From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: Clark Williams <williams@redhat.com>,
linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
Sandipan Das <sandipan@linux.ibm.com>,
Jiri Olsa <jolsa@redhat.com>,
Maynard Johnson <maynard@us.ibm.com>,
"Naveen N . Rao" <naveen.n.rao@linux.vnet.ibm.com>,
Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>,
Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>,
Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 11/27] perf powerpc: Fix callchain ip filtering
Date: Wed, 25 Jul 2018 14:59:45 -0300 [thread overview]
Message-ID: <20180725180001.15108-12-acme@kernel.org> (raw)
In-Reply-To: <20180725180001.15108-1-acme@kernel.org>
From: Sandipan Das <sandipan@linux.ibm.com>
For powerpc64, redundant entries in the callchain are filtered out by
determining the state of the return address and the stack frame using
DWARF debug information.
For making these filtering decisions we must analyze the debug
information for the location corresponding to the program counter value,
i.e. the first entry in the callchain, and not the LR value; otherwise,
perf may filter out either the second or the third entry in the
callchain incorrectly.
This can be observed on a powerpc64le system running Fedora 27 as shown
below.
Case 1 - Attaching a probe at inet_pton+0x8 (binary offset 0x15af28).
Return address is still in LR and a new stack frame is not yet
allocated. The LR value, i.e. the second entry, should not be
filtered out.
# objdump -d /usr/lib64/libc-2.26.so | less
...
000000000010eb10 <gaih_inet.constprop.7>:
...
10fa48: 78 bb e4 7e mr r4,r23
10fa4c: 0a 00 60 38 li r3,10
10fa50: d9 b4 04 48 bl 15af28 <inet_pton+0x8>
10fa54: 00 00 00 60 nop
10fa58: ac f4 ff 4b b 10ef04 <gaih_inet.constprop.7+0x3f4>
...
0000000000110450 <getaddrinfo>:
...
1105a8: 54 00 ff 38 addi r7,r31,84
1105ac: 58 00 df 38 addi r6,r31,88
1105b0: 69 e5 ff 4b bl 10eb18 <gaih_inet.constprop.7+0x8>
1105b4: 78 1b 71 7c mr r17,r3
1105b8: 50 01 7f e8 ld r3,336(r31)
...
000000000015af20 <inet_pton>:
15af20: 0b 00 4c 3c addis r2,r12,11
15af24: e0 c1 42 38 addi r2,r2,-15904
15af28: a6 02 08 7c mflr r0
15af2c: f0 ff c1 fb std r30,-16(r1)
15af30: f8 ff e1 fb std r31,-8(r1)
...
# perf probe -x /usr/lib64/libc-2.26.so -a inet_pton+0x8
# perf record -e probe_libc:inet_pton -g ping -6 -c 1 ::1
# perf script
Before:
ping 4507 [002] 514985.546540: probe_libc:inet_pton: (7fffa7dbaf28)
7fffa7dbaf28 __GI___inet_pton+0x8 (/usr/lib64/libc-2.26.so)
7fffa7d705b4 getaddrinfo+0x164 (/usr/lib64/libc-2.26.so)
13fb52d70 _init+0xbfc (/usr/bin/ping)
7fffa7c836a0 generic_start_main.isra.0+0x140 (/usr/lib64/libc-2.26.so)
7fffa7c83898 __libc_start_main+0xb8 (/usr/lib64/libc-2.26.so)
0 [unknown] ([unknown])
After:
ping 4507 [002] 514985.546540: probe_libc:inet_pton: (7fffa7dbaf28)
7fffa7dbaf28 __GI___inet_pton+0x8 (/usr/lib64/libc-2.26.so)
7fffa7d6fa54 gaih_inet.constprop.7+0xf44 (/usr/lib64/libc-2.26.so)
7fffa7d705b4 getaddrinfo+0x164 (/usr/lib64/libc-2.26.so)
13fb52d70 _init+0xbfc (/usr/bin/ping)
7fffa7c836a0 generic_start_main.isra.0+0x140 (/usr/lib64/libc-2.26.so)
7fffa7c83898 __libc_start_main+0xb8 (/usr/lib64/libc-2.26.so)
0 [unknown] ([unknown])
Case 2 - Attaching a probe at _int_malloc+0x180 (binary offset 0x9cf10).
