Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1271299 > unrolled thread

[PATCH 0/3] perf tools DWARF libunwind: Add callchain order support

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support Jiri Olsa <jolsa@kernel.org> - 2015-11-17 16:10 +0100
    [PATCH 3/3] perf test: Add callchain order setup for DWARF unwinder test Jiri Olsa <jolsa@kernel.org> - 2015-11-17 16:10 +0100
      [tip:perf/core] perf test:   Add callchain order setup for DWARF unwinder test tip-bot for Jiri Olsa <tipbot@zytor.com> - 2015-11-26 09:30 +0100
    Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support "Wangnan (F)" <wangnan0@huawei.com> - 2015-11-18 05:30 +0100
    [PATCH 4/3] perf tools: Add callchain order support for libdw DWARF  unwinder Jiri Olsa <jolsa@redhat.com> - 2015-11-19 12:30 +0100
      Re: [PATCH 4/3] perf tools: Add callchain order support for libdw  DWARF unwinder "Wangnan (F)" <wangnan0@huawei.com> - 2015-11-19 13:20 +0100
      Re: [PATCH 4/3] perf tools: Add callchain order support for libdw  DWARF unwinder "Wangnan (F)" <wangnan0@huawei.com> - 2015-11-19 13:20 +0100
        [PATCHv2 4/3] perf tools: Add callchain order support for libdw  DWARF unwinder Jiri Olsa <jolsa@redhat.com> - 2015-11-19 14:10 +0100
          [tip:perf/core] perf callchain:   Add order support for libdw DWARF unwinder tip-bot for Jiri Olsa <tipbot@zytor.com> - 2015-11-26 09:30 +0100
    Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order  support Arnaldo Carvalho de Melo <acme@kernel.org> - 2015-11-19 20:20 +0100
      Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order  support Jiri Olsa <jolsa@redhat.com> - 2015-11-20 09:40 +0100
      Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support Milian Wolff <milian.wolff@kdab.com> - 2015-11-22 20:20 +0100

#1271299 — [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support

hi,
as reported by Milian, currently for DWARF unwind (both libdw
and libunwind) we display callchain in callee order only.

Adding the support to follow callchain order setup to libunwind
DWARF unwinder, so we could get following output for report:

  $ perf record --call-graph dwarf ls
  ...
  $ perf report --no-children --stdio

    39.26%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---__strcoll_l
                    mpsort_with_tmp
                    mpsort_with_tmp
                    sort_files
                    main
                    __libc_start_main
                    _start
                    0

  $ perf report -g caller --no-children --stdio
    ...
    39.26%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---0
                    _start
                    __libc_start_main
                    main
                    sort_files
                    mpsort_with_tmp
                    mpsort_with_tmp
                    __strcoll_l

Tested on x86_64. The change is in generic code only,
so it should not affect other archs. Still it would be
nice to have some confirmation.. Wang Nan? ;-)

It'd be nice to have this for libdw unwind as well,
but it looks like it's out of reach for perf code.. Jan?

Also available in:
  git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
  perf/callchain_1

thanks,
jirka


Cc: Jan Kratochvil <jkratoch@redhat.com>
---
Jiri Olsa (3):
      perf tools: Move initial entry call into get_entries function
      perf tools: Add callchain order support for libunwind DWARF unwinder
      perf test: Add callchain order setup for DWARF unwinder test

 tools/perf/tests/dwarf-unwind.c    | 22 +++++++++++++++++++---
 tools/perf/util/unwind-libunwind.c | 60 +++++++++++++++++++++++++++++++++++++++---------------------
 2 files changed, 58 insertions(+), 24 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1271301 — [PATCH 3/3] perf test: Add callchain order setup for DWARF unwinder test

Adding callchain order setup for DWARF unwinder test. The test
now runs unwinder for both callee and caller orders.

Link: http://lkml.kernel.org/n/tip-zz3q3vrre3re8efdw3c89aef@git.kernel.org
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 tools/perf/tests/dwarf-unwind.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/tools/perf/tests/dwarf-unwind.c b/tools/perf/tests/dwarf-unwind.c
index 07221793a3ac..05d9ba387c43 100644
--- a/tools/perf/tests/dwarf-unwind.c
+++ b/tools/perf/tests/dwarf-unwind.c
@@ -51,6 +51,12 @@ static int unwind_entry(struct unwind_entry *entry, void *arg)
 		"krava_1",
 		"test__dwarf_unwind"
 	};
+	/*
+	 * The funcs[MAX_STACK] array index, based on the
+	 * callchain order setup.
+	 */
+	int idx = callchain_param.order == ORDER_CALLER ?
+		  MAX_STACK - *cnt - 1 : *cnt;
 
 	if (*cnt >= MAX_STACK) {
 		pr_debug("failed: crossed the max stack value %d\n", MAX_STACK);
@@ -63,8 +69,10 @@ static int unwind_entry(struct unwind_entry *entry, void *arg)
 		return -1;
 	}
 
