[PATCH 1/2] Support perf script -F brstack,dso and brstacksym,dso

[PATCH 1/2] Support perf script -F brstack,dso and brstacksym,dso

[Mark Santaniello]

Perf script can report the dso for "addr" and "ip" fields.
This adds the same support for the "brstack" and "brstacksym" fields.
This can be helpful for AutoFDO: we can ignore LBR entries unless the
source and target address are both in the target module we are about to
build.

I built a small test akin to "while(1) { do_nothing(); }" where the
do_nothing function is loaded from a dso:

  ~/src/burncpu$ cat burncpu.cpp
  #include <dlfcn.h>

  int main() {
    void* handle = dlopen("./dso.so", RTLD_LAZY);
    if (!handle) return -1;

    typedef void (*fp)();
    fp do_nothing = (fp) dlsym(handle, "do_nothing");

    while(1) {
      do_nothing();
    }
  }

  ~/src/burncpu$ cat dso.cpp
  extern "C" void do_nothing() {}

  ~/src/burncpu$ cat build.sh
  #!/bin/bash
  g++ -shared dso.cpp -o dso.so
  g++ burncpu.cpp -o burncpu -ldl

I sampled the execution with perf record -b.  Using the new perf script
functionality I can easily find cases where there was a transition from one
dso to another:

  $ ./perf record -a -b -- sleep 5
  [ perf record: Woken up 55 times to write data ]
  [ perf record: Captured and wrote 18.815 MB perf.data (43593 samples) ]

  $ ./perf script -F brstack,dso | sed 's/\/0 /\/0\n/g' \
  > | grep burncpu | grep dso.so | head -n 1
  0x7f967139b6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

  $ ./perf script -F brstacksym,dso | sed 's/\/0 /\/0\n/g' \
  > | grep burncpu | grep dso.so | head -n 1
  do_nothing+0x5(/tmp/burncpu/dso.so)/main+0x44(/tmp/burncpu/exe)/P/-/-/0

Signed-off-by: Mark Santaniello <marksan@fb.com>
---
 tools/perf/builtin-script.c | 61 ++++++++++++++++++++++++++++++++++++---------
 1 file changed, 49 insertions(+), 12 deletions(-)

diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 4761b0d..6a7033b 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -298,10 +298,11 @@ static int perf_evsel__check_attr(struct perf_evsel *evsel,
 		       "selected.\n");
 		return -EINVAL;
 	}
-	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR)) {
-		pr_err("Display of DSO requested but neither sample IP nor "
-			   "sample address\nis selected. Hence, no addresses to convert "
-		       "to DSO.\n");
+	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR) &&
+	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM)) {
+		pr_err("Display of DSO requested but none of sample IP, sample address, "
+		       "brstack\nor brstacksym are selected. Hence, no addresses to "
+		       "convert to DSO.\n");
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(SRCLINE) && !PRINT_FIELD(IP)) {
@@ -514,18 +515,43 @@ mispred_str(struct branch_entry *br)
 	return br->flags.predicted ? 'P' : 'M';
 }
 
-static void print_sample_brstack(struct perf_sample *sample)
+static void print_sample_brstack(struct perf_sample *sample,
+				 struct thread *thread,
+				 struct perf_event_attr *attr)
 {
 	struct branch_stack *br = sample->branch_stack;
-	u64 i;
+	struct addr_location alf, alt;
+	u64 i, from, to;
 
 	if (!(br && br->nr))
 		return;
 