Return address in still in LR and a new stack frame has already
been allocated but not used. The caller's caller, i.e. the third
entry, is invalid and should be filtered out and not the second
one.
# objdump -d /usr/lib64/libc-2.26.so | less
...
000000000009cd90 <_int_malloc>:
9cd90: 17 00 4c 3c addis r2,r12,23
9cd94: 70 a3 42 38 addi r2,r2,-23696
9cd98: 26 00 80 7d mfcr r12
9cd9c: f8 ff e1 fb std r31,-8(r1)
9cda0: 17 00 e4 3b addi r31,r4,23
9cda4: d8 ff 61 fb std r27,-40(r1)
9cda8: 78 23 9b 7c mr r27,r4
9cdac: 1f 00 bf 2b cmpldi cr7,r31,31
9cdb0: f0 ff c1 fb std r30,-16(r1)
9cdb4: b0 ff c1 fa std r22,-80(r1)
9cdb8: 78 1b 7e 7c mr r30,r3
9cdbc: 08 00 81 91 stw r12,8(r1)
9cdc0: 11 ff 21 f8 stdu r1,-240(r1)
9cdc4: 4c 01 9d 41 bgt cr7,9cf10 <_int_malloc+0x180>
9cdc8: 20 00 a4 2b cmpldi cr7,r4,32
...
9cf08: 00 00 00 60 nop
9cf0c: 00 00 42 60 ori r2,r2,0
9cf10: e4 06 ff 7b rldicr r31,r31,0,59
9cf14: 40 f8 a4 7f cmpld cr7,r4,r31
9cf18: 68 05 9d 41 bgt cr7,9d480 <_int_malloc+0x6f0>
...
000000000009e3c0 <tcache_init.part.4>:
...
9e420: 40 02 80 38 li r4,576
9e424: 78 fb e3 7f mr r3,r31
9e428: 71 e9 ff 4b bl 9cd98 <_int_malloc+0x8>
9e42c: 00 00 a3 2f cmpdi cr7,r3,0
9e430: 78 1b 7e 7c mr r30,r3
...
000000000009f7a0 <__libc_malloc>:
...
9f8f8: 00 00 89 2f cmpwi cr7,r9,0
9f8fc: 1c ff 9e 40 bne cr7,9f818 <__libc_malloc+0x78>
9f900: c9 ea ff 4b bl 9e3c8 <tcache_init.part.4+0x8>
9f904: 00 00 00 60 nop
9f908: e8 90 22 e9 ld r9,-28440(r2)
...
# perf probe -x /usr/lib64/libc-2.26.so -a _int_malloc+0x180
# perf record -e probe_libc:_int_malloc -g ./test-malloc
# perf script
Before:
test-malloc 6554 [009] 515975.797403: probe_libc:_int_malloc: (7fffa6e6cf10)
7fffa6e6cf10 _int_malloc+0x180 (/usr/lib64/libc-2.26.so)
7fffa6dd0000 [unknown] (/usr/lib64/libc-2.26.so)
7fffa6e6f904 malloc+0x164 (/usr/lib64/libc-2.26.so)
7fffa6e6f9fc malloc+0x25c (/usr/lib64/libc-2.26.so)
100006b4 main+0x38 (/home/testuser/test-malloc)
7fffa6df36a0 generic_start_main.isra.0+0x140 (/usr/lib64/libc-2.26.so)
7fffa6df3898 __libc_start_main+0xb8 (/usr/lib64/libc-2.26.so)
0 [unknown] ([unknown])
After:
test-malloc 6554 [009] 515975.797403: probe_libc:_int_malloc: (7fffa6e6cf10)
7fffa6e6cf10 _int_malloc+0x180 (/usr/lib64/libc-2.26.so)
7fffa6e6e42c tcache_init.part.4+0x6c (/usr/lib64/libc-2.26.so)
7fffa6e6f904 malloc+0x164 (/usr/lib64/libc-2.26.so)
7fffa6e6f9fc malloc+0x25c (/usr/lib64/libc-2.26.so)
100006b4 main+0x38 (/home/sandipan/test-malloc)
7fffa6df36a0 generic_start_main.isra.0+0x140 (/usr/lib64/libc-2.26.so)
7fffa6df3898 __libc_start_main+0xb8 (/usr/lib64/libc-2.26.so)
0 [unknown] ([unknown])
Signed-off-by: Sandipan Das <sandipan@linux.ibm.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Maynard Johnson <maynard@us.ibm.com>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Fixes: a60335ba3298 ("perf tools powerpc: Adjust callchain based on DWARF debug info")
Link: http://lkml.kernel.org/r/24bb726d91ed173aebc972ec3f41a2ef2249434e.1530724939.git.sandipan@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/arch/powerpc/util/skip-callchain-idx.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/perf/arch/powerpc/util/skip-callchain-idx.c b/tools/perf/arch/powerpc/util/skip-callchain-idx.c
index ef5d59a5742e..264fc0158100 100644
--- a/tools/perf/arch/powerpc/util/skip-callchain-idx.c
+++ b/tools/perf/arch/powerpc/util/skip-callchain-idx.c
@@ -246,7 +246,7 @@ int arch_skip_callchain_idx(struct thread *thread, struct ip_callchain *chain)
if (!chain || chain->nr < 3)
return skip_slot;
- ip = chain->ips[2];
+ ip = chain->ips[1];
thread__find_symbol(thread, PERF_RECORD_MISC_USER, ip, &al);
--
2.14.4
