Skip to contentSkip to conversation

perf tools: Add traceevent plugins support

PATCH Superseded Jiri Olsa 5 participants 45 messages Nov 28, 2013 – Dec 4, 2013 perf events

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Jiri Olsa

author
#0 lore raw

v2 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

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 (29):
perf tools: Remove stackprotector feature check
perf tools: Include test-stackprotector-all.c in test-all tools lib traceevent: Add plugin support
tools lib traceevent: Add plugin build support
tools lib traceevent: Add traceevent_host_bigendian 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 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 udis86 disassembler feature check
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/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            |  10 ++
 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                                 |  50 +++++++++
 tools/perf/util/util.h                                 |   1 +
 34 files changed, 2220 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 1/29 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 80ea6d8..38f388c2 100644
@@ -161,7 +161,6 @@ CORE_FEATURE_TESTS =			\
 	libslang			\
 	libunwind			\
 	on-exit				\
-	stackprotector			\
 	stackprotector-all		\
 	timerfd
 
@@ -229,10 +228,6 @@ 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 8dffb62..fc78699 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 -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 -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

Re: [PATCH 01/29] perf tools: Remove stackprotector feature check

Em Thu, Nov 28, 2013 at 12:33:00PM +0100, Jiri Olsa escreveu:

Quoted 6 lines

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.

Doesn't applies to my perf/core branch.

- Arnaldo

author
#3 lore raw

PATCH v3 1/29 perf tools: Remove stackprotector feature check

Quoted 9 lines · Arnaldo Carvalho de Melo

Em Thu, Nov 28, 2013 at 12:33:00PM +0100, Jiri Olsa escreveu:

Quoted 6 lines

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.

Doesn't applies to my perf/core branch.

yep, that's that libunwind stuff messing up ;-)

attaching v3, I could resend whole rebased patchset if needed.. I also updated my perf/core_plugins branch

thanks,
jirka

---
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 80ea6d8..38f388c2 100644
@@ -161,7 +161,6 @@ CORE_FEATURE_TESTS =			\
 	libslang			\
 	libunwind			\
 	on-exit				\
-	stackprotector			\
 	stackprotector-all		\
 	timerfd
 
@@ -229,10 +228,6 @@ 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 8dffb62..fc78699 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 -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 -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
Acked-by on v3 1/29
#4 lore raw

Re: [PATCHv3 01/29] perf tools: Remove stackprotector feature check

Quoted 38 lines · Jiri Olsa
Quoted 9 lines · Arnaldo Carvalho de Melo

Em Thu, Nov 28, 2013 at 12:33:00PM +0100, Jiri Olsa escreveu:

Quoted 6 lines

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.

Doesn't applies to my perf/core branch.

yep, that's that libunwind stuff messing up ;-)

attaching v3, I could resend whole rebased patchset if needed.. I also updated my perf/core_plugins branch

thanks,
jirka

---
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]>
Quoted 6 lines
---
 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

This should also speed up the feature tests by about 10-20 msecs in the test-all case.

Thanks,

Ingo

author
#5 lore raw

PATCH 2/29 perf tools: Include test-stackprotector-all.c in test-all

In reply to Jiri Olsa

We're missing test-stackprotector-all.c in test-all check, adding it.

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/feature-checks/test-all.c | 5 +++++
 1 file changed, 5 insertions(+)
tools/perf/config/feature-checks/test-all.c +5 -0
index 59e7a70..9b8a544 100644
@@ -85,6 +85,10 @@
 # include "test-timerfd.c"
 #undef main
 
+#define main main_test_stackprotector_all
+# include "test-stackprotector-all.c"
+#undef main
+
 int main(int argc, char *argv[])
 {
 	main_test_libpython();
@@ -106,6 +110,7 @@ int main(int argc, char *argv[])
 	main_test_backtrace();
 	main_test_libnuma();
 	main_test_timerfd();
+	main_test_stackprotector_all();
 
 	return 0;
 }
Signature · 1 line
1.8.3.1
#6 lore raw

[tip:perf/core] perf tools: Include test-stackprotector-all.c in test-all

perf tools: Include test-stackprotector-all.c in test-all
Commit-ID:  bc32358b569247ceabba14dab297070a8fb255c4
Gitweb:     http://git.kernel.org/tip/bc32358b569247ceabba14dab297070a8fb255c4
Author:     Jiri Olsa <[email protected]>
AuthorDate: Thu, 28 Nov 2013 12:33:01 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 2 Dec 2013 16:19:26 -0300

perf tools: Include test-stackprotector-all.c in test-all

We're missing test-stackprotector-all.c in test-all check, adding it.

