Skip to contentSkip to conversation

perf report: Fix arm64 gap between kernel start and module end

PATCH Applied Kemeng Shi 5 participants 9 messages Mar 30, 2020 – Feb 9, 2021

Subsystems (2)
lore
Share & tools
lore

Kemeng Shi

author
#0 lore raw

PATCH 1/1 perf report: Fix arm64 gap between kernel start and module end

During execution of command 'perf report' in my arm64 virtual machine, the error message is showed:

failed to process sample

__symbol__inc_addr_samples(860): ENOMEM! sym->name=__this_module,

    start=0x1477100, addr=0x147dbd8, end=0x80002000, func: 0

The error is caused with path:
cmd_report
__cmd_report
perf_session__process_events
__perf_session__process_events

    ordered_events__flush
     __ordered_events__flush
      oe->deliver (ordered_events__deliver_event)
       perf_session__deliver_event
        machines__deliver_event
         perf_evlist__deliver_sample
          tool->sample (process_sample_event)
           hist_entry_iter__add
            iter->add_entry_cb(hist_iter__report_callback)
             hist_entry__inc_addr_samples
              symbol__inc_addr_samples
               __symbol__inc_addr_samples
                h = annotated_source__histogram(src, evidx) (NULL)

annotated_source__histogram failed is caused with path:
...
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__hists

    annotated_source__alloc_histograms
     src->histograms = calloc(nr_hists, sizeof_sym_hist) (failed)

Calloc failed as the symbol__size(sym) is too huge. As show in error
message: start=0x1477100, end=0x80002000, size of symbol is about 2G.

This is the same problem as 'perf annotate: Fix s390 gap between kernel end and module start (b9c0a64901d5bd)'. Perf gets symbol information from /proc/kallsyms in __dso__load_kallsyms. A part of symbol in /proc/kallsyms from my virtual machine is as follows:
#cat /proc/kallsyms | sort
...
ffff000001475080 d rpfilter_mt_reg [ip6t_rpfilter]
ffff000001475100 d $d [ip6t_rpfilter]
ffff000001475100 d __this_module [ip6t_rpfilter]
ffff000080080000 t _head
ffff000080080000 T _text
ffff000080080040 t pe_header
...

Take line 'ffff000001475100 d __this_module [ip6t_rpfilter]' as example. The start and end of symbol are both set to ffff000001475100 in dso__load_all_kallsyms. Then symbols__fixup_end will set the end of symbol to next big address to ffff000001475100 in /proc/kallsyms, ffff000080080000 in this example. Then sizeof of symbol will be about 2G and cause the problem.

The start of module in my machine is
ffff000000a62000 t $x [dm_mod]

The start of kernel in my machine is
ffff000080080000 t _head

There is a big gap between end of module and begin of kernel if a samll amount of memory is used by module. And the last symbol in module will have a large address range as caotaining the big gap.

Give that the module and kernel text segment sequence may change in the future, fix this by limiting range of last symbol in module and kernel to 4K in arch arm64.

---
 tools/perf/arch/arm64/util/Build     |  1 +
 tools/perf/arch/arm64/util/machine.c | 26 ++++++++++++++++++++++++++
 2 files changed, 27 insertions(+)
 create mode 100644 tools/perf/arch/arm64/util/machine.c
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o
 libperf-$(CONFIG_DWARF)     += dwarf-regs.o
 libperf-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
 libperf-$(CONFIG_LIBDW_DWARF_UNWIND) += unwind-libdw.o
tools/perf/arch/arm64/util/machine.c new file +26 -0
new file mode 100644
index 000000000..a25be2431
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include "debug.h"
+#include "symbol.h"
+
+/* On arm64, kernel text segment start at high memory address,
+ * for example 0xffff 0000 8xxx xxxx. Modules start at a low memory
+ * address, like 0xffff 0000 00ax xxxx. When only samll amount of
+ * memory is used by modules, gap between end of module's text segment
+ * and start of kernel text segment may be reach 2G.
+ * Therefore do not fill this gap and do not assign it to the kernel dso map.
+ */
+
+#define SYMBOL_LIMIT (1 << 12) /* 4K */
+
+void arch__symbols__fixup_end(struct symbol *p, struct symbol *c)
+{
+	if ((strchr(p->name, '[') && strchr(c->name, '[') == NULL) ||
+			(strchr(p->name, '[') == NULL && strchr(c->name, '[')))
+		/* Limit range of last symbol in module and kernel */
+		p->end += SYMBOL_LIMIT;
+	else
+		p->end = c->start;
+	pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
+}
Signature · 1 line
2.19.1
#1 lore raw