 	for (i = 0; i < br->nr; i++) {
-		printf(" 0x%"PRIx64"/0x%"PRIx64"/%c/%c/%c/%d ",
-			br->entries[i].from,
-			br->entries[i].to,
+		from = br->entries[i].from;
+		to   = br->entries[i].to;
+
+		if (PRINT_FIELD(DSO)) {
+			memset(&alf, 0, sizeof(alf));
+			memset(&alt, 0, sizeof(alt));
+			thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, from, &alf);
+			thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, to, &alt);
+		}
+
+		printf("0x%"PRIx64, from);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
+
+		printf("/0x%"PRIx64, to);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
+
+		printf("/%c/%c/%c/%d ",
 			mispred_str( br->entries + i),
 			br->entries[i].flags.in_tx? 'X' : '-',
 			br->entries[i].flags.abort? 'A' : '-',
@@ -534,7 +560,8 @@ static void print_sample_brstack(struct perf_sample *sample)
 }
 
 static void print_sample_brstacksym(struct perf_sample *sample,
-				    struct thread *thread)
+				    struct thread *thread,
+				    struct perf_event_attr *attr)
 {
 	struct branch_stack *br = sample->branch_stack;
 	struct addr_location alf, alt;
@@ -559,8 +586,18 @@ static void print_sample_brstacksym(struct perf_sample *sample,
 			alt.sym = map__find_symbol(alt.map, alt.addr);
 
 		symbol__fprintf_symname_offs(alf.sym, &alf, stdout);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
 		putchar('/');
 		symbol__fprintf_symname_offs(alt.sym, &alt, stdout);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
 		printf("/%c/%c/%c/%d ",
 			mispred_str( br->entries + i),
 			br->entries[i].flags.in_tx? 'X' : '-',
@@ -1187,9 +1224,9 @@ static void process_event(struct perf_script *script,
 		print_sample_iregs(sample, attr);
 
 	if (PRINT_FIELD(BRSTACK))
-		print_sample_brstack(sample);
+		print_sample_brstack(sample, thread, attr);
 	else if (PRINT_FIELD(BRSTACKSYM))
-		print_sample_brstacksym(sample, thread);
+		print_sample_brstacksym(sample, thread, attr);
 
 	if (perf_evsel__is_bpf_output(evsel) && PRINT_FIELD(BPF_OUTPUT))
 		print_sample_bpf_output(sample);
-- 
2.9.3

[PATCH 2/2] Support perf script -F brstackoff,dso

[Mark Santaniello]

The idea here is to make AutoFDO easier in cloud environment with ASLR.
It's easiest to show how this is useful by example. I built a small test
akin to "while(1) { do_nothing(); }" where the do_nothing function is
loaded from a dso:

  ~/src/burncpu$ cat burncpu.cpp
  #include <dlfcn.h>

  int main() {
    void* handle = dlopen("./dso.so", RTLD_LAZY);
    if (!handle) return -1;

    typedef void (*fp)();
    fp do_nothing = (fp) dlsym(handle, "do_nothing");

    while(1) {
      do_nothing();
    }
  }

  ~/src/burncpu$ cat dso.cpp
  extern "C" void do_nothing() {}

  ~/src/burncpu$ cat build.sh
  #!/bin/bash
  g++ -shared dso.cpp -o dso.so
  g++ burncpu.cpp -o burncpu -ldl

I sampled the execution of this program with perf record -b.

Using the existing "brstack,dso", we get absolute addresses that are
affected by ASLR, and could be different on different hosts. The address
does not uniquely identify a branch/target in the binary:
  $ ./perf script -F brstack,dso | sed 's/\/0 /\/0\n/g' \
  > | grep burncpu | grep dso.so | head -n 1
  0x7f967139b6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

Using the existing "brstacksym,dso" is a little better, because the
symbol plus offset and dso name *does* uniquely identify a branch/target
in the binary.  Ultimately, however, AutoFDO wants a simple offset into
the binary, so we'd have to undo all the work perf did to symbolize in
the first place:
  $ ./perf script -F brstacksym,dso | sed 's/\/0 /\/0\n/g' \
  > | grep burncpu | grep dso.so | head -n 1
  do_nothing+0x5(/tmp/burncpu/dso.so)/main+0x44(/tmp/burncpu/exe)/P/-/-/0