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/feature-checks/test-all.c | 5 +++++
 1 file changed, 5 insertions(+)
tools/perf/config/feature-checks/test-all.c +5 -0
index 59e7a70..9b8a544 100644
@@ -85,6 +85,10 @@
 # include "test-timerfd.c"
 #undef main
 
+#define main main_test_stackprotector_all
+# include "test-stackprotector-all.c"
+#undef main
+
 int main(int argc, char *argv[])
 {
 	main_test_libpython();
@@ -106,6 +110,7 @@ int main(int argc, char *argv[])
 	main_test_backtrace();
 	main_test_libnuma();
 	main_test_timerfd();
+	main_test_stackprotector_all();
 
 	return 0;
 }
author
#7 lore raw

PATCH 3/29 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
author
#8 lore raw

PATCH 4/29 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
author
#9 lore raw

PATCH 5/29 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
author
#10 lore raw

PATCH 6/29 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 1526798..474485a 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
#11 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

Em Thu, Nov 28, 2013 at 12:33:05PM +0100, Jiri Olsa escreveu:

Quoted 5 lines

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

Quoted 7 lines

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')

The lines after one that starts with

---

are ignored by 'git am', please ident them with at least one space, I'm doing that this time.

Quoted 115 lines
script.jbd2.new renamed from script.jbd2.old +2 -2
-     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 1526798..474485a 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
new file mode 100644
index 0000000..5d85de7
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2010 Red Hat Inc, Steven Rostedt <[email protected]>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation;
+ * version 2.1 of the License (not later!)
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this program; if not,  see <http://www.gnu.org/licenses>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "event-parse.h"
+
+#define MINORBITS	20
+#define MINORMASK	((1U << MINORBITS) - 1)
+
+#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)
+{
+	unsigned int dev = args[0];
+
+	trace_seq_printf(s, "%d:%d", MAJOR(dev), MINOR(dev));
+	return 0;
+}
+
+unsigned long long process_jiffies_to_msecs(struct trace_seq *s,
+					    unsigned long long *args)
+{
+	unsigned long long jiffies = args[0];
+
+	trace_seq_printf(s, "%lld", jiffies);
+	return jiffies;
+}
+
+int PEVENT_PLUGIN_LOADER(struct pevent *pevent)
+{
+	pevent_register_print_function(pevent,
+				       process_jbd2_dev_to_name,
+				       PEVENT_FUNC_ARG_STRING,
+				       "jbd2_dev_to_name",
+				       PEVENT_FUNC_ARG_INT,
+				       PEVENT_FUNC_ARG_VOID);
+
+	pevent_register_print_function(pevent,
+				       process_jiffies_to_msecs,
+				       PEVENT_FUNC_ARG_LONG,
+				       "jiffies_to_msecs",
+				       PEVENT_FUNC_ARG_LONG,
+				       PEVENT_FUNC_ARG_VOID);
+	return 0;
+}
Signature · 1 line
1.8.3.1
#12 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

Em Fri, Nov 29, 2013 at 03:42:35PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 10 lines

Em Thu, Nov 28, 2013 at 12:33:05PM +0100, Jiri Olsa escreveu:

Quoted 6 lines

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')

The lines after one that starts with

Quoted 7 lines

---

are ignored by 'git am', please ident them with at least one space, I'm doing that this time.

script.jbd2.new

Quoted 2 lines
script.jbd2.new renamed from script.jbd2.old +0 -0

Also I just tried:

[acme@ssdandy linux]$ git log --oneline | head -7 b0496b02fcea tools lib traceevent: Add hrtimer plugin 1ec7a71e0b2c tools lib traceevent: Add jbd2 plugin c63100c95c6e tools lib traceevent: Add traceevent_host_bigendian function d54b335656b7 tools lib traceevent: Add plugin build support e796a0d73bcc tools lib traceevent: Add plugin support 28c803842250 perf tools: Include test-stackprotector-all.c in test-all 1a69f6ce513d perf timechart: Move wake_events list to 'struct timechart' [acme@ssdandy linux]$

[acme@ssdandy linux]$ time make O=/tmp/build/perf -C tools/perf/ install <SNIP successful build>

[acme@ssdandy linux]$ ls -la ~/.traceevent
ls: cannot access /home/acme/.traceevent: No such file or directory [acme@ssdandy linux]$

Where did the plugins go? At least two should be there by now, no?

- Arnaldo

author
#13 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

Quoted 31 lines · Arnaldo Carvalho de Melo

Em Fri, Nov 29, 2013 at 03:42:35PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 10 lines

Em Thu, Nov 28, 2013 at 12:33:05PM +0100, Jiri Olsa escreveu:

Quoted 6 lines

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')

The lines after one that starts with

Quoted 7 lines

---

are ignored by 'git am', please ident them with at least one space, I'm doing that this time.

script.jbd2.new

Quoted 2 lines
script.jbd2.new renamed from script.jbd2.old +0 -0

Also I just tried:

[acme@ssdandy linux]$ git log --oneline | head -7 b0496b02fcea tools lib traceevent: Add hrtimer plugin 1ec7a71e0b2c tools lib traceevent: Add jbd2 plugin c63100c95c6e tools lib traceevent: Add traceevent_host_bigendian function d54b335656b7 tools lib traceevent: Add plugin build support e796a0d73bcc tools lib traceevent: Add plugin support 28c803842250 perf tools: Include test-stackprotector-all.c in test-all 1a69f6ce513d perf timechart: Move wake_events list to 'struct timechart' [acme@ssdandy linux]$

Quoted 8 lines

[acme@ssdandy linux]$ time make O=/tmp/build/perf -C tools/perf/ install <SNIP successful build>

[acme@ssdandy linux]$ ls -la ~/.traceevent
ls: cannot access /home/acme/.traceevent: No such file or directory [acme@ssdandy linux]$

Where did the plugins go? At least two should be there by now, no?

At this stage (patches applied) you'd need to run the make for traceevent directory.

The perf installation link is done within this patch:
perf tools: Add build and install plugins targets

which I presume you haven't applied yet?

jirka

#14 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

Em Sat, Nov 30, 2013 at 05:49:21PM +0100, Jiri Olsa escreveu:

Quoted 23 lines · Arnaldo Carvalho de Melo
Quoted 13 lines

Em Fri, Nov 29, 2013 at 03:42:35PM -0300, Arnaldo Carvalho de Melo escreveu:
Also I just tried:

[acme@ssdandy linux]$ git log --oneline | head -7 b0496b02fcea tools lib traceevent: Add hrtimer plugin 1ec7a71e0b2c tools lib traceevent: Add jbd2 plugin c63100c95c6e tools lib traceevent: Add traceevent_host_bigendian function d54b335656b7 tools lib traceevent: Add plugin build support e796a0d73bcc tools lib traceevent: Add plugin support 28c803842250 perf tools: Include test-stackprotector-all.c in test-all 1a69f6ce513d perf timechart: Move wake_events list to 'struct timechart' [acme@ssdandy linux]$

Quoted 8 lines

[acme@ssdandy linux]$ time make O=/tmp/build/perf -C tools/perf/ install <SNIP successful build>

[acme@ssdandy linux]$ ls -la ~/.traceevent
ls: cannot access /home/acme/.traceevent: No such file or directory [acme@ssdandy linux]$

Where did the plugins go? At least two should be there by now, no?

Quoted 7 lines

At this stage (patches applied) you'd need to run the make for traceevent directory.

The perf installation link is done within this patch:
perf tools: Add build and install plugins targets

which I presume you haven't applied yet?

No I haven't, and I expected that I could test it straight away, so I suggest you reorder the patches so that, right after the first plugin is introduced a:

make -C tools/perf install

in all its variants (in place, with O=, etc) allows one to test it.