Em Mon, Mar 30, 2020 at 03:41:11PM +0800, Kemeng Shi escreveu:

tools/perf/arch/arm64/util/Build libperf-y += header.o · line 5

Quoted 9 lines
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o

You made the patch against an old perf codebase, right? This libperf-y above was changed to perf-y here:

commit 5ff328836dfde0cef9f28c8b8791a90a36d7a183

Author: Jiri Olsa <[email protected]>
Date:   Wed Feb 13 13:32:39 2019 +0100

perf tools: Rename build libperf to perf

----

I'm fixing this up, please check my perf/core branch later to see that all is working as intended.

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core

Thanks,

- Arnaldo

#2 lore raw

Em Mon, Mar 30, 2020 at 10:11:29AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 18 lines

Em Mon, Mar 30, 2020 at 03:41:11PM +0800, Kemeng Shi escreveu:

tools/perf/arch/arm64/util/Build libperf-y += header.o · line 5

Quoted 9 lines
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o

You made the patch against an old perf codebase, right? This libperf-y above was changed to perf-y here:

commit 5ff328836dfde0cef9f28c8b8791a90a36d7a183

Author: Jiri Olsa <[email protected]>
Date:   Wed Feb 13 13:32:39 2019 +0100
Quoted 8 lines

perf tools: Rename build libperf to perf

----

I'm fixing this up, please check my perf/core branch later to see that all is working as intended.

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core

Here it is:

From 829b915d7d7eeafbe4af76dce19ccbdc743a43c8 Mon Sep 17 00:00:00 2001
From: Kemeng Shi <[email protected]>
Date: Mon, 30 Mar 2020 15:41:11 +0800
Subject: [PATCH 1/1] perf symbols: Fix arm64 gap between kernel start and
module end

During execution of command 'perf report' in my arm64 virtual machine, this error message is showed:

failed to process sample

__symbol__inc_addr_samples(860): ENOMEM! sym->name=__this_module,

    start=0x1477100, addr=0x147dbd8, end=0x80002000, func: 0

The error is caused with path:
cmd_report
__cmd_report
perf_session__process_events
__perf_session__process_events

    ordered_events__flush
     __ordered_events__flush
      oe->deliver (ordered_events__deliver_event)
       perf_session__deliver_event
        machines__deliver_event
         perf_evlist__deliver_sample
          tool->sample (process_sample_event)
           hist_entry_iter__add
            iter->add_entry_cb(hist_iter__report_callback)
             hist_entry__inc_addr_samples
              symbol__inc_addr_samples
               __symbol__inc_addr_samples
                h = annotated_source__histogram(src, evidx) (NULL)

annotated_source__histogram failed is caused with path:
...
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__hists

    annotated_source__alloc_histograms
     src->histograms = calloc(nr_hists, sizeof_sym_hist) (failed)

Calloc failed as the symbol__size(sym) is too huge. As show in error
message: start=0x1477100, end=0x80002000, size of symbol is about 2G.

This is the same problem as 'perf annotate: Fix s390 gap between kernel end and module start (b9c0a64901d5bd)'. Perf gets symbol information from /proc/kallsyms in __dso__load_kallsyms. A part of symbol in /proc/kallsyms from my virtual machine is as follows:
#cat /proc/kallsyms | sort
...
ffff000001475080 d rpfilter_mt_reg [ip6t_rpfilter]
ffff000001475100 d $d [ip6t_rpfilter]
ffff000001475100 d __this_module [ip6t_rpfilter]
ffff000080080000 t _head
ffff000080080000 T _text
ffff000080080040 t pe_header
...

