perf tools: Add libdw DWARF unwind support
Subsystems (2)
cover perf tools: Add libdw DWARF unwind support
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.
We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
Examples:
- compile in libdw unwinder if present:
$ make NO_LIBUNWIND=1
- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1
- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1
Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldw
real 0m35.746s user 0m24.211s sys 0m11.214s
The libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unw
real 1m41.149s user 1m9.694s sys 0m30.595s
In example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwind
Kudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)
thanks,
jirka
---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Make perf_event__synthesize_mmap_events global perf tests x86: Introduce perf_regs_load function
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwind max_stack processing
perf tools: Do not report zero address in unwind
perf tools: Add mask into struct regs_dump
perf tools: Separate libunwind code to special object perf tools: Rename unwind__arch_reg_id into libunwind__arch_reg_id perf tools: Introduce HAVE_DWARF_UNWIND_SUPPORT macro perf tools: Separate perf_reg_value function in perf_regs object perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: Add NO_LIBDW_DWARF_UNWIND make test
tools/perf/Makefile.perf | 19 ++++++-
tools/perf/arch/arm/Makefile | 2 +-
tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +-
tools/perf/arch/x86/Makefile | 9 +++-
tools/perf/arch/x86/include/perf_regs.h | 6 +++
tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++
tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++
tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++
tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +-
tools/perf/builtin-inject.c | 1 -
tools/perf/builtin-record.c | 10 ++--
tools/perf/config/Makefile | 94 +++++++++++++++++++++++---------
tools/perf/config/feature-checks/Makefile | 6 ++-
tools/perf/config/feature-checks/test-all.c | 5 ++
tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++
tools/perf/tests/builtin-test.c | 8 +++
tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++
tools/perf/tests/make | 63 +++++++++++-----------
tools/perf/tests/sample-parsing.c | 17 +++---
tools/perf/tests/tests.h | 9 ++++
tools/perf/util/event.c | 12 ++---
tools/perf/util/event.h | 12 ++++-
tools/perf/util/evsel.c | 13 ++---
tools/perf/util/machine.c | 4 +-
tools/perf/util/perf_regs.c | 19 +++++++
tools/perf/util/perf_regs.h | 13 +++++
tools/perf/util/session.c | 5 +-
tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
tools/perf/util/unwind-libdw.h | 21 ++++++++
tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++-----------
tools/perf/util/unwind.h | 11 ++--
31 files changed, 849 insertions(+), 135 deletions(-)
rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%)
create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c
create mode 100644 tools/perf/arch/x86/tests/regs_load.S
create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c
rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%)
create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c
create mode 100644 tools/perf/tests/dwarf-unwind.c
create mode 100644 tools/perf/util/perf_regs.c
create mode 100644 tools/perf/util/unwind-libdw.c
create mode 100644 tools/perf/util/unwind-libdw.h
rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)PATCH 1/16 perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables
Instead of explicitly adding same value into FEATURE_CHECK_(C|LD)FLAGS-all variables we can do that automatically.
--- tools/perf/config/Makefile | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
tools/perf/config/Makefile +11 -3
@@ -53,9 +53,6 @@ else FEATURE_CHECK_LDFLAGS-libunwind = $(LIBUNWIND_LDFLAGS) FEATURE_CHECK_CFLAGS-libunwind-debug-frame = $(LIBUNWIND_CFLAGS) FEATURE_CHECK_LDFLAGS-libunwind-debug-frame = $(LIBUNWIND_LDFLAGS) - # and the flags for the test-all case - FEATURE_CHECK_CFLAGS-all += $(LIBUNWIND_CFLAGS) - FEATURE_CHECK_LDFLAGS-all += $(LIBUNWIND_LDFLAGS) endif ifeq ($(NO_PERF_REGS),0)
@@ -152,6 +149,17 @@ CORE_FEATURE_TESTS = \ stackprotector-all \ timerfd +# Set FEATURE_CHECK_(C|LD)FLAGS-all for all CORE_FEATURE_TESTS features. +# If in the future we need per-feature checks/flags for features not +# mentioned in this list we need to refactor this ;-). +set_test_all_flags = $(eval $(set_test_all_flags_code)) +define set_test_all_flags_code + FEATURE_CHECK_CFLAGS-all += $(FEATURE_CHECK_CFLAGS-$(1)) + FEATURE_CHECK_LDFLAGS-all += $(FEATURE_CHECK_LDFLAGS-$(1)) +endef + +$(foreach feat,$(CORE_FEATURE_TESTS),$(call set_test_all_flags,$(feat))) + # # So here we detect whether test-all was rebuilt, to be able # to skip the print-out of the long features list if the file
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf tools: Automate setup of FEATURE_CHECK_(C|LD )FLAGS-all variables
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables
Commit-ID: c4eb6c0e7aa3a5106a3382880bc41c696e72adb4 Gitweb: http://git.kernel.org/tip/c4eb6c0e7aa3a5106a3382880bc41c696e72adb4 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:18 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Mon, 13 Jan 2014 10:06:24 -0300
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables
Instead of explicitly adding same value into FEATURE_CHECK_(C|LD)FLAGS-all variables we can do that automatically.
--- tools/perf/config/Makefile | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
tools/perf/config/Makefile +11 -3
@@ -53,9 +53,6 @@ else FEATURE_CHECK_LDFLAGS-libunwind = $(LIBUNWIND_LDFLAGS) FEATURE_CHECK_CFLAGS-libunwind-debug-frame = $(LIBUNWIND_CFLAGS) FEATURE_CHECK_LDFLAGS-libunwind-debug-frame = $(LIBUNWIND_LDFLAGS) - # and the flags for the test-all case - FEATURE_CHECK_CFLAGS-all += $(LIBUNWIND_CFLAGS) - FEATURE_CHECK_LDFLAGS-all += $(LIBUNWIND_LDFLAGS) endif ifeq ($(NO_PERF_REGS),0)
@@ -152,6 +149,17 @@ CORE_FEATURE_TESTS = \ stackprotector-all \ timerfd +# Set FEATURE_CHECK_(C|LD)FLAGS-all for all CORE_FEATURE_TESTS features. +# If in the future we need per-feature checks/flags for features not +# mentioned in this list we need to refactor this ;-). +set_test_all_flags = $(eval $(set_test_all_flags_code)) +define set_test_all_flags_code + FEATURE_CHECK_CFLAGS-all += $(FEATURE_CHECK_CFLAGS-$(1)) + FEATURE_CHECK_LDFLAGS-all += $(FEATURE_CHECK_LDFLAGS-$(1)) +endef + +$(foreach feat,$(CORE_FEATURE_TESTS),$(call set_test_all_flags,$(feat))) + # # So here we detect whether test-all was rebuilt, to be able # to skip the print-out of the long features list if the file
PATCH 2/16 perf tools: Fix machine initialization
In reply to Jiri OlsaThe id_hdr_size field was not properly initialized, setting it to zero.
--- tools/perf/util/machine.c | 1 + 1 file changed, 1 insertion(+)
tools/perf/util/machine.c +1 -0
@@ -27,6 +27,7 @@ int machine__init(struct machine *machine, const char *root_dir, pid_t pid) machine->pid = pid; machine->symbol_filter =NULL ; + machine->id_hdr_size =0 ; machine->root_dir =strdup (root_dir);if (machine->root_dir ==NULL )
Signature · 1 line
1.8.3.1
Re: [PATCH 02/16] perf tools: Fix machine initialization
Em Tue, Jan 07, 2014 at 01:47:19PM +0100, Jiri Olsa escreveu:
Quoted 2 lines
The
id_hdr_sizefield was not properly initialized, setting it to zero.
Humm, is there any place where 'machine' doesn't come from calloc/zalloc?
- Arnaldo
Quoted 28 lines
Signed-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jean Pihet <[email protected]>--- tools/perf/util/machine.c | 1 + 1 file changed, 1 insertion(+)tools/perf/util/machine.c +1
-0@@ -27,6 +27,7 @@ int machine__init(struct machine *machine, const char *root_dir, pid_t pid) machine->pid = pid; machine->symbol_filter =NULL ; + machine->id_hdr_size =0 ; machine->root_dir =strdup (root_dir);if (machine->root_dir ==NULL )Signature · 1 line
1.8.3.1
Re: [PATCH 02/16] perf tools: Fix machine initialization
Quoted 5 lines · Arnaldo Carvalho de Melo
Em Tue, Jan 07, 2014 at 01:47:19PM +0100, Jiri Olsa escreveu:
Quoted 2 lines
The
id_hdr_sizefield was not properly initialized, setting it to zero.Humm, is there any place where 'machine' doesn't come from calloc/zalloc?
machine__new_host, but I used machine allocated on stack in the tests/dwarf-unwind.c and I thought this was the last member not being initialized..
now I can see vmlinux_maps array does not get initialized as well..
jirka
Quoted 31 lines
- Arnaldo
Quoted 28 lines
Signed-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jean Pihet <[email protected]>--- tools/perf/util/machine.c | 1 + 1 file changed, 1 insertion(+)tools/perf/util/machine.c +1
-0@@ -27,6 +27,7 @@ int machine__init(struct machine *machine, const char *root_dir, pid_t pid) machine->pid = pid; machine->symbol_filter =NULL ; + machine->id_hdr_size =0 ; machine->root_dir =strdup (root_dir);if (machine->root_dir ==NULL )Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf machine: Fix id_hdr_size initialization
In reply to Jiri Olsaperf machine: Fix id_hdr_size initialization
Commit-ID: 14bd6d20fef603060474701967085442b716b6a9 Gitweb: http://git.kernel.org/tip/14bd6d20fef603060474701967085442b716b6a9 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:19 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Mon, 13 Jan 2014 10:06:24 -0300
perf machine: Fix id_hdr_size initialization
The id_hdr_size field was not properly initialized, set it to zero, as the machine struct may have come from some non zeroing allocation routine or from the stack without any field being initialized.
--- tools/perf/util/machine.c | 1 + 1 file changed, 1 insertion(+)
tools/perf/util/machine.c +1 -0
@@ -27,6 +27,7 @@ int machine__init(struct machine *machine, const char *root_dir, pid_t pid) machine->pid = pid; machine->symbol_filter =NULL ; + machine->id_hdr_size =0 ; machine->root_dir =strdup (root_dir);if (machine->root_dir ==NULL )
PATCH 3/16 perf tools: Make perf_event__synthesize_mmap_events global
In reply to Jiri OlsaMaking perf_event__synthesize_mmap_events global, it's used in following patch from test code.
--- tools/perf/util/event.c | 12 ++++++------ tools/perf/util/event.h | 7 +++++++ 2 files changed, 13 insertions(+), 6 deletions(-)
tools/perf/util/event.c +6 -6
@@ -175,12 +175,12 @@ out:return tgid; } -static int perf_event__synthesize_mmap_events (struct perf_tool *tool, -union perf_event *event, -pid_t pid,pid_t tgid, -perf_event__handler_t process, -struct machine *machine, -bool mmap_data) +int perf_event__synthesize_mmap_events (struct perf_tool *tool, +union perf_event *event, +pid_t pid,pid_t tgid, +perf_event__handler_t process, +struct machine *machine, +bool mmap_data) {char filename[PATH_MAX ];FILE *fp;
tools/perf/util/event.h +7 -0
@@ -266,6 +266,13 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type,const struct perf_sample *sample, bool swapped); +int perf_event__synthesize_mmap_events (struct perf_tool *tool, +union perf_event *event, +pid_t pid,pid_t tgid, +perf_event__handler_t process, +struct machine *machine, +bool mmap_data); +size_t perf_event__fprintf_comm (union perf_event *event,FILE *fp);size_t perf_event__fprintf_mmap (union perf_event *event,FILE *fp);size_t perf_event__fprintf_mmap2 (union perf_event *event,FILE *fp);
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf tools: Make perf_event__synthesize_mmap_events global
perf tools: Make perf_event__synthesize_mmap_events global
Commit-ID: a18382b68f8bf1a8d43e3cb08b3479cb768913ea Gitweb: http://git.kernel.org/tip/a18382b68f8bf1a8d43e3cb08b3479cb768913ea Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:20 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Mon, 13 Jan 2014 10:06:24 -0300
perf tools: Make perf_event__synthesize_mmap_events global
Making perf_event__synthesize_mmap_events global, it will be used in following patch from test code.
--- tools/perf/util/event.c | 12 ++++++------ tools/perf/util/event.h | 7 +++++++ 2 files changed, 13 insertions(+), 6 deletions(-)
tools/perf/util/event.c +6 -6
@@ -175,12 +175,12 @@ out:return tgid; } -static int perf_event__synthesize_mmap_events (struct perf_tool *tool, -union perf_event *event, -pid_t pid,pid_t tgid, -perf_event__handler_t process, -struct machine *machine, -bool mmap_data) +int perf_event__synthesize_mmap_events (struct perf_tool *tool, +union perf_event *event, +pid_t pid,pid_t tgid, +perf_event__handler_t process, +struct machine *machine, +bool mmap_data) {char filename[PATH_MAX ];FILE *fp;
tools/perf/util/event.h +7 -0
@@ -266,6 +266,13 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type,const struct perf_sample *sample, bool swapped); +int perf_event__synthesize_mmap_events (struct perf_tool *tool, +union perf_event *event, +pid_t pid,pid_t tgid, +perf_event__handler_t process, +struct machine *machine, +bool mmap_data); +size_t perf_event__fprintf_comm (union perf_event *event,FILE *fp);size_t perf_event__fprintf_mmap (union perf_event *event,FILE *fp);size_t perf_event__fprintf_mmap2 (union perf_event *event,FILE *fp);
PATCH 4/16 perf tests x86: Introduce perf_regs_load function
In reply to Jiri OlsaIntroducing perf_regs_load function, which is going to be used for dwarf unwind test in following patches.