- Arnaldo

author
#15 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

Quoted 32 lines · Arnaldo Carvalho de Melo

Em Sat, Nov 30, 2013 at 05:49:21PM +0100, Jiri Olsa escreveu:

Quoted 23 lines · Arnaldo Carvalho de Melo
Quoted 13 lines

Em Fri, Nov 29, 2013 at 03:42:35PM -0300, Arnaldo Carvalho de Melo escreveu:
Also I just tried:

[acme@ssdandy linux]$ git log --oneline | head -7 b0496b02fcea tools lib traceevent: Add hrtimer plugin 1ec7a71e0b2c tools lib traceevent: Add jbd2 plugin c63100c95c6e tools lib traceevent: Add traceevent_host_bigendian function d54b335656b7 tools lib traceevent: Add plugin build support e796a0d73bcc tools lib traceevent: Add plugin support 28c803842250 perf tools: Include test-stackprotector-all.c in test-all 1a69f6ce513d perf timechart: Move wake_events list to 'struct timechart' [acme@ssdandy linux]$

Quoted 8 lines

[acme@ssdandy linux]$ time make O=/tmp/build/perf -C tools/perf/ install <SNIP successful build>

[acme@ssdandy linux]$ ls -la ~/.traceevent
ls: cannot access /home/acme/.traceevent: No such file or directory [acme@ssdandy linux]$

Where did the plugins go? At least two should be there by now, no?

Quoted 7 lines

At this stage (patches applied) you'd need to run the make for traceevent directory.

The perf installation link is done within this patch:
perf tools: Add build and install plugins targets

which I presume you haven't applied yet?

Quoted 7 lines

No I haven't, and I expected that I could test it straight away, so I suggest you reorder the patches so that, right after the first plugin is introduced a:

make -C tools/perf install

in all its variants (in place, with O=, etc) allows one to test it.

ok, I will send v3

jirka

author
#16 lore raw

Re: [PATCH 06/29] tools lib traceevent: Add jbd2 plugin

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

Em Thu, Nov 28, 2013 at 12:33:05PM +0100, Jiri Olsa escreveu:

Quoted 5 lines

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

Quoted 7 lines

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')

Quoted 6 lines

The lines after one that starts with

---

are ignored by 'git am', please ident them with at least one space, I'm doing that this time.

oops, ok

thanks,
jirka

author
#17 lore raw

PATCH 7/29 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 474485a..23a0ff7 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
author
#18 lore raw

PATCH 8/29 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 23a0ff7..8e546e4 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
author
#19 lore raw

PATCH 9/29 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 8e546e4..aa95783 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
author
#20 lore raw

PATCH 10/29 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 aa95783..2c7c5e6 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
author
#21 lore raw

PATCH 11/29 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 2c7c5e6..726e9bb 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
author
#22 lore raw

PATCH 12/29 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 726e9bb..63ef6c8 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
author
#23 lore raw

PATCH 13/29 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 63ef6c8..4356a6e 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
author
#24 lore raw

PATCH 14/29 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 4356a6e..5c178ea 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
author
#25 lore raw

PATCH 15/29 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 5c178ea..19afd03 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
author
#26 lore raw

PATCH 16/29 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
author
#27 lore raw

PATCH 17/29 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
author
#28 lore raw

PATCH 18/29 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
author
#29 lore raw

PATCH 19/29 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
author
#30 lore raw

PATCH 20/29 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 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/plugin_function.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
tools/lib/traceevent/plugin_function.c +13 -2
index 87acf9c..328d17d 100644
@@ -44,10 +44,16 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 		free(stack->stack[pos]);
 	else {
 		if (!stack->stack)
-			stack->stack = malloc_or_die(sizeof(char *) * STK_BLK);
+			stack->stack = malloc(sizeof(char *) * STK_BLK);
 		else
 			stack->stack = realloc(stack->stack, sizeof(char *) *
 					       (stack->size + STK_BLK));
+
+		if (!stack->stack) {
+			warning("could not allocate plugin memory\n");
+			return;
+		}
+
 		for (i = stack->size; i < stack->size + STK_BLK; i++)
 			stack->stack[i] = NULL;
 		stack->size += STK_BLK;
@@ -67,7 +73,12 @@ static int add_and_get_index(const char *parent, const char *child, int cpu)
 		if (fstack)
 			fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
 		else
-			fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+			fstack = malloc(sizeof(*fstack) * (cpu + 1));
+
+		if (!fstack) {
+			warning("could not allocate plugin memory\n");
+			return 0;
+		}
 