Take line 'ffff000001475100 d __this_module [ip6t_rpfilter]' as example. The start and end of symbol are both set to ffff000001475100 in dso__load_all_kallsyms. Then symbols__fixup_end will set the end of symbol to next big address to ffff000001475100 in /proc/kallsyms, ffff000080080000 in this example. Then sizeof of symbol will be about 2G and cause the problem.

The start of module in my machine is
ffff000000a62000 t $x [dm_mod]

The start of kernel in my machine is
ffff000080080000 t _head

There is a big gap between end of module and begin of kernel if a samll amount of memory is used by module. And the last symbol in module will have a large address range as caotaining the big gap.

Give that the module and kernel text segment sequence may change in the future, fix this by limiting range of last symbol in module and kernel to 4K in arch arm64.

Cc Alexander Shishkin <[email protected]>
Cc Hu Shiyuan <[email protected]>
Cc Ian Rogers <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Nick Desaulniers <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Richter <[email protected]>
[ refreshed the patch on current codebase ]
---
 tools/perf/arch/arm64/util/Build     |  1 +
 tools/perf/arch/arm64/util/machine.c | 26 ++++++++++++++++++++++++++
 2 files changed, 27 insertions(+)
 create mode 100644 tools/perf/arch/arm64/util/machine.c
tools/perf/arch/arm64/util/Build +1 -0
index 789956f76d85..5c13438c7bd4 100644
@@ -1,4 +1,5 @@
 perf-y += header.o
+perf-y += machine.o
 perf-y += perf_regs.o
 perf-$(CONFIG_DWARF)     += dwarf-regs.o
 perf-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
tools/perf/arch/arm64/util/machine.c new file +26 -0
new file mode 100644
index 000000000000..94745131e89a
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include "debug.h"
+#include "symbol.h"
+
+/* On arm64, kernel text segment start at high memory address,
+ * for example 0xffff 0000 8xxx xxxx. Modules start at a low memory
+ * address, like 0xffff 0000 00ax xxxx. When only samll amount of
+ * memory is used by modules, gap between end of module's text segment
+ * and start of kernel text segment may be reach 2G.
+ * Therefore do not fill this gap and do not assign it to the kernel dso map.
+ */
+
+#define SYMBOL_LIMIT (1 << 12) /* 4K */
+
+void arch__symbols__fixup_end(struct symbol *p, struct symbol *c)
+{
+	if ((strchr(p->name, '[') && strchr(c->name, '[') == NULL) ||
+			(strchr(p->name, '[') == NULL && strchr(c->name, '[')))
+		/* Limit range of last symbol in module and kernel */
+		p->end += SYMBOL_LIMIT;
+	else
+		p->end = c->start;
+	pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
+}
Signature · 1 line
2.25.1
#3 lore raw

cover perf tools: fix compilation on Arm

Hi Arnaldo,

This patch seems to break the build for me on your perf/core branch but adding the header fixes it.

Thanks
James

James Clark (1):
perf tools: fix compilation on Arm

 tools/perf/arch/arm64/util/machine.c | 1 +
 1 file changed, 1 insertion(+)
Signature · 1 line
2.17.1
#4 lore raw

PATCH 1/1 perf tools: fix compilation on Arm

The commit "perf report: Fix arm64 gap between kernel start and module end" introduces the following build error with GCC 7.3.0 on Arm:

"error: implicit declaration of function ‘strchr’"

Adding the required header fixes the issue.

---
 tools/perf/arch/arm64/util/machine.c | 1 +
 1 file changed, 1 insertion(+)
tools/perf/arch/arm64/util/machine.c +1 -0
index 94745131e89a..d41b27e781d3 100644
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 
 #include <stdio.h>
+#include <string.h>
 #include "debug.h"
 #include "symbol.h"
 
Signature · 1 line
2.17.1
#5 lore raw
In reply to Arnaldo Carvalho de Melo

Em Mon, Mar 30, 2020 at 10:18:10AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 162 lines

Em Mon, Mar 30, 2020 at 10:11:29AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 18 lines

Em Mon, Mar 30, 2020 at 03:41:11PM +0800, Kemeng Shi escreveu:

tools/perf/arch/arm64/util/Build libperf-y += header.o · line 5

Quoted 9 lines
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o

You made the patch against an old perf codebase, right? This libperf-y above was changed to perf-y here:

commit 5ff328836dfde0cef9f28c8b8791a90a36d7a183

Author: Jiri Olsa <[email protected]>
Date:   Wed Feb 13 13:32:39 2019 +0100
Quoted 8 lines

perf tools: Rename build libperf to perf

----

I'm fixing this up, please check my perf/core branch later to see that all is working as intended.