It takes single argument as a pointer to the regs dump buffer and populates it with current registers values.
--- tools/perf/arch/x86/Makefile | 1 + tools/perf/arch/x86/include/perf_regs.h | 2 + tools/perf/arch/x86/tests/regs_load.S | 92 +++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+) create mode 100644 tools/perf/arch/x86/tests/regs_load.S
tools/perf/arch/x86/Makefile +1 -0
@@ -4,6 +4,7 @@ LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o
tools/perf/arch/x86/include/perf_regs.h +2 -0
@@ -5,6 +5,8 @@#include "../../util/types.h" #include <asm/perf_regs.h> +void perf_regs_load (u64 *regs); +#ifndef HAVE_ARCH_X86_64_SUPPORT #define PERF_REGS_MASK ((1ULL << PERF_REG_X86_32_MAX) - 1)#else
tools/perf/arch/x86/tests/regs_load.S new file +92 -0
92 lines not on this page · raw message
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf tests x86: Introduce perf_regs_load function
perf tests x86: Introduce perf_regs_load function
Commit-ID: 3c8b06f981091f91ee603768855e9739a8938296 Gitweb: http://git.kernel.org/tip/3c8b06f981091f91ee603768855e9739a8938296 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:21 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:47 -0300
perf tests x86: Introduce perf_regs_load function
Introducing perf_regs_load function, which is going to be used for dwarf unwind test in following patches.
It takes single argument as a pointer to the regs dump buffer and populates it with current registers values.
--- tools/perf/arch/x86/Makefile | 1 + tools/perf/arch/x86/include/perf_regs.h | 2 + tools/perf/arch/x86/tests/regs_load.S | 92 +++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+)
tools/perf/arch/x86/Makefile +1 -0
@@ -4,6 +4,7 @@ LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o
tools/perf/arch/x86/include/perf_regs.h +2 -0
@@ -5,6 +5,8 @@#include "../../util/types.h" #include <asm/perf_regs.h> +void perf_regs_load (u64 *regs); +#ifndef HAVE_ARCH_X86_64_SUPPORT #define PERF_REGS_MASK ((1ULL << PERF_REG_X86_32_MAX) - 1)#else
tools/perf/arch/x86/tests/regs_load.S new file +92 -0
92 lines not on this page · raw message
PATCH 5/16 perf tests x86: Add dwarf unwind test
In reply to Jiri OlsaAdding dwarf unwind test, that setups live machine data over the perf test thread and does the remote unwind.
At this moment this test fails due to bug in the max_stack processing in unwind__get_entries function. This is fixed in following patch.
Need to use -fno-optimize-sibling-calls for test compilation, otherwise 'krava_*' function calls are optimized into jumps and ommited from the stack unwind.
So far it's enable only for x86.
--- tools/perf/Makefile.perf | 8 ++ tools/perf/arch/x86/Makefile | 1 + tools/perf/arch/x86/include/perf_regs.h | 4 + tools/perf/arch/x86/tests/dwarf-unwind.c | 58 +++++++++++++ tools/perf/tests/builtin-test.c | 8 ++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++ tools/perf/tests/tests.h | 9 ++ 7 files changed, 232 insertions(+) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c
tools/perf/Makefile.perf +8 -0
@@ -407,6 +407,11 @@ endif LIB_OBJS += $(OUTPUT)tests/code-reading.o LIB_OBJS += $(OUTPUT)tests/sample-parsing.o LIB_OBJS += $(OUTPUT)tests/parse-no-sample-id-all.o +ifndef NO_LIBUNWIND +ifeq ($(ARCH),x86) +LIB_OBJS += $(OUTPUT)tests/dwarf-unwind.o +endif +endif BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o BUILTIN_OBJS += $(OUTPUT)builtin-bench.o
@@ -654,6 +659,9 @@ $(OUTPUT)tests/python-use.o: tests/python-use.c $(OUTPUT)PERF-CFLAGS -DPYTHON='"$(PYTHON_WORD)"' \ $< +$(OUTPUT)tests/dwarf-unwind.o: tests/dwarf-unwind.c + $(QUIET_CC)$(CC) -o $@ -c $(CFLAGS) -fno-optimize-sibling-calls $< + $(OUTPUT)util/config.o: util/config.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $<
tools/perf/arch/x86/Makefile +1 -0
@@ -5,6 +5,7 @@ endif ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o
tools/perf/arch/x86/include/perf_regs.h +4 -0
@@ -9,12 +9,16 @@ void perf_regs_load(u64 *regs);#ifndef HAVE_ARCH_X86_64_SUPPORT #define PERF_REGS_MASK ((1ULL << PERF_REG_X86_32_MAX) - 1) +#define PERF_REGS_MAX PERF_REG_X86_32_MAX +#define PERF_SAMPLE_REGS_ABI PERF_SAMPLE_REGS_ABI_32#else #define REG_NOSUPPORT ((1ULL << PERF_REG_X86_DS) | \ (1ULL << PERF_REG_X86_ES) | \ (1ULL << PERF_REG_X86_FS) | \ (1ULL << PERF_REG_X86_GS))#define PERF_REGS_MASK (((1ULL << PERF_REG_X86_64_MAX) - 1) & ~REG_NOSUPPORT) +#define PERF_REGS_MAX PERF_REG_X86_64_MAX +#define PERF_SAMPLE_REGS_ABI PERF_SAMPLE_REGS_ABI_64#endif #define PERF_REG_IP PERF_REG_X86_IP#define PERF_REG_SP PERF_REG_X86_SP
tools/perf/arch/x86/tests/dwarf-unwind.c new file +58 -0
58 lines not on this page · raw message
tools/perf/tests/builtin-test.c +8 -0
@@ -115,6 +115,14 @@ static struct test { .desc = "Test parsing with no sample_id_all bit set", .func = test__parse_no_sample_id_all, }, +#if defined(__x86_64__) || defined(__i386__) +#ifdef HAVE_LIBUNWIND_SUPPORT + { + .desc = "Test dwarf unwind", + .func = test__dwarf_unwind, + }, +#endif +#endif { .func = NULL, },
tools/perf/tests/dwarf-unwind.c new file +144 -0
144 lines not on this page · raw message
tools/perf/tests/tests.h +9 -0
@@ -40,5 +40,14 @@ int test__code_reading(void);int test__sample_parsing (void );int test__keep_tracking (void );int test__parse_no_sample_id_all (void ); +int test__dwarf_unwind (void ); +#if defined(__x86_64__) || defined(__i386__) +#ifdef HAVE_LIBUNWIND_SUPPORT +struct thread ; +struct perf_sample ; +int test__arch_unwind_sample (struct perf_sample *sample, +struct thread *thread); +#endif +#endif #endif /* TESTS_H */
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf tests x86: Add dwarf unwind test
perf tests x86: Add dwarf unwind test
Commit-ID: aa16b81fe916378ef6474530c59f719c36cd6ec4 Gitweb: http://git.kernel.org/tip/aa16b81fe916378ef6474530c59f719c36cd6ec4 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:22 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:47 -0300
perf tests x86: Add dwarf unwind test
Adding dwarf unwind test, that setups live machine data over the perf test thread and does the remote unwind.
At this moment this test fails due to bug in the max_stack processing in unwind__get_entries function. This is fixed in following patch.
Need to use -fno-optimize-sibling-calls for test compilation, otherwise 'krava_*' function calls are optimized into jumps and ommited from the stack unwind.
So far it's enabled only for x86.
--- tools/perf/Makefile.perf | 8 ++ tools/perf/arch/x86/Makefile | 1 + tools/perf/arch/x86/include/perf_regs.h | 4 + tools/perf/arch/x86/tests/dwarf-unwind.c | 58 +++++++++++++ tools/perf/tests/builtin-test.c | 8 ++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++ tools/perf/tests/tests.h | 9 ++ 7 files changed, 232 insertions(+)
tools/perf/Makefile.perf +8 -0
@@ -408,6 +408,11 @@ endif LIB_OBJS += $(OUTPUT)tests/code-reading.o LIB_OBJS += $(OUTPUT)tests/sample-parsing.o LIB_OBJS += $(OUTPUT)tests/parse-no-sample-id-all.o +ifndef NO_LIBUNWIND +ifeq ($(ARCH),x86) +LIB_OBJS += $(OUTPUT)tests/dwarf-unwind.o +endif +endif BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o BUILTIN_OBJS += $(OUTPUT)builtin-bench.o
@@ -655,6 +660,9 @@ $(OUTPUT)tests/python-use.o: tests/python-use.c $(OUTPUT)PERF-CFLAGS -DPYTHON='"$(PYTHON_WORD)"' \ $< +$(OUTPUT)tests/dwarf-unwind.o: tests/dwarf-unwind.c + $(QUIET_CC)$(CC) -o $@ -c $(CFLAGS) -fno-optimize-sibling-calls $< + $(OUTPUT)util/config.o: util/config.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $<
tools/perf/arch/x86/Makefile +1 -0
@@ -5,6 +5,7 @@ endif ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o
tools/perf/arch/x86/include/perf_regs.h +4 -0
@@ -9,12 +9,16 @@ void perf_regs_load(u64 *regs);#ifndef HAVE_ARCH_X86_64_SUPPORT #define PERF_REGS_MASK ((1ULL << PERF_REG_X86_32_MAX) - 1) +#define PERF_REGS_MAX PERF_REG_X86_32_MAX +#define PERF_SAMPLE_REGS_ABI PERF_SAMPLE_REGS_ABI_32#else #define REG_NOSUPPORT ((1ULL << PERF_REG_X86_DS) | \ (1ULL << PERF_REG_X86_ES) | \ (1ULL << PERF_REG_X86_FS) | \ (1ULL << PERF_REG_X86_GS))#define PERF_REGS_MASK (((1ULL << PERF_REG_X86_64_MAX) - 1) & ~REG_NOSUPPORT) +#define PERF_REGS_MAX PERF_REG_X86_64_MAX +#define PERF_SAMPLE_REGS_ABI PERF_SAMPLE_REGS_ABI_64#endif #define PERF_REG_IP PERF_REG_X86_IP#define PERF_REG_SP PERF_REG_X86_SP
tools/perf/arch/x86/tests/dwarf-unwind.c new file +58 -0
58 lines not on this page · raw message
tools/perf/tests/builtin-test.c +8 -0
@@ -115,6 +115,14 @@ static struct test { .desc = "Test parsing with no sample_id_all bit set", .func = test__parse_no_sample_id_all, }, +#if defined(__x86_64__) || defined(__i386__) +#ifdef HAVE_LIBUNWIND_SUPPORT + { + .desc = "Test dwarf unwind", + .func = test__dwarf_unwind, + }, +#endif +#endif { .func = NULL, },
tools/perf/tests/dwarf-unwind.c new file +144 -0
144 lines not on this page · raw message
tools/perf/tests/tests.h +9 -0
@@ -40,5 +40,14 @@ int test__code_reading(void);int test__sample_parsing (void );int test__keep_tracking (void );int test__parse_no_sample_id_all (void ); +int test__dwarf_unwind (void ); +#if defined(__x86_64__) || defined(__i386__) +#ifdef HAVE_LIBUNWIND_SUPPORT +struct thread ; +struct perf_sample ; +int test__arch_unwind_sample (struct perf_sample *sample, +struct thread *thread); +#endif +#endif #endif /* TESTS_H */
PATCH 6/16 perf tools: Fix dwarf unwind max_stack processing
In reply to Jiri OlsaThe 'unwind__get_entries' function currently returns 'max_stack + 1' entries (instead of exact max_stack entries), because max_stack value does not get decremented for the first entry.
This fix makes dwarf-unwind test pass.
--- tools/perf/util/unwind.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/unwind.c +1 -1
@@ -611,5 +611,5 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (ret)return -ENOMEM ; -return get_entries (&ui, cb, arg, max_stack); +return --max_stack >0 ?get_entries (&ui, cb, arg, max_stack) :0 ; }
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf tools: Fix dwarf unwind max_stack processing
perf tools: Fix dwarf unwind max_stack processing
Commit-ID: b42dc32d4f91e4c0f34b628fdb012eb423da9e69 Gitweb: http://git.kernel.org/tip/b42dc32d4f91e4c0f34b628fdb012eb423da9e69 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:23 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:47 -0300
perf tools: Fix dwarf unwind max_stack processing
The 'unwind__get_entries' function currently returns 'max_stack + 1' entries (instead of exact max_stack entries), because max_stack value does not get decremented for the first entry.
This fix makes dwarf-unwind test pass.
--- tools/perf/util/unwind.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/unwind.c +1 -1
@@ -595,5 +595,5 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (ret)return -ENOMEM ; -return get_entries (&ui, cb, arg, max_stack); +return --max_stack >0 ?get_entries (&ui, cb, arg, max_stack) :0 ; }
PATCH 7/16 perf tools: Do not report zero address in unwind
In reply to Jiri OlsaWe are not interested in zero addresses in callchain, do not report them.
