Skip to contentSkip to conversation

perf fixes

GIT PULL Ingo Molnar 4 participants 10 messages Oct 26, 2013 – Oct 29, 2013 perf events

lore
Share & tools
lore
#0 lore raw

Linus,

Please pull the latest perf-urgent-for-linus git tree from:

git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-urgent-for-linus

# HEAD: e4f8eaad70ea186b8da290c99239dce721a34f88 Merge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent

The tree contains three fixes:

- Two tooling fixes

- Reversal of the new 'MMAP2' extended mmap record ABI, introduced in this merge window. (Patches were proposed to fix it but it was all a bit late and we felt it's safer to just delay the ABI one more kernel release and do it right.)

Thanks,

Ingo

------------------>
Arnaldo Carvalho de Melo (1):
perf scripting perl: Fix build error on Fedora 12

Masami Hiramatsu (1):
perf probe: Fix to initialize fname always before use it

Stephane Eranian (1):
perf: Disable PERF_RECORD_MMAP2 support

 kernel/events/core.c                               |  4 +++
 tools/perf/util/event.c                            | 30 ++++++++++------------
 tools/perf/util/evsel.c                            |  1 -
 tools/perf/util/probe-finder.c                     |  2 +-
 .../perf/util/scripting-engines/trace-event-perl.c |  2 +-
 5 files changed, 19 insertions(+), 20 deletions(-)
kernel/events/core.c +4 -0
index d49a9d2..953c143 100644
@@ -6767,6 +6767,10 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr,
 	if (ret)
 		return -EFAULT;
 
+	/* disabled for now */
+	if (attr->mmap2)
+		return -EINVAL;
+
 	if (attr->__reserved_1)
 		return -EINVAL;
 
tools/perf/util/event.c +13 -17
index 9b393e7..63df031 100644
@@ -187,7 +187,7 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		return -1;
 	}
 
-	event->header.type = PERF_RECORD_MMAP2;
+	event->header.type = PERF_RECORD_MMAP;
 	/*
 	 * Just like the kernel, see __perf_event_mmap in kernel/perf_event.c
 	 */
@@ -198,7 +198,6 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		char prot[5];
 		char execname[PATH_MAX];
 		char anonstr[] = "//anon";
-		unsigned int ino;
 		size_t size;
 		ssize_t n;
 
@@ -209,13 +208,10 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		strcpy(execname, "");
 
 		/* 00400000-0040c000 r-xp 00000000 fd:01 41038  /bin/cat */
-		n = sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %x:%x %u %s\n",
-		       &event->mmap2.start, &event->mmap2.len, prot,
-		       &event->mmap2.pgoff, &event->mmap2.maj,
-		       &event->mmap2.min,
-		       &ino, execname);
-
-		event->mmap2.ino = (u64)ino;
+		n = sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %*x:%*x %*u %s\n",
+		       &event->mmap.start, &event->mmap.len, prot,
+		       &event->mmap.pgoff,
+		       execname);
 
 		if (n != 8)
 			continue;
@@ -227,15 +223,15 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 			strcpy(execname, anonstr);
 
 		size = strlen(execname) + 1;
-		memcpy(event->mmap2.filename, execname, size);
+		memcpy(event->mmap.filename, execname, size);
 		size = PERF_ALIGN(size, sizeof(u64));
-		event->mmap2.len -= event->mmap.start;
-		event->mmap2.header.size = (sizeof(event->mmap2) -
-					(sizeof(event->mmap2.filename) - size));
-		memset(event->mmap2.filename + size, 0, machine->id_hdr_size);
-		event->mmap2.header.size += machine->id_hdr_size;
-		event->mmap2.pid = tgid;
-		event->mmap2.tid = pid;
+		event->mmap.len -= event->mmap.start;
+		event->mmap.header.size = (sizeof(event->mmap) -
+					(sizeof(event->mmap.filename) - size));
+		memset(event->mmap.filename + size, 0, machine->id_hdr_size);
+		event->mmap.header.size += machine->id_hdr_size;
+		event->mmap.pid = tgid;
+		event->mmap.tid = pid;
 
 		if (process(tool, event, &synth_sample, machine) != 0) {
 			rc = -1;
tools/perf/util/evsel.c +0 -1
index 0ce9feb..9f1ef9b 100644
@@ -678,7 +678,6 @@ void perf_evsel__config(struct perf_evsel *evsel,
 		attr->sample_type	|= PERF_SAMPLE_WEIGHT;
 
 	attr->mmap  = track;
-	attr->mmap2 = track && !perf_missing_features.mmap2;
 	attr->comm  = track;
 