-	pr_debug("got: %s 0x%" PRIx64 "\n", symbol, entry->ip);
-	return strcmp((const char *) symbol, funcs[(*cnt)++]);
+	(*cnt)++;
+	pr_debug("got: %s 0x%" PRIx64 ", expecting %s\n",
+		 symbol, entry->ip, funcs[idx]);
+	return strcmp((const char *) symbol, funcs[idx]);
 }
 
 __attribute__ ((noinline))
@@ -105,8 +113,16 @@ static int compare(void *p1, void *p2)
 	/* Any possible value should be 'thread' */
 	struct thread *thread = *(struct thread **)p1;
 
-	if (global_unwind_retval == -INT_MAX)
+	if (global_unwind_retval == -INT_MAX) {
+		/* Call unwinder twice for both callchain orders. */
+		callchain_param.order = ORDER_CALLER;
+
 		global_unwind_retval = unwind_thread(thread);
+		if (!global_unwind_retval) {
+			callchain_param.order = ORDER_CALLEE;
+			global_unwind_retval = unwind_thread(thread);
+		}
+	}
 
 	return p1 - p2;
 }
-- 
2.4.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1278075 — [tip:perf/core] perf test: Add callchain order setup for DWARF unwinder test

Commit-ID:  8dc0564d809e3903834950e2d12f6d1d2fcff708
Gitweb:     http://git.kernel.org/tip/8dc0564d809e3903834950e2d12f6d1d2fcff708
Author:     Jiri Olsa <jolsa@kernel.org>
AuthorDate: Tue, 17 Nov 2015 16:05:39 +0100
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Mon, 23 Nov 2015 18:31:01 -0300

perf test: Add callchain order setup for DWARF unwinder test

Adding callchain order setup for DWARF unwinder test. The test now runs
unwinder for both callee and caller orders.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Milian Wolff <milian.wolff@kdab.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1447772739-18471-4-git-send-email-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/tests/dwarf-unwind.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/tools/perf/tests/dwarf-unwind.c b/tools/perf/tests/dwarf-unwind.c
index 01f0b61..b2357e8 100644
--- a/tools/perf/tests/dwarf-unwind.c
+++ b/tools/perf/tests/dwarf-unwind.c
@@ -51,6 +51,12 @@ static int unwind_entry(struct unwind_entry *entry, void *arg)
 		"krava_1",
 		"test__dwarf_unwind"
 	};
+	/*
+	 * The funcs[MAX_STACK] array index, based on the
+	 * callchain order setup.
+	 */
+	int idx = callchain_param.order == ORDER_CALLER ?
+		  MAX_STACK - *cnt - 1 : *cnt;
 
 	if (*cnt >= MAX_STACK) {
 		pr_debug("failed: crossed the max stack value %d\n", MAX_STACK);
@@ -63,8 +69,10 @@ static int unwind_entry(struct unwind_entry *entry, void *arg)
 		return -1;
 	}
 
-	pr_debug("got: %s 0x%" PRIx64 "\n", symbol, entry->ip);
-	return strcmp((const char *) symbol, funcs[(*cnt)++]);
+	(*cnt)++;
+	pr_debug("got: %s 0x%" PRIx64 ", expecting %s\n",
+		 symbol, entry->ip, funcs[idx]);
+	return strcmp((const char *) symbol, funcs[idx]);
 }
 
 __attribute__ ((noinline))
@@ -105,8 +113,16 @@ static int compare(void *p1, void *p2)
 	/* Any possible value should be 'thread' */
 	struct thread *thread = *(struct thread **)p1;
 
-	if (global_unwind_retval == -INT_MAX)
+	if (global_unwind_retval == -INT_MAX) {
+		/* Call unwinder twice for both callchain orders. */
+		callchain_param.order = ORDER_CALLER;
+
 		global_unwind_retval = unwind_thread(thread);
+		if (!global_unwind_retval) {
+			callchain_param.order = ORDER_CALLEE;
+			global_unwind_retval = unwind_thread(thread);
+		}
+	}
 
 	return p1 - p2;
 }
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1271842

Hi Jiri,


On 2015/11/17 23:05, Jiri Olsa wrote:
> hi,
> as reported by Milian, currently for DWARF unwind (both libdw
> and libunwind) we display callchain in callee order only.
>
> Adding the support to follow callchain order setup to libunwind
> DWARF unwinder, so we could get following output for report:
>
>    $ perf record --call-graph dwarf ls
>    ...
>    $ perf report --no-children --stdio
>
>      39.26%  ls       libc-2.21.so      [.] __strcoll_l
>                   |
>                   ---__strcoll_l
>                      mpsort_with_tmp
>                      mpsort_with_tmp
>                      sort_files
>                      main
>                      __libc_start_main
>                      _start
>                      0
>
>    $ perf report -g caller --no-children --stdio
>      ...
>      39.26%  ls       libc-2.21.so      [.] __strcoll_l
>                   |
>                   ---0
>                      _start
>                      __libc_start_main
>                      main
>                      sort_files
>                      mpsort_with_tmp
>                      mpsort_with_tmp
>                      __strcoll_l
>
> Tested on x86_64. The change is in generic code only,
> so it should not affect other archs. Still it would be
> nice to have some confirmation.. Wang Nan? ;-)
>
> It'd be nice to have this for libdw unwind as well,
> but it looks like it's out of reach for perf code.. Jan?
>
> Also available in:
>    git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
>    perf/callchain_1


Thanks for notifying me about this. I have tested it in my environment.

