b35d0 <-- ffffffff811adb9b
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b3520 <-- ffffffff811b35e8
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b2720 <-- ffffffff811b3549
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81297e10 <-- ffffffff811b356c
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81298f40 <-- ffffffff81297e2c
  -              ls 10781 [001] 32667.291382: ftrace:function:  ffffffff81076160 <-- ffffffff811afbf0
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff811c3eb0 <-- ffffffff811afbfc
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff8164e100 <-- ffffffff811c3ed8
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811a5d10 <-- ffffffff811c3f53
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811e8e70 <-- ffffffff811a5d58
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811f38e0 <-- ffffffff811a5d63
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811a9ff0 <-- ffffffff811a5d6b
  -              ls 10781 [001] 32667.291386: ftrace:function:  ffffffff811a9fa0 <-- ffffffff811aa015
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff810851c0 <-- ffffffff811aa053
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff81090e00 <-- ffffffff81085211
  +              ls 10781 [001] 32667.291379: ftrace:function: would_dump <-- setup_new_exec
  +              ls 10781 [001] 32667.291379: ftrace:function:    inode_permission <-- would_dump
  +              ls 10781 [001] 32667.291380: ftrace:function:       __inode_permission <-- inode_permission
  +              ls 10781 [001] 32667.291380: ftrace:function:          generic_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:          security_inode_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:             cap_inode_permission <-- security_inode_permission
  +              ls 10781 [001] 32667.291382: ftrace:function: flush_signal_handlers <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function: do_close_on_exec <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function:    _raw_spin_lock <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:    filp_close <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:       dnotify_flush <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       locks_remove_posix <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       fput <-- filp_close
  +              ls 10781 [001] 32667.291386: ftrace:function:          file_sb_list_del <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:          task_work_add <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:             kick_process <-- task_work_add

Removing options support as it's not backported yet. Currently this plugin supports 2 options:
'parent' to display parent function 'indent' to show function call indents

Enabling both of them by default.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile          |   1 +
 tools/lib/traceevent/plugin_function.c | 151 +++++++++++++++++++++++++++++++++
 2 files changed, 152 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_function.c
tools/lib/traceevent/Makefile +1 -0
index 1c319a8..21f9b8f 100644
@@ -218,6 +218,7 @@ PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
+PLUGIN_OBJS += plugin_function.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_function.c new file +151 -0

151 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#36 lore raw

[tip:perf/core] tools lib traceevent: Add function plugin

tools lib traceevent: Add function plugin
Commit-ID:  07a180a0bfea7029b8cf84a7d8e856539e5c69f5
Gitweb:     http://git.kernel.org/tip/07a180a0bfea7029b8cf84a7d8e856539e5c69f5
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:32 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:32:00 -0300

tools lib traceevent: Add function plugin

Backporting function plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function and parent function fields resolving for ftrace:function tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e ftrace:function ls')

--- script.function.old
+++ script.function.new

  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811adb80 <-- ffffffff811afc48
  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811b35d0 <-- ffffffff811adb9b
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b3520 <-- ffffffff811b35e8
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b2720 <-- ffffffff811b3549
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81297e10 <-- ffffffff811b356c
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81298f40 <-- ffffffff81297e2c
  -              ls 10781 [001] 32667.291382: ftrace:function:  ffffffff81076160 <-- ffffffff811afbf0
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff811c3eb0 <-- ffffffff811afbfc
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff8164e100 <-- ffffffff811c3ed8
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811a5d10 <-- ffffffff811c3f53
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811e8e70 <-- ffffffff811a5d58
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811f38e0 <-- ffffffff811a5d63
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811a9ff0 <-- ffffffff811a5d6b
  -              ls 10781 [001] 32667.291386: ftrace:function:  ffffffff811a9fa0 <-- ffffffff811aa015
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff810851c0 <-- ffffffff811aa053
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff81090e00 <-- ffffffff81085211
  +              ls 10781 [001] 32667.291379: ftrace:function: would_dump <-- setup_new_exec
  +              ls 10781 [001] 32667.291379: ftrace:function:    inode_permission <-- would_dump
  +              ls 10781 [001] 32667.291380: ftrace:function:       __inode_permission <-- inode_permission
  +              ls 10781 [001] 32667.291380: ftrace:function:          generic_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:          security_inode_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:             cap_inode_permission <-- security_inode_permission
  +              ls 10781 [001] 32667.291382: ftrace:function: flush_signal_handlers <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function: do_close_on_exec <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function:    _raw_spin_lock <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:    filp_close <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:       dnotify_flush <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       locks_remove_posix <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       fput <-- filp_close
  +              ls 10781 [001] 32667.291386: ftrace:function:          file_sb_list_del <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:          task_work_add <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:             kick_process <-- task_work_add

Removing options support as it's not backported yet. Currently this plugin supports 2 options:
'parent' to display parent function 'indent' to show function call indents

Enabling both of them by default.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile          |   1 +
 tools/lib/traceevent/plugin_function.c | 151 +++++++++++++++++++++++++++++++++
 2 files changed, 152 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 1c319a8..21f9b8f 100644
@@ -218,6 +218,7 @@ PLUGIN_OBJS += plugin_kmem.o
 PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
+PLUGIN_OBJS += plugin_function.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_function.c new file +151 -0

151 lines not on this page · raw message

author
#37 lore raw

PATCH v3 19/28 tools lib traceevent: Add xen plugin

In reply to Jiri Olsa

Adding xen plugin.

This plugin adds fields resolving for
following tracepoint events:
xen:xen_mc_entry
xen:xen_mc_extend_args

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'xen:*' ls')

--- script.xen.old
+++ script.xen.new

  -         swapper     0 [002]   136.267492: xen:xen_mc_entry: [FAILED TO PARSE] op=3 nargs=2 args=ARRAY[18, 00, 00, 00, 00, 00, 00, 00, 00, e0, d4, 4b, 04, 88, ff, ff, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
  +         swapper     0 [002]   136.267492: xen:xen_mc_entry: op 3(stack_switch) args [18, 0, 0, 0, 0, 0]
  -            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: [FAILED TO PARSE] op=1 args=16 res=1
  +            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: extending op 1(mmu_update) by 16 bytes res ???

NOTE We still do not handle the 'sizeof' and fail
to parse following xen tracepoints:
xen:xen_mmu_set_pte
xen:xen_mmu_set_pte_atomic
xen:xen_mmu_set_domain_pte
xen:xen_mmu_set_pte_at
xen:xen_mmu_set_pmd
xen:xen_mmu_set_pud
xen:xen_mmu_set_pgd
xen:xen_mmu_ptep_modify_prot_start
xen:xen_mmu_ptep_modify_prot_commit

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
Cc Jeremy Fitzhardinge <[email protected]>
---
 tools/lib/traceevent/Makefile     |   1 +
 tools/lib/traceevent/plugin_xen.c | 130 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 131 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_xen.c
tools/lib/traceevent/Makefile +1 -0
index 21f9b8f..54af60a 100644
@@ -219,6 +219,7 @@ PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
+PLUGIN_OBJS += plugin_xen.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_xen.c new file +130 -0

130 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#38 lore raw

[tip:perf/core] tools lib traceevent: Add xen plugin

tools lib traceevent: Add xen plugin
Commit-ID:  40c03ad5925855e47e4b6294323107b2b180ae3d
Gitweb:     http://git.kernel.org/tip/40c03ad5925855e47e4b6294323107b2b180ae3d
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:33 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:34:04 -0300

tools lib traceevent: Add xen plugin

Adding xen plugin.

This plugin adds fields resolving for following tracepoint events:

xen:xen_mc_entry
xen:xen_mc_extend_args

The diff of 'perf script' output generated by old and new code: (data was generated by 'perf record -e 'xen:*' ls')

--- script.xen.old
+++ script.xen.new

  -         swapper     0 [002]   136.267492: xen:xen_mc_entry: [FAILED TO PARSE] op=3 nargs=2 args=ARRAY[18, 00, 00, 00, 00, 00, 00, 00, 00, e0, d4, 4b, 04, 88, ff, ff, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
  +         swapper     0 [002]   136.267492: xen:xen_mc_entry: op 3(stack_switch) args [18, 0, 0, 0, 0, 0]
  -            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: [FAILED TO PARSE] op=1 args=16 res=1
  +            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: extending op 1(mmu_update) by 16 bytes res ???