 	/*
tools/perf/util/probe-finder.c +1 -1
index c09e0a9..f069273 100644
@@ -1357,10 +1357,10 @@ int debuginfo__find_probe_point(struct debuginfo *self, unsigned long addr,
 			goto post;
 		}
 
+		fname = dwarf_decl_file(&spdie);
 		if (addr == (unsigned long)baseaddr) {
 			/* Function entry - Relative line number is 0 */
 			lineno = baseline;
-			fname = dwarf_decl_file(&spdie);
 			goto post;
 		}
 
tools/perf/util/scripting-engines/trace-event-perl.c +1 -1
index a85e4ae..c0c9795 100644
@@ -282,7 +282,7 @@ static void perl_process_tracepoint(union perf_event *perf_event __maybe_unused,
 
 	event = find_cache_event(evsel);
 	if (!event)
-		die("ug! no event found for type %" PRIu64, evsel->attr.config);
+		die("ug! no event found for type %" PRIu64, (u64)evsel->attr.config);
 
 	pid = raw_field_value(event, "common_pid", data);
 
#1 lore raw
Quoted 4 lines · Ingo Molnar

- Reversal of the new 'MMAP2' extended mmap record ABI, introduced in this merge window. (Patches were proposed to fix it but it was all a bit late and we felt it's safer to just delay the ABI one more kernel release and do it right.)

commit 3090ffb5a2515990182f3f55b0688a7817325488

Author: Stephane Eranian <[email protected]>
Date:   Thu Oct 17 19:32:15 2013 +0200

perf: Disable PERF_RECORD_MMAP2 support

The commit above breaks "perf top" on my system, because it causes only [kernel] symbols to show up in the list. When I revert the commit I get the usual mixture of userspace and kernel symbols in the list.

Signature · 1 line
Markus
#2 lore raw
Quoted 7 lines · Markus Trippelsdorf
Quoted 4 lines · Ingo Molnar

- Reversal of the new 'MMAP2' extended mmap record ABI, introduced in this merge window. (Patches were proposed to fix it but it was all a bit late and we felt it's safer to just delay the ABI one more kernel release and do it right.)

commit 3090ffb5a2515990182f3f55b0688a7817325488

Quoted 8 lines
Author: Stephane Eranian <[email protected]>
Date:   Thu Oct 17 19:32:15 2013 +0200

perf: Disable PERF_RECORD_MMAP2 support

The commit above breaks "perf top" on my system, because it causes only [kernel] symbols to show up in the list. When I revert the commit I get the usual mixture of userspace and kernel symbols in the list.

Forgot to mention that this happens on an otherwise idle machine. If I run CPU intensive apps they'll show up eventually.

Example (current Linus tree):

PerfTop: 593 irqs/sec kernel:80.6% exact: 0.0% [4000Hz cycles], (all, 4 CPUs)
----------------------------------------------------------------------------------------------

    57.75%  [kernel]          [k] amd_e400_idle
    28.66%  [unknown]         [.] 0x0000000000450bc8
     1.51%  [kernel]          [k] kallsyms_expand_symbol.const
     1.22%  [kernel]          [k] cpu_startup_entry
     1.13%  [kernel]          [k] number.isra.1
     0.77%  [kernel]          [k] clear_page_c
     0.74%  [kernel]          [k] memcpy
     0.46%  [kernel]          [k] task_waking_fair
     0.45%  [kernel]          [k] _raw_spin_lock_irqsave
     0.42%  [kernel]          [k] _raw_spin_unlock_irqrestore
     0.39%  [kernel]          [k] format_decode
     0.34%  [kernel]          [k] do_sys_poll
     0.31%  [kernel]          [k] current_kernel_time
     0.30%  [kernel]          [k] vsnprintf
     0.29%  [kernel]          [k] native_apic_mem_write
     0.28%  [kernel]          [k] select_task_rq_fair
     0.26%  [kernel]          [k] swiotlb_free_coherent
     0.26%  [kernel]          [k] irq_entries_start
     0.2n%  [kernel]          [k] strnlen