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core

Here it is:

From 829b915d7d7eeafbe4af76dce19ccbdc743a43c8 Mon Sep 17 00:00:00 2001
From: Kemeng Shi <[email protected]>
Date: Mon, 30 Mar 2020 15:41:11 +0800
Subject: [PATCH 1/1] perf symbols: Fix arm64 gap between kernel start and
module end

During execution of command 'perf report' in my arm64 virtual machine, this error message is showed:

failed to process sample

__symbol__inc_addr_samples(860): ENOMEM! sym->name=__this_module,

    start=0x1477100, addr=0x147dbd8, end=0x80002000, func: 0

The error is caused with path:
cmd_report
__cmd_report
perf_session__process_events
__perf_session__process_events

    ordered_events__flush
     __ordered_events__flush
      oe->deliver (ordered_events__deliver_event)
       perf_session__deliver_event
        machines__deliver_event
         perf_evlist__deliver_sample
          tool->sample (process_sample_event)
           hist_entry_iter__add
            iter->add_entry_cb(hist_iter__report_callback)
             hist_entry__inc_addr_samples
              symbol__inc_addr_samples
               __symbol__inc_addr_samples
                h = annotated_source__histogram(src, evidx) (NULL)

annotated_source__histogram failed is caused with path:
...
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__hists

    annotated_source__alloc_histograms
     src->histograms = calloc(nr_hists, sizeof_sym_hist) (failed)

Calloc failed as the symbol__size(sym) is too huge. As show in error
message: start=0x1477100, end=0x80002000, size of symbol is about 2G.

This is the same problem as 'perf annotate: Fix s390 gap between kernel end and module start (b9c0a64901d5bd)'. Perf gets symbol information from /proc/kallsyms in __dso__load_kallsyms. A part of symbol in /proc/kallsyms from my virtual machine is as follows:
#cat /proc/kallsyms | sort
...
ffff000001475080 d rpfilter_mt_reg [ip6t_rpfilter]
ffff000001475100 d $d [ip6t_rpfilter]
ffff000001475100 d __this_module [ip6t_rpfilter]
ffff000080080000 t _head
ffff000080080000 T _text
ffff000080080040 t pe_header
...

Take line 'ffff000001475100 d __this_module [ip6t_rpfilter]' as example. The start and end of symbol are both set to ffff000001475100 in dso__load_all_kallsyms. Then symbols__fixup_end will set the end of symbol to next big address to ffff000001475100 in /proc/kallsyms, ffff000080080000 in this example. Then sizeof of symbol will be about 2G and cause the problem.

The start of module in my machine is
ffff000000a62000 t $x [dm_mod]

The start of kernel in my machine is
ffff000080080000 t _head

There is a big gap between end of module and begin of kernel if a samll amount of memory is used by module. And the last symbol in module will have a large address range as caotaining the big gap.

Give that the module and kernel text segment sequence may change in the future, fix this by limiting range of last symbol in module and kernel to 4K in arch arm64.