It works well for me except a small behavior changing. Please see below.

Before applying these patch set:

# perf report --no-children --stdio --call-graph=callee
# Overhead  Command  Shared Object     Symbol
# ........  .......  ................  .........................
#
     96.61%  a.out    [vdso]            [.] __vdso_gettimeofday
               |
               ---__vdso_gettimeofday
                  funcc
                  funcb
                  funca
                  main
                  __libc_start_main
                  _start

      3.38%  a.out    a.out             [.] funcc
               |
               ---funcc
                  |
                   --2.70%-- funcb
                             funca
                             main
                             __libc_start_main
                             _start

      0.02%  pref_re  [kernel.vmlinux]  [k] sched_clock
               |
               ---sched_clock
                  perf_event_nmi_handler
                  nmi_handle
      ...

And caller:

# ./perf report --no-children --stdio --call-graph=caller
# Overhead  Command  Shared Object     Symbol
# ........  .......  ................  .........................
#
     96.61%  a.out    [vdso]            [.] __vdso_gettimeofday
               |
               ---__vdso_gettimeofday
                  funcc
                  funcb
                  funca
                  main
                  __libc_start_main
                  _start

      3.38%  a.out    a.out             [.] funcc
               |
               ---funcc
                  |
                   --2.70%-- funcb
                             funca
                             main
                             __libc_start_main
                             _start

      0.02%  pref_re  [kernel.vmlinux]  [k] sched_clock
               |
               ---return_from_execve
                  sys_execve
                  do_execveat_common.isra.27


The user code part of output are identical so I confirm the bug.

After applying this patchset:

# ./perf report --no-children --stdio --call-graph=callee
# Overhead  Command  Shared Object     Symbol
# ........  .......  ................  .........................
#
     96.61%  a.out    [vdso]            [.] __vdso_gettimeofday
               |
               ---__vdso_gettimeofday
                  funcc
                  funcb
                  funca
                  main
                  __libc_start_main
                  _start

      3.38%  a.out    a.out             [.] funcc
               |
               ---funcc
                  |
                  |--2.70%-- funcb
                  |          funca
                  |          main
                  |          __libc_start_main
                  |          _start
                  |
                   --0.68%-- 0
      0.02%  pref_re  [kernel.vmlinux]  [k] sched_clock
               |
               ---sched_clock
                  perf_event_nmi_handler
      ...

And caller:

# ./perf report --no-children --stdio --call-graph=caller
# Overhead  Command  Shared Object     Symbol
# ........  .......  ................  .........................
#
     96.61%  a.out    [vdso]            [.] __vdso_gettimeofday
               |
               ---_start
                  __libc_start_main
                  main
                  funca
                  funcb
                  funcc
                  __vdso_gettimeofday

      3.38%  a.out    a.out             [.] funcc
               |
               |--2.70%-- _start
               |          __libc_start_main
               |          main
               |          funca
               |          funcb
               |          funcc
               |
                --0.68%-- 0
                          funcc

      0.02%  pref_re  [kernel.vmlinux]  [k] sched_clock
               |
               ---return_from_execve
                  sys_execve
     ...

It fixes the bug. However, do you see the extra "0.68%-- 0" in the tree?

I give a message on patch 2/3, please have a look. I think this change
would be okay for me if we treat the old behavior as a bug (for example:
sum of all branches not equal to the overhead of itself). However, the
original code explicitly avoid generating '0' entry so I think we
should make it clear.

Thank you.


> thanks,
> jirka
>
>
> Cc: Jan Kratochvil <jkratoch@redhat.com>
> ---
> Jiri Olsa (3):
>        perf tools: Move initial entry call into get_entries function
>        perf tools: Add callchain order support for libunwind DWARF unwinder
>        perf test: Add callchain order setup for DWARF unwinder test
>
>   tools/perf/tests/dwarf-unwind.c    | 22 +++++++++++++++++++---
>   tools/perf/util/unwind-libunwind.c | 60 +++++++++++++++++++++++++++++++++++++++---------------------
>   2 files changed, 58 insertions(+), 24 deletions(-)


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273049 — [PATCH 4/3] perf tools: Add callchain order support for libdw DWARF unwinder

On Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa wrote:

SNIP

> Tested on x86_64. The change is in generic code only,
> so it should not affect other archs. Still it would be
> nice to have some confirmation.. Wang Nan? ;-)
> 
> It'd be nice to have this for libdw unwind as well,
> but it looks like it's out of reach for perf code.. Jan?
> 
> Also available in:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
>   perf/callchain_1

adding also libdw support.. test with 'make NO_LIBUNWIND=1'

Also available in:
  git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
  perf/callchain_3

thanks,
jirka


---
As reported by Milian, currently for DWARF unwind (both libdw
and libunwind) we display callchain in callee order only.