With the commit above reverted:

PerfTop: 907 irqs/sec kernel:73.3% exact: 0.0% [4000Hz cycles], (all, 4 CPUs)
-----------------------------------------------------------------------------------------------

    30.89%  [kernel]                    [k] amd_e400_idle
     6.88%  [unknown]                   [.] 0x00007f346f2ef3a6
     5.22%  libc-2.18.90.so             [.] strcmp
     3.65%  libc-2.18.90.so             [.] strstr
     2.65%  libbfd-2.24.51.20131018.so  [.] d_print_comp.part.8
     2.61%  perf                        [.] symbols__insert
     2.18%  [kernel]                    [k] cpu_startup_entry
     1.82%  perf                        [.] symbol_filter
     1.73%  perf                        [.] rb_next
     1.58%  libc-2.18.90.so             [.] critical_factorization
     1.34%  libelf-0.156.so             [.] gelf_getsym
     1.28%  libQtGui.so.4.8.5           [.] qt_alphargbblit_quint32(QRasterBuffer*, int, int, unsigned int, unsigned int const*, int, int, int, QClipData
     1.25%  libc-2.18.90.so             [.] __libc_calloc
     1.17%  libc-2.18.90.so             [.] _int_malloc
     1.15%  libc-2.18.90.so             [.] strchr
     0.97%  xmobar                      [.] 0x000000000000c922
     0.97%  [kernel]                    [k] number.isra.1
     0.96%  libpthread-2.18.90.so       [.] pthread_rwlock_rdlock
     0.96%  libc-2.18.90.so             [.] memcpy@@GLIBC_2.14
     0.94%  perf                        [.] dso__load_sym
     0.93%  libc-2.18.90.so             [.] memchr
     0.84%  libbfd-2.24.51.20131018.so  [.] d_name
     0.75%  [kernel]                    [k] clear_page_c
     0.73%  libpthread-2.18.90.so       [.] pthread_rwlock_unlock
     0.66%  libc-2.18.90.so             [.] memset
Signature · 1 line
Markus
#3 lore raw
Quoted 11 lines · ------------------------------ Markus Trippelsdorf
Quoted 10 lines · Markus Trippelsdorf
Quoted 4 lines · Ingo Molnar

- Reversal of the new 'MMAP2' extended mmap record ABI, introduced in this merge window. (Patches were proposed to fix it but it was all a bit late and we felt it's safer to just delay the ABI one more kernel release and do it right.)

commit 3090ffb5a2515990182f3f55b0688a7817325488

Author: Stephane Eranian <[email protected]>
Date:   Thu Oct 17 19:32:15 2013 +0200
Quoted 5 lines
Quoted 5 lines

perf: Disable PERF_RECORD_MMAP2 support

The commit above breaks "perf top" on my system, because it causes only [kernel] symbols to show up in the list. When I revert the commit I get the usual mixture of userspace and kernel symbols in the list.

The following patch fixes the issue:

tools/perf/util/event.c +1 -5
index 63df031fc9c7..811f3950654d 100644
@@ -199,7 +199,6 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		char execname[PATH_MAX];
 		char anonstr[] = "//anon";
 		size_t size;
-		ssize_t n;
 
 		if (fgets(bf, sizeof(bf), fp) == NULL)
 			break;
@@ -208,14 +207,11 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		strcpy(execname, "");
 
 		/* 00400000-0040c000 r-xp 00000000 fd:01 41038  /bin/cat */
-		n = sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %*x:%*x %*u %s\n",
+		sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %*x:%*x %*u %s\n",
 		       &event->mmap.start, &event->mmap.len, prot,
 		       &event->mmap.pgoff,
 		       execname);
 
-		if (n != 8)
-			continue;
-
 		if (prot[2] != 'x')
 			continue;
 
Signature · 1 line
Markus
#4 lore raw

Em Mon, Oct 28, 2013 at 10:34:30AM +0100, Markus Trippelsdorf escreveu:

Quoted 13 lines · Markus Trippelsdorf
Quoted 8 lines · Markus Trippelsdorf
Quoted 7 lines · Ingo Molnar
Quoted 1 line

- Reversal of the new 'MMAP2' extended mmap record ABI, introduced

commit 3090ffb5a2515990182f3f55b0688a7817325488
Author: Stephane Eranian <[email protected]>

perf: Disable PERF_RECORD_MMAP2 support

The following patch fixes the issue:

tools/perf/util/event.c

Quoted 10 lines
tools/perf/util/event.c +1 -4
 		/* 00400000-0040c000 r-xp 00000000 fd:01 41038  /bin/cat */
-		n = sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %*x:%*x %*u %s\n",
+		sscanf(bf, "%"PRIx64"-%"PRIx64" %s %"PRIx64" %*x:%*x %*u %s\n",
 		       &event->mmap.start, &event->mmap.len, prot,
 		       &event->mmap.pgoff,
 		       execname);
 
-		if (n != 8)
-			continue;
-

Yeah, this one was added in the patch that introduced MMAP2 support in the tooling side, it was unrelated to what the patch needed, i.e. should have come in a separate patch, checking sscanf number of itens successfully matched/assigned.

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

- Arnaldo

tools/perf/util/event.c +1 -1
index 63df031fc9c7..49096ea58a15 100644
@@ -213,7 +213,7 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		       &event->mmap.pgoff,
 		       execname);
 
-		if (n != 8)
+		if (n != 5)
 			continue;
 
 		if (prot[2] != 'x')
#5 lore raw

Em Mon, Oct 28, 2013 at 09:34:11AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 3 lines

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

Does the wording on this commit log sounds about right to you?

- Arnaldo

commit 2fd869f08aec5a8e4cbf01bc3fa345c4e53342d7

Author: Arnaldo Carvalho de Melo <[email protected]>
Date:   Mon Oct 28 09:38:12 2013 -0300

perf tools: Fix up /proc/PID/maps parsing

When introducing support for MMAP2 we considered more parts of each map representation in /proc/PID/maps, and when disabling it we forgot to reduce the number of expected parsed/assigned entries in the sscanf call, fix it to expect the right number of desired fields, 5.

tools/perf/util/event.c +1 -1
index 63df031fc9c7..49096ea58a15 100644
@@ -213,7 +213,7 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool,
 		       &event->mmap.pgoff,
 		       execname);
 
-		if (n != 8)
+		if (n != 5)
 			continue;
 
 		if (prot[2] != 'x')
#6 lore raw
Quoted 6 lines · Arnaldo Carvalho de Melo

Em Mon, Oct 28, 2013 at 09:34:11AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 3 lines

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

Does the wording on this commit log sounds about right to you?

Yes. Your patch is fine. Thanks.

Signature · 1 line
Markus
#7 lore raw
In reply to Arnaldo Carvalho de Melo
Quoted 7 lines · Arnaldo Carvalho de Melo

Em Mon, Oct 28, 2013 at 09:34:11AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 3 lines

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

Does the wording on this commit log sounds about right to you?

Thanks for fixing this issue.

Quoted 1 line

- Arnaldo

#8 lore raw
Quoted 10 lines · Stephane Eranian
Quoted 7 lines · Arnaldo Carvalho de Melo

Em Mon, Oct 28, 2013 at 09:34:11AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 3 lines

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

Does the wording on this commit log sounds about right to you?

Thanks for fixing this issue.

Thanks guys for testing it - I just sent the pending fixes to Linus so it should soon be resolved upstream as well.

Thanks,

Ingo

#9 lore raw
In reply to Stephane Eranian

Em Tue, Oct 29, 2013 at 10:50:30AM +0100, Stephane Eranian escreveu:

Quoted 9 lines · Arnaldo Carvalho de Melo
Quoted 6 lines

Em Mon, Oct 28, 2013 at 09:34:11AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 3 lines

And then, when reverting it, Stephane forgot to match the number number of expected entries to be matched/assigned from 8 to 5, can you try with the following patch instead?

Does the wording on this commit log sounds about right to you?

Thanks for fixing this issue.

You're welcome! :-)

- Arnaldo