NOTE We still do not handle the 'sizeof' and fail to parse following xen tracepoints:

xen:xen_mmu_set_pte
xen:xen_mmu_set_pte_atomic
xen:xen_mmu_set_domain_pte
xen:xen_mmu_set_pte_at
xen:xen_mmu_set_pmd
xen:xen_mmu_set_pud
xen:xen_mmu_set_pgd
xen:xen_mmu_ptep_modify_prot_start
xen:xen_mmu_ptep_modify_prot_commit

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Jeremy Fitzhardinge <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile     |   1 +
 tools/lib/traceevent/plugin_xen.c | 130 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 131 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 21f9b8f..54af60a 100644
@@ -219,6 +219,7 @@ PLUGIN_OBJS += plugin_kvm.o
 PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
+PLUGIN_OBJS += plugin_xen.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_xen.c new file +130 -0

130 lines not on this page · raw message

author
#39 lore raw

PATCH v3 20/28 tools lib traceevent: Add scsi plugin

In reply to Jiri Olsa

Adding scsi plugin.

This plugin adds fields resolving functions for following tracepoint events:
scsi:scsi_dispatch_cmd_start scsi:scsi_dispatch_cmd_error scsi:scsi_dispatch_cmd_done scsi:scsi_dispatch_cmd_timeout

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'scsi:scsi_dispatch_cmd*' -a)

  -         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 result=0 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) result=(driver=DRIVER_OK host=DID_OK message=COMMAND_COMPLETE status=SAM_STAT_GOOD)
  -     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 opcode=42 cmd_len=10 data_sglen=1 prot_sglen=0 prot_op=0 cmnd=*
  +     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: host_no=0 channel=0 id=0 lun=0 data_sgl=1 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(WRITE_10 lba=570899168 txlen=8 protect=0 raw=2a 00 22 07 3a e0 00 00 08 00)
  -     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 rtn=4182 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) rtn=4182

NOTE I couldn't generate scsi_dispatch_cmd_timeout tracepoint,
but it's similar to the rest, so I believe it's ok.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
Cc Martin K. Petersen <[email protected]>
---
 tools/lib/traceevent/Makefile      |   1 +
 tools/lib/traceevent/plugin_scsi.c | 423 +++++++++++++++++++++++++++++++++++++
 2 files changed, 424 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_scsi.c
tools/lib/traceevent/Makefile +1 -0
index 54af60a..671f969 100644
@@ -220,6 +220,7 @@ PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
 PLUGIN_OBJS += plugin_xen.o
+PLUGIN_OBJS += plugin_scsi.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_scsi.c new file +423 -0

423 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#40 lore raw

[tip:perf/core] tools lib traceevent: Add scsi plugin

tools lib traceevent: Add scsi plugin
Commit-ID:  04010929fa33ad3a446b9fd50b82321c1c3cf239
Gitweb:     http://git.kernel.org/tip/04010929fa33ad3a446b9fd50b82321c1c3cf239
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:34 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:34:51 -0300

tools lib traceevent: Add scsi plugin

Adding scsi plugin.

This plugin adds fields resolving functions for following tracepoint events:

scsi:scsi_dispatch_cmd_start
scsi:scsi_dispatch_cmd_error
scsi:scsi_dispatch_cmd_done
scsi:scsi_dispatch_cmd_timeout

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'scsi:scsi_dispatch_cmd*' -a)

  -         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 result=0 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) result=(driver=DRIVER_OK host=DID_OK message=COMMAND_COMPLETE status=SAM_STAT_GOOD)
  -     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 opcode=42 cmd_len=10 data_sglen=1 prot_sglen=0 prot_op=0 cmnd=*
  +     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: host_no=0 channel=0 id=0 lun=0 data_sgl=1 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(WRITE_10 lba=570899168 txlen=8 protect=0 raw=2a 00 22 07 3a e0 00 00 08 00)
  -     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 rtn=4182 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) rtn=4182

NOTE I couldn't generate scsi_dispatch_cmd_timeout tracepoint,
but it's similar to the rest, so I believe it's ok.

Cc "Martin K. Petersen" <[email protected]>
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Martin K. Petersen <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile                      |   1 +
 .../lib/traceevent/plugin_scsi.c                   | 187 ++++++++++++++++++---
 2 files changed, 162 insertions(+), 26 deletions(-)
tools/lib/traceevent/Makefile +1 -0
index 54af60a..671f969 100644
@@ -220,6 +220,7 @@ PLUGIN_OBJS += plugin_mac80211.o
 PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
 PLUGIN_OBJS += plugin_xen.o
+PLUGIN_OBJS += plugin_scsi.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_scsi.c renamed from drivers/scsi/scsi_trace.c +161 -26

243 lines not on this page · raw message

author
#41 lore raw

PATCH v3 21/28 tools lib traceevent: Add cfg80211 plugin

In reply to Jiri Olsa

Adding cfg80211 plugin.

This plugin adds handler for __le16_to_cpup function t properly parse following tracepoint events:
cfg80211:cfg80211_tx_mlme_mgmt cfg80211:cfg80211_rx_mlme_mgmt cfg80211:cfg80211_rx_unprot_mlme_mgmt

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'cfg80211:*' -a')

--- script.cfg80211.old
+++ script.cfg80211.new

  -        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[c0, 00, 00, 00, 00, 3a, 98, a0, 30, 51, 10, 0b, a9, c6, f4, 74, 00, 3a, 98, a0, 30, 51, 00, 00, 03, 00]
  +        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: netdev:wlan0(3), ftype:0xc0
  -   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[b0, 00, da, 00, 10, 0b, a9, c6, f4, 74, d8, 24, bd, a1, 26, 31, d8, 24, bd, a1, 26, 31, 10, b7, 00, 00, 02, 00, 00, 00]
  +   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: netdev:wlan0(3), ftype:0xb0
Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/Makefile          |  1 +
 tools/lib/traceevent/plugin_cfg80211.c | 24 ++++++++++++++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100644 tools/lib/traceevent/plugin_cfg80211.c
tools/lib/traceevent/Makefile +1 -0
index 671f969..0d9cbb4 100644
@@ -221,6 +221,7 @@ PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
 PLUGIN_OBJS += plugin_xen.o
 PLUGIN_OBJS += plugin_scsi.o
+PLUGIN_OBJS += plugin_cfg80211.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_cfg80211.c new file +24 -0
new file mode 100644
index 0000000..dcab8e8
@@ -0,0 +1,24 @@
+#include <stdio.h>
+#include <string.h>
+#include <inttypes.h>
+#include <endian.h>
+#include "event-parse.h"
+
+static unsigned long long
+process___le16_to_cpup(struct trace_seq *s,
+		       unsigned long long *args)
+{
+	uint16_t *val = (uint16_t *) args[0];
+	return val ? (long long) le16toh(*val) : 0;
+}
+
+int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
+{
+	pevent_register_print_function(pevent,
+				       process___le16_to_cpup,
+				       PEVENT_FUNC_ARG_INT,
+				       "__le16_to_cpup",
+				       PEVENT_FUNC_ARG_PTR,
+				       PEVENT_FUNC_ARG_VOID);
+	return 0;
+}
Signature · 1 line
1.8.3.1
#42 lore raw

[tip:perf/core] tools lib traceevent: Add cfg80211 plugin

tools lib traceevent: Add cfg80211 plugin
Commit-ID:  943714737374635a134dc3de59a7f062cea53fb3
Gitweb:     http://git.kernel.org/tip/943714737374635a134dc3de59a7f062cea53fb3
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:35 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:35:27 -0300

tools lib traceevent: Add cfg80211 plugin

Adding cfg80211 plugin.