 		/* Account for holes in the cpu count */
 		for (i = cpus + 1; i <= cpu; i++)
Signature · 1 line
1.8.3.1
maintainer
#31 lore raw

Re: [PATCH 20/29] tools lib traceevent: Remove malloc_or_die from plugin_function.c

Hi Jiri,

Quoted 23 lines · Jiri Olsa

Removing malloc_or_die calls from plugin_function.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/plugin_function.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
tools/lib/traceevent/plugin_function.c +0 -0
index 87acf9c..328d17d 100644

tools/lib/traceevent/plugin_function.c static void add_child(struct func_stack *stack, const char *child, int pos) · line 50

Quoted 8 lines
tools/lib/traceevent/plugin_function.c +1 -1
@@ -44,10 +44,16 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 		free(stack->stack[pos]);
 	else {
 		if (!stack->stack)
-			stack->stack = malloc_or_die(sizeof(char *) * STK_BLK);
+			stack->stack = malloc(sizeof(char *) * STK_BLK);
 		else
 			stack->stack = realloc(stack->stack, sizeof(char *) *
 					       (stack->size + STK_BLK));

I think single realloc() can handle both cases. And this code has a problem that it overwrites stack->stack to NULL in case of error so that we cannot point original region anymore. You'd better to use a temp variable IMHO.

Quoted 10 lines
tools/lib/traceevent/plugin_function.c +6 -0
+
+		if (!stack->stack) {
+			warning("could not allocate plugin memory\n");
+			return;
+		}
+
 		for (i = stack->size; i < stack->size + STK_BLK; i++)
 			stack->stack[i] = NULL;
 		stack->size += STK_BLK;

tools/lib/traceevent/plugin_function.c static int add_and_get_index(const char *parent, const char *child, int cpu) · line 76

Quoted 5 lines
tools/lib/traceevent/plugin_function.c +1 -1
@@ -67,7 +73,12 @@ static int add_and_get_index(const char *parent, const char *child, int cpu)
 		if (fstack)
 			fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
 		else
-			fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+			fstack = malloc(sizeof(*fstack) * (cpu + 1));

Ditto.

Thanks,
Namhyung

Quoted 8 lines
tools/lib/traceevent/plugin_function.c +5 -0
+
+		if (!fstack) {
+			warning("could not allocate plugin memory\n");
+			return 0;
+		}
 
 		/* Account for holes in the cpu count */
 		for (i = cpus + 1; i <= cpu; i++)
author
#32 lore raw

PATCH v3 20/29 tools lib traceevent: Remove malloc_or_die from plugin_function.c

Quoted 31 lines · Namhyung Kim

Hi Jiri,

Quoted 28 lines · Jiri Olsa

Removing malloc_or_die calls from plugin_function.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/plugin_function.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
tools/lib/traceevent/plugin_function.c +1 -1
index 87acf9c..328d17d 100644
@@ -44,10 +44,16 @@ static void add_child(struct func_stack *stack, const char *child, int pos)
 		free(stack->stack[pos]);
 	else {
 		if (!stack->stack)
-			stack->stack = malloc_or_die(sizeof(char *) * STK_BLK);
+			stack->stack = malloc(sizeof(char *) * STK_BLK);
Quoted 8 lines

tools/lib/traceevent/plugin_function.c static void add_child(struct func_stack *stack, const char *child, int pos) · line 50

Quoted 3 lines
tools/lib/traceevent/plugin_function.c +0 -0
 		else
 			stack->stack = realloc(stack->stack, sizeof(char *) *
 					       (stack->size + STK_BLK));

I think single realloc() can handle both cases. And this code has a problem that it overwrites stack->stack to NULL in case of error so that we cannot point original region anymore. You'd better to use a temp variable IMHO.