--- tools/perf/util/unwind.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/unwind.c +1 -1
@@ -579,7 +579,7 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,unw_word_t ip;unw_get_reg (&c,UNW_REG_IP , &ip); - ret =entry (ip, ui->thread, ui->machine, cb, arg); + ret = ip ?entry (ip, ui->thread, ui->machine, cb, arg) :0 ; }unw_destroy_addr_space (addr_space);
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Do not report zero address in unwind
perf callchain: Do not report zero address in unwind
Commit-ID: 1cf0382af98f6365b01b59453fe18dffe3c73d2f Gitweb: http://git.kernel.org/tip/1cf0382af98f6365b01b59453fe18dffe3c73d2f Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:24 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:48 -0300
perf callchain: Do not report zero address in unwind
We are not interested in zero addresses in callchain, do not report them.
--- tools/perf/util/unwind.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/unwind.c +1 -1
@@ -563,7 +563,7 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,unw_word_t ip;unw_get_reg (&c,UNW_REG_IP , &ip); - ret =entry (ip, ui->thread, ui->machine, cb, arg); + ret = ip ?entry (ip, ui->thread, ui->machine, cb, arg) :0 ; }unw_destroy_addr_space (addr_space);
PATCH 8/16 perf tools: Add mask into struct regs_dump
In reply to Jiri OlsaAdding mask info into struct regs_dump to make the registers information compact.
The mask was always passed along, so logically the mask info fits more into the struct regs_dump.
--- tools/perf/arch/x86/tests/dwarf-unwind.c | 1 + tools/perf/builtin-inject.c | 1 - tools/perf/tests/dwarf-unwind.c | 2 +- tools/perf/tests/sample-parsing.c | 17 ++++++++--------- tools/perf/util/event.h | 5 +++-- tools/perf/util/evsel.c | 13 +++++++------ tools/perf/util/machine.c | 3 +-- tools/perf/util/session.c | 5 +++-- tools/perf/util/unwind.c | 20 ++++++++------------ tools/perf/util/unwind.h | 2 -- 10 files changed, 32 insertions(+), 37 deletions(-)
tools/perf/arch/x86/tests/dwarf-unwind.c +1 -0
@@ -53,6 +53,7 @@ int test__arch_unwind_sample(struct perf_sample *sample,perf_regs_load (buf); regs->abi =PERF_SAMPLE_REGS_ABI ; regs->regs = buf; + regs->mask =PERF_REGS_MASK ;return sample_ustack (sample, thread, buf); }
tools/perf/builtin-inject.c +0 -1
@@ -312,7 +312,6 @@ found: sample_sw.period = sample->period; sample_sw.time = sample->time;perf_event__synthesize_sample (event_sw, evsel->attr.sample_type, - evsel->attr.sample_regs_user, evsel->attr.read_format, &sample_sw, false);build_id__mark_dso_hit (tool, event_sw, &sample_sw, evsel, machine);
tools/perf/tests/dwarf-unwind.c +1 -1
@@ -72,7 +72,7 @@ static int unwind_thread(struct thread *thread, struct machine *machine) } err =unwind__get_entries (unwind_entry, &cnt, machine, thread, -PERF_REGS_MASK , &sample,MAX_STACK ); + &sample,MAX_STACK );if (err)pr_debug ("unwind failed\n" );else if (cnt !=MAX_STACK ) {
tools/perf/tests/sample-parsing.c +8 -9
@@ -22,8 +22,8 @@ }while (0 ) static bool samples_same(const struct perf_sample *s1, -const struct perf_sample *s2,u64 type,u64 regs_user, -u64 read_format) +const struct perf_sample *s2, +u64 type,u64 read_format) {size_t i;
@@ -95,8 +95,9 @@ static bool samples_same(const struct perf_sample *s1, }if (type &PERF_SAMPLE_REGS_USER ) { -size_t sz =hweight_long (regs_user) *sizeof (u64); +size_t sz =hweight_long (s1->user_regs.mask) *sizeof (u64); +COMP (user_regs.mask);COMP (user_regs.abi); if (s1->user_regs.abi && (!s1->user_regs.regs || !s2->user_regs.regs ||
@@ -174,6 +175,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) .branch_stack = &branch_stack.branch_stack, .user_regs = { .abi = PERF_SAMPLE_REGS_ABI_64, + .mask = sample_regs_user, .regs = user_regs, }, .user_stack = {
@@ -201,8 +203,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) sample.read.one.id =99 ; } - sz =perf_event__sample_event_size (&sample, sample_type, - sample_regs_user, read_format); + sz =perf_event__sample_event_size (&sample, sample_type, read_format); bufsz = sz +4096 ;/* Add a bit for overrun checking */ event =malloc (bufsz);if (!event) {
@@ -215,8 +216,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) event->header.misc =0 ; event->header.size = sz; - err =perf_event__synthesize_sample (event, sample_type, - sample_regs_user, read_format, + err =perf_event__synthesize_sample (event, sample_type, read_format, &sample, false);if (err) { pr_debug("%s failed for sample_type %#"PRIx64", error %d\n",
@@ -244,8 +244,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) goto out_free; } -if (!samples_same (&sample, &sample_out, sample_type, - sample_regs_user, read_format)) { +if (!samples_same (&sample, &sample_out, sample_type, read_format)) {pr_debug ("parsing failed for sample_type %#" PRIx64"\n" , sample_type); goto out_free;
tools/perf/util/event.h +3 -2
@@ -85,6 +85,7 @@ struct sample_event {struct regs_dump {u64 abi; +u64 mask;u64 *regs; };
@@ -260,9 +261,9 @@ int perf_event__preprocess_sample(const union perf_event *event,const char *perf_event__name (unsigned int id);size_t perf_event__sample_event_size (const struct perf_sample *sample,u64 type, -u64 sample_regs_user,u64 read_format); +u64 read_format);int perf_event__synthesize_sample (union perf_event *event,u64 type, -u64 sample_regs_user,u64 read_format, +u64 read_format,const struct perf_sample *sample,bool swapped);
tools/perf/util/evsel.c +7 -6
@@ -1396,10 +1396,11 @@ int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event, array++;if (data->user_regs.abi) { -u64 regs_user = evsel->attr.sample_regs_user; +u64 mask = evsel->attr.sample_regs_user; - sz =hweight_long (regs_user) *sizeof (u64); + sz =hweight_long (mask) *sizeof (u64);OVERFLOW_CHECK (array, sz, max_size); + data->user_regs.mask = mask; data->user_regs.regs = (u64 *)array; array = (void *)array + sz; }
@@ -1451,7 +1452,7 @@ int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event, }size_t perf_event__sample_event_size (const struct perf_sample *sample,u64 type, -u64 sample_regs_user,u64 read_format) +u64 read_format) {size_t sz, result =sizeof (struct sample_event );
@@ -1517,7 +1518,7 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type,if (type &PERF_SAMPLE_REGS_USER ) {if (sample->user_regs.abi) { result +=sizeof (u64); - sz =hweight_long (sample_regs_user) *sizeof (u64); + sz =hweight_long (sample->user_regs.mask) *sizeof (u64); result += sz; }else { result +=sizeof (u64);
@@ -1546,7 +1547,7 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type, }int perf_event__synthesize_sample (union perf_event *event,u64 type, -u64 sample_regs_user,u64 read_format, +u64 read_format,const struct perf_sample *sample,bool swapped) {
@@ -1687,7 +1688,7 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type,if (type &PERF_SAMPLE_REGS_USER ) {if (sample->user_regs.abi) { *array++ = sample->user_regs.abi; - sz =hweight_long (sample_regs_user) *sizeof (u64); + sz =hweight_long (sample->user_regs.mask) *sizeof (u64);memcpy (array, sample->user_regs.regs, sz); array = (void *)array + sz; }else {
tools/perf/util/machine.c +1 -2
@@ -1361,8 +1361,7 @@ int machine__resolve_callchain(struct machine *machine,return 0 ;return unwind__get_entries (unwind_entry, &callchain_cursor, machine, - thread, evsel->attr.sample_regs_user, - sample, max_stack); + thread, sample, max_stack); }
tools/perf/util/session.c +3 -2
@@ -702,11 +702,12 @@ static void regs_dump__printf(u64 mask, u64 *regs) } } -static void regs_user__printf (struct perf_sample *sample,u64 mask) +static void regs_user__printf (struct perf_sample *sample) {struct regs_dump *user_regs = &sample->user_regs;if (user_regs->regs) { +u64 mask = user_regs->mask;printf ("... user regs: mask 0x%" PRIx64"\n" , mask);regs_dump__printf (mask, user_regs->regs); }
@@ -806,7 +807,7 @@ static void dump_sample(struct perf_evsel *evsel, union perf_event *event,branch_stack__printf (sample);if (sample_type &PERF_SAMPLE_REGS_USER ) -regs_user__printf (sample, evsel->attr.sample_regs_user); +regs_user__printf (sample);if (sample_type &PERF_SAMPLE_STACK_USER )stack_user__printf (&sample->user_stack);
tools/perf/util/unwind.c +8 -12
@@ -85,7 +85,6 @@ struct unwind_info {struct perf_sample *sample;struct machine *machine;struct thread *thread; -u64 sample_uregs; };#define dw_read (ptr, type, end) ({ \
@@ -407,16 +406,16 @@ static int access_dso_mem(struct unwind_info *ui, unw_word_t addr,return !(size ==sizeof (*data)); } -static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id, -u64 sample_regs) +static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id) {int i, idx =0 ; +u64 mask = regs->mask; -if (!(sample_regs & (1 << id))) +if (!(mask & (1 << id)))return -EINVAL ;for (i =0 ; i < id; i++) { -if (sample_regs & (1 << i)) +if (mask & (1 << i)) idx++; }
@@ -440,8 +439,7 @@ static int access_mem(unw_addr_space_t __maybe_unused as,return 0 ; } - ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP , - ui->sample_uregs); + ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP );if (ret)return ret;
@@ -491,7 +489,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as,if (id <0 )return -EINVAL ; - ret =reg_value (valp, &ui->sample->user_regs, id, ui->sample_uregs); + ret =reg_value (valp, &ui->sample->user_regs, id);if (ret) {pr_err ("unwind: can't read reg %d\n" , regnum);return ret;
@@ -588,13 +586,11 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread, -u64 sample_uregs,struct perf_sample *data, -int max_stack) +struct perf_sample *data,int max_stack) {unw_word_t ip;struct unwind_info ui = { .sample = data, - .sample_uregs = sample_uregs, .thread = thread, .machine = machine, };
@@ -603,7 +599,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (!data->user_regs.regs)return -EINVAL ; - ret =reg_value (&ip, &data->user_regs,PERF_REG_IP , sample_uregs); + ret =reg_value (&ip, &data->user_regs,PERF_REG_IP );if (ret)return ret;
tools/perf/util/unwind.h +0 -2
@@ -17,7 +17,6 @@ typedef int (*unwind_entry_cb_t)(struct unwind_entry *entry, void *arg);int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread, -u64 sample_uregs,struct perf_sample *data,int max_stack);int unwind__arch_reg_id (int regnum);#else
@@ -26,7 +25,6 @@ unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,void *arg__maybe_unused , struct machine *machine __maybe_unused, struct thread *thread __maybe_unused, - u64 sample_uregs __maybe_unused, struct perf_sample *data __maybe_unused, int max_stack __maybe_unused) {
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Add mask into struct regs_dump
perf callchain: Add mask into struct regs_dump
Commit-ID: 352ea45a7229df8f5ae83c0757f6d426ba0f41b5 Gitweb: http://git.kernel.org/tip/352ea45a7229df8f5ae83c0757f6d426ba0f41b5 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:25 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:48 -0300
perf callchain: Add mask into struct regs_dump
Adding mask info into struct regs_dump to make the registers information compact.
The mask was always passed along, so logically the mask info fits more into the struct regs_dump.