This plugin adds handler for __le16_to_cpup function t properly parse following tracepoint events:
cfg80211:cfg80211_tx_mlme_mgmt cfg80211:cfg80211_rx_mlme_mgmt cfg80211:cfg80211_rx_unprot_mlme_mgmt

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'cfg80211:*' -a')

--- script.cfg80211.old
+++ script.cfg80211.new

  -        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[c0, 00, 00, 00, 00, 3a, 98, a0, 30, 51, 10, 0b, a9, c6, f4, 74, 00, 3a, 98, a0, 30, 51, 00, 00, 03, 00]
  +        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: netdev:wlan0(3), ftype:0xc0
  -   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[b0, 00, da, 00, 10, 0b, a9, c6, f4, 74, d8, 24, bd, a1, 26, 31, d8, 24, bd, a1, 26, 31, 10, b7, 00, 00, 02, 00, 00, 00]
  +   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: netdev:wlan0(3), ftype:0xb0
Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/Makefile          |  1 +
 tools/lib/traceevent/plugin_cfg80211.c | 24 ++++++++++++++++++++++++
 2 files changed, 25 insertions(+)
tools/lib/traceevent/Makefile +1 -0
index 671f969..0d9cbb4 100644
@@ -221,6 +221,7 @@ PLUGIN_OBJS += plugin_sched_switch.o
 PLUGIN_OBJS += plugin_function.o
 PLUGIN_OBJS += plugin_xen.o
 PLUGIN_OBJS += plugin_scsi.o
+PLUGIN_OBJS += plugin_cfg80211.o
 
 PLUGINS := $(PLUGIN_OBJS:.o=.so)
 
tools/lib/traceevent/plugin_cfg80211.c new file +24 -0
new file mode 100644
index 0000000..dcab8e8
@@ -0,0 +1,24 @@
+#include <stdio.h>
+#include <string.h>
+#include <inttypes.h>
+#include <endian.h>
+#include "event-parse.h"
+
+static unsigned long long
+process___le16_to_cpup(struct trace_seq *s,
+		       unsigned long long *args)
+{
+	uint16_t *val = (uint16_t *) args[0];
+	return val ? (long long) le16toh(*val) : 0;
+}
+
+int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
+{
+	pevent_register_print_function(pevent,
+				       process___le16_to_cpup,
+				       PEVENT_FUNC_ARG_INT,
+				       "__le16_to_cpup",
+				       PEVENT_FUNC_ARG_PTR,
+				       PEVENT_FUNC_ARG_VOID);
+	return 0;
+}
author
#43 lore raw

PATCH v3 22/28 tools lib traceevent: Remove malloc_or_die from event-plugin.c

In reply to Jiri Olsa

Removing malloc_or_die calls from event-plugin.c, replacing them with standard malloc and error path.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/event-plugin.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
tools/lib/traceevent/event-plugin.c +16 -3
index d272d87..125f567 100644
@@ -47,7 +47,11 @@ load_plugin(struct pevent *pevent, const char *path,
 	char *plugin;
 	void *handle;
 
-	plugin = malloc_or_die(strlen(path) + strlen(file) + 2);
+	plugin = malloc(strlen(path) + strlen(file) + 2);
+	if (!plugin) {
+		warning("could not allocate plugin memory\n");
+		return;
+	}
 
 	strcpy(plugin, path);
 	strcat(plugin, "/");
@@ -71,7 +75,12 @@ load_plugin(struct pevent *pevent, const char *path,
 		goto out_free;
 	}
 
-	list = malloc_or_die(sizeof(*list));
+	list = malloc(sizeof(*list));
+	if (!list) {
+		warning("could not allocate plugin memory\n");
+		goto out_free;
+	}
+
 	list->next = *plugin_list;
 	list->handle = handle;
 	list->name = plugin;
@@ -163,7 +172,11 @@ load_plugins(struct pevent *pevent, const char *suffix,
 	if (!home)
 		return;
 
-	path = malloc_or_die(strlen(home) + strlen(LOCAL_PLUGIN_DIR) + 2);
+	path = malloc(strlen(home) + strlen(LOCAL_PLUGIN_DIR) + 2);
+	if (!path) {
+		warning("could not allocate plugin memory\n");
+		return;
+	}
 
 	strcpy(path, home);
 	strcat(path, "/");
Signature · 1 line
1.8.3.1
#44 lore raw

Re: [PATCH 22/28] tools lib traceevent: Remove malloc_or_die from event-plugin.c

Quoted 23 lines · On Tue, 3 Dec 2013 14:09:36 +0100 Jiri Olsa <[email protected]> wrote:

Removing malloc_or_die calls from event-plugin.c, replacing them with standard malloc and error path.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/event-plugin.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
tools/lib/traceevent/event-plugin.c +0 -0
index d272d87..125f567 100644

tools/lib/traceevent/event-plugin.c load_plugin(struct pevent *pevent, const char *path, · line 53

Quoted 8 lines
tools/lib/traceevent/event-plugin.c +4 -1
@@ -47,7 +47,11 @@ load_plugin(struct pevent *pevent, const char *path,
 	char *plugin;
 	void *handle;
 
-	plugin = malloc_or_die(strlen(path) + strlen(file) + 2);
+	plugin = malloc(strlen(path) + strlen(file) + 2);
+	if (!plugin) {
+		warning("could not allocate plugin memory\n");
+		return;

This should be changed to return an error code. Yes it will require other places to change for that update as well.

-- Steve

Quoted 4 lines
tools/lib/traceevent/event-plugin.c +1 -0
+	}
 
 	strcpy(plugin, path);
 	strcat(plugin, "/");
author
#45 lore raw

Re: [PATCH 22/28] tools lib traceevent: Remove malloc_or_die from event-plugin.c

Quoted 29 lines · Steven Rostedt
Quoted 26 lines · On Tue, 3 Dec 2013 14:09:36 +0100 Jiri Olsa <[email protected]> wrote:

Removing malloc_or_die calls from event-plugin.c, replacing them with standard malloc and error path.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/event-plugin.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
tools/lib/traceevent/event-plugin.c +0 -0
index d272d87..125f567 100644
@@ -47,7 +47,11 @@ load_plugin(struct pevent *pevent, const char *path,
 	char *plugin;
 	void *handle;
 
Quoted 8 lines

tools/lib/traceevent/event-plugin.c load_plugin(struct pevent *pevent, const char *path, · line 53

Quoted 5 lines
tools/lib/traceevent/event-plugin.c +4 -1
-	plugin = malloc_or_die(strlen(path) + strlen(file) + 2);
+	plugin = malloc(strlen(path) + strlen(file) + 2);
+	if (!plugin) {
+		warning("could not allocate plugin memory\n");
+		return;

This should be changed to return an error code. Yes it will require other places to change for that update as well.

Any chance this could be a separated feature? ;-)

AFACIS there's no technical problem with the current code. If the load_plugin fails (due to ENOMEM or interface error) it's not added on the plugin_list, which is the output/handle of plugin interface (and there's warning ;-)).

I think we need some complex/unified error handling for the whole library and add that globally.

jirka

#46 lore raw

Re: [PATCH 22/28] tools lib traceevent: Remove malloc_or_die from event-plugin.c

Em Wed, Dec 04, 2013 at 10:29:34AM +0100, Jiri Olsa escreveu:

Quoted 22 lines · Steven Rostedt
Quoted 10 lines · On Tue, 3 Dec 2013 14:09:36 +0100 Jiri Olsa <[email protected]> wrote:
Quoted 8 lines

Removing malloc_or_die calls from event-plugin.c, replacing them with standard malloc and error path.

+++ b/tools/lib/traceevent/event-plugin.c

(quoted diff) +0 -0
@@ -47,7 +47,11 @@ load_plugin(struct pevent *pevent, const char *path,
Quoted 8 lines

tools/lib/traceevent/event-plugin.c load_plugin(struct pevent *pevent, const char *path, · line 53

Quoted 5 lines
tools/lib/traceevent/event-plugin.c +4 -1
-	plugin = malloc_or_die(strlen(path) + strlen(file) + 2);
+	plugin = malloc(strlen(path) + strlen(file) + 2);
+	if (!plugin) {
+		warning("could not allocate plugin memory\n");
+		return;

This should be changed to return an error code. Yes it will require other places to change for that update as well.

Any chance this could be a separated feature? ;-)

Quoted 7 lines

AFACIS there's no technical problem with the current code. If the load_plugin fails (due to ENOMEM or interface error) it's not added on the plugin_list, which is the output/handle of plugin interface (and there's warning ;-)).