right you are, please check v3 attached..

thanks,
jirka

---
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 | 25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)
tools/lib/traceevent/plugin_function.c +16 -9
index 87acf9c..b43cade 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 +70,11 @@ 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));
+		fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
+		if (!fstack) {
+			warning("could not allocate plugin memory\n");
+			return 0;
+		}
 
 		/* Account for holes in the cpu count */
 		for (i = cpus + 1; i <= cpu; i++)
Signature · 1 line
1.8.3.1
maintainer
#33 lore raw

Re: [PATCHv3 20/29] tools lib traceevent: Remove malloc_or_die from plugin_function.c

Quoted 3 lines

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

[SNIP]

tools/lib/traceevent/plugin_function.c static int add_and_get_index(const char *parent, const char *child, int cpu) · line 73

Quoted 7 lines
        if (cpu > cpus) {
tools/lib/traceevent/plugin_function.c +1 -4
-               if (fstack)
-                       fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
-               else
-                       fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+               fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));

This code also has same problem..

Thanks,
Namhyung

Quoted 10 lines
tools/lib/traceevent/plugin_function.c +4 -1
+               if (!fstack) {
+                       warning("could not allocate plugin memory\n");
+                       return 0;
+               }
 
                /* Account for holes in the cpu count */
                for (i = cpus + 1; i <= cpu; i++)
--

1.8.3.1

author
#34 lore raw

PATCH v3 20/29 tools lib traceevent: Remove malloc_or_die from plugin_function.c

Quoted 6 lines · Namhyung Kim
Quoted 3 lines

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

[SNIP]

Quoted 1 line
Quoted 9 lines

Review

Quoted 6 lines
        if (cpu > cpus) {
(quoted diff) +1 -4
-               if (fstack)
-                       fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
-               else
-                       fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+               fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));

This code also has same problem..

and v3.1 ;-) thanks

jirka

---
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
maintainer
#35 lore raw

Re: [PATCHv3 20/29] tools lib traceevent: Remove malloc_or_die from plugin_function.c

Quoted 14 lines · Jiri Olsa
Quoted 13 lines · Namhyung Kim
Quoted 3 lines

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

[SNIP]

Quoted 7 lines
        if (cpu > cpus) {
tools/lib/traceevent/plugin_function.c +1 -4
-               if (fstack)
-                       fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
-               else
-                       fstack = malloc_or_die(sizeof(*fstack) * (cpu + 1));
+               fstack = realloc(fstack, sizeof(*fstack) * (cpu + 1));
Quoted 4 lines
Quoted 2 lines

This code also has same problem..

and v3.1 ;-) thanks

Thank you for your effort.

Now the whole series looks all good to me (with other changes fixed).

Thanks,
Namhyung

author
#36 lore raw

PATCH 21/29 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
author
#37 lore raw

PATCH 22/29 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
author
#38 lore raw

PATCH 23/29 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 19afd03..0d9cbb4 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
author
#39 lore raw

PATCH 24/29 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 38f388c2..b7756f0 100644
@@ -597,3 +597,11 @@ 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
author
#40 lore raw

PATCH 25/29 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 abc7ac3..4c894f4d 100644
@@ -711,7 +711,7 @@ $(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 b7756f0..021c438 100644
@@ -417,6 +417,14 @@ 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 fc78699..ed68814 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
author
#41 lore raw

PATCH 26/29 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 | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/util.h |  1 +
 2 files changed, 51 insertions(+)
tools/perf/util/util.c +50 -0
index 28a0a89..861231c 100644
@@ -6,6 +6,9 @@
 #endif
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+#include <linux/kernel.h>
 
 /*
  * XXX We need to find a better place for these things...
@@ -413,3 +416,50 @@ 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 c8f362d..e54bb10 100644
@@ -307,4 +307,5 @@ 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
author
#42 lore raw

PATCH 27/29 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 4c894f4d..be336f8 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 1cd0357..0af4470 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
author
#43 lore raw

PATCH 28/29 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 e9f345e2..47bf590 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
author
#44 lore raw

PATCH 29/29 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