--- tools/perf/arch/x86/tests/dwarf-unwind.c | 1 + tools/perf/builtin-inject.c | 1 - tools/perf/tests/dwarf-unwind.c | 2 +- tools/perf/tests/sample-parsing.c | 17 ++++++++--------- tools/perf/util/event.h | 5 +++-- tools/perf/util/evsel.c | 13 +++++++------ tools/perf/util/machine.c | 3 +-- tools/perf/util/session.c | 5 +++-- tools/perf/util/unwind.c | 20 ++++++++------------ tools/perf/util/unwind.h | 2 -- 10 files changed, 32 insertions(+), 37 deletions(-)
tools/perf/arch/x86/tests/dwarf-unwind.c +1 -0
@@ -53,6 +53,7 @@ int test__arch_unwind_sample(struct perf_sample *sample,perf_regs_load (buf); regs->abi =PERF_SAMPLE_REGS_ABI ; regs->regs = buf; + regs->mask =PERF_REGS_MASK ;return sample_ustack (sample, thread, buf); }
tools/perf/builtin-inject.c +0 -1
@@ -312,7 +312,6 @@ found: sample_sw.period = sample->period; sample_sw.time = sample->time;perf_event__synthesize_sample (event_sw, evsel->attr.sample_type, - evsel->attr.sample_regs_user, evsel->attr.read_format, &sample_sw, false);build_id__mark_dso_hit (tool, event_sw, &sample_sw, evsel, machine);
tools/perf/tests/dwarf-unwind.c +1 -1
@@ -72,7 +72,7 @@ static int unwind_thread(struct thread *thread, struct machine *machine) } err =unwind__get_entries (unwind_entry, &cnt, machine, thread, -PERF_REGS_MASK , &sample,MAX_STACK ); + &sample,MAX_STACK );if (err)pr_debug ("unwind failed\n" );else if (cnt !=MAX_STACK ) {
tools/perf/tests/sample-parsing.c +8 -9
@@ -22,8 +22,8 @@ }while (0 ) static bool samples_same(const struct perf_sample *s1, -const struct perf_sample *s2,u64 type,u64 regs_user, -u64 read_format) +const struct perf_sample *s2, +u64 type,u64 read_format) {size_t i;
@@ -95,8 +95,9 @@ static bool samples_same(const struct perf_sample *s1, }if (type &PERF_SAMPLE_REGS_USER ) { -size_t sz =hweight_long (regs_user) *sizeof (u64); +size_t sz =hweight_long (s1->user_regs.mask) *sizeof (u64); +COMP (user_regs.mask);COMP (user_regs.abi); if (s1->user_regs.abi && (!s1->user_regs.regs || !s2->user_regs.regs ||
@@ -174,6 +175,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) .branch_stack = &branch_stack.branch_stack, .user_regs = { .abi = PERF_SAMPLE_REGS_ABI_64, + .mask = sample_regs_user, .regs = user_regs, }, .user_stack = {
@@ -201,8 +203,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) sample.read.one.id =99 ; } - sz =perf_event__sample_event_size (&sample, sample_type, - sample_regs_user, read_format); + sz =perf_event__sample_event_size (&sample, sample_type, read_format); bufsz = sz +4096 ;/* Add a bit for overrun checking */ event =malloc (bufsz);if (!event) {
@@ -215,8 +216,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) event->header.misc =0 ; event->header.size = sz; - err =perf_event__synthesize_sample (event, sample_type, - sample_regs_user, read_format, + err =perf_event__synthesize_sample (event, sample_type, read_format, &sample, false);if (err) { pr_debug("%s failed for sample_type %#"PRIx64", error %d\n",
@@ -244,8 +244,7 @@ static int do_test(u64 sample_type, u64 sample_regs_user, u64 read_format) goto out_free; } -if (!samples_same (&sample, &sample_out, sample_type, - sample_regs_user, read_format)) { +if (!samples_same (&sample, &sample_out, sample_type, read_format)) {pr_debug ("parsing failed for sample_type %#" PRIx64"\n" , sample_type); goto out_free;
tools/perf/util/event.h +3 -2
@@ -85,6 +85,7 @@ struct sample_event {struct regs_dump {u64 abi; +u64 mask;u64 *regs; };
@@ -259,9 +260,9 @@ int perf_event__preprocess_sample(const union perf_event *event,const char *perf_event__name (unsigned int id);size_t perf_event__sample_event_size (const struct perf_sample *sample,u64 type, -u64 sample_regs_user,u64 read_format); +u64 read_format);int perf_event__synthesize_sample (union perf_event *event,u64 type, -u64 sample_regs_user,u64 read_format, +u64 read_format,const struct perf_sample *sample,bool swapped);
tools/perf/util/evsel.c +7 -6
@@ -1396,10 +1396,11 @@ int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event, array++;if (data->user_regs.abi) { -u64 regs_user = evsel->attr.sample_regs_user; +u64 mask = evsel->attr.sample_regs_user; - sz =hweight_long (regs_user) *sizeof (u64); + sz =hweight_long (mask) *sizeof (u64);OVERFLOW_CHECK (array, sz, max_size); + data->user_regs.mask = mask; data->user_regs.regs = (u64 *)array; array = (void *)array + sz; }
@@ -1451,7 +1452,7 @@ int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event, }size_t perf_event__sample_event_size (const struct perf_sample *sample,u64 type, -u64 sample_regs_user,u64 read_format) +u64 read_format) {size_t sz, result =sizeof (struct sample_event );
@@ -1517,7 +1518,7 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type,if (type &PERF_SAMPLE_REGS_USER ) {if (sample->user_regs.abi) { result +=sizeof (u64); - sz =hweight_long (sample_regs_user) *sizeof (u64); + sz =hweight_long (sample->user_regs.mask) *sizeof (u64); result += sz; }else { result +=sizeof (u64);
@@ -1546,7 +1547,7 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type, }int perf_event__synthesize_sample (union perf_event *event,u64 type, -u64 sample_regs_user,u64 read_format, +u64 read_format,const struct perf_sample *sample,bool swapped) {
@@ -1687,7 +1688,7 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type,if (type &PERF_SAMPLE_REGS_USER ) {if (sample->user_regs.abi) { *array++ = sample->user_regs.abi; - sz =hweight_long (sample_regs_user) *sizeof (u64); + sz =hweight_long (sample->user_regs.mask) *sizeof (u64);memcpy (array, sample->user_regs.regs, sz); array = (void *)array + sz; }else {
tools/perf/util/machine.c +1 -2
@@ -1383,8 +1383,7 @@ int machine__resolve_callchain(struct machine *machine,return 0 ;return unwind__get_entries (unwind_entry, &callchain_cursor, machine, - thread, evsel->attr.sample_regs_user, - sample, max_stack); + thread, sample, max_stack); }
tools/perf/util/session.c +3 -2
@@ -702,11 +702,12 @@ static void regs_dump__printf(u64 mask, u64 *regs) } } -static void regs_user__printf (struct perf_sample *sample,u64 mask) +static void regs_user__printf (struct perf_sample *sample) {struct regs_dump *user_regs = &sample->user_regs;if (user_regs->regs) { +u64 mask = user_regs->mask;printf ("... user regs: mask 0x%" PRIx64"\n" , mask);regs_dump__printf (mask, user_regs->regs); }
@@ -806,7 +807,7 @@ static void dump_sample(struct perf_evsel *evsel, union perf_event *event,branch_stack__printf (sample);if (sample_type &PERF_SAMPLE_REGS_USER ) -regs_user__printf (sample, evsel->attr.sample_regs_user); +regs_user__printf (sample);if (sample_type &PERF_SAMPLE_STACK_USER )stack_user__printf (&sample->user_stack);
tools/perf/util/unwind.c +8 -12
@@ -86,7 +86,6 @@ struct unwind_info {struct perf_sample *sample;struct machine *machine;struct thread *thread; -u64 sample_uregs; };#define dw_read (ptr, type, end) ({ \
@@ -391,16 +390,16 @@ static int access_dso_mem(struct unwind_info *ui, unw_word_t addr,return !(size ==sizeof (*data)); } -static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id, -u64 sample_regs) +static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id) {int i, idx =0 ; +u64 mask = regs->mask; -if (!(sample_regs & (1 << id))) +if (!(mask & (1 << id)))return -EINVAL ;for (i =0 ; i < id; i++) { -if (sample_regs & (1 << i)) +if (mask & (1 << i)) idx++; }
@@ -424,8 +423,7 @@ static int access_mem(unw_addr_space_t __maybe_unused as,return 0 ; } - ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP , - ui->sample_uregs); + ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP );if (ret)return ret;
@@ -475,7 +473,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as,if (id <0 )return -EINVAL ; - ret =reg_value (valp, &ui->sample->user_regs, id, ui->sample_uregs); + ret =reg_value (valp, &ui->sample->user_regs, id);if (ret) {pr_err ("unwind: can't read reg %d\n" , regnum);return ret;
@@ -572,13 +570,11 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread, -u64 sample_uregs,struct perf_sample *data, -int max_stack) +struct perf_sample *data,int max_stack) {unw_word_t ip;struct unwind_info ui = { .sample = data, - .sample_uregs = sample_uregs, .thread = thread, .machine = machine, };
@@ -587,7 +583,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (!data->user_regs.regs)return -EINVAL ; - ret =reg_value (&ip, &data->user_regs,PERF_REG_IP , sample_uregs); + ret =reg_value (&ip, &data->user_regs,PERF_REG_IP );if (ret)return ret;
tools/perf/util/unwind.h +0 -2
@@ -17,7 +17,6 @@ typedef int (*unwind_entry_cb_t)(struct unwind_entry *entry, void *arg);int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread, -u64 sample_uregs,struct perf_sample *data,int max_stack);int unwind__arch_reg_id (int regnum);#else
@@ -26,7 +25,6 @@ unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,void *arg__maybe_unused , struct machine *machine __maybe_unused, struct thread *thread __maybe_unused, - u64 sample_uregs __maybe_unused, struct perf_sample *data __maybe_unused, int max_stack __maybe_unused) {
PATCH 9/16 perf tools: Separate libunwind code to special object
In reply to Jiri OlsaWe are going to add libdw library support to do dwarf post unwind.
Making the code ready by moving libunwind dwarf post unwind stuff into separate object.
--- tools/perf/Makefile.perf | 2 +- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/unwind-libunwind.c | 48 +++ tools/perf/arch/arm/util/unwind.c | 48 --- tools/perf/arch/x86/Makefile | 2 +- tools/perf/arch/x86/util/unwind-libunwind.c | 111 +++++ tools/perf/arch/x86/util/unwind.c | 111 ----- tools/perf/util/unwind-libunwind.c | 611 ++++++++++++++++++++++++++++ tools/perf/util/unwind.c | 611 ---------------------------- 9 files changed, 773 insertions(+), 773 deletions(-) create mode 100644 tools/perf/arch/arm/util/unwind-libunwind.c delete mode 100644 tools/perf/arch/arm/util/unwind.c create mode 100644 tools/perf/arch/x86/util/unwind-libunwind.c delete mode 100644 tools/perf/arch/x86/util/unwind.c create mode 100644 tools/perf/util/unwind-libunwind.c delete mode 100644 tools/perf/util/unwind.c
tools/perf/Makefile.perf +1 -1
@@ -480,7 +480,7 @@ endif # NO_DWARF endif # NO_LIBELF ifndef NO_LIBUNWIND - LIB_OBJS += $(OUTPUT)util/unwind.o + LIB_OBJS += $(OUTPUT)util/unwind-libunwind.o endif LIB_OBJS += $(OUTPUT)tests/keep-tracking.o
tools/perf/arch/arm/Makefile +1 -1
@@ -3,5 +3,5 @@ PERF_HAVE_DWARF_REGS := 1 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o endif
tools/perf/arch/arm/util/unwind-libunwind.c new file +48 -0
48 lines not on this page · raw message
tools/perf/arch/arm/util/unwind.c deleted +0 -48
48 lines not on this page · raw message
tools/perf/arch/x86/Makefile +1 -1
@@ -3,7 +3,7 @@ PERF_HAVE_DWARF_REGS := 1 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif
tools/perf/arch/x86/util/unwind-libunwind.c new file +111 -0
111 lines not on this page · raw message
tools/perf/arch/x86/util/unwind.c deleted +0 -111
111 lines not on this page · raw message
tools/perf/util/unwind-libunwind.c new file +611 -0
611 lines not on this page · raw message
tools/perf/util/unwind.c deleted +0 -611
611 lines not on this page · raw message
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Separate libunwind code to special object
perf callchain: Separate libunwind code to special object
Commit-ID: 436aa749bb193385f6a3b741ddc6bfba61770a6e Gitweb: http://git.kernel.org/tip/436aa749bb193385f6a3b741ddc6bfba61770a6e Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:26 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:48 -0300
perf callchain: Separate libunwind code to special object
We are going to add libdw library support to do dwarf post unwind.
Making the code ready by moving libunwind dwarf post unwind stuff into separate object.
---
tools/perf/Makefile.perf | 2 +-
tools/perf/arch/arm/Makefile | 2 +-
tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 0
tools/perf/arch/x86/Makefile | 2 +-
tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 0
tools/perf/util/{unwind.c => unwind-libunwind.c} | 0
6 files changed, 3 insertions(+), 3 deletions(-)tools/perf/Makefile.perf +1 -1
@@ -481,7 +481,7 @@ endif # NO_DWARF endif # NO_LIBELF ifndef NO_LIBUNWIND - LIB_OBJS += $(OUTPUT)util/unwind.o + LIB_OBJS += $(OUTPUT)util/unwind-libunwind.o endif LIB_OBJS += $(OUTPUT)tests/keep-tracking.o
tools/perf/arch/arm/Makefile +1 -1
@@ -3,5 +3,5 @@ PERF_HAVE_DWARF_REGS := 1 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o endif
tools/perf/arch/arm/util/unwind-libunwind.c renamed from tools/perf/arch/arm/util/unwind.c +0 -0
tools/perf/arch/x86/Makefile +1 -1
@@ -3,7 +3,7 @@ PERF_HAVE_DWARF_REGS := 1 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif
tools/perf/arch/x86/util/unwind-libunwind.c renamed from tools/perf/arch/x86/util/unwind.c +0 -0
tools/perf/util/unwind-libunwind.c renamed from tools/perf/util/unwind.c +0 -0
PATCH 10/16 perf tools: Rename unwind__arch_reg_id into libunwind__arch_reg_id
In reply to Jiri OlsaRenaming unwind__arch_reg_id into libunwind__arch_reg_id, so it's clear it's specific to libunwind.