With the new "brstackoff,dso" we get what we need: a simple offset into a
specific dso/binary that uniquely identifies a branch/target:
  $ ./perf script -F brstackoff,dso | sed 's/\/0 /\/0\n/g' \
  > | grep burncpu | grep dso.so | head -n 1
  0x6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

Signed-off-by: Mark Santaniello <marksan@fb.com>
---
 tools/perf/builtin-script.c | 56 +++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 52 insertions(+), 4 deletions(-)

diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 6a7033b..1effc64 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -85,6 +85,7 @@ enum perf_output_field {
 	PERF_OUTPUT_INSN	    = 1U << 21,
 	PERF_OUTPUT_INSNLEN	    = 1U << 22,
 	PERF_OUTPUT_BRSTACKINSN	    = 1U << 23,
+	PERF_OUTPUT_BRSTACKOFF	    = 1U << 24,
 };
 
 struct output_option {
@@ -115,6 +116,7 @@ struct output_option {
 	{.str = "insn", .field = PERF_OUTPUT_INSN},
 	{.str = "insnlen", .field = PERF_OUTPUT_INSNLEN},
 	{.str = "brstackinsn", .field = PERF_OUTPUT_BRSTACKINSN},
+	{.str = "brstackoff", .field = PERF_OUTPUT_BRSTACKOFF},
 };
 
 /* default set to maintain compatibility with current format */
@@ -299,10 +301,9 @@ static int perf_evsel__check_attr(struct perf_evsel *evsel,
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR) &&
-	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM)) {
-		pr_err("Display of DSO requested but none of sample IP, sample address, "
-		       "brstack\nor brstacksym are selected. Hence, no addresses to "
-		       "convert to DSO.\n");
+	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM) && !PRINT_FIELD(BRSTACKOFF)) {
+		pr_err("Display of DSO requested but no address to convert.  Select\n"
+		       "sample IP, sample address, brstack, brstacksym, or brstackoff.\n");
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(SRCLINE) && !PRINT_FIELD(IP)) {
@@ -606,6 +607,51 @@ static void print_sample_brstacksym(struct perf_sample *sample,
 	}
 }
 
+static void print_sample_brstackoff(struct perf_sample *sample,
+				    struct thread *thread,
+				    struct perf_event_attr *attr)
+{
+	struct branch_stack *br = sample->branch_stack;
+	struct addr_location alf, alt;
+	u64 i, from, to;
+
+	if (!(br && br->nr))
+		return;
+
+	for (i = 0; i < br->nr; i++) {
+
+		memset(&alf, 0, sizeof(alf));
+		memset(&alt, 0, sizeof(alt));
+		from = br->entries[i].from;
+		to   = br->entries[i].to;
+
+		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, from, &alf);
+		if (alf.map && !alf.map->dso->adjust_symbols)
+			from = map__map_ip(alf.map, from);
+
+		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, to, &alt);
+		if (alt.map && !alt.map->dso->adjust_symbols)
+			to = map__map_ip(alt.map, to);
+
+		printf("0x%"PRIx64, from);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
+		printf("/0x%"PRIx64, to);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
+		printf("/%c/%c/%c/%d ",
+			mispred_str(br->entries + i),
+			br->entries[i].flags.in_tx ? 'X' : '-',
+			br->entries[i].flags.abort ? 'A' : '-',
+			br->entries[i].flags.cycles);
+	}
+}
 #define MAXBB 16384UL
 
 static int grab_bb(u8 *buffer, u64 start, u64 end,
@@ -1227,6 +1273,8 @@ static void process_event(struct perf_script *script,
 		print_sample_brstack(sample, thread, attr);
 	else if (PRINT_FIELD(BRSTACKSYM))
 		print_sample_brstacksym(sample, thread, attr);
+	else if (PRINT_FIELD(BRSTACKOFF))
+		print_sample_brstackoff(sample, thread, attr);
 
 	if (perf_evsel__is_bpf_output(evsel) && PRINT_FIELD(BPF_OUTPUT))
 		print_sample_bpf_output(sample);
-- 
2.9.3