Adding the support to follow callchain order setup to libdw
DWARF unwinder, so we could get following output for report:

  $ perf record --call-graph dwarf ls
  ...

  $ perf report --no-children --stdio

    13.63%  ls       [kernel.vmlinux]  [k] __rb_insert_augmented
                 |
                 ---__rb_insert_augmented
                    __vma_link_rb
                    vma_link
                    do_brk
                    vm_brk
                    load_elf_binary
                    search_binary_handler
                    do_execveat_common.isra.29
                    sys_execve
                    return_from_execve

  $ perf report --stdio --no-children -g caller

    13.63%  ls       [kernel.vmlinux]  [k] __rb_insert_augmented
                 |
                 ---return_from_execve
                    sys_execve
                    do_execveat_common.isra.29
                    search_binary_handler
                    load_elf_binary
                    vm_brk
                    do_brk
                    vma_link
                    __vma_link_rb
                    __rb_insert_augmented

Reported-by: Milian Wolff <milian.wolff@kdab.com>
Link: http://lkml.kernel.org/n/tip-zk99ay9s71whl9qd9if28u8z@git.kernel.org
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 tools/perf/util/unwind-libdw.c | 53 ++++++++++++++++++++++++++++++------------
 tools/perf/util/unwind-libdw.h |  2 ++
 2 files changed, 40 insertions(+), 15 deletions(-)

diff --git a/tools/perf/util/unwind-libdw.c b/tools/perf/util/unwind-libdw.c
index 2dcfe9a7c8d0..db8142ba7cb9 100644
--- a/tools/perf/util/unwind-libdw.c
+++ b/tools/perf/util/unwind-libdw.c
@@ -11,6 +11,7 @@
 #include <linux/types.h>
 #include "event.h"
 #include "perf_regs.h"
+#include "callchain.h"
 
 static char *debuginfo_path;
 
@@ -52,25 +53,28 @@ static int report_module(u64 ip, struct unwind_info *ui)
 	return __report_module(&al, ip, ui);
 }
 
+/*
+ * Store all entries within entries array,
+ * we will process it after we finish unwind.
+ */
 static int entry(u64 ip, struct unwind_info *ui)
 
 {
-	struct unwind_entry e;
+	struct unwind_entry *e = &ui->entries[ui->idx++];
 	struct addr_location al;
 
 	if (__report_module(&al, ip, ui))
 		return -1;
 
-	e.ip  = ip;
-	e.map = al.map;
-	e.sym = al.sym;
+	e->ip  = ip;
+	e->map = al.map;
+	e->sym = al.sym;
 
 	pr_debug("unwind: %s:ip = 0x%" PRIx64 " (0x%" PRIx64 ")\n",
 		 al.sym ? al.sym->name : "''",
 		 ip,
 		 al.map ? al.map->map_ip(al.map, ip) : (u64) 0);
-
-	return ui->cb(&e, ui->arg);
+	return 0;
 }
 
 static pid_t next_thread(Dwfl *dwfl, void *arg, void **thread_argp)
@@ -168,7 +172,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 			struct perf_sample *data,
 			int max_stack)
 {
-	struct unwind_info ui = {
+	struct unwind_info *ui, ui_buf = {
 		.sample		= data,
 		.thread		= thread,
 		.machine	= thread->mg->machine,
@@ -177,35 +181,54 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 		.max_stack	= max_stack,
 	};
 	Dwarf_Word ip;
-	int err = -EINVAL;
+	int err = -EINVAL, i;
 
 	if (!data->user_regs.regs)
 		return -EINVAL;
 
-	ui.dwfl = dwfl_begin(&offline_callbacks);
-	if (!ui.dwfl)
+	ui = zalloc(sizeof(ui_buf) + sizeof(ui_buf.entries[0]) * max_stack);
+	if (!ui)
+		return -ENOMEM;
+
+	*ui = ui_buf;
+
+	ui->dwfl = dwfl_begin(&offline_callbacks);
+	if (!ui->dwfl)
 		goto out;
 
 	err = perf_reg_value(&ip, &data->user_regs, PERF_REG_IP);
 	if (err)
 		goto out;
 
-	err = report_module(ip, &ui);
+	err = report_module(ip, ui);
 	if (err)
 		goto out;
 
-	if (!dwfl_attach_state(ui.dwfl, EM_NONE, thread->tid, &callbacks, &ui))
+	if (!dwfl_attach_state(ui->dwfl, EM_NONE, thread->tid, &callbacks, ui))
 		goto out;
 