--- tools/perf/arch/arm/util/unwind-libunwind.c | 2 +- tools/perf/arch/x86/util/unwind-libunwind.c | 4 ++-- tools/perf/util/unwind-libunwind.c | 2 +- tools/perf/util/unwind.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-)
tools/perf/arch/arm/util/unwind-libunwind.c +1 -1
@@ -4,7 +4,7 @@#include "perf_regs.h" #include "../../util/unwind.h" -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {switch (regnum) {case UNW_ARM_R0 :
tools/perf/arch/x86/util/unwind-libunwind.c +2 -2
@@ -5,7 +5,7 @@#include "../../util/unwind.h" #ifdef HAVE_ARCH_X86_64_SUPPORT -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {int id;
@@ -69,7 +69,7 @@ int unwind__arch_reg_id(int regnum) return id; }#else -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {int id;
tools/perf/util/unwind-libunwind.c +1 -1
@@ -485,7 +485,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as,return 0 ; } - id =unwind__arch_reg_id (regnum); + id =libunwind__arch_reg_id (regnum);if (id <0 )return -EINVAL ;
tools/perf/util/unwind.h +1 -1
@@ -18,7 +18,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,struct machine *machine, struct thread *thread, struct perf_sample *data, int max_stack); -int unwind__arch_reg_id (int regnum); +int libunwind__arch_reg_id (int regnum);#else static inline int unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Rename unwind__arch_reg_id into libunwind__arch_reg_id
perf callchain: Rename unwind__arch_reg_id into libunwind__arch_reg_id
Commit-ID: ea3da69df959dfa1c807586c061bfcc4ec7e5e17 Gitweb: http://git.kernel.org/tip/ea3da69df959dfa1c807586c061bfcc4ec7e5e17 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:27 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:48 -0300
perf callchain: Rename unwind__arch_reg_id into libunwind__arch_reg_id
Renaming unwind__arch_reg_id into libunwind__arch_reg_id, so it's clear it's specific to libunwind.
--- tools/perf/arch/arm/util/unwind-libunwind.c | 2 +- tools/perf/arch/x86/util/unwind-libunwind.c | 4 ++-- tools/perf/util/unwind-libunwind.c | 2 +- tools/perf/util/unwind.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-)
tools/perf/arch/arm/util/unwind-libunwind.c +1 -1
@@ -4,7 +4,7 @@#include "perf_regs.h" #include "../../util/unwind.h" -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {switch (regnum) {case UNW_ARM_R0 :
tools/perf/arch/x86/util/unwind-libunwind.c +2 -2
@@ -5,7 +5,7 @@#include "../../util/unwind.h" #ifdef HAVE_ARCH_X86_64_SUPPORT -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {int id;
@@ -69,7 +69,7 @@ int unwind__arch_reg_id(int regnum) return id; }#else -int unwind__arch_reg_id (int regnum) +int libunwind__arch_reg_id (int regnum) {int id;
tools/perf/util/unwind-libunwind.c +1 -1
@@ -469,7 +469,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as,return 0 ; } - id =unwind__arch_reg_id (regnum); + id =libunwind__arch_reg_id (regnum);if (id <0 )return -EINVAL ;
tools/perf/util/unwind.h +1 -1
@@ -18,7 +18,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,struct machine *machine, struct thread *thread, struct perf_sample *data, int max_stack); -int unwind__arch_reg_id (int regnum); +int libunwind__arch_reg_id (int regnum);#else static inline int unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
PATCH 11/16 perf tools: Introduce HAVE_DWARF_UNWIND_SUPPORT macro
In reply to Jiri OlsaIntroducing global macro HAVE_DWARF_UNWIND_SUPPORT to indicate we have dwarf unwind support. Any library providing the dwarf post unwind support will enable this macro.
--- tools/perf/builtin-record.c | 10 +++++----- tools/perf/config/Makefile | 2 +- tools/perf/tests/builtin-test.c | 2 +- tools/perf/tests/tests.h | 2 +- tools/perf/util/unwind.h | 7 +++++-- 5 files changed, 13 insertions(+), 10 deletions(-)
tools/perf/builtin-record.c +5 -5
@@ -650,7 +650,7 @@ error:return ret; } -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT static int get_stack_size (char *str,unsigned long *_size) {char *endptr;
@@ -676,7 +676,7 @@ static int get_stack_size(char *str, unsigned long *_size) max_size, str); return -1; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ int record_parse_callchain (const char *arg,struct record_opts *opts) {
@@ -705,7 +705,7 @@ int record_parse_callchain(const char *arg, struct record_opts *opts) "needed for -g fp\n");break ; -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT /* Dwarf style */ }else if (!strncmp (name,"dwarf" ,sizeof ("dwarf" ))) {const unsigned long default_stack_dump_size =8192 ;
@@ -721,7 +721,7 @@ int record_parse_callchain(const char *arg, struct record_opts *opts) ret =get_stack_size (tok, &size); opts->stack_dump_size = size; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ }else {pr_err ("callchain: Unknown --call-graph option " "value: %s\n" , arg);
@@ -808,7 +808,7 @@ static struct record record = {#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: " -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT const char record_callchain_help[] =CALLCHAIN_HELP "fp dwarf" ;#else const char record_callchain_help[] =CALLCHAIN_HELP "fp" ;
tools/perf/config/Makefile +1 -1
@@ -337,7 +337,7 @@ ifndef NO_LIBUNWIND CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME endif - CFLAGS += -DHAVE_LIBUNWIND_SUPPORT + CFLAGS += -DHAVE_DWARF_UNWIND_SUPPORT -DHAVE_LIBUNWIND_SUPPORT EXTLIBS += $(LIBUNWIND_LIBS) CFLAGS += $(LIBUNWIND_CFLAGS) LDFLAGS += $(LIBUNWIND_LDFLAGS)
tools/perf/tests/builtin-test.c +1 -1
@@ -116,7 +116,7 @@ static struct test { .func = test__parse_no_sample_id_all, }, #if defined(__x86_64__) || defined(__i386__) -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT { .desc = "Test dwarf unwind", .func = test__dwarf_unwind,
tools/perf/tests/tests.h +1 -1
@@ -43,7 +43,7 @@ int test__parse_no_sample_id_all(void);int test__dwarf_unwind (void );#if defined(__x86_64__) || defined(__i386__) -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT struct thread ;struct perf_sample ; int test__arch_unwind_sample(struct perf_sample *sample,
tools/perf/util/unwind.h +5 -2
@@ -13,12 +13,15 @@ struct unwind_entry {typedef int (*unwind_entry_cb_t )(struct unwind_entry *entry,void *arg); -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread,struct perf_sample *data,int max_stack); +/* libunwind specific */ +#ifdef HAVE_LIBUNWIND_SUPPORT int libunwind__arch_reg_id (int regnum); +#endif #else static inline int unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
@@ -30,5 +33,5 @@ unwind__get_entries(unwind_entry_cb_t cb __maybe_unused, {return 0 ; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ #endif /* __UNWIND_H */
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Introduce HAVE_DWARF_UNWIND_SUPPORT macro
perf callchain: Introduce HAVE_DWARF_UNWIND_SUPPORT macro
Commit-ID: 9ff125d132001c02d32a193a9423be0690526e11 Gitweb: http://git.kernel.org/tip/9ff125d132001c02d32a193a9423be0690526e11 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:28 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:48 -0300
perf callchain: Introduce HAVE_DWARF_UNWIND_SUPPORT macro
Introducing global macro HAVE_DWARF_UNWIND_SUPPORT to indicate we have dwarf unwind support. Any library providing the dwarf post unwind support will enable this macro.
--- tools/perf/builtin-record.c | 10 +++++----- tools/perf/config/Makefile | 2 +- tools/perf/tests/builtin-test.c | 2 +- tools/perf/tests/tests.h | 2 +- tools/perf/util/unwind.h | 7 +++++-- 5 files changed, 13 insertions(+), 10 deletions(-)
tools/perf/builtin-record.c +5 -5
@@ -649,7 +649,7 @@ error:return ret; } -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT static int get_stack_size (char *str,unsigned long *_size) {char *endptr;
@@ -675,7 +675,7 @@ static int get_stack_size(char *str, unsigned long *_size) max_size, str); return -1; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ int record_parse_callchain (const char *arg,struct record_opts *opts) {
@@ -704,7 +704,7 @@ int record_parse_callchain(const char *arg, struct record_opts *opts) "needed for -g fp\n");break ; -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT /* Dwarf style */ }else if (!strncmp (name,"dwarf" ,sizeof ("dwarf" ))) {const unsigned long default_stack_dump_size =8192 ;
@@ -720,7 +720,7 @@ int record_parse_callchain(const char *arg, struct record_opts *opts) ret =get_stack_size (tok, &size); opts->stack_dump_size = size; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ }else {pr_err ("callchain: Unknown --call-graph option " "value: %s\n" , arg);
@@ -823,7 +823,7 @@ static struct record record = {#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: " -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT const char record_callchain_help[] =CALLCHAIN_HELP "fp dwarf" ;#else const char record_callchain_help[] =CALLCHAIN_HELP "fp" ;
tools/perf/config/Makefile +1 -1
@@ -338,7 +338,7 @@ ifndef NO_LIBUNWIND CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME endif - CFLAGS += -DHAVE_LIBUNWIND_SUPPORT + CFLAGS += -DHAVE_DWARF_UNWIND_SUPPORT -DHAVE_LIBUNWIND_SUPPORT EXTLIBS += $(LIBUNWIND_LIBS) CFLAGS += $(LIBUNWIND_CFLAGS) LDFLAGS += $(LIBUNWIND_LDFLAGS)
tools/perf/tests/builtin-test.c +1 -1
@@ -116,7 +116,7 @@ static struct test { .func = test__parse_no_sample_id_all, }, #if defined(__x86_64__) || defined(__i386__) -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT { .desc = "Test dwarf unwind", .func = test__dwarf_unwind,
tools/perf/tests/tests.h +1 -1
@@ -43,7 +43,7 @@ int test__parse_no_sample_id_all(void);int test__dwarf_unwind (void );#if defined(__x86_64__) || defined(__i386__) -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT struct thread ;struct perf_sample ; int test__arch_unwind_sample(struct perf_sample *sample,
tools/perf/util/unwind.h +5 -2
@@ -13,12 +13,15 @@ struct unwind_entry {typedef int (*unwind_entry_cb_t )(struct unwind_entry *entry,void *arg); -#ifdef HAVE_LIBUNWIND_SUPPORT +#ifdef HAVE_DWARF_UNWIND_SUPPORT int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct machine *machine,struct thread *thread,struct perf_sample *data,int max_stack); +/* libunwind specific */ +#ifdef HAVE_LIBUNWIND_SUPPORT int libunwind__arch_reg_id (int regnum); +#endif #else static inline int unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
@@ -30,5 +33,5 @@ unwind__get_entries(unwind_entry_cb_t cb __maybe_unused, {return 0 ; } -#endif /* HAVE_LIBUNWIND_SUPPORT */ +#endif /* HAVE_DWARF_UNWIND_SUPPORT */ #endif /* __UNWIND_H */
PATCH 12/16 perf tools: Separate perf_reg_value function in perf_regs object
In reply to Jiri OlsaMaking perf_reg_value function global (formely reg_value), because it's going to be used globaly across all code providing the dwarf post unwind feature.
Changing its prototype to be generic:
-int reg_value(unw_word_t *valp, struct regs_dump *regs, int id) +int perf_reg_value(u64 *valp, struct regs_dump *regs, int id);
Changing the valp type from libunwind specific
'unw_word_t' to u64.