Re: [PATCH 2/2] Support perf script -F brstackoff,dso

[Arnaldo Carvalho de Melo]

Em Mon, Jun 19, 2017 at 09:38:25AM -0700, Mark Santaniello escreveu:
> With the new "brstackoff,dso" we get what we need: a simple offset into a
> specific dso/binary that uniquely identifies a branch/target:

You forgot to update tools/perf/Documentation/perf-script.txt about
'brstackoff', I added a line using the above explanation, please check
later and improve/expand it.

Thanks,

- Arnaldo

Re: [PATCH 2/2] Support perf script -F brstackoff,dso

[Mark Santaniello]

On 6/19/17, 11:12 AM, "Arnaldo Carvalho de Melo" <acme@kernel.org> wrote:

    Em Mon, Jun 19, 2017 at 09:38:25AM -0700, Mark Santaniello escreveu:
    > With the new "brstackoff,dso" we get what we need: a simple offset into a
    > specific dso/binary that uniquely identifies a branch/target:
    
    You forgot to update tools/perf/Documentation/perf-script.txt about
    'brstackoff', I added a line using the above explanation, please check
    later and improve/expand it.
    
    Thanks,
    
    - Arnaldo

Arnaldo,

Thanks very much.  Will do.

-Mark    

[tip:perf/core] perf script: Support -F brstackoff,dso

[tip-bot for Mark Santaniello]

Commit-ID:  106dacd86f042968e0bb974490fcb9cd017cd03a
Gitweb:     http://git.kernel.org/tip/106dacd86f042968e0bb974490fcb9cd017cd03a
Author:     Mark Santaniello <marksan@fb.com>
AuthorDate: Mon, 19 Jun 2017 09:38:25 -0700
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Mon, 19 Jun 2017 22:05:46 -0300

perf script: Support -F brstackoff,dso

The idea here is to make AutoFDO easier in cloud environment with ASLR.
It's easiest to show how this is useful by example. I built a small test
akin to "while(1) { do_nothing(); }" where the do_nothing function is
loaded from a dso:

  $ cat burncpu.cpp
  #include <dlfcn.h>

  int main() {
    void* handle = dlopen("./dso.so", RTLD_LAZY);
    if (!handle) return -1;

    typedef void (*fp)();
    fp do_nothing = (fp) dlsym(handle, "do_nothing");

    while(1) {
      do_nothing();
    }
  }

  $ cat dso.cpp
  extern "C" void do_nothing() {}

  $ cat build.sh
  #!/bin/bash
  g++ -shared dso.cpp -o dso.so
  g++ burncpu.cpp -o burncpu -ldl

I sampled the execution of this program with perf record -b.

Using the existing "brstack,dso", we get absolute addresses that are
affected by ASLR, and could be different on different hosts. The address
does not uniquely identify a branch/target in the binary:

  $ perf script -F brstack,dso | sed 's/\/0 /\/0\n/g' | grep burncpu | grep dso.so | head -n 1
  0x7f967139b6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

Using the existing "brstacksym,dso" is a little better, because the
symbol plus offset and dso name *does* uniquely identify a branch/target
in the binary.  Ultimately, however, AutoFDO wants a simple offset into
the binary, so we'd have to undo all the work perf did to symbolize in
the first place:

  $ perf script -F brstacksym,dso | sed 's/\/0 /\/0\n/g' | grep burncpu | grep dso.so | head -n 1
  do_nothing+0x5(/tmp/burncpu/dso.so)/main+0x44(/tmp/burncpu/exe)/P/-/-/0

With the new "brstackoff,dso" we get what we need: a simple offset into a
specific dso/binary that uniquely identifies a branch/target:
  $ perf script -F brstackoff,dso | sed 's/\/0 /\/0\n/g' | grep burncpu | grep dso.so | head -n 1
  0x6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