I think we need some complex/unified error handling for the whole library and add that globally.

Yeah, I also have some reservations about adding more globals (the trace_event stuff just to associate a plugins list with pevent, probably pevent should have that list, haven't looked 100% at the code), but in general probably this patchset is OK, I think we should take it in and then address those other problems in a followup patchset, d'accord?

- Arnaldo

#47 lore raw

Re: [PATCH 22/28] tools lib traceevent: Remove malloc_or_die from event-plugin.c

Quoted 5 lines · On Wed, 4 Dec 2013 10:29:32 -0300 Arnaldo Carvalho de Melo…

Yeah, I also have some reservations about adding more globals (the trace_event stuff just to associate a plugins list with pevent, probably pevent should have that list, haven't looked 100% at the code), but in general probably this patchset is OK, I think we should take it in and then address those other problems in a followup patchset, d'accord?

Yep, agreed. If we can associate it to the pevent descriptor then that would be best.

Yes, take this in now, and we can address the other issues in a followup. But these issues need to be addressed before we make the libtraecevent library public.

-- Steve

#48 lore raw

[tip:perf/core] tools lib traceevent: Remove malloc_or_die from event-plugin.c

In reply to Jiri Olsa
tools lib traceevent: Remove malloc_or_die from event-plugin.c
Commit-ID:  f9bb36afb25d3bfda1c9276a55985b710c8a91ae
Gitweb:     http://git.kernel.org/tip/f9bb36afb25d3bfda1c9276a55985b710c8a91ae
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:36 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:35:48 -0300

tools lib traceevent: Remove malloc_or_die from event-plugin.c

Removing malloc_or_die calls from event-plugin.c, replacing them with standard malloc and error path.

---
 tools/lib/traceevent/event-plugin.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
tools/lib/traceevent/event-plugin.c +16 -3
index d272d87..125f567 100644
@@ -47,7 +47,11 @@ load_plugin(struct pevent *pevent, const char *path,
 	char *plugin;
 	void *handle;
 
-	plugin = malloc_or_die(strlen(path) + strlen(file) + 2);
+	plugin = malloc(strlen(path) + strlen(file) + 2);
+	if (!plugin) {
+		warning("could not allocate plugin memory\n");
+		return;
+	}
 
 	strcpy(plugin, path);
 	strcat(plugin, "/");
@@ -71,7 +75,12 @@ load_plugin(struct pevent *pevent, const char *path,
 		goto out_free;
 	}
 
-	list = malloc_or_die(sizeof(*list));
+	list = malloc(sizeof(*list));
+	if (!list) {
+		warning("could not allocate plugin memory\n");
+		goto out_free;
+	}
+
 	list->next = *plugin_list;
 	list->handle = handle;
 	list->name = plugin;
@@ -163,7 +172,11 @@ load_plugins(struct pevent *pevent, const char *suffix,
 	if (!home)
 		return;
 
-	path = malloc_or_die(strlen(home) + strlen(LOCAL_PLUGIN_DIR) + 2);
+	path = malloc(strlen(home) + strlen(LOCAL_PLUGIN_DIR) + 2);
+	if (!path) {
+		warning("could not allocate plugin memory\n");
+		return;
+	}
 
 	strcpy(path, home);
 	strcat(path, "/");
author
#49 lore raw

PATCH v3 23/28 tools lib traceevent: Use static functions in jbd2 plugin

In reply to Jiri Olsa

There's no need for following functions to be global:
process_jbd2_dev_to_name process_jiffies_to_msecs

Making them static.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/plugin_jbd2.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
tools/lib/traceevent/plugin_jbd2.c +6 -4
index 5d85de7..2f93f81 100644
@@ -29,8 +29,9 @@
 #define MAJOR(dev)	((unsigned int) ((dev) >> MINORBITS))
 #define MINOR(dev)	((unsigned int) ((dev) & MINORMASK))
 
-unsigned long long process_jbd2_dev_to_name(struct trace_seq *s,
-					    unsigned long long *args)
+static unsigned long long
+process_jbd2_dev_to_name(struct trace_seq *s,
+			 unsigned long long *args)
 {
 	unsigned int dev = args[0];
 
@@ -38,8 +39,9 @@ unsigned long long process_jbd2_dev_to_name(struct trace_seq *s,
 	return 0;
 }
 
-unsigned long long process_jiffies_to_msecs(struct trace_seq *s,
-					    unsigned long long *args)
+static unsigned long long
+process_jiffies_to_msecs(struct trace_seq *s,
+			 unsigned long long *args)
 {
 	unsigned long long jiffies = args[0];
 
Signature · 1 line
1.8.3.1
#50 lore raw

[tip:perf/core] tools lib traceevent: Use static functions in jbd2 plugin

tools lib traceevent: Use static functions in jbd2 plugin
Commit-ID:  f04015ecb5eb0bf25260de585f0c5f60e84f073a
Gitweb:     http://git.kernel.org/tip/f04015ecb5eb0bf25260de585f0c5f60e84f073a
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:37 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:36:24 -0300

tools lib traceevent: Use static functions in jbd2 plugin

There's no need for following functions to be global:

process_jbd2_dev_to_name
process_jiffies_to_msecs

Make them static.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/plugin_jbd2.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
tools/lib/traceevent/plugin_jbd2.c +6 -4
index 5d85de7..2f93f81 100644
@@ -29,8 +29,9 @@
 #define MAJOR(dev)	((unsigned int) ((dev) >> MINORBITS))
 #define MINOR(dev)	((unsigned int) ((dev) & MINORMASK))
 
-unsigned long long process_jbd2_dev_to_name(struct trace_seq *s,
-					    unsigned long long *args)
+static unsigned long long
+process_jbd2_dev_to_name(struct trace_seq *s,
+			 unsigned long long *args)
 {
 	unsigned int dev = args[0];
 
@@ -38,8 +39,9 @@ unsigned long long process_jbd2_dev_to_name(struct trace_seq *s,
 	return 0;
 }
 
-unsigned long long process_jiffies_to_msecs(struct trace_seq *s,
-					    unsigned long long *args)
+static unsigned long long
+process_jiffies_to_msecs(struct trace_seq *s,
+			 unsigned long long *args)
 {
 	unsigned long long jiffies = args[0];
 
author
#51 lore raw

PATCH v3 24/28 tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler

In reply to Jiri Olsa

The pevent_print_func_field function encompasses all the functionality used in the hrtimer_start handler. Changing the handler to use this function.

This also unifies the function field output with the hrtimer_expire_entry handler.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/plugin_hrtimer.c | 19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)
tools/lib/traceevent/plugin_hrtimer.c +2 -17
index e41d4cf..0b0ebf3 100644
@@ -48,10 +48,6 @@ static int timer_start_handler(struct trace_seq *s,
 			       struct pevent_record *record,
 			       struct event_format *event, void *context)
 {
-	struct pevent *pevent = event->pevent;
-	struct format_field *fn = pevent_find_field(event, "function");
-	void *data = record->data;
-
 	trace_seq_printf(s, "hrtimer=");
 
 	if (pevent_print_num_field(s, "0x%llx", event, "timer",
@@ -59,19 +55,8 @@ static int timer_start_handler(struct trace_seq *s,
 		pevent_print_num_field(s, "0x%llx", event, "hrtimer",
 				       record, 1);
 
-	if (!fn) {
-		trace_seq_printf(s, " function=MISSING");
-	} else {
-		unsigned long long function;
-		const char *func;
-
-		if (pevent_read_number_field(fn, data, &function))
-			trace_seq_printf(s, " function=INVALID");
-
-		func = pevent_find_function(pevent, function);
-
-		trace_seq_printf(s, " function=%s", func);
-	}
+	pevent_print_func_field(s, " function=%s", event, "function",
+				record, 0);
 
 	trace_seq_printf(s, " expires=");
 	pevent_print_num_field(s, "%llu", event, "expires", record, 1);
Signature · 1 line
1.8.3.1
#52 lore raw

[tip:perf/core] tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler

tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler
Commit-ID:  b32cea650a16092a654baa2095c78fa917bad4da
Gitweb:     http://git.kernel.org/tip/b32cea650a16092a654baa2095c78fa917bad4da
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:38 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:36:50 -0300

tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler

The pevent_print_func_field function encompasses all the functionality used in the hrtimer_start handler. Change the handler to use this function.