--- tools/perf/Makefile.perf | 1 + tools/perf/util/perf_regs.c | 19 +++++++++++++++++++ tools/perf/util/perf_regs.h | 13 +++++++++++++ tools/perf/util/unwind-libunwind.c | 38 ++++++++++++-------------------------- 4 files changed, 45 insertions(+), 26 deletions(-) create mode 100644 tools/perf/util/perf_regs.c
tools/perf/Makefile.perf +1 -0
@@ -537,6 +537,7 @@ ifeq ($(NO_PERF_REGS),0) ifeq ($(ARCH),x86) LIB_H += arch/x86/include/perf_regs.h endif + LIB_OBJS += $(OUTPUT)util/perf_regs.o endif ifndef NO_LIBNUMA
tools/perf/util/perf_regs.c new file +19 -0
@@ -0,0 +1,19 @@ +#include <errno.h> +#include "perf_regs.h" + +int perf_reg_value (u64 *valp,struct regs_dump *regs,int id) +{ +int i, idx =0 ; +u64 mask = regs->mask; + +if (!(mask & (1 << id))) +return -EINVAL ; + +for (i =0 ; i < id; i++) { +if (mask & (1 << i)) + idx++; + } + + *valp = regs->regs[idx]; +return 0 ; +}
tools/perf/util/perf_regs.h +13 -0
@@ -1,8 +1,14 @@#ifndef __PERF_REGS_H#define __PERF_REGS_H +#include "types.h" +#include "event.h" +#ifdef HAVE_PERF_REGS_SUPPORT #include <perf_regs.h> + +int perf_reg_value (u64 *valp,struct regs_dump *regs,int id); +#else #define PERF_REGS_MASK 0
@@ -10,5 +16,12 @@ static inline const char *perf_reg_name(int id __maybe_unused) {return NULL ; } + +static inline int perf_reg_value (u64 *valp__maybe_unused , +struct regs_dump *regs__maybe_unused , +int id__maybe_unused ) +{ +return 0 ; +}#endif /* HAVE_PERF_REGS_SUPPORT */ #endif /* __PERF_REGS_H */
tools/perf/util/unwind-libunwind.c +12 -26
@@ -406,30 +406,13 @@ static int access_dso_mem(struct unwind_info *ui, unw_word_t addr,return !(size ==sizeof (*data)); } -static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id) -{ -int i, idx =0 ; -u64 mask = regs->mask; - -if (!(mask & (1 << id))) -return -EINVAL ; - -for (i =0 ; i < id; i++) { -if (mask & (1 << i)) - idx++; - } - - *valp = regs->regs[idx]; -return 0 ; -} -static int access_mem (unw_addr_space_t __maybe_unused as,unw_word_t addr,unw_word_t *valp,int __write,void *arg) {struct unwind_info *ui = arg;struct stack_dump *stack = &ui->sample->user_stack; -unw_word_t start, end; +u64 start, end;int offset;int ret;
@@ -439,7 +422,7 @@ static int access_mem(unw_addr_space_t __maybe_unused as,return 0 ; } - ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP ); + ret =perf_reg_value (&start, &ui->sample->user_regs,PERF_REG_SP );if (ret)return ret;
@@ -452,8 +435,9 @@ static int access_mem(unw_addr_space_t __maybe_unused as,if (addr < start || addr +sizeof (unw_word_t) >= end) { ret =access_dso_mem (ui, addr, valp);if (ret) { -pr_debug ("unwind: access_mem %p not inside range %p-%p\n" , - (void *)addr, (void *)start, (void *)end); +pr_debug ("unwind: access_mem %p not inside range" +" 0x%" PRIx64"-0x%" PRIx64"\n" , + (void *) addr, start, end); *valp =0 ;return ret; }
@@ -462,8 +446,8 @@ static int access_mem(unw_addr_space_t __maybe_unused as, offset = addr - start; *valp = *(unw_word_t *)&stack->data[offset]; -pr_debug ("unwind: access_mem addr %p, val %lx, offset %d\n" , - (void *)addr, (unsigned long )*valp, offset); +pr_debug ("unwind: access_mem addr %p val %lx, offset %d\n" , + (void *) addr, (unsigned long )*valp, offset);return 0 ; }
@@ -473,6 +457,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as, {struct unwind_info *ui = arg;int id, ret; +u64 val;/* Don't support write, I suspect we don't need it. */ if (__write) {
@@ -489,12 +474,13 @@ static int access_reg(unw_addr_space_t __maybe_unused as,if (id <0 )return -EINVAL ; - ret =reg_value (valp, &ui->sample->user_regs, id); + ret =perf_reg_value (&val, &ui->sample->user_regs, id);if (ret) {pr_err ("unwind: can't read reg %d\n" , regnum);return ret; } + *valp = (unw_word_t ) val;pr_debug ("unwind: reg %d, val %lx\n" , regnum, (unsigned long )*valp);return 0 ; }
@@ -588,7 +574,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg, struct machine *machine, struct thread *thread, struct perf_sample *data, int max_stack) { - unw_word_t ip; + u64 ip; struct unwind_info ui = { .sample = data, .thread = thread,
@@ -599,7 +585,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (!data->user_regs.regs)return -EINVAL ; - ret =reg_value (&ip, &data->user_regs,PERF_REG_IP ); + ret =perf_reg_value (&ip, &data->user_regs,PERF_REG_IP );if (ret)return ret;
Signature · 1 line
1.8.3.1
tip-bot for Jiri Olsa
merge bot[tip:perf/core] perf callchain: Separate perf_reg_value function in perf_regs object
perf callchain: Separate perf_reg_value function in perf_regs object
Commit-ID: c9b951c4d12f0b2e9a07dd459c554bc05628d092 Gitweb: http://git.kernel.org/tip/c9b951c4d12f0b2e9a07dd459c554bc05628d092 Author: Jiri Olsa <[email protected]> AuthorDate: Tue, 7 Jan 2014 13:47:29 +0100 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Tue, 18 Feb 2014 09:34:49 -0300
perf callchain: Separate perf_reg_value function in perf_regs object
Making perf_reg_value function global (formely reg_value), because it's going to be used globaly across all code providing the dwarf post unwind feature.
Changing its prototype to be generic:
-int reg_value(unw_word_t *valp, struct regs_dump *regs, int id) +int perf_reg_value(u64 *valp, struct regs_dump *regs, int id);
Changing the valp type from libunwind specific 'unw_word_t' to u64.
--- tools/perf/Makefile.perf | 1 + tools/perf/util/perf_regs.c | 19 +++++++++++++++++++ tools/perf/util/perf_regs.h | 13 +++++++++++++ tools/perf/util/unwind-libunwind.c | 38 ++++++++++++-------------------------- 4 files changed, 45 insertions(+), 26 deletions(-)
tools/perf/Makefile.perf +1 -0
@@ -538,6 +538,7 @@ ifeq ($(NO_PERF_REGS),0) ifeq ($(ARCH),x86) LIB_H += arch/x86/include/perf_regs.h endif + LIB_OBJS += $(OUTPUT)util/perf_regs.o endif ifndef NO_LIBNUMA
tools/perf/util/perf_regs.c new file +19 -0
@@ -0,0 +1,19 @@ +#include <errno.h> +#include "perf_regs.h" + +int perf_reg_value (u64 *valp,struct regs_dump *regs,int id) +{ +int i, idx =0 ; +u64 mask = regs->mask; + +if (!(mask & (1 << id))) +return -EINVAL ; + +for (i =0 ; i < id; i++) { +if (mask & (1 << i)) + idx++; + } + + *valp = regs->regs[idx]; +return 0 ; +}
tools/perf/util/perf_regs.h +13 -0
@@ -1,8 +1,14 @@#ifndef __PERF_REGS_H#define __PERF_REGS_H +#include "types.h" +#include "event.h" +#ifdef HAVE_PERF_REGS_SUPPORT #include <perf_regs.h> + +int perf_reg_value (u64 *valp,struct regs_dump *regs,int id); +#else #define PERF_REGS_MASK 0
@@ -10,5 +16,12 @@ static inline const char *perf_reg_name(int id __maybe_unused) {return NULL ; } + +static inline int perf_reg_value (u64 *valp__maybe_unused , +struct regs_dump *regs__maybe_unused , +int id__maybe_unused ) +{ +return 0 ; +}#endif /* HAVE_PERF_REGS_SUPPORT */ #endif /* __PERF_REGS_H */
tools/perf/util/unwind-libunwind.c +12 -26
@@ -390,30 +390,13 @@ static int access_dso_mem(struct unwind_info *ui, unw_word_t addr,return !(size ==sizeof (*data)); } -static int reg_value (unw_word_t *valp,struct regs_dump *regs,int id) -{ -int i, idx =0 ; -u64 mask = regs->mask; - -if (!(mask & (1 << id))) -return -EINVAL ; - -for (i =0 ; i < id; i++) { -if (mask & (1 << i)) - idx++; - } - - *valp = regs->regs[idx]; -return 0 ; -} -static int access_mem (unw_addr_space_t __maybe_unused as,unw_word_t addr,unw_word_t *valp,int __write,void *arg) {struct unwind_info *ui = arg;struct stack_dump *stack = &ui->sample->user_stack; -unw_word_t start, end; +u64 start, end;int offset;int ret;
@@ -423,7 +406,7 @@ static int access_mem(unw_addr_space_t __maybe_unused as,return 0 ; } - ret =reg_value (&start, &ui->sample->user_regs,PERF_REG_SP ); + ret =perf_reg_value (&start, &ui->sample->user_regs,PERF_REG_SP );if (ret)return ret;
@@ -436,8 +419,9 @@ static int access_mem(unw_addr_space_t __maybe_unused as,if (addr < start || addr +sizeof (unw_word_t) >= end) { ret =access_dso_mem (ui, addr, valp);if (ret) { -pr_debug ("unwind: access_mem %p not inside range %p-%p\n" , - (void *)addr, (void *)start, (void *)end); +pr_debug ("unwind: access_mem %p not inside range" +" 0x%" PRIx64"-0x%" PRIx64"\n" , + (void *) addr, start, end); *valp =0 ;return ret; }
@@ -446,8 +430,8 @@ static int access_mem(unw_addr_space_t __maybe_unused as, offset = addr - start; *valp = *(unw_word_t *)&stack->data[offset]; -pr_debug ("unwind: access_mem addr %p, val %lx, offset %d\n" , - (void *)addr, (unsigned long )*valp, offset); +pr_debug ("unwind: access_mem addr %p val %lx, offset %d\n" , + (void *) addr, (unsigned long )*valp, offset);return 0 ; }
@@ -457,6 +441,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as, {struct unwind_info *ui = arg;int id, ret; +u64 val;/* Don't support write, I suspect we don't need it. */ if (__write) {
@@ -473,12 +458,13 @@ static int access_reg(unw_addr_space_t __maybe_unused as,if (id <0 )return -EINVAL ; - ret =reg_value (valp, &ui->sample->user_regs, id); + ret =perf_reg_value (&val, &ui->sample->user_regs, id);if (ret) {pr_err ("unwind: can't read reg %d\n" , regnum);return ret; } + *valp = (unw_word_t ) val;pr_debug ("unwind: reg %d, val %lx\n" , regnum, (unsigned long )*valp);return 0 ; }
@@ -572,7 +558,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg, struct machine *machine, struct thread *thread, struct perf_sample *data, int max_stack) { - unw_word_t ip; + u64 ip; struct unwind_info ui = { .sample = data, .thread = thread,
@@ -583,7 +569,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,if (!data->user_regs.regs)return -EINVAL ; - ret =reg_value (&ip, &data->user_regs,PERF_REG_IP ); + ret =perf_reg_value (&ip, &data->user_regs,PERF_REG_IP );if (ret)return ret;
PATCH 13/16 perf tools: Add feature check for libdw dwarf unwind
In reply to Jiri OlsaAdding feature check for libdw dwarf unwind.
--- tools/perf/config/Makefile | 22 ++++++++++++++-------- tools/perf/config/feature-checks/Makefile | 6 +++++- tools/perf/config/feature-checks/test-all.c | 5 +++++ .../feature-checks/test-libdw-dwarf-unwind.c | 13 +++++++++++++ 4 files changed, 37 insertions(+), 9 deletions(-) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c
tools/perf/config/Makefile +14 -8
@@ -59,6 +59,18 @@ ifeq ($(NO_PERF_REGS),0) CFLAGS += -DHAVE_PERF_REGS_SUPPORT endif +ifndef NO_LIBELF + # for linking with debug library, run like: + # make DEBUG=1 LIBDW_DIR=/opt/libdw/ + ifdef LIBDW_DIR + LIBDW_CFLAGS := -I$(LIBDW_DIR)/include + LIBDW_LDFLAGS := -L$(LIBDW_DIR)/lib + + FEATURE_CHECK_CFLAGS-libdw-dwarf-unwind := $(LIBDW_CFLAGS) + FEATURE_CHECK_LDFLAGS-libdw-dwarf-unwind := $(LIBDW_LDFLAGS) -ldw + endif +endif + # include ARCH specific config -include $(src-perf)/arch/$(ARCH)/Makefile
@@ -147,7 +159,8 @@ CORE_FEATURE_TESTS = \ libunwind \ on-exit \ stackprotector-all \ - timerfd + timerfd \ + libdw-dwarf-unwind # Set FEATURE_CHECK_(C|LD)FLAGS-all for all CORE_FEATURE_TESTS features. # If in the future we need per-feature checks/flags for features not
@@ -281,13 +294,6 @@ else msg := $(error No gnu/libc-version.h found, please install glibc-dev[el]/glibc-static); endif else - # for linking with debug library, run like: - # make DEBUG=1 LIBDW_DIR=/opt/libdw/ - ifdef LIBDW_DIR - LIBDW_CFLAGS := -I$(LIBDW_DIR)/include - LIBDW_LDFLAGS := -L$(LIBDW_DIR)/lib - endif - ifneq ($(feature-dwarf), 1) msg := $(warning No libdw.h found or old libdw.h found or elfutils is older than 0.138, disables dwarf support. Please install new elfutils-devel/libdw-dev); NO_DWARF := 1
tools/perf/config/feature-checks/Makefile +5 -1
@@ -26,7 +26,8 @@ FILES= \ test-libunwind-debug-frame.bin \ test-on-exit.bin \ test-stackprotector-all.bin \ - test-timerfd.bin + test-timerfd.bin \ + test-libdw-dwarf-unwind.bin CC := $(CC) -MD
@@ -140,6 +141,9 @@ test-backtrace.bin: test-timerfd.bin: $(BUILD) +test-libdw-dwarf-unwind.bin: + $(BUILD) + -include *.d ###############################
tools/perf/config/feature-checks/test-all.c +5 -0
@@ -89,6 +89,10 @@# include "test-stackprotector-all.c" #undef main +#define main main_test_libdw_dwarf_unwind +# include "test-libdw-dwarf-unwind.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_libdw_dwarf_unwind ();return 0 ; }
tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c new file +13 -0
@@ -0,0 +1,13 @@ + +#include <elfutils/libdwfl.h> + +int main (void ) +{ +/* +* This function is guarded via: __nonnull_attribute__ (1, 2). +* Passing '1' as arguments value. This code is never executed, +* only compiled. +*/ +dwfl_thread_getframes ((void *)1 , (void *)1 ,NULL ); +return 0 ; +}
Signature · 1 line
1.8.3.1
PATCH 14/16 perf tools: Add libdw DWARF post unwind support
In reply to Jiri OlsaAdding libdw DWARF post unwind support, which is part of elfutils-devel/libdw-dev package from version 0.158.