-	err = dwfl_getthread_frames(ui.dwfl, thread->tid, frame_callback, &ui);
+	err = dwfl_getthread_frames(ui->dwfl, thread->tid, frame_callback, ui);
 
-	if (err && !ui.max_stack)
+	if (err && !ui->max_stack)
 		err = 0;
 
+	/*
+	 * Display what we got based on the order setup.
+	 */
+	for (i = 0; i < ui->idx && !err; i++) {
+		int j = i;
+
+		if (callchain_param.order == ORDER_CALLER)
+			j = ui->idx - i - 1;
+
+		err = ui->entries[j].ip ? ui->cb(&ui->entries[j], ui->arg) : 0;
+	}
+
  out:
 	if (err)
 		pr_debug("unwind: failed with '%s'\n", dwfl_errmsg(-1));
 
-	dwfl_end(ui.dwfl);
+	dwfl_end(ui->dwfl);
+	free(ui);
 	return 0;
 }
diff --git a/tools/perf/util/unwind-libdw.h b/tools/perf/util/unwind-libdw.h
index 417a1426f3ad..58328669ed16 100644
--- a/tools/perf/util/unwind-libdw.h
+++ b/tools/perf/util/unwind-libdw.h
@@ -16,6 +16,8 @@ struct unwind_info {
 	unwind_entry_cb_t	cb;
 	void			*arg;
 	int			max_stack;
+	int			idx;
+	struct unwind_entry	entries[];
 };
 
 #endif /* __PERF_UNWIND_LIBDW_H */
-- 
2.4.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273083 — Re: [PATCH 4/3] perf tools: Add callchain order support for libdw DWARF unwinder

On 2015/11/19 20:10, Wangnan (F) wrote:
>
>
> On 2015/11/19 19:22, Jiri Olsa wrote:
>> On Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa wrote:
>>

[SNIP]

> In addition, just for your information, I find libdw can unwind call
> chain I described yesterday in [1].

Please forget it. The problem is still their even with libdw. I forgot
to add -fomit-frame-pointer.

Thank you.
>
> Thank you.
>
> [1] https://lkml.kernel.org/g/564C26C4.2040603@huawei.com


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273086 — Re: [PATCH 4/3] perf tools: Add callchain order support for libdw DWARF unwinder

On 2015/11/19 19:22, Jiri Olsa wrote:
> On Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa wrote:
>
> SNIP
>
>> Tested on x86_64. The change is in generic code only,
>> so it should not affect other archs. Still it would be
>> nice to have some confirmation.. Wang Nan? ;-)
>>
>> It'd be nice to have this for libdw unwind as well,
>> but it looks like it's out of reach for perf code.. Jan?
>>
>> Also available in:
>>    git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
>>    perf/callchain_1
> adding also libdw support.. test with 'make NO_LIBUNWIND=1'
>
> Also available in:
>    git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
>    perf/callchain_3
>
> thanks,
> jirka
>
>
> ---
> As reported by Milian, currently for DWARF unwind (both libdw
> and libunwind) we display callchain in callee order only.
>
> Adding the support to follow callchain order setup to libdw
> DWARF unwinder, so we could get following output for report:
>
>    $ perf record --call-graph dwarf ls
>    ...
>
>    $ perf report --no-children --stdio
>
>      13.63%  ls       [kernel.vmlinux]  [k] __rb_insert_augmented
>                   |
>                   ---__rb_insert_augmented
>                      __vma_link_rb
>                      vma_link
>                      do_brk
>                      vm_brk
>                      load_elf_binary
>                      search_binary_handler
>                      do_execveat_common.isra.29
>                      sys_execve
>                      return_from_execve
>
>    $ perf report --stdio --no-children -g caller
>
>      13.63%  ls       [kernel.vmlinux]  [k] __rb_insert_augmented
>                   |
>                   ---return_from_execve
>                      sys_execve
>                      do_execveat_common.isra.29
>                      search_binary_handler
>                      load_elf_binary
>                      vm_brk
>                      do_brk
>                      vma_link
>                      __vma_link_rb
>                      __rb_insert_augmented
>
> Reported-by: Milian Wolff <milian.wolff@kdab.com>
> Link: http://lkml.kernel.org/n/tip-zk99ay9s71whl9qd9if28u8z@git.kernel.org
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>

Tested-by: Wang Nan <wangnan0@huawei.com>

And I suggest you put a user space call graph to the commit message
to show the result of this patch. What you show now is not related
to this patch, right?

In addition, just for your information, I find libdw can unwind call
chain I described yesterday in [1].

Thank you.