Cc Alexander Shishkin <[email protected]>
Cc Hu Shiyuan <[email protected]>
Cc Ian Rogers <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Nick Desaulniers <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Richter <[email protected]>
[ refreshed the patch on current codebase ]
---
 tools/perf/arch/arm64/util/Build     |  1 +
 tools/perf/arch/arm64/util/machine.c | 26 ++++++++++++++++++++++++++
 2 files changed, 27 insertions(+)
 create mode 100644 tools/perf/arch/arm64/util/machine.c
tools/perf/arch/arm64/util/Build +1 -0
index 789956f76d85..5c13438c7bd4 100644
@@ -1,4 +1,5 @@
 perf-y += header.o
+perf-y += machine.o
 perf-y += perf_regs.o
 perf-$(CONFIG_DWARF)     += dwarf-regs.o
 perf-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
tools/perf/arch/arm64/util/machine.c new file +16 -0
new file mode 100644
index 000000000000..94745131e89a
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include "debug.h"
+#include "symbol.h"
+
+/* On arm64, kernel text segment start at high memory address,
+ * for example 0xffff 0000 8xxx xxxx. Modules start at a low memory
+ * address, like 0xffff 0000 00ax xxxx. When only samll amount of
+ * memory is used by modules, gap between end of module's text segment
+ * and start of kernel text segment may be reach 2G.
+ * Therefore do not fill this gap and do not assign it to the kernel dso map.
+ */
+
+#define SYMBOL_LIMIT (1 << 12) /* 4K */
+

tools/perf/arch/arm64/util/machine.c line 26

Quoted 10 lines
tools/perf/arch/arm64/util/machine.c new file +10 -0
+void arch__symbols__fixup_end(struct symbol *p, struct symbol *c)
+{
+	if ((strchr(p->name, '[') && strchr(c->name, '[') == NULL) ||
+			(strchr(p->name, '[') == NULL && strchr(c->name, '[')))
+		/* Limit range of last symbol in module and kernel */
+		p->end += SYMBOL_LIMIT;
+	else
+		p->end = c->start;
+	pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
+}

This breaks the build on powerpc 32-bit, I'm fixing it using the proper "%#" PRIx64 " style.

- Arnaldo

72 13.69 ubuntu:18.04-x-powerpc : FAIL powerpc-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0

    arch/powerpc/util/machine.c: In function 'arch__symbols__fixup_end':
    arch/powerpc/util/machine.c:23:12: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'u64 {aka long long unsigned int}' [-Werror=format=]
      pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
                ^
    /git/linux/tools/perf/util/debug.h:18:21: note: in definition of macro 'pr_fmt'
     #define pr_fmt(fmt) fmt
                         ^~~
    /git/linux/tools/perf/util/debug.h:33:29: note: in expansion of macro 'pr_debugN'
     #define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
                                 ^~~~~~~~~
    /git/linux/tools/perf/util/debug.h:33:42: note: in expansion of macro 'pr_fmt'
     #define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
                                              ^~~~~~
    arch/powerpc/util/machine.c:23:2: note: in expansion of macro 'pr_debug4'
      pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
      ^~~~~~~~~
    cc1: all warnings being treated as errors
    /git/linux/tools/build/Makefile.build:139: recipe for target 'util' failed
    make[5]: *** [util] Error 2
    /git/linux/tools/build/Makefile.build:139: recipe for target 'powerpc' failed
    make[4]: *** [powerpc] Error 2
    /git/linux/tools/build/Makefile.build:139: recipe for target 'arch' failed
    make[3]: *** [arch] Error 2

73 30.47 ubuntu:18.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0

#6 lore raw

Em Tue, Feb 09, 2021 at 09:40:09AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 176 lines

Em Mon, Mar 30, 2020 at 10:18:10AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 162 lines

Em Mon, Mar 30, 2020 at 10:11:29AM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 18 lines

Em Mon, Mar 30, 2020 at 03:41:11PM +0800, Kemeng Shi escreveu:

tools/perf/arch/arm64/util/Build libperf-y += header.o · line 5

Quoted 9 lines
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o

You made the patch against an old perf codebase, right? This libperf-y above was changed to perf-y here:

commit 5ff328836dfde0cef9f28c8b8791a90a36d7a183

Author: Jiri Olsa <[email protected]>
Date:   Wed Feb 13 13:32:39 2019 +0100
Quoted 8 lines

perf tools: Rename build libperf to perf

----

I'm fixing this up, please check my perf/core branch later to see that all is working as intended.