This also unifies the function field output with the hrtimer_expire_entry handler.

---
 tools/lib/traceevent/plugin_hrtimer.c | 19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)
tools/lib/traceevent/plugin_hrtimer.c +2 -17
index e41d4cf..0b0ebf3 100644
@@ -48,10 +48,6 @@ static int timer_start_handler(struct trace_seq *s,
 			       struct pevent_record *record,
 			       struct event_format *event, void *context)
 {
-	struct pevent *pevent = event->pevent;
-	struct format_field *fn = pevent_find_field(event, "function");
-	void *data = record->data;
-
 	trace_seq_printf(s, "hrtimer=");
 
 	if (pevent_print_num_field(s, "0x%llx", event, "timer",
@@ -59,19 +55,8 @@ static int timer_start_handler(struct trace_seq *s,
 		pevent_print_num_field(s, "0x%llx", event, "hrtimer",
 				       record, 1);
 
-	if (!fn) {
-		trace_seq_printf(s, " function=MISSING");
-	} else {
-		unsigned long long function;
-		const char *func;
-
-		if (pevent_read_number_field(fn, data, &function))
-			trace_seq_printf(s, " function=INVALID");
-
-		func = pevent_find_function(pevent, function);
-
-		trace_seq_printf(s, " function=%s", func);
-	}
+	pevent_print_func_field(s, " function=%s", event, "function",
+				record, 0);
 
 	trace_seq_printf(s, " expires=");
 	pevent_print_num_field(s, "%llu", event, "expires", record, 1);
author
#53 lore raw

PATCH v3 25/28 tools lib traceevent: Several cleanups for function plugin

In reply to Jiri Olsa

Several cleanups suggested by Namhyung:
* removing index field from struct func_stack as it's not needed
* renaming get_index into add_and_get_index
* using '%*X' format string capability instead of the loop

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/plugin_function.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)
tools/lib/traceevent/plugin_function.c +4 -6
index 8deb22e..87acf9c 100644
@@ -25,7 +25,6 @@
 #include "event-utils.h"
 
 static struct func_stack {
-	int index;
 	int size;
 	char **stack;
 } *fstack;
@@ -57,7 +56,7 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 	stack->stack[pos] = strdup(child);
 }
 
-static int get_index(const char *parent, const char *child, int cpu)
+static int add_and_get_index(const char *parent, const char *child, int cpu)
 {
 	int i;
 
@@ -97,7 +96,7 @@ static int function_handler(struct trace_seq *s, struct pevent_record *record,
 	unsigned long long pfunction;
 	const char *func;
 	const char *parent;
-	int i, index;
+	int index;
 
 	if (pevent_get_field_val(s, event, "ip", record, &function, 1))
 		return trace_seq_putc(s, '!');
@@ -109,10 +108,9 @@ static int function_handler(struct trace_seq *s, struct pevent_record *record,
 
 	parent = pevent_find_function(pevent, pfunction);
 
-	index = get_index(parent, func, record->cpu);
+	index = add_and_get_index(parent, func, record->cpu);
 
-	for (i = 0; i < index; i++)
-		trace_seq_printf(s, "   ");
+	trace_seq_printf(s, "%*s", index*3, "");
 
 	if (func)
 		trace_seq_printf(s, "%s", func);
Signature · 1 line
1.8.3.1
#54 lore raw

[tip:perf/core] tools lib traceevent: Several cleanups for function plugin

tools lib traceevent: Several cleanups for function plugin
Commit-ID:  39956e78010645ee9d121e6a6eb6d9892e3fc92e
Gitweb:     http://git.kernel.org/tip/39956e78010645ee9d121e6a6eb6d9892e3fc92e
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:39 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:37:24 -0300

tools lib traceevent: Several cleanups for function plugin

Several cleanups suggested by Namhyung:

* Remove index field from struct func_stack as it's not needed.

* Rename get_index into add_and_get_index.

* Use '%*X' format string capability instead of the loop

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/plugin_function.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)
tools/lib/traceevent/plugin_function.c +4 -6
index 8deb22e..87acf9c 100644
@@ -25,7 +25,6 @@
 #include "event-utils.h"
 
 static struct func_stack {
-	int index;
 	int size;
 	char **stack;
 } *fstack;
@@ -57,7 +56,7 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 	stack->stack[pos] = strdup(child);
 }
 
-static int get_index(const char *parent, const char *child, int cpu)
+static int add_and_get_index(const char *parent, const char *child, int cpu)
 {
 	int i;
 
@@ -97,7 +96,7 @@ static int function_handler(struct trace_seq *s, struct pevent_record *record,
 	unsigned long long pfunction;
 	const char *func;
 	const char *parent;
-	int i, index;
+	int index;
 
 	if (pevent_get_field_val(s, event, "ip", record, &function, 1))
 		return trace_seq_putc(s, '!');
@@ -109,10 +108,9 @@ static int function_handler(struct trace_seq *s, struct pevent_record *record,
 
 	parent = pevent_find_function(pevent, pfunction);
 
-	index = get_index(parent, func, record->cpu);
+	index = add_and_get_index(parent, func, record->cpu);
 
-	for (i = 0; i < index; i++)
-		trace_seq_printf(s, "   ");
+	trace_seq_printf(s, "%*s", index*3, "");
 
 	if (func)
 		trace_seq_printf(s, "%s", func);
author
#55 lore raw

PATCH v3 26/28 tools lib traceevent: Remove malloc_or_die from plugin_function.c

In reply to Jiri Olsa

Removing malloc_or_die calls from plugin_function.c, replacing them and factoring the code with standard realloc and error path.

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/plugin_function.c | 29 ++++++++++++++++++++---------
 1 file changed, 20 insertions(+), 9 deletions(-)
tools/lib/traceevent/plugin_function.c +20 -9
index 87acf9c..aad92ad 100644
@@ -43,11 +43,17 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 	if (pos < stack->size)
 		free(stack->stack[pos]);
 	else {
-		if (!stack->stack)
-			stack->stack = malloc_or_die(sizeof(char *) * STK_BLK);
-		else
-			stack->stack = realloc(stack->stack, sizeof(char *) *
-					       (stack->size + STK_BLK));
+		char **ptr;
+
+		ptr = realloc(stack->stack, sizeof(char *) *
+			      (stack->size + STK_BLK));
+		if (!ptr) {
+			warning("could not allocate plugin memory\n");
+			return;
+		}
+
+		stack->stack = ptr;
+
 		for (i = stack->size; i < stack->size + STK_BLK; i++)
 			stack->stack[i] = NULL;
 		stack->size += STK_BLK;
@@ -64,10 +70,15 @@ static int add_and_get_index(const char *parent, const char *child, int cpu)
 		return 0;
 
 	if (cpu > cpus) {
-		if (fstack)
-			fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
-		else
-			fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+		struct func_stack *ptr;
+
+		ptr = realloc(fstack, sizeof(*fstack) * (cpu + 1));
+		if (!ptr) {
+			warning("could not allocate plugin memory\n");
+			return 0;
+		}
+
+		fstack = ptr;
 
 		/* Account for holes in the cpu count */
 		for (i = cpus + 1; i <= cpu; i++)
Signature · 1 line
1.8.3.1
#56 lore raw

[tip:perf/core] tools lib traceevent: Remove malloc_or_die from plugin_function.c

tools lib traceevent: Remove malloc_or_die from plugin_function.c
Commit-ID:  d8e56c98b7ef96a31a64c69df24ab5d80f90e055
Gitweb:     http://git.kernel.org/tip/d8e56c98b7ef96a31a64c69df24ab5d80f90e055
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:40 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:37:58 -0300

tools lib traceevent: Remove malloc_or_die from plugin_function.c

Removing malloc_or_die calls from plugin_function.c, replacing them and factoring the code with standard realloc and error path.