[1] https://lkml.kernel.org/g/564C26C4.2040603@huawei.com

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273117 — [PATCHv2 4/3] perf tools: Add callchain order support for libdw DWARF unwinder

On Thu, Nov 19, 2015 at 08:10:45PM +0800, Wangnan (F) wrote:

SNIP

> >                     load_elf_binary
> >                     vm_brk
> >                     do_brk
> >                     vma_link
> >                     __vma_link_rb
> >                     __rb_insert_augmented
> >
> >Reported-by: Milian Wolff <milian.wolff@kdab.com>
> >Link: http://lkml.kernel.org/n/tip-zk99ay9s71whl9qd9if28u8z@git.kernel.org
> >Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> 
> Tested-by: Wang Nan <wangnan0@huawei.com>
> 
> And I suggest you put a user space call graph to the commit message
> to show the result of this patch. What you show now is not related
> to this patch, right?

oopsie.. sry ;-) attached v2

thanks,
jirka


---
As reported by Milian, currently for DWARF unwind (both libdw
and libunwind) we display callchain in callee order only.

Adding the support to follow callchain order setup to libdw
DWARF unwinder, so we could get following output for report:

  $ perf record --call-graph dwarf ls
  ...

  $ perf report --no-children --stdio

    21.12%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---__strcoll_l
                    mpsort_with_tmp
                    mpsort_with_tmp
                    mpsort_with_tmp
                    sort_files
                    main
                    __libc_start_main
                    _start

  $ perf report --stdio --no-children -g caller

    21.12%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---_start
                    __libc_start_main
                    main
                    sort_files
                    mpsort_with_tmp
                    mpsort_with_tmp
                    mpsort_with_tmp
                    __strcoll_l

Tested-by: Wang Nan <wangnan0@huawei.com>
Reported-by: Milian Wolff <milian.wolff@kdab.com>
Link: http://lkml.kernel.org/n/tip-zk99ay9s71whl9qd9if28u8z@git.kernel.org
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 tools/perf/util/unwind-libdw.c | 53 ++++++++++++++++++++++++++++++------------
 tools/perf/util/unwind-libdw.h |  2 ++
 2 files changed, 40 insertions(+), 15 deletions(-)

diff --git a/tools/perf/util/unwind-libdw.c b/tools/perf/util/unwind-libdw.c
index 2dcfe9a7c8d0..db8142ba7cb9 100644
--- a/tools/perf/util/unwind-libdw.c
+++ b/tools/perf/util/unwind-libdw.c
@@ -11,6 +11,7 @@
 #include <linux/types.h>
 #include "event.h"
 #include "perf_regs.h"
+#include "callchain.h"
 
 static char *debuginfo_path;
 
@@ -52,25 +53,28 @@ static int report_module(u64 ip, struct unwind_info *ui)
 	return __report_module(&al, ip, ui);
 }
 
+/*
+ * Store all entries within entries array,
+ * we will process it after we finish unwind.
+ */
 static int entry(u64 ip, struct unwind_info *ui)
 
 {
-	struct unwind_entry e;
+	struct unwind_entry *e = &ui->entries[ui->idx++];
 	struct addr_location al;
 
 	if (__report_module(&al, ip, ui))
 		return -1;
 
-	e.ip  = ip;
-	e.map = al.map;
-	e.sym = al.sym;
+	e->ip  = ip;
+	e->map = al.map;
+	e->sym = al.sym;
 
 	pr_debug("unwind: %s:ip = 0x%" PRIx64 " (0x%" PRIx64 ")\n",
 		 al.sym ? al.sym->name : "''",
 		 ip,
 		 al.map ? al.map->map_ip(al.map, ip) : (u64) 0);
-
-	return ui->cb(&e, ui->arg);
+	return 0;
 }
 
 static pid_t next_thread(Dwfl *dwfl, void *arg, void **thread_argp)
@@ -168,7 +172,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 			struct perf_sample *data,
 			int max_stack)
 {
-	struct unwind_info ui = {
+	struct unwind_info *ui, ui_buf = {
 		.sample		= data,
 		.thread		= thread,
 		.machine	= thread->mg->machine,
@@ -177,35 +181,54 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 		.max_stack	= max_stack,
 	};
 	Dwarf_Word ip;
-	int err = -EINVAL;
+	int err = -EINVAL, i;
 
 	if (!data->user_regs.regs)
 		return -EINVAL;
 
-	ui.dwfl = dwfl_begin(&offline_callbacks);
-	if (!ui.dwfl)
+	ui = zalloc(sizeof(ui_buf) + sizeof(ui_buf.entries[0]) * max_stack);
+	if (!ui)
+		return -ENOMEM;
+
+	*ui = ui_buf;
+
+	ui->dwfl = dwfl_begin(&offline_callbacks);
+	if (!ui->dwfl)
 		goto out;
 
 	err = perf_reg_value(&ip, &data->user_regs, PERF_REG_IP);
 	if (err)
 		goto out;
 
-	err = report_module(ip, &ui);
+	err = report_module(ip, ui);
 	if (err)
 		goto out;
 