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core

Here it is:

From 829b915d7d7eeafbe4af76dce19ccbdc743a43c8 Mon Sep 17 00:00:00 2001
From: Kemeng Shi <[email protected]>
Date: Mon, 30 Mar 2020 15:41:11 +0800
Subject: [PATCH 1/1] perf symbols: Fix arm64 gap between kernel start and
module end

During execution of command 'perf report' in my arm64 virtual machine, this error message is showed:

failed to process sample

__symbol__inc_addr_samples(860): ENOMEM! sym->name=__this_module,

    start=0x1477100, addr=0x147dbd8, end=0x80002000, func: 0

The error is caused with path:
cmd_report
__cmd_report
perf_session__process_events
__perf_session__process_events

    ordered_events__flush
     __ordered_events__flush
      oe->deliver (ordered_events__deliver_event)
       perf_session__deliver_event
        machines__deliver_event
         perf_evlist__deliver_sample
          tool->sample (process_sample_event)
           hist_entry_iter__add
            iter->add_entry_cb(hist_iter__report_callback)
             hist_entry__inc_addr_samples
              symbol__inc_addr_samples
               __symbol__inc_addr_samples
                h = annotated_source__histogram(src, evidx) (NULL)

annotated_source__histogram failed is caused with path:
...
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__hists

    annotated_source__alloc_histograms
     src->histograms = calloc(nr_hists, sizeof_sym_hist) (failed)

Calloc failed as the symbol__size(sym) is too huge. As show in error
message: start=0x1477100, end=0x80002000, size of symbol is about 2G.

This is the same problem as 'perf annotate: Fix s390 gap between kernel end and module start (b9c0a64901d5bd)'. Perf gets symbol information from /proc/kallsyms in __dso__load_kallsyms. A part of symbol in /proc/kallsyms from my virtual machine is as follows:
#cat /proc/kallsyms | sort
...
ffff000001475080 d rpfilter_mt_reg [ip6t_rpfilter]
ffff000001475100 d $d [ip6t_rpfilter]
ffff000001475100 d __this_module [ip6t_rpfilter]
ffff000080080000 t _head
ffff000080080000 T _text
ffff000080080040 t pe_header
...

Take line 'ffff000001475100 d __this_module [ip6t_rpfilter]' as example. The start and end of symbol are both set to ffff000001475100 in dso__load_all_kallsyms. Then symbols__fixup_end will set the end of symbol to next big address to ffff000001475100 in /proc/kallsyms, ffff000080080000 in this example. Then sizeof of symbol will be about 2G and cause the problem.

The start of module in my machine is
ffff000000a62000 t $x [dm_mod]

The start of kernel in my machine is
ffff000080080000 t _head

There is a big gap between end of module and begin of kernel if a samll amount of memory is used by module. And the last symbol in module will have a large address range as caotaining the big gap.

Give that the module and kernel text segment sequence may change in the future, fix this by limiting range of last symbol in module and kernel to 4K in arch arm64.

Cc Alexander Shishkin <[email protected]>
Cc Hu Shiyuan <[email protected]>
Cc Ian Rogers <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Nick Desaulniers <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Richter <[email protected]>
[ refreshed the patch on current codebase ]
---
 tools/perf/arch/arm64/util/Build     |  1 +
 tools/perf/arch/arm64/util/machine.c | 26 ++++++++++++++++++++++++++
 2 files changed, 27 insertions(+)
 create mode 100644 tools/perf/arch/arm64/util/machine.c
tools/perf/arch/arm64/util/Build +1 -0
index 789956f76d85..5c13438c7bd4 100644
@@ -1,4 +1,5 @@
 perf-y += header.o
+perf-y += machine.o
 perf-y += perf_regs.o
 perf-$(CONFIG_DWARF)     += dwarf-regs.o
 perf-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
tools/perf/arch/arm64/util/machine.c new file +16 -0
new file mode 100644
index 000000000000..94745131e89a
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include "debug.h"
+#include "symbol.h"
+
+/* On arm64, kernel text segment start at high memory address,
+ * for example 0xffff 0000 8xxx xxxx. Modules start at a low memory
+ * address, like 0xffff 0000 00ax xxxx. When only samll amount of
+ * memory is used by modules, gap between end of module's text segment
+ * and start of kernel text segment may be reach 2G.
+ * Therefore do not fill this gap and do not assign it to the kernel dso map.
+ */
+
+#define SYMBOL_LIMIT (1 << 12) /* 4K */
+