Signed-off-by: Mark Santaniello <marksan@fb.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20170619163825.2012979-2-marksan@fb.com
[ Updated documentation about 'brstackoff' using text from above ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-script.txt |  4 ++-
 tools/perf/builtin-script.c              | 56 +++++++++++++++++++++++++++++---
 2 files changed, 55 insertions(+), 5 deletions(-)

diff --git a/tools/perf/Documentation/perf-script.txt b/tools/perf/Documentation/perf-script.txt
index 3eca8c0..e2468ed 100644
--- a/tools/perf/Documentation/perf-script.txt
+++ b/tools/perf/Documentation/perf-script.txt
@@ -116,7 +116,7 @@ OPTIONS
 --fields::
         Comma separated list of fields to print. Options are:
         comm, tid, pid, time, cpu, event, trace, ip, sym, dso, addr, symoff,
-        srcline, period, iregs, brstack, brstacksym, flags, bpf-output, brstackinsn,
+        srcline, period, iregs, brstack, brstacksym, flags, bpf-output, brstackinsn, brstackoff,
         callindent, insn, insnlen. Field list can be prepended with the type, trace, sw or hw,
         to indicate to which event type the field list applies.
         e.g., -F sw:comm,tid,time,ip,sym  and -F trace:time,cpu,trace
@@ -211,6 +211,8 @@ OPTIONS
 	is printed. This is the full execution path leading to the sample. This is only supported when the
 	sample was recorded with perf record -b or -j any.
 
+	The brstackoff field will print an offset into a specific dso/binary.
+
 -k::
 --vmlinux=<file>::
         vmlinux pathname
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 3c21089..db5261c 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -85,6 +85,7 @@ enum perf_output_field {
 	PERF_OUTPUT_INSN	    = 1U << 21,
 	PERF_OUTPUT_INSNLEN	    = 1U << 22,
 	PERF_OUTPUT_BRSTACKINSN	    = 1U << 23,
+	PERF_OUTPUT_BRSTACKOFF	    = 1U << 24,
 };
 
 struct output_option {
@@ -115,6 +116,7 @@ struct output_option {
 	{.str = "insn", .field = PERF_OUTPUT_INSN},
 	{.str = "insnlen", .field = PERF_OUTPUT_INSNLEN},
 	{.str = "brstackinsn", .field = PERF_OUTPUT_BRSTACKINSN},
+	{.str = "brstackoff", .field = PERF_OUTPUT_BRSTACKOFF},
 };
 
 /* default set to maintain compatibility with current format */
@@ -299,10 +301,9 @@ static int perf_evsel__check_attr(struct perf_evsel *evsel,
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR) &&
-	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM)) {
-		pr_err("Display of DSO requested but none of sample IP, sample address, "
-		       "brstack\nor brstacksym are selected. Hence, no addresses to "
-		       "convert to DSO.\n");
+	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM) && !PRINT_FIELD(BRSTACKOFF)) {
+		pr_err("Display of DSO requested but no address to convert.  Select\n"
+		       "sample IP, sample address, brstack, brstacksym, or brstackoff.\n");
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(SRCLINE) && !PRINT_FIELD(IP)) {
@@ -606,6 +607,51 @@ static void print_sample_brstacksym(struct perf_sample *sample,
 	}
 }
 