The new code is contained in unwin-libdw.c object, and implements unwind__get_entries unwind interface function.
New Makefile variable NO_LIBDW_DWARF_UNWIND was added to control its compilation, and is marked as disabled now. It's factored with the rest of the Makefile unwind build code in the next patch.
Arch specific code was added for x86.
--- tools/perf/Makefile.perf | 11 ++ tools/perf/arch/x86/Makefile | 3 + tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++ tools/perf/config/Makefile | 1 + tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++ 6 files changed, 297 insertions(+) create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h
tools/perf/Makefile.perf +11 -0
@@ -55,6 +55,12 @@ include config/utilities.mak # Define NO_LIBAUDIT if you do not want libaudit support # # Define NO_LIBBIONIC if you do not want bionic support +# +# Define NO_LIBDW_DWARF_UNWIND if you do not want libdw support +# for dwarf backtrace post unwind. + +# temporarily disabled +NO_LIBDW_DWARF_UNWIND := 1 ifeq ($(srctree),) srctree := $(patsubst %/,%,$(dir $(shell pwd)))
@@ -479,6 +485,11 @@ ifndef NO_DWARF endif # NO_DWARF endif # NO_LIBELF +ifndef NO_LIBDW_DWARF_UNWIND + LIB_OBJS += $(OUTPUT)util/unwind-libdw.o + LIB_H += util/unwind-libdw.h +endif + ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)util/unwind-libunwind.o endif
tools/perf/arch/x86/Makefile +3 -0
@@ -7,6 +7,9 @@ LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif +ifndef NO_LIBDW_DWARF_UNWIND +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libdw.o +endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o LIB_H += arch/$(ARCH)/util/tsc.h
tools/perf/arch/x86/util/unwind-libdw.c new file +51 -0
51 lines not on this page · raw message
tools/perf/config/Makefile +1 -0
@@ -276,6 +276,7 @@ ifdef NO_LIBELF NO_DWARF := 1 NO_DEMANGLE := 1 NO_LIBUNWIND := 1 + NO_LIBDW_DWARF_UNWIND := 1 else ifeq ($(feature-libelf), 0) ifeq ($(feature-glibc), 1)
tools/perf/util/unwind-libdw.c new file +210 -0
210 lines not on this page · raw message
tools/perf/util/unwind-libdw.h new file +21 -0
@@ -0,0 +1,21 @@ +#ifndef __PERF_UNWIND_LIBDW_H +#define __PERF_UNWIND_LIBDW_H + +#include <elfutils/libdwfl.h> +#include "event.h" +#include "thread.h" +#include "unwind.h" + +bool libdw__arch_set_initial_registers (Dwfl_Thread *thread,void *arg); + +struct unwind_info { +Dwfl *dwfl; +struct perf_sample *sample; +struct machine *machine; +struct thread *thread; +unwind_entry_cb_t cb; +void *arg; +int max_stack; +}; + +#endif /* __PERF_UNWIND_LIBDW_H */
Signature · 1 line
1.8.3.1
PATCH 15/16 perf tools: Setup default dwarf post unwinder
In reply to Jiri OlsaFactor NO_LIBDW_DWARF_UNWIND makefile variable and code that selects default DWARf post unwinder based on detected features (libdw and libunwind support)
If both are detected the libunwind is selected as default. Simple 'make' will try to add:
- libunwind unwinder if present
- libdw unwinder if present
- disable dwarf unwind if non of libunwind and libdw libraries are present
Examples:
- compile in libdw unwinder if present:
$ make NO_LIBUNWIND=1
- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1
- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1
--- tools/perf/Makefile.perf | 5 +--- tools/perf/arch/x86/Makefile | 6 +++-- tools/perf/config/Makefile | 59 ++++++++++++++++++++++++++++++++------------ 3 files changed, 48 insertions(+), 22 deletions(-)
tools/perf/Makefile.perf +1 -4
@@ -59,9 +59,6 @@ include config/utilities.mak # Define NO_LIBDW_DWARF_UNWIND if you do not want libdw support # for dwarf backtrace post unwind. -# temporarily disabled -NO_LIBDW_DWARF_UNWIND := 1 - ifeq ($(srctree),) srctree := $(patsubst %/,%,$(dir $(shell pwd))) srctree := $(patsubst %/,%,$(dir $(srctree)))
@@ -413,7 +410,7 @@ endif LIB_OBJS += $(OUTPUT)tests/code-reading.o LIB_OBJS += $(OUTPUT)tests/sample-parsing.o LIB_OBJS += $(OUTPUT)tests/parse-no-sample-id-all.o -ifndef NO_LIBUNWIND +ifndef NO_DWARF_UNWIND ifeq ($(ARCH),x86) LIB_OBJS += $(OUTPUT)tests/dwarf-unwind.o endif
tools/perf/arch/x86/Makefile +4 -2
@@ -4,12 +4,14 @@ LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o endif ifndef NO_LIBUNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libunwind.o -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o -LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o endif ifndef NO_LIBDW_DWARF_UNWIND LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/unwind-libdw.o endif +ifndef NO_DWARF_UNWIND +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/regs_load.o +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/tests/dwarf-unwind.o +endif LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/tsc.o LIB_H += arch/$(ARCH)/util/tsc.h
tools/perf/config/Makefile +43 -16
74 lines not on this page · raw message
Signature · 1 line
1.8.3.1
PATCH 16/16 perf tests: Add NO_LIBDW_DWARF_UNWIND make test
In reply to Jiri OlsaAdding make test for NO_LIBDW_DWARF_UNWIND option, plus updating minimal build test with it.
--- tools/perf/tests/make | 63 +++++++++++++++++++++++++++------------------------ 1 file changed, 33 insertions(+), 30 deletions(-)
tools/perf/tests/make +33 -30
@@ -14,41 +14,43 @@ endif has = $(shell which $1 2>/dev/null) # standard single make variable specified -make_clean_all := clean all -make_python_perf_so := python/perf.so -make_debug := DEBUG=1 -make_no_libperl := NO_LIBPERL=1 -make_no_libpython := NO_LIBPYTHON=1 -make_no_scripts := NO_LIBPYTHON=1 NO_LIBPERL=1 -make_no_newt := NO_NEWT=1 -make_no_slang := NO_SLANG=1 -make_no_gtk2 := NO_GTK2=1 -make_no_ui := NO_NEWT=1 NO_SLANG=1 NO_GTK2=1 -make_no_demangle := NO_DEMANGLE=1 -make_no_libelf := NO_LIBELF=1 -make_no_libunwind := NO_LIBUNWIND=1 -make_no_backtrace := NO_BACKTRACE=1 -make_no_libnuma := NO_LIBNUMA=1 -make_no_libaudit := NO_LIBAUDIT=1 -make_no_libbionic := NO_LIBBIONIC=1 -make_tags := tags -make_cscope := cscope -make_help := help -make_doc := doc -make_perf_o := perf.o -make_util_map_o := util/map.o -make_install := install -make_install_bin := install-bin -make_install_doc := install-doc -make_install_man := install-man -make_install_html := install-html -make_install_info := install-info -make_install_pdf := install-pdf +make_clean_all := clean all +make_python_perf_so := python/perf.so +make_debug := DEBUG=1 +make_no_libperl := NO_LIBPERL=1 +make_no_libpython := NO_LIBPYTHON=1 +make_no_scripts := NO_LIBPYTHON=1 NO_LIBPERL=1 +make_no_newt := NO_NEWT=1 +make_no_slang := NO_SLANG=1 +make_no_gtk2 := NO_GTK2=1 +make_no_ui := NO_NEWT=1 NO_SLANG=1 NO_GTK2=1 +make_no_demangle := NO_DEMANGLE=1 +make_no_libelf := NO_LIBELF=1 +make_no_libunwind := NO_LIBUNWIND=1 +make_no_libdw_dwarf_unwind := NO_LIBDW_DWARF_UNWIND=1 +make_no_backtrace := NO_BACKTRACE=1 +make_no_libnuma := NO_LIBNUMA=1 +make_no_libaudit := NO_LIBAUDIT=1 +make_no_libbionic := NO_LIBBIONIC=1 +make_tags := tags +make_cscope := cscope +make_help := help +make_doc := doc +make_perf_o := perf.o +make_util_map_o := util/map.o +make_install := install +make_install_bin := install-bin +make_install_doc := install-doc +make_install_man := install-man +make_install_html := install-html +make_install_info := install-info +make_install_pdf := install-pdf # all the NO_* variable combined make_minimal := NO_LIBPERL=1 NO_LIBPYTHON=1 NO_NEWT=1 NO_GTK2=1 make_minimal += NO_DEMANGLE=1 NO_LIBELF=1 NO_LIBUNWIND=1 NO_BACKTRACE=1 make_minimal += NO_LIBNUMA=1 NO_LIBAUDIT=1 NO_LIBBIONIC=1 +make_minimal += NO_LIBDW_DWARF_UNWIND=1 # $(run) contains all available tests run := make_pure
@@ -65,6 +67,7 @@ run += make_no_ui run += make_no_demangle run += make_no_libelf run += make_no_libunwind +run += make_no_libdw_dwarf_unwind run += make_no_backtrace run += make_no_libnuma run += make_no_libaudit
Signature · 1 line
1.8.3.1
Hi Jiri,
Quoted 6 lines · Jiri Olsa
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
That is great!
Quoted 24 lines
Examples:
- compile in libdw unwinder if present:$ make NO_LIBUNWIND=1- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldwreal 0m35.746s user 0m24.211s sys 0m11.214sThe libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unw
Quoted 8 lines
real 1m41.149s user 1m9.694s sys 0m30.595sIn example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Nice improvement!
Quoted 11 lines
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwindKudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)
Yes sure. I will review the patches and test/benchmark them on ARMv7/v8. Let me do that after the current tasks.
Quoted 3 lines
thanks,
jirka
Regards,
Jean
Quoted 75 lines
Signed-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jan Kratochvil <[email protected]>Cc Jean Pihet <[email protected]>---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Makeperf_event__synthesize_mmap_eventsglobal perf tests x86: Introduceperf_regs_loadfunction
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwindmax_stackprocessing
perf tools: Do not report zero address in unwind
perf tools: Add mask into structregs_dump
perf tools: Separate libunwind code to special object perf tools: Renameunwind__arch_reg_idintolibunwind__arch_reg_idperf tools: IntroduceHAVE_DWARF_UNWIND_SUPPORTmacro perf tools: Separateperf_reg_valuefunction inperf_regsobject perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: AddNO_LIBDW_DWARF_UNWINDmake testtools/perf/Makefile.perf | 19 ++++++- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +- tools/perf/arch/x86/Makefile | 9 +++- tools/perf/arch/x86/include/perf_regs.h | 6 +++ tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++ tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++ tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++ tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +- tools/perf/builtin-inject.c | 1 - tools/perf/builtin-record.c | 10 ++-- tools/perf/config/Makefile | 94 +++++++++++++++++++++++--------- tools/perf/config/feature-checks/Makefile | 6 ++- tools/perf/config/feature-checks/test-all.c | 5 ++ tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++ tools/perf/tests/builtin-test.c | 8 +++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/make | 63 +++++++++++----------- tools/perf/tests/sample-parsing.c | 17 +++--- tools/perf/tests/tests.h | 9 ++++ tools/perf/util/event.c | 12 ++--- tools/perf/util/event.h | 12 ++++- tools/perf/util/evsel.c | 13 ++--- tools/perf/util/machine.c | 4 +- tools/perf/util/perf_regs.c | 19 +++++++ tools/perf/util/perf_regs.h | 13 +++++ tools/perf/util/session.c | 5 +- tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++++++ tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++----------- tools/perf/util/unwind.h | 11 ++-- 31 files changed, 849 insertions(+), 135 deletions(-) rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/arch/x86/tests/regs_load.S create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c create mode 100644 tools/perf/util/perf_regs.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)
ping ;-)
thanks,
jirka
Quoted 128 lines · Jiri Olsa
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
Examples:
- compile in libdw unwinder if present:$ make NO_LIBUNWIND=1- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldwreal 0m35.746s user 0m24.211s sys 0m11.214sThe libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unwreal 1m41.149s user 1m9.694s sys 0m30.595sIn example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwindKudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)thanks,
jirkaSigned-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jan Kratochvil <[email protected]>Cc Jean Pihet <[email protected]>---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Makeperf_event__synthesize_mmap_eventsglobal perf tests x86: Introduceperf_regs_loadfunction
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwindmax_stackprocessing
perf tools: Do not report zero address in unwind
perf tools: Add mask into structregs_dump
perf tools: Separate libunwind code to special object perf tools: Renameunwind__arch_reg_idintolibunwind__arch_reg_idperf tools: IntroduceHAVE_DWARF_UNWIND_SUPPORTmacro perf tools: Separateperf_reg_valuefunction inperf_regsobject perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: AddNO_LIBDW_DWARF_UNWINDmake testtools/perf/Makefile.perf | 19 ++++++- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +- tools/perf/arch/x86/Makefile | 9 +++- tools/perf/arch/x86/include/perf_regs.h | 6 +++ tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++ tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++ tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++ tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +- tools/perf/builtin-inject.c | 1 - tools/perf/builtin-record.c | 10 ++-- tools/perf/config/Makefile | 94 +++++++++++++++++++++++--------- tools/perf/config/feature-checks/Makefile | 6 ++- tools/perf/config/feature-checks/test-all.c | 5 ++ tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++ tools/perf/tests/builtin-test.c | 8 +++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/make | 63 +++++++++++----------- tools/perf/tests/sample-parsing.c | 17 +++--- tools/perf/tests/tests.h | 9 ++++ tools/perf/util/event.c | 12 ++--- tools/perf/util/event.h | 12 ++++- tools/perf/util/evsel.c | 13 ++--- tools/perf/util/machine.c | 4 +- tools/perf/util/perf_regs.c | 19 +++++++ tools/perf/util/perf_regs.h | 13 +++++ tools/perf/util/session.c | 5 +- tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++++++ tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++----------- tools/perf/util/unwind.h | 11 ++-- 31 files changed, 849 insertions(+), 135 deletions(-) rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/arch/x86/tests/regs_load.S create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c create mode 100644 tools/perf/util/perf_regs.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)
Hi Jiri,
Let me start on it next week. Is that OK?