tools/perf/arch/arm64/util/machine.c line 26

Quoted 10 lines
tools/perf/arch/arm64/util/machine.c new file +10 -0
+void arch__symbols__fixup_end(struct symbol *p, struct symbol *c)
+{
+	if ((strchr(p->name, '[') && strchr(c->name, '[') == NULL) ||
+			(strchr(p->name, '[') == NULL && strchr(c->name, '[')))
+		/* Limit range of last symbol in module and kernel */
+		p->end += SYMBOL_LIMIT;
+	else
+		p->end = c->start;
+	pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
+}

This breaks the build on powerpc 32-bit, I'm fixing it using the proper "%#" PRIx64 " style.

Quoted 1 line

- Arnaldo

Sorry, the build broke on powerpc 32-bit, but the above for arm also has the same problem, I'll fix it there as well even it not breaking the build as that is 64-bit specific (ARM case).

- Arnaldo

Quoted 27 lines

72 13.69 ubuntu:18.04-x-powerpc : FAIL powerpc-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0

    arch/powerpc/util/machine.c: In function 'arch__symbols__fixup_end':
    arch/powerpc/util/machine.c:23:12: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'u64 {aka long long unsigned int}' [-Werror=format=]
      pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
                ^
    /git/linux/tools/perf/util/debug.h:18:21: note: in definition of macro 'pr_fmt'
     #define pr_fmt(fmt) fmt
                         ^~~
    /git/linux/tools/perf/util/debug.h:33:29: note: in expansion of macro 'pr_debugN'
     #define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
                                 ^~~~~~~~~
    /git/linux/tools/perf/util/debug.h:33:42: note: in expansion of macro 'pr_fmt'
     #define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
                                              ^~~~~~
    arch/powerpc/util/machine.c:23:2: note: in expansion of macro 'pr_debug4'
      pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
      ^~~~~~~~~
    cc1: all warnings being treated as errors
    /git/linux/tools/build/Makefile.build:139: recipe for target 'util' failed
    make[5]: *** [util] Error 2
    /git/linux/tools/build/Makefile.build:139: recipe for target 'powerpc' failed
    make[4]: *** [powerpc] Error 2
    /git/linux/tools/build/Makefile.build:139: recipe for target 'arch' failed
    make[3]: *** [arch] Error 2

73 30.47 ubuntu:18.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0

Signature · 1 line
- Arnaldo
author
#7 lore raw
In reply to Arnaldo Carvalho de Melo
Quoted 14 lines · Arnaldo Carvalho de Melo

Em Mon, Mar 30, 2020 at 03:41:11PM +0800, Kemeng Shi escreveu:

tools/perf/arch/arm64/util/Build libperf-y += header.o · line 5

Quoted 9 lines
tools/perf/arch/arm64/util/Build +1 -0
index 393b9895c..37cbfa5e9 100644
@@ -2,6 +2,7 @@ libperf-y += header.o
 libperf-y += tsc.o
 libperf-y += sym-handling.o
 libperf-y += kvm-stat.o
+libperf-y += machine.o

You made the patch against an old perf codebase, right? This libperf-y above was changed to perf-y here:

I'm sorry. I checked the problem in https://github.com/torvalds/linux.git,
but I forget to make patch against it.

Quoted 8 lines

commit 5ff328836dfde0cef9f28c8b8791a90a36d7a183

Author: Jiri Olsa <[email protected]>
Date:   Wed Feb 13 13:32:39 2019 +0100

perf tools: Rename build libperf to perf

----

Quoted 9 lines

I'm fixing this up, please check my perf/core branch later to see that all is working as intended.