+static void print_sample_brstackoff(struct perf_sample *sample,
+				    struct thread *thread,
+				    struct perf_event_attr *attr)
+{
+	struct branch_stack *br = sample->branch_stack;
+	struct addr_location alf, alt;
+	u64 i, from, to;
+
+	if (!(br && br->nr))
+		return;
+
+	for (i = 0; i < br->nr; i++) {
+
+		memset(&alf, 0, sizeof(alf));
+		memset(&alt, 0, sizeof(alt));
+		from = br->entries[i].from;
+		to   = br->entries[i].to;
+
+		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, from, &alf);
+		if (alf.map && !alf.map->dso->adjust_symbols)
+			from = map__map_ip(alf.map, from);
+
+		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, to, &alt);
+		if (alt.map && !alt.map->dso->adjust_symbols)
+			to = map__map_ip(alt.map, to);
+
+		printf("0x%"PRIx64, from);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
+		printf("/0x%"PRIx64, to);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
+		printf("/%c/%c/%c/%d ",
+			mispred_str(br->entries + i),
+			br->entries[i].flags.in_tx ? 'X' : '-',
+			br->entries[i].flags.abort ? 'A' : '-',
+			br->entries[i].flags.cycles);
+	}
+}
 #define MAXBB 16384UL
 
 static int grab_bb(u8 *buffer, u64 start, u64 end,
@@ -1227,6 +1273,8 @@ static void process_event(struct perf_script *script,
 		print_sample_brstack(sample, thread, attr);
 	else if (PRINT_FIELD(BRSTACKSYM))
 		print_sample_brstacksym(sample, thread, attr);
+	else if (PRINT_FIELD(BRSTACKOFF))
+		print_sample_brstackoff(sample, thread, attr);
 
 	if (perf_evsel__is_bpf_output(evsel) && PRINT_FIELD(BPF_OUTPUT))
 		print_sample_bpf_output(sample);

[tip:perf/core] perf script: Support -F brstack,dso and brstacksym,dso

[tip-bot for Mark Santaniello]

Commit-ID:  55b9b50811ca459e4688543b688b7b2b85ec5ea8
Gitweb:     http://git.kernel.org/tip/55b9b50811ca459e4688543b688b7b2b85ec5ea8
Author:     Mark Santaniello <marksan@fb.com>
AuthorDate: Mon, 19 Jun 2017 09:38:24 -0700
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Mon, 19 Jun 2017 22:05:40 -0300

perf script: Support -F brstack,dso and brstacksym,dso

Perf script can report the dso for "addr" and "ip" fields.

This adds the same support for the "brstack" and "brstacksym" fields.
This can be helpful for AutoFDO: we can ignore LBR entries unless the
source and target address are both in the target module we are about to
build.

I built a small test akin to "while(1) { do_nothing(); }" where the
do_nothing function is loaded from a dso:

  $ cat burncpu.cpp
  #include <dlfcn.h>

  int main() {
    void* handle = dlopen("./dso.so", RTLD_LAZY);
    if (!handle) return -1;

    typedef void (*fp)();
    fp do_nothing = (fp) dlsym(handle, "do_nothing");

    while(1) {
      do_nothing();
    }
  }

  $ cat dso.cpp
  extern "C" void do_nothing() {}

  $ cat build.sh
  #!/bin/bash
  g++ -shared dso.cpp -o dso.so
  g++ burncpu.cpp -o burncpu -ldl

I sampled the execution with perf record -b.  Using the new perf script
functionality I can easily find cases where there was a transition from one
dso to another:

  $ perf record -a -b -- sleep 5
  [ perf record: Woken up 55 times to write data ]
  [ perf record: Captured and wrote 18.815 MB perf.data (43593 samples) ]

  $ perf script -F brstack,dso | sed 's/\/0 /\/0\n/g' | grep burncpu | grep dso.so | head -n 1
  0x7f967139b6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0

  $ perf script -F brstacksym,dso | sed 's/\/0 /\/0\n/g' | grep burncpu | grep dso.so | head -n 1
  do_nothing+0x5(/tmp/burncpu/dso.so)/main+0x44(/tmp/burncpu/exe)/P/-/-/0