Thx for the reminder, I had not entirely forgotten it though ;p
Regards,
Jean
Quoted 134 lines · Jiri Olsa
ping ;-)
thanks,
jirkaQuoted 128 lines · Jiri Olsa
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
Examples:
- compile in libdw unwinder if present:$ make NO_LIBUNWIND=1- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldwreal 0m35.746s user 0m24.211s sys 0m11.214sThe libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unwreal 1m41.149s user 1m9.694s sys 0m30.595sIn example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwindKudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)thanks,
jirkaSigned-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jan Kratochvil <[email protected]>Cc Jean Pihet <[email protected]>---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Makeperf_event__synthesize_mmap_eventsglobal perf tests x86: Introduceperf_regs_loadfunction
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwindmax_stackprocessing
perf tools: Do not report zero address in unwind
perf tools: Add mask into structregs_dump
perf tools: Separate libunwind code to special object perf tools: Renameunwind__arch_reg_idintolibunwind__arch_reg_idperf tools: IntroduceHAVE_DWARF_UNWIND_SUPPORTmacro perf tools: Separateperf_reg_valuefunction inperf_regsobject perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: AddNO_LIBDW_DWARF_UNWINDmake testtools/perf/Makefile.perf | 19 ++++++- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +- tools/perf/arch/x86/Makefile | 9 +++- tools/perf/arch/x86/include/perf_regs.h | 6 +++ tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++ tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++ tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++ tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +- tools/perf/builtin-inject.c | 1 - tools/perf/builtin-record.c | 10 ++-- tools/perf/config/Makefile | 94 +++++++++++++++++++++++--------- tools/perf/config/feature-checks/Makefile | 6 ++- tools/perf/config/feature-checks/test-all.c | 5 ++ tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++ tools/perf/tests/builtin-test.c | 8 +++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/make | 63 +++++++++++----------- tools/perf/tests/sample-parsing.c | 17 +++--- tools/perf/tests/tests.h | 9 ++++ tools/perf/util/event.c | 12 ++--- tools/perf/util/event.h | 12 ++++- tools/perf/util/evsel.c | 13 ++--- tools/perf/util/machine.c | 4 +- tools/perf/util/perf_regs.c | 19 +++++++ tools/perf/util/perf_regs.h | 13 +++++ tools/perf/util/session.c | 5 +- tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++++++ tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++----------- tools/perf/util/unwind.h | 11 ++-- 31 files changed, 849 insertions(+), 135 deletions(-) rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/arch/x86/tests/regs_load.S create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c create mode 100644 tools/perf/util/perf_regs.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)
Quoted 3 lines · Jean Pihet
Hi Jiri,
Let me start on it next week. Is that OK?
great ;-) thanks
jirka
Quoted 141 lines
Thx for the reminder, I had not entirely forgotten it though ;p
Regards,
JeanQuoted 134 lines · Jiri Olsa
ping ;-)
thanks,
jirkaQuoted 128 lines · Jiri Olsa
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
Examples:
- compile in libdw unwinder if present:$ make NO_LIBUNWIND=1- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldwreal 0m35.746s user 0m24.211s sys 0m11.214sThe libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unwreal 1m41.149s user 1m9.694s sys 0m30.595sIn example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwindKudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)thanks,
jirkaSigned-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jan Kratochvil <[email protected]>Cc Jean Pihet <[email protected]>---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Makeperf_event__synthesize_mmap_eventsglobal perf tests x86: Introduceperf_regs_loadfunction
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwindmax_stackprocessing
perf tools: Do not report zero address in unwind
perf tools: Add mask into structregs_dump
perf tools: Separate libunwind code to special object perf tools: Renameunwind__arch_reg_idintolibunwind__arch_reg_idperf tools: IntroduceHAVE_DWARF_UNWIND_SUPPORTmacro perf tools: Separateperf_reg_valuefunction inperf_regsobject perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: AddNO_LIBDW_DWARF_UNWINDmake testtools/perf/Makefile.perf | 19 ++++++- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +- tools/perf/arch/x86/Makefile | 9 +++- tools/perf/arch/x86/include/perf_regs.h | 6 +++ tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++ tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++ tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++ tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +- tools/perf/builtin-inject.c | 1 - tools/perf/builtin-record.c | 10 ++-- tools/perf/config/Makefile | 94 +++++++++++++++++++++++--------- tools/perf/config/feature-checks/Makefile | 6 ++- tools/perf/config/feature-checks/test-all.c | 5 ++ tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++ tools/perf/tests/builtin-test.c | 8 +++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/make | 63 +++++++++++----------- tools/perf/tests/sample-parsing.c | 17 +++--- tools/perf/tests/tests.h | 9 ++++ tools/perf/util/event.c | 12 ++--- tools/perf/util/event.h | 12 ++++- tools/perf/util/evsel.c | 13 ++--- tools/perf/util/machine.c | 4 +- tools/perf/util/perf_regs.c | 19 +++++++ tools/perf/util/perf_regs.h | 13 +++++ tools/perf/util/session.c | 5 +- tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++++++ tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++----------- tools/perf/util/unwind.h | 11 ++-- 31 files changed, 849 insertions(+), 135 deletions(-) rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/arch/x86/tests/regs_load.S create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c create mode 100644 tools/perf/util/perf_regs.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)
Quoted 5 lines · Jean Pihet
Hi Jiri,
Let me start on it next week. Is that OK?
Thx for the reminder, I had not entirely forgotten it though ;p
hi,
rebased patchset to latest acme's perf/core is in:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwind
aaand ping ;-)
thanks,
jirka
Hi Jiri,
Quoted 10 lines · Jiri Olsa
Quoted 5 lines · Jean Pihet
Hi Jiri,
Let me start on it next week. Is that OK?
Thx for the reminder, I had not entirely forgotten it though ;p
hi,
rebased patchset to latest acme's perf/core is in:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwind
Great, that was needed.
Quoted 1 line
aaand ping ;-)
I checked that the build still works ok on ARM. I am now trying with libdw. I can give an Ack as soon as that is done, then implement the ARM specific code.
Thanks,
Jean
Quoted 3 lines
thanks,
jirka
Jean Pihet
Acked-byHi Jiri,
Quoted 13 lines · Jean Pihet
Hi Jiri,
Quoted 10 lines · Jiri Olsa
Quoted 5 lines · Jean Pihet
Hi Jiri,
Let me start on it next week. Is that OK?
Thx for the reminder, I had not entirely forgotten it though ;p
hi,
rebased patchset to latest acme's perf/core is in:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwind
Quoted 5 lines
Great, that was needed.
Quoted 1 line
aaand ping ;-)
I checked that the build still works ok on ARM. I am now trying with libdw. I can give an Ack as soon as that is done, then implement the ARM specific code.
The build with and without libunwind and libdw has been tested succesfully on ARM, including the paths to the desired libraries. Of course I am getting a link error on 'libdw__arch_set_initial_registers' because the ARM support is not yet in.
Here is an ack on the series:
Next on my plate: implement the ARM specifics.
Thanks!
Jean
Quoted 4 lines
Quoted 4 lines
thanks,
jirka
Em Fri, Jan 17, 2014 at 01:24:31PM +0100, Jiri Olsa escreveu:
Quoted 1 line
ping ;-)
I'll try and test this this afternoon to possibly get it in the next pull req, after the one I'll do shortly
Quoted 132 lines
thanks,
jirkaQuoted 128 lines · Jiri Olsa
hi,
adding libdw DWARF unwind support, which is part of just released elfutils package 0.158.We can now compile perf to have either libunwind DWARF unwind, (which is still default) or the new one libdw unwind.
Examples:
- compile in libdw unwinder if present:$ make NO_LIBUNWIND=1- compile in libdw (with libdw installation directory) unwinder if present:
$ make LIBDW_DIR=/opt/elfutils/ NO_LIBUNWIND=1- disable post dwarf unwind completely:
$ make NO_LIBUNWIND=1 NO_LIBDW_DWARF_UNWIND=1Performance wise the libdw unwinder seems to be faster. Following results are from 2GB perf.data file with 249156 samples:
The libdw unwind perf:
$ time perf report -i ~/perf.data > perf.data.ldwreal 0m35.746s user 0m24.211s sys 0m11.214sThe libunwind unwind perf:
$ time perf report -i ~/perf.data > perf.data.unwreal 1m41.149s user 1m9.694s sys 0m30.595sIn example above I've got 2206 differences (perf.data.ldw and perf.data.unw files) in the callchains output out of the 249156 samples which is below 1%. Together with some other tips for speed increase it's on my TODO list for future.
Git tree:
https://git.kernel.org/cgit/linux/kernel/git/jolsa/perf.git/
perf/core_libdw_unwindKudos to Jan Kratochvil for helping me with the new interface.
Jean,
I haven't tested on arm.. so not sure I broke anything there. Any chance you could test for me, and maybe make the automated test I added work there? ;-)thanks,
jirkaSigned-off-by Jiri Olsa <[email protected]>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 David Ahern <[email protected]>Cc Jan Kratochvil <[email protected]>Cc Jean Pihet <[email protected]>---
Jiri Olsa (16):
perf tools: Automate setup of FEATURE_CHECK_(C|LD)FLAGS-all variables perf tools: Fix machine initialization
perf tools: Makeperf_event__synthesize_mmap_eventsglobal perf tests x86: Introduceperf_regs_loadfunction
perf tests x86: Add dwarf unwind test
perf tools: Fix dwarf unwindmax_stackprocessing
perf tools: Do not report zero address in unwind
perf tools: Add mask into structregs_dump
perf tools: Separate libunwind code to special object perf tools: Renameunwind__arch_reg_idintolibunwind__arch_reg_idperf tools: IntroduceHAVE_DWARF_UNWIND_SUPPORTmacro perf tools: Separateperf_reg_valuefunction inperf_regsobject perf tools: Add feature check for libdw dwarf unwind
perf tools: Add libdw DWARF post unwind support
perf tools: Setup default dwarf post unwinder
perf tests: AddNO_LIBDW_DWARF_UNWINDmake testtools/perf/Makefile.perf | 19 ++++++- tools/perf/arch/arm/Makefile | 2 +- tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} | 2 +- tools/perf/arch/x86/Makefile | 9 +++- tools/perf/arch/x86/include/perf_regs.h | 6 +++ tools/perf/arch/x86/tests/dwarf-unwind.c | 59 ++++++++++++++++++++ tools/perf/arch/x86/tests/regs_load.S | 92 ++++++++++++++++++++++++++++++++ tools/perf/arch/x86/util/unwind-libdw.c | 51 ++++++++++++++++++ tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} | 4 +- tools/perf/builtin-inject.c | 1 - tools/perf/builtin-record.c | 10 ++-- tools/perf/config/Makefile | 94 +++++++++++++++++++++++--------- tools/perf/config/feature-checks/Makefile | 6 ++- tools/perf/config/feature-checks/test-all.c | 5 ++ tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c | 13 +++++ tools/perf/tests/builtin-test.c | 8 +++ tools/perf/tests/dwarf-unwind.c | 144 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/make | 63 +++++++++++----------- tools/perf/tests/sample-parsing.c | 17 +++--- tools/perf/tests/tests.h | 9 ++++ tools/perf/util/event.c | 12 ++--- tools/perf/util/event.h | 12 ++++- tools/perf/util/evsel.c | 13 ++--- tools/perf/util/machine.c | 4 +- tools/perf/util/perf_regs.c | 19 +++++++ tools/perf/util/perf_regs.h | 13 +++++ tools/perf/util/session.c | 5 +- tools/perf/util/unwind-libdw.c | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/unwind-libdw.h | 21 ++++++++ tools/perf/util/{unwind.c => unwind-libunwind.c} | 50 ++++++----------- tools/perf/util/unwind.h | 11 ++-- 31 files changed, 849 insertions(+), 135 deletions(-) rename tools/perf/arch/arm/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/arch/x86/tests/dwarf-unwind.c create mode 100644 tools/perf/arch/x86/tests/regs_load.S create mode 100644 tools/perf/arch/x86/util/unwind-libdw.c rename tools/perf/arch/x86/util/{unwind.c => unwind-libunwind.c} (95%) create mode 100644 tools/perf/config/feature-checks/test-libdw-dwarf-unwind.c create mode 100644 tools/perf/tests/dwarf-unwind.c create mode 100644 tools/perf/util/perf_regs.c create mode 100644 tools/perf/util/unwind-libdw.c create mode 100644 tools/perf/util/unwind-libdw.h rename tools/perf/util/{unwind.c => unwind-libunwind.c} (92%)