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/plugin_function.c | 29 ++++++++++++++++++++---------
 1 file changed, 20 insertions(+), 9 deletions(-)
tools/lib/traceevent/plugin_function.c +20 -9
index 87acf9c..aad92ad 100644
@@ -43,11 +43,17 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 	if (pos < stack->size)
 		free(stack->stack[pos]);
 	else {
-		if (!stack->stack)
-			stack->stack = malloc_or_die(sizeof(char *) * STK_BLK);
-		else
-			stack->stack = realloc(stack->stack, sizeof(char *) *
-					       (stack->size + STK_BLK));
+		char **ptr;
+
+		ptr = realloc(stack->stack, sizeof(char *) *
+			      (stack->size + STK_BLK));
+		if (!ptr) {
+			warning("could not allocate plugin memory\n");
+			return;
+		}
+
+		stack->stack = ptr;
+
 		for (i = stack->size; i < stack->size + STK_BLK; i++)
 			stack->stack[i] = NULL;
 		stack->size += STK_BLK;
@@ -64,10 +70,15 @@ static int add_and_get_index(const char *parent, const char *child, int cpu)
 		return 0;
 
 	if (cpu > cpus) {
-		if (fstack)
-			fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
-		else
-			fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+		struct func_stack *ptr;
+
+		ptr = realloc(fstack, sizeof(*fstack) * (cpu + 1));
+		if (!ptr) {
+			warning("could not allocate plugin memory\n");
+			return 0;
+		}
+
+		fstack = ptr;
 
 		/* Account for holes in the cpu count */
 		for (i = cpus + 1; i <= cpu; i++)
author
#57 lore raw

PATCH v3 27/28 tools lib traceevent: Update kvm plugin with is_writable_pte helper

In reply to Jiri Olsa

Adding is_writable_pte print helper function, so the kvmmmu:fast_page_fault print format gets resolved properly.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

--- script.kvm.old
+++ script.kvm.new
- qemu-system-x86 3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
+ qemu-system-x86 3290 [002] 10708.755312: kvmmmu:fast_page_fault: vcpu 0 gva f40ce640 error_code P|W sptep 0xffff88019f1e3670 old 0x140cec75 new 140cec77 spurious 0 fixed 1

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/lib/traceevent/plugin_kvm.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
tools/lib/traceevent/plugin_kvm.c +17 -0
index be9d9c6..a0e282c 100644
@@ -389,6 +389,16 @@ static int kvm_mmu_get_page_handler(struct trace_seq *s,
 	return kvm_mmu_print_role(s, record, event, context);
 }
 
+#define PT_WRITABLE_SHIFT 1
+#define PT_WRITABLE_MASK (1ULL << PT_WRITABLE_SHIFT)
+
+static unsigned long long
+process_is_writable_pte(struct trace_seq *s, unsigned long long *args)
+{
+	unsigned long pte = args[0];
+	return pte & PT_WRITABLE_MASK;
+}
+
 int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
 {
 	init_disassembler();
@@ -415,5 +425,12 @@ int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
 	pevent_register_event_handler(pevent, -1, "kvmmmu",
 			"kvm_mmu_prepare_zap_page", kvm_mmu_print_role,
 			NULL);
+
+	pevent_register_print_function(pevent,
+				       process_is_writable_pte,
+				       PEVENT_FUNC_ARG_INT,
+				       "is_writable_pte",
+				       PEVENT_FUNC_ARG_LONG,
+				       PEVENT_FUNC_ARG_VOID);
 	return 0;
 }
Signature · 1 line
1.8.3.1
#58 lore raw

[tip:perf/core] tools lib traceevent: Update kvm plugin with is_writable_pte helper

tools lib traceevent: Update kvm plugin with is_writable_pte helper
Commit-ID:  6d65894bc028d0342829ea1e64c9e9efad571124
Gitweb:     http://git.kernel.org/tip/6d65894bc028d0342829ea1e64c9e9efad571124
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 3 Dec 2013 14:09:41 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Wed, 4 Dec 2013 15:38:14 -0300

tools lib traceevent: Update kvm plugin with is_writable_pte helper

Adding is_writable_pte print helper function, so the kvmmmu:fast_page_fault print format gets resolved properly.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

--- script.kvm.old
+++ script.kvm.new
- qemu-system-x86 3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
+ qemu-system-x86 3290 [002] 10708.755312: kvmmmu:fast_page_fault: vcpu 0 gva f40ce640 error_code P|W sptep 0xffff88019f1e3670 old 0x140cec75 new 140cec77 spurious 0 fixed 1

Cc Corey Ashford <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 tools/lib/traceevent/plugin_kvm.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
tools/lib/traceevent/plugin_kvm.c +17 -0
index be9d9c6..a0e282c 100644
@@ -389,6 +389,16 @@ static int kvm_mmu_get_page_handler(struct trace_seq *s,
 	return kvm_mmu_print_role(s, record, event, context);
 }
 
+#define PT_WRITABLE_SHIFT 1
+#define PT_WRITABLE_MASK (1ULL << PT_WRITABLE_SHIFT)
+
+static unsigned long long
+process_is_writable_pte(struct trace_seq *s, unsigned long long *args)
+{
+	unsigned long pte = args[0];
+	return pte & PT_WRITABLE_MASK;
+}
+
 int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
 {
 	init_disassembler();
@@ -415,5 +425,12 @@ int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
 	pevent_register_event_handler(pevent, -1, "kvmmmu",
 			"kvm_mmu_prepare_zap_page", kvm_mmu_print_role,
 			NULL);
+
+	pevent_register_print_function(pevent,
+				       process_is_writable_pte,
+				       PEVENT_FUNC_ARG_INT,
+				       "is_writable_pte",
+				       PEVENT_FUNC_ARG_LONG,
+				       PEVENT_FUNC_ARG_VOID);
 	return 0;
 }
author
#59 lore raw

PATCH v3 28/28 perf tools: Add udis86 disassembler feature check

In reply to Jiri Olsa

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e kvm:kvm_emulate_insn -a')

--- script.kvm.old
+++ script.kvm.new
- qemu-system-x86 15519 [003] 5332.470049: kvm:kvm_emulate_insn: 0:ffffffff8103c596:89 b7 00 80 5f ff (prot64)
+ qemu-system-x86 15519 [003] 5332.470049: kvm:kvm_emulate_insn: 0:ffffffff8103c596: mov %esi, -0xa08000(%rdi)

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/Makefile.perf                       |  2 +-
 tools/perf/config/Makefile                     |  8 ++++++++
 tools/perf/config/feature-checks/Makefile      |  6 +++++-
 tools/perf/config/feature-checks/test-all.c    |  5 +++++
 tools/perf/config/feature-checks/test-udis86.c | 11 +++++++++++
 5 files changed, 30 insertions(+), 2 deletions(-)
 create mode 100644 tools/perf/config/feature-checks/test-udis86.c
tools/perf/Makefile.perf +1 -1
index ca3b87d..be336f8 100644
@@ -712,7 +712,7 @@ $(LIB_FILE): $(LIB_OBJS)
 TE_SOURCES = $(wildcard $(TRACE_EVENT_DIR)*.[ch])
 
 LIBTRACEEVENT_FLAGS  = $(QUIET_SUBDIR1) O=$(OUTPUT)
-LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS)"
+LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS) $(LIBTRACEEVENT_CFLAGS)"
 LIBTRACEEVENT_FLAGS += plugin_dir=$(plugindir_SQ)
 
 $(LIBTRACEEVENT): $(TE_SOURCES) $(OUTPUT)PERF-CFLAGS
tools/perf/config/Makefile +8 -0
index bae1072..5e3fb25 100644
@@ -413,6 +413,14 @@ else
   msg := $(warning No timerfd support. Disables 'perf kvm stat live');
 endif
 
+$(call feature_check,udis86)
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif
+
 disable-python = $(eval $(disable-python_code))
 define disable-python_code
   CFLAGS += -DNO_LIBPYTHON
tools/perf/config/feature-checks/Makefile +5 -1
index b8bb749..c550b4f 100644
@@ -26,7 +26,8 @@ FILES=					\
 	test-libunwind-debug-frame	\
 	test-on-exit			\
 	test-stackprotector-all		\
-	test-timerfd
+	test-timerfd			\
+	test-udis86
 
 CC := $(CC) -MD
 
@@ -140,6 +141,9 @@ test-backtrace:
 test-timerfd:
 	$(BUILD)
 