Signed-off-by: Mark Santaniello <marksan@fb.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20170619163825.2012979-1-marksan@fb.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-script.c | 61 ++++++++++++++++++++++++++++++++++++---------
 1 file changed, 49 insertions(+), 12 deletions(-)

diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index afa84de..3c21089 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -298,10 +298,11 @@ static int perf_evsel__check_attr(struct perf_evsel *evsel,
 		       "selected.\n");
 		return -EINVAL;
 	}
-	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR)) {
-		pr_err("Display of DSO requested but neither sample IP nor "
-			   "sample address\nis selected. Hence, no addresses to convert "
-		       "to DSO.\n");
+	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR) &&
+	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM)) {
+		pr_err("Display of DSO requested but none of sample IP, sample address, "
+		       "brstack\nor brstacksym are selected. Hence, no addresses to "
+		       "convert to DSO.\n");
 		return -EINVAL;
 	}
 	if (PRINT_FIELD(SRCLINE) && !PRINT_FIELD(IP)) {
@@ -514,18 +515,43 @@ mispred_str(struct branch_entry *br)
 	return br->flags.predicted ? 'P' : 'M';
 }
 
-static void print_sample_brstack(struct perf_sample *sample)
+static void print_sample_brstack(struct perf_sample *sample,
+				 struct thread *thread,
+				 struct perf_event_attr *attr)
 {
 	struct branch_stack *br = sample->branch_stack;
-	u64 i;
+	struct addr_location alf, alt;
+	u64 i, from, to;
 
 	if (!(br && br->nr))
 		return;
 
 	for (i = 0; i < br->nr; i++) {
-		printf(" 0x%"PRIx64"/0x%"PRIx64"/%c/%c/%c/%d ",
-			br->entries[i].from,
-			br->entries[i].to,
+		from = br->entries[i].from;
+		to   = br->entries[i].to;
+
+		if (PRINT_FIELD(DSO)) {
+			memset(&alf, 0, sizeof(alf));
+			memset(&alt, 0, sizeof(alt));
+			thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, from, &alf);
+			thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, to, &alt);
+		}
+
+		printf("0x%"PRIx64, from);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
+
+		printf("/0x%"PRIx64, to);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
+
+		printf("/%c/%c/%c/%d ",
 			mispred_str( br->entries + i),
 			br->entries[i].flags.in_tx? 'X' : '-',
 			br->entries[i].flags.abort? 'A' : '-',
@@ -534,7 +560,8 @@ static void print_sample_brstack(struct perf_sample *sample)
 }
 
 static void print_sample_brstacksym(struct perf_sample *sample,
-				    struct thread *thread)
+				    struct thread *thread,
+				    struct perf_event_attr *attr)
 {
 	struct branch_stack *br = sample->branch_stack;
 	struct addr_location alf, alt;
@@ -559,8 +586,18 @@ static void print_sample_brstacksym(struct perf_sample *sample,
 			alt.sym = map__find_symbol(alt.map, alt.addr);
 
 		symbol__fprintf_symname_offs(alf.sym, &alf, stdout);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alf.map, stdout);
+			printf(")");
+		}
 		putchar('/');
 		symbol__fprintf_symname_offs(alt.sym, &alt, stdout);
+		if (PRINT_FIELD(DSO)) {
+			printf("(");
+			map__fprintf_dsoname(alt.map, stdout);
+			printf(")");
+		}
 		printf("/%c/%c/%c/%d ",
 			mispred_str( br->entries + i),
 			br->entries[i].flags.in_tx? 'X' : '-',
@@ -1187,9 +1224,9 @@ static void process_event(struct perf_script *script,
 		print_sample_iregs(sample, attr);
 
 	if (PRINT_FIELD(BRSTACK))
-		print_sample_brstack(sample);
+		print_sample_brstack(sample, thread, attr);
 	else if (PRINT_FIELD(BRSTACKSYM))
-		print_sample_brstacksym(sample, thread);
+		print_sample_brstacksym(sample, thread, attr);
 
 	if (perf_evsel__is_bpf_output(evsel) && PRINT_FIELD(BPF_OUTPUT))
 		print_sample_bpf_output(sample);