-	if (!dwfl_attach_state(ui.dwfl, EM_NONE, thread->tid, &callbacks, &ui))
+	if (!dwfl_attach_state(ui->dwfl, EM_NONE, thread->tid, &callbacks, ui))
 		goto out;
 
-	err = dwfl_getthread_frames(ui.dwfl, thread->tid, frame_callback, &ui);
+	err = dwfl_getthread_frames(ui->dwfl, thread->tid, frame_callback, ui);
 
-	if (err && !ui.max_stack)
+	if (err && !ui->max_stack)
 		err = 0;
 
+	/*
+	 * Display what we got based on the order setup.
+	 */
+	for (i = 0; i < ui->idx && !err; i++) {
+		int j = i;
+
+		if (callchain_param.order == ORDER_CALLER)
+			j = ui->idx - i - 1;
+
+		err = ui->entries[j].ip ? ui->cb(&ui->entries[j], ui->arg) : 0;
+	}
+
  out:
 	if (err)
 		pr_debug("unwind: failed with '%s'\n", dwfl_errmsg(-1));
 
-	dwfl_end(ui.dwfl);
+	dwfl_end(ui->dwfl);
+	free(ui);
 	return 0;
 }
diff --git a/tools/perf/util/unwind-libdw.h b/tools/perf/util/unwind-libdw.h
index 417a1426f3ad..58328669ed16 100644
--- a/tools/perf/util/unwind-libdw.h
+++ b/tools/perf/util/unwind-libdw.h
@@ -16,6 +16,8 @@ struct unwind_info {
 	unwind_entry_cb_t	cb;
 	void			*arg;
 	int			max_stack;
+	int			idx;
+	struct unwind_entry	entries[];
 };
 
 #endif /* __PERF_UNWIND_LIBDW_H */
-- 
2.4.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1278073 — [tip:perf/core] perf callchain: Add order support for libdw DWARF unwinder

Commit-ID:  8bd508b001629a5d836987d9a0702a6bfc4fc705
Gitweb:     http://git.kernel.org/tip/8bd508b001629a5d836987d9a0702a6bfc4fc705
Author:     Jiri Olsa <jolsa@redhat.com>
AuthorDate: Thu, 19 Nov 2015 14:01:19 +0100
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Mon, 23 Nov 2015 18:31:13 -0300

perf callchain: Add order support for libdw DWARF unwinder

As reported by Milian, currently for DWARF unwind (both libdw and
libunwind) we display callchain in callee order only.

Adding the support to follow callchain order setup to libdw DWARF
unwinder, so we could get following output for report:

  $ perf record --call-graph dwarf ls
  ...

  $ perf report --no-children --stdio

    21.12%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---__strcoll_l
                    mpsort_with_tmp
                    mpsort_with_tmp
                    mpsort_with_tmp
                    sort_files
                    main
                    __libc_start_main
                    _start

  $ perf report --stdio --no-children -g caller

    21.12%  ls       libc-2.21.so      [.] __strcoll_l
                 |
                 ---_start
                    __libc_start_main
                    main
                    sort_files
                    mpsort_with_tmp
                    mpsort_with_tmp
                    mpsort_with_tmp
                    __strcoll_l

Reported-and-Tested-by: Milian Wolff <milian.wolff@kdab.com>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Wang Nan <wangnan0@huawei.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jan Kratochvil <jkratoch@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20151119130119.GA26617@krava.brq.redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/unwind-libdw.c | 53 ++++++++++++++++++++++++++++++------------
 tools/perf/util/unwind-libdw.h |  2 ++
 2 files changed, 40 insertions(+), 15 deletions(-)

diff --git a/tools/perf/util/unwind-libdw.c b/tools/perf/util/unwind-libdw.c
index 2dcfe9a..db8142b 100644
--- a/tools/perf/util/unwind-libdw.c
+++ b/tools/perf/util/unwind-libdw.c
@@ -11,6 +11,7 @@
 #include <linux/types.h>
 #include "event.h"
 #include "perf_regs.h"
+#include "callchain.h"
 
 static char *debuginfo_path;
 
@@ -52,25 +53,28 @@ static int report_module(u64 ip, struct unwind_info *ui)
 	return __report_module(&al, ip, ui);
 }
 
+/*
+ * Store all entries within entries array,
+ * we will process it after we finish unwind.
+ */
 static int entry(u64 ip, struct unwind_info *ui)
 
 {
-	struct unwind_entry e;
+	struct unwind_entry *e = &ui->entries[ui->idx++];
 	struct addr_location al;
 
 	if (__report_module(&al, ip, ui))
 		return -1;
 