next prev parent reply other threads:[~2018-07-25 17:59 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-25 17:59 [GIT PULL 00/27] perf/core improvements and fixes Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 01/27] perf tests: Check that complex event name is parsed correctly Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 02/27] perf hists: Clarify callchain disabling when available Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 03/27] perf cs-etm: Introduce invalid address macro Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 04/27] perf cs-etm: Bail out immediately for instruction sample failure Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 05/27] Revert "perf list: Add s390 support for detailed/verbose PMU event description" Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 06/27] perf list: Add s390 support for detailed PMU event description Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 07/27] perf json: Add s390 transaction counter definition Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 08/27] perf stat: Add transaction flag (-T) support for s390 Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 09/27] perf kvm: Fix subcommands on s390 Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 10/27] perf list: Add missing documentation for --desc and --debug options Arnaldo Carvalho de Melo
2018-07-25 17:59 ` Arnaldo Carvalho de Melo [this message]
2018-07-25 17:59 ` [PATCH 12/27] perf powerpc: Fix callchain ip filtering when return address is in a register Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 13/27] perf tests: Fix record+probe_libc_inet_pton.sh for powerpc64 Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 14/27] perf tests: Fix record+probe_libc_inet_pton.sh to ensure cleanups Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 15/27] perf tests: Fix record+probe_libc_inet_pton.sh when event exists Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 16/27] tools include: Grab copies of arm64 dependent unistd.h files Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 17/27] perf arm64: Generate system call table from asm/unistd.h Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 18/27] perf trace arm64: Use generated syscall table Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 19/27] perf script: Show correct offsets for DWARF-based unwinding Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 20/27] perf tools: Synthesize GROUP_DESC feature in pipe mode Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 21/27] perf machine: Add threads__get_last_match function Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 22/27] perf machine: Add threads__set_last_match function Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 23/27] perf machine: Use last_match threads cache only in single thread mode Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 24/27] perf tools: Fix struct comm_str removal crash Arnaldo Carvalho de Melo
2018-07-25 17:59 ` [PATCH 25/27] perf tools: Use perf_evsel__match instead of open coded equivalent Arnaldo Carvalho de Melo
2018-07-25 18:00 ` [PATCH 26/27] perf stat: Get rid of extra clock display function Arnaldo Carvalho de Melo
2018-07-25 18:00 ` [PATCH 27/27] perf test: Fix subtest number when showing results Arnaldo Carvalho de Melo
2018-07-25 20:34 ` [GIT PULL 00/27] perf/core improvements and fixes Ingo Molnar
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180725180001.15108-12-acme@kernel.org \
--to=acme@kernel.org \
--cc=acme@redhat.com \
--cc=jolsa@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=maynard@us.ibm.com \
--cc=mingo@kernel.org \
--cc=naveen.n.rao@linux.vnet.ibm.com \
--cc=ravi.bangoria@linux.vnet.ibm.com \
--cc=sandipan@linux.ibm.com \
--cc=sukadev@linux.vnet.ibm.com \
--cc=williams@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).