+test-udis86:
+	$(BUILD) -DHAVE_UDIS86 -ludis86
+
 -include *.d
 
 ###############################
tools/perf/config/feature-checks/test-all.c +5 -0
index 9b8a544..df087a5 100644
@@ -89,6 +89,10 @@
 # include "test-stackprotector-all.c"
 #undef main
 
+#define main main_test_udis86
+# include "test-udis86.c"
+#undef main
+
 int main(int argc, char *argv[])
 {
 	main_test_libpython();
@@ -111,6 +115,7 @@ int main(int argc, char *argv[])
 	main_test_libnuma();
 	main_test_timerfd();
 	main_test_stackprotector_all();
+	main_test_udis86();
 
 	return 0;
 }
tools/perf/config/feature-checks/test-udis86.c new file +11 -0
new file mode 100644
index 0000000..dbc2c26
@@ -0,0 +1,11 @@
+
+/*
+ * test for udis86 functions used by plugin_kvm.c
+ */
+
+#include <udis86.h>
+
+int main(void)
+{
+	return puts("hi");
+}
Signature · 1 line
1.8.3.1
#60 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

http://bit.ly/1hyrN52

Wow, that was easy, but yeah, could be made easier 8-) ;-P

- Arnaldo

#61 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

Em Wed, Dec 04, 2013 at 03:50:24PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 15 lines

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 3 lines

http://bit.ly/1hyrN52

Wow, that was easy, but yeah, could be made easier 8-) ;-P

Jokes aside, I'm dropping this one for now, as it breaks feature detection, now everytime I run make I get:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

config/Makefile:421: No udis86 support.

---

Please try to fix it and resubmit. And yes, I don't have udis86-devel installed in this test system.

- Arnaldo

author
#62 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

On Wed, Dec 04, 2013 at 03:53:02PM -0300, Arnaldo Carvalho de Melo wrote:

SNIP

Quoted 22 lines

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]
Quoted 6 lines

config/Makefile:421: No udis86 support.

---

Please try to fix it and resubmit. And yes, I don't have udis86-devel installed in this test system.

hum, I did not make it as the feature check with visible state '[ on ]', but I guess check for packages should be on the list.. I'll repost

jirka

author
#63 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

In reply to Arnaldo Carvalho de Melo
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 1 line

http://bit.ly/1hyrN52

nice :-)

Quoted 2 lines

Wow, that was easy, but yeah, could be made easier 8-) ;-P

so something like:
No udis86 found. Please install udis86-devel.

or:
No udis86 found, disabling kvm tracepoints instruction disassembly. Please install udis86-devel.

jirka

author
#64 lore raw

PATCH v4 28/28 perf tools: Add udis86 disassembler feature check

Quoted 21 lines · Jiri Olsa
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

Review

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

(quoted diff) +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 1 line

http://bit.ly/1hyrN52

nice :-)

Quoted 1 line
Quoted 7 lines
Quoted 1 line

Wow, that was easy, but yeah, could be made easier 8-) ;-P

so something like:
No udis86 found. Please install udis86-devel.

or:
No udis86 found, disabling kvm tracepoints instruction disassembly. Please install udis86-devel.

took the first one, v4 attached, perf/core_plugins is updated

thanks,
jirka

---
Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e kvm:kvm_emulate_insn -a')

--- script.kvm.old
+++ script.kvm.new
- qemu-system-x86 15519 [003] 5332.470049: kvm:kvm_emulate_insn: 0:ffffffff8103c596:89 b7 00 80 5f ff (prot64)
+ qemu-system-x86 15519 [003] 5332.470049: kvm:kvm_emulate_insn: 0:ffffffff8103c596: mov %esi, -0xa08000(%rdi)

Cc Corey Ashford <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc David Ahern <[email protected]>
---
 tools/perf/Makefile.perf                       |  4 ++--
 tools/perf/config/Makefile                     | 11 ++++++++++-
 tools/perf/config/feature-checks/Makefile      |  8 ++++++--
 tools/perf/config/feature-checks/test-all.c    |  5 +++++
 tools/perf/config/feature-checks/test-udis86.c | 11 +++++++++++
 5 files changed, 34 insertions(+), 5 deletions(-)
 create mode 100644 tools/perf/config/feature-checks/test-udis86.c
tools/perf/Makefile.perf +2 -2
index ca3b87d..5c4c8bd 100644
@@ -712,7 +712,7 @@ $(LIB_FILE): $(LIB_OBJS)
 TE_SOURCES = $(wildcard $(TRACE_EVENT_DIR)*.[ch])
 
 LIBTRACEEVENT_FLAGS  = $(QUIET_SUBDIR1) O=$(OUTPUT)
-LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS)"
+LIBTRACEEVENT_FLAGS += CFLAGS="-g -Wall $(EXTRA_CFLAGS) $(LIBTRACEEVENT_CFLAGS)"
 LIBTRACEEVENT_FLAGS += plugin_dir=$(plugindir_SQ)
 
 $(LIBTRACEEVENT): $(TE_SOURCES) $(OUTPUT)PERF-CFLAGS
@@ -793,7 +793,7 @@ cscope:
 
 ### Detect prefix changes
 TRACK_CFLAGS = $(subst ','\'',$(CFLAGS)):\
-             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ):$(plugindir_SQ)
+             $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ):$(plugindir_SQ):$(LIBTRACEEVENT_CFLAGS)
 
 $(OUTPUT)PERF-CFLAGS: .FORCE-PERF-CFLAGS
 	@FLAGS='$(TRACK_CFLAGS)'; \
tools/perf/config/Makefile +10 -1
index bae1072..ddae5c4 100644
@@ -142,7 +142,8 @@ CORE_FEATURE_TESTS =			\
 	libunwind			\
 	on-exit				\
 	stackprotector-all		\
-	timerfd
+	timerfd				\
+	udis86
 
 #
 # So here we detect whether test-all was rebuilt, to be able
@@ -413,6 +414,14 @@ else
   msg := $(warning No timerfd support. Disables 'perf kvm stat live');
 endif
 
+$(call feature_check,udis86)
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 found. Please install udis86-devel.)
+endif
+
 disable-python = $(eval $(disable-python_code))
 define disable-python_code
   CFLAGS += -DNO_LIBPYTHON
tools/perf/config/feature-checks/Makefile +6 -2
index b8bb749..3ed2667 100644
@@ -26,7 +26,8 @@ FILES=					\
 	test-libunwind-debug-frame	\
 	test-on-exit			\
 	test-stackprotector-all		\
-	test-timerfd
+	test-timerfd			\
+	test-udis86
 
 CC := $(CC) -MD
 
@@ -37,7 +38,7 @@ BUILD = $(CC) $(CFLAGS) $(LDFLAGS) -o $(OUTPUT)$@ [email protected]
 ###############################
 
 test-all:
-	$(BUILD) -Werror -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl
+	$(BUILD) -Werror -fstack-protector-all -O2 -Werror -D_FORTIFY_SOURCE=2 -ldw -lelf -lnuma $(LIBUNWIND_LIBS) -lelf -laudit -I/usr/include/slang -lslang $(shell pkg-config --libs --cflags gtk+-2.0 2>/dev/null) $(FLAGS_PERL_EMBED) $(FLAGS_PYTHON_EMBED) -DPACKAGE='"perf"' -lbfd -ldl -ludis86
 
 test-hello:
 	$(BUILD)
@@ -140,6 +141,9 @@ test-backtrace:
 test-timerfd:
 	$(BUILD)
 
+test-udis86:
+	$(BUILD) -DHAVE_UDIS86 -ludis86
+
 -include *.d
 
 ###############################
tools/perf/config/feature-checks/test-all.c +5 -0
index 9b8a544..df087a5 100644
@@ -89,6 +89,10 @@
 # include "test-stackprotector-all.c"
 #undef main
 
+#define main main_test_udis86
+# include "test-udis86.c"
+#undef main
+
 int main(int argc, char *argv[])
 {
 	main_test_libpython();
@@ -111,6 +115,7 @@ int main(int argc, char *argv[])
 	main_test_libnuma();
 	main_test_timerfd();
 	main_test_stackprotector_all();
+	main_test_udis86();
 
 	return 0;
 }
tools/perf/config/feature-checks/test-udis86.c new file +11 -0
new file mode 100644
index 0000000..dbc2c26
@@ -0,0 +1,11 @@
+
+/*
+ * test for udis86 functions used by plugin_kvm.c
+ */
+
+#include <udis86.h>
+
+int main(void)
+{
+	return puts("hi");
+}
Signature · 1 line
1.8.3.1
#65 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

In reply to Jiri Olsa
Quoted 21 lines · Jiri Olsa
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 1 line

http://bit.ly/1hyrN52

nice :-)

Quoted 1 line
Quoted 7 lines
Quoted 1 line

Wow, that was easy, but yeah, could be made easier 8-) ;-P

so something like:
No udis86 found. Please install udis86-devel.

or:
No udis86 found, disabling kvm tracepoints instruction disassembly. Please install udis86-devel.