-	e.ip  = ip;
-	e.map = al.map;
-	e.sym = al.sym;
+	e->ip  = ip;
+	e->map = al.map;
+	e->sym = al.sym;
 
 	pr_debug("unwind: %s:ip = 0x%" PRIx64 " (0x%" PRIx64 ")\n",
 		 al.sym ? al.sym->name : "''",
 		 ip,
 		 al.map ? al.map->map_ip(al.map, ip) : (u64) 0);
-
-	return ui->cb(&e, ui->arg);
+	return 0;
 }
 
 static pid_t next_thread(Dwfl *dwfl, void *arg, void **thread_argp)
@@ -168,7 +172,7 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 			struct perf_sample *data,
 			int max_stack)
 {
-	struct unwind_info ui = {
+	struct unwind_info *ui, ui_buf = {
 		.sample		= data,
 		.thread		= thread,
 		.machine	= thread->mg->machine,
@@ -177,35 +181,54 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 		.max_stack	= max_stack,
 	};
 	Dwarf_Word ip;
-	int err = -EINVAL;
+	int err = -EINVAL, i;
 
 	if (!data->user_regs.regs)
 		return -EINVAL;
 
-	ui.dwfl = dwfl_begin(&offline_callbacks);
-	if (!ui.dwfl)
+	ui = zalloc(sizeof(ui_buf) + sizeof(ui_buf.entries[0]) * max_stack);
+	if (!ui)
+		return -ENOMEM;
+
+	*ui = ui_buf;
+
+	ui->dwfl = dwfl_begin(&offline_callbacks);
+	if (!ui->dwfl)
 		goto out;
 
 	err = perf_reg_value(&ip, &data->user_regs, PERF_REG_IP);
 	if (err)
 		goto out;
 
-	err = report_module(ip, &ui);
+	err = report_module(ip, ui);
 	if (err)
 		goto out;
 
-	if (!dwfl_attach_state(ui.dwfl, EM_NONE, thread->tid, &callbacks, &ui))
+	if (!dwfl_attach_state(ui->dwfl, EM_NONE, thread->tid, &callbacks, ui))
 		goto out;
 
-	err = dwfl_getthread_frames(ui.dwfl, thread->tid, frame_callback, &ui);
+	err = dwfl_getthread_frames(ui->dwfl, thread->tid, frame_callback, ui);
 
-	if (err && !ui.max_stack)
+	if (err && !ui->max_stack)
 		err = 0;
 
+	/*
+	 * Display what we got based on the order setup.
+	 */
+	for (i = 0; i < ui->idx && !err; i++) {
+		int j = i;
+
+		if (callchain_param.order == ORDER_CALLER)
+			j = ui->idx - i - 1;
+
+		err = ui->entries[j].ip ? ui->cb(&ui->entries[j], ui->arg) : 0;
+	}
+
  out:
 	if (err)
 		pr_debug("unwind: failed with '%s'\n", dwfl_errmsg(-1));
 
-	dwfl_end(ui.dwfl);
+	dwfl_end(ui->dwfl);
+	free(ui);
 	return 0;
 }
diff --git a/tools/perf/util/unwind-libdw.h b/tools/perf/util/unwind-libdw.h
index 417a142..5832866 100644
--- a/tools/perf/util/unwind-libdw.h
+++ b/tools/perf/util/unwind-libdw.h
@@ -16,6 +16,8 @@ struct unwind_info {
 	unwind_entry_cb_t	cb;
 	void			*arg;
 	int			max_stack;
+	int			idx;
+	struct unwind_entry	entries[];
 };
 
 #endif /* __PERF_UNWIND_LIBDW_H */
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273422 — Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support

Em Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa escreveu:
> hi,
> as reported by Milian, currently for DWARF unwind (both libdw
> and libunwind) we display callchain in callee order only.

Applied the latest versions, please check my perf/core branch to check
that all is there as you expect.

- Arnaldo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1273834 — Re: [PATCH 0/3] perf tools DWARF libunwind: Add callchain order support

On Thu, Nov 19, 2015 at 04:10:37PM -0300, Arnaldo Carvalho de Melo wrote:
> Em Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa escreveu:
> > hi,
> > as reported by Milian, currently for DWARF unwind (both libdw
> > and libunwind) we display callchain in callee order only.
> 
> Applied the latest versions, please check my perf/core branch to check
> that all is there as you expect.

looks ok, thanks

jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1274925

[Multipart message — attachments visible in raw view] — view raw

On Donnerstag, 19. November 2015 16:10:37 CET Arnaldo Carvalho de Melo wrote:
> Em Tue, Nov 17, 2015 at 04:05:36PM +0100, Jiri Olsa escreveu:
> > hi,
> > as reported by Milian, currently for DWARF unwind (both libdw
> > and libunwind) we display callchain in callee order only.
> 
> Applied the latest versions, please check my perf/core branch to check
> that all is there as you expect.

Sorry for the late reply. I've just tested the current code in perf/core - 
works like a charm! Many thanks to Jiri for improving my initial take on this 
problem and also getting it fixed for libdw. Much appreciated!

Cheers

-- 
Milian Wolff | milian.wolff@kdab.com | Software Engineer
KDAB (Deutschland) GmbH&Co KG, a KDAB Group company
Tel: +49-30-521325470
KDAB - The Qt Experts

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web