1)

So, the first problem I can see is in the output:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

config/Makefile:421: No udis86 support.

such a 'no XYZ support' message should only occur if a feature test failed. _If_ we decide that 'udis86' support is required for a full perf build then there should be a new line saying something like:

... udis86: [ OFF ]

2)

The other problem I see is that I think we should start adding the install-suggestions to the feature-check lines themselves.

On .deb based systems it should say something like:

...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

... udis86: [ OFF ] # apt-get install udis86-dev

On RPM based systems it should say something like:

...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

... udis86: [ OFF ] # yum install udis86-devel

Or so. I think we can maintain package suggestion strings for these two main distro variants.

The package install strings should probably be maintained together with the list of feature names in some table/list format. Such lookup tables can be implemented either in Make via:

http://www.gnu.org/savannah-checkouts/gnu/make/manual/html_node/Computed-Names.html

Or via a shell command in the makefile, using Bash associative arrays or such.

Thanks,

Ingo

author
#66 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

Quoted 58 lines · Ingo Molnar
Quoted 21 lines · Jiri Olsa
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 1 line

http://bit.ly/1hyrN52

nice :-)

Quoted 1 line
Quoted 7 lines
Quoted 1 line

Wow, that was easy, but yeah, could be made easier 8-) ;-P

so something like:
No udis86 found. Please install udis86-devel.

or:
No udis86 found, disabling kvm tracepoints instruction disassembly. Please install udis86-devel.

1)

So, the first problem I can see is in the output:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]
Quoted 7 lines

config/Makefile:421: No udis86 support.

such a 'no XYZ support' message should only occur if a feature test failed. _If_ we decide that 'udis86' support is required for a full perf build then there should be a new line saying something like:

... udis86: [ OFF ]

hum, just sent v4 that added udis86 into above list ;-)

So.. do we consider 'full perf build' to be build with all possible libraries linked in?

then we need to cleanup above list and remove:
on-exit
timerfd
libelf-getphdrnum
libelf-mmap
stackprotector-all
fortify-source
backtrace

or maybe have verbose list where we display everything (for V=1 ?) and by default display only library checks

Quoted 23 lines

2)

The other problem I see is that I think we should start adding the install-suggestions to the feature-check lines themselves.

On .deb based systems it should say something like:

...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

... udis86: [ OFF ] # apt-get install udis86-dev

On RPM based systems it should say something like:

...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

... udis86: [ OFF ] # yum install udis86-devel

Or so. I think we can maintain package suggestion strings for these two main distro variants.

Quoted 8 lines

The package install strings should probably be maintained together with the list of feature names in some table/list format. Such lookup tables can be implemented either in Make via:

http://www.gnu.org/savannah-checkouts/gnu/make/manual/html_node/Computed-Names.html

Or via a shell command in the makefile, using Bash associative arrays or such.

sounds good

jirka

#67 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

Quoted 65 lines · Jiri Olsa
Quoted 64 lines · Ingo Molnar
Quoted 21 lines · Jiri Olsa
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Tue, Dec 03, 2013 at 02:09:42PM +0100, Jiri Olsa escreveu:

tools/perf/config/Makefile else · line 422

Quoted 10 lines

Adding udis86 disassembler feature check which support is needed for kvm:kvm_emulate_insn tracepoint.

+$(call feature_check,udis86)

tools/perf/config/Makefile +6 -0
+ifeq ($(feature-udis86), 1)
+  LIBTRACEEVENT_CFLAGS += -DHAVE_UDIS86
+  EXTLIBS += -ludis86
+else
+  msg := $(warning No udis86 support.);
+endif

That is really an incomplete message, what package should I install? Perhaps we should add this there then:

Quoted 1 line

http://bit.ly/1hyrN52

nice :-)

Quoted 1 line
Quoted 7 lines
Quoted 1 line

Wow, that was easy, but yeah, could be made easier 8-) ;-P

so something like:
No udis86 found. Please install udis86-devel.

or:
No udis86 found, disabling kvm tracepoints instruction disassembly. Please install udis86-devel.

1)

So, the first problem I can see is in the output:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

config/Makefile:421: No udis86 support.

such a 'no XYZ support' message should only occur if a feature test failed. _If_ we decide that 'udis86' support is required for a full perf build then there should be a new line saying something like:

Quoted 6 lines
Quoted 1 line

... udis86: [ OFF ]

hum, just sent v4 that added udis86 into above list ;-)

So.. do we consider 'full perf build' to be build with all possible libraries linked in?

So my worry is, how widely is libudis86 available in distros? What is the package name on Debian/Ubuntu for example?

Thanks,

Ingo

author
#68 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

On Thu, Dec 05, 2013 at 11:45:46AM +0100, Ingo Molnar wrote:

SNIP

Quoted 35 lines
Quoted 35 lines
Quoted 34 lines

1)

So, the first problem I can see is in the output:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]

config/Makefile:421: No udis86 support.

such a 'no XYZ support' message should only occur if a feature test failed. _If_ we decide that 'udis86' support is required for a full perf build then there should be a new line saying something like:

... udis86: [ OFF ]

Quoted 7 lines
Quoted 4 lines

hum, just sent v4 that added udis86 into above list ;-)

So.. do we consider 'full perf build' to be build with all possible libraries linked in?

So my worry is, how widely is libudis86 available in distros? What is the package name on Debian/Ubuntu for example?

I did not find any.. :-\ I can see Fedora and Gentoo

thanks,
jirka

#69 lore raw

Re: [PATCH 28/28] perf tools: Add udis86 disassembler feature check

Quoted 41 lines · Jiri Olsa

On Thu, Dec 05, 2013 at 11:45:46AM +0100, Ingo Molnar wrote:

SNIP

Quoted 37 lines
Quoted 37 lines
Quoted 27 lines

1)

So, the first problem I can see is in the output:

Auto-detecting system features:

...                     backtrace: [ on  ]
...                         dwarf: [ on  ]
...                fortify-source: [ on  ]
...                         glibc: [ on  ]
...                          gtk2: [ on  ]
...                  gtk2-infobar: [ on  ]
...                      libaudit: [ on  ]
...                        libbfd: [ on  ]
...                        libelf: [ on  ]
...             libelf-getphdrnum: [ on  ]
...                   libelf-mmap: [ on  ]
...                       libnuma: [ on  ]
...                       libperl: [ on  ]
...                     libpython: [ on  ]
...             libpython-version: [ on  ]
...                      libslang: [ on  ]
...                     libunwind: [ on  ]
...                       on-exit: [ on  ]
...            stackprotector-all: [ on  ]
...                       timerfd: [ on  ]
Quoted 7 lines

config/Makefile:421: No udis86 support.

such a 'no XYZ support' message should only occur if a feature test failed. _If_ we decide that 'udis86' support is required for a full perf build then there should be a new line saying something like:

... udis86: [ OFF ]

hum, just sent v4 that added udis86 into above list ;-)

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

So.. do we consider 'full perf build' to be build with all possible libraries linked in?

So my worry is, how widely is libudis86 available in distros? What is the package name on Debian/Ubuntu for example?

I did not find any.. :-\ I can see Fedora and Gentoo

That's a problem really - all perf developers who are using Ubuntu will see the failure, without being able to do anything about it.

So maybe we need an intermediate feature-mode for such convenience features that don't really impact measurement capabilities, those features could be probed but failures would not be displayed in such a highly prominent fashion.

Thanks,

Ingo