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core

Thanks,

- Arnaldo

I test it and it works fine, thanks for fixing up.

merge bot Applied
#8 lore raw

[tip: perf/urgent] perf symbols: Fix arm64 gap between kernel start and module end

In reply to Kemeng Shi
The following commit has been merged into the perf/urgent branch of tip:

The following commit has been merged into the perf/urgent branch of tip:

Commit-ID:     78886f3ed37e89a06c76b95be873573e27900979
Gitweb:        https://git.kernel.org/tip/78886f3ed37e89a06c76b95be873573e27900979
Author:        Kemeng Shi <[email protected]>
AuthorDate:    Mon, 30 Mar 2020 15:41:11 +08:00
Committer:     Arnaldo Carvalho de Melo <[email protected]>
CommitterDate: Fri, 03 Apr 2020 09:37:55 -03:00

perf symbols: Fix arm64 gap between kernel start and module end

During execution of command 'perf report' in my arm64 virtual machine, this error message is showed:

failed to process sample

__symbol__inc_addr_samples(860): ENOMEM! sym->name=__this_module,

    start=0x1477100, addr=0x147dbd8, end=0x80002000, func: 0

The error is caused with path:
cmd_report
__cmd_report
perf_session__process_events
__perf_session__process_events

    ordered_events__flush
     __ordered_events__flush
      oe->deliver (ordered_events__deliver_event)
       perf_session__deliver_event
        machines__deliver_event
         perf_evlist__deliver_sample
          tool->sample (process_sample_event)
           hist_entry_iter__add
            iter->add_entry_cb(hist_iter__report_callback)
             hist_entry__inc_addr_samples
              symbol__inc_addr_samples
               __symbol__inc_addr_samples
                h = annotated_source__histogram(src, evidx) (NULL)

annotated_source__histogram failed is caused with path:
...
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__hists

    annotated_source__alloc_histograms
     src->histograms = calloc(nr_hists, sizeof_sym_hist) (failed)

Calloc failed as the symbol__size(sym) is too huge. As show in error
message: start=0x1477100, end=0x80002000, size of symbol is about 2G.

This is the same problem as 'perf annotate: Fix s390 gap between kernel end and module start (b9c0a64901d5bd)'. Perf gets symbol information from /proc/kallsyms in __dso__load_kallsyms. A part of symbol in /proc/kallsyms from my virtual machine is as follows:
#cat /proc/kallsyms | sort
...
ffff000001475080 d rpfilter_mt_reg [ip6t_rpfilter]
ffff000001475100 d $d [ip6t_rpfilter]
ffff000001475100 d __this_module [ip6t_rpfilter]
ffff000080080000 t _head
ffff000080080000 T _text
ffff000080080040 t pe_header
...

Take line 'ffff000001475100 d __this_module [ip6t_rpfilter]' as example. The start and end of symbol are both set to ffff000001475100 in dso__load_all_kallsyms. Then symbols__fixup_end will set the end of symbol to next big address to ffff000001475100 in /proc/kallsyms, ffff000080080000 in this example. Then sizeof of symbol will be about 2G and cause the problem.

The start of module in my machine is
ffff000000a62000 t $x [dm_mod]

The start of kernel in my machine is
ffff000080080000 t _head

There is a big gap between end of module and begin of kernel if a samll amount of memory is used by module. And the last symbol in module will have a large address range as caotaining the big gap.

Give that the module and kernel text segment sequence may change in the future, fix this by limiting range of last symbol in module and kernel to 4K in arch arm64.

Cc Alexander Shishkin <[email protected]>
Cc Hewenliang <[email protected]>
Cc Hu Shiyuan <[email protected]>
Cc Ian Rogers <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Nick Desaulniers <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Richter <[email protected]>
[ refreshed the patch on current codebase, added string.h include as strchr() is used ]
---
 tools/perf/arch/arm64/util/Build     |  1 +-
 tools/perf/arch/arm64/util/machine.c | 27 +++++++++++++++++++++++++++-
 2 files changed, 28 insertions(+)
 create mode 100644 tools/perf/arch/arm64/util/machine.c
tools/perf/arch/arm64/util/Build +1 -0
index 789956f..5c13438 100644
@@ -1,4 +1,5 @@
 perf-y += header.o
+perf-y += machine.o
 perf-y += perf_regs.o
 perf-$(CONFIG_DWARF)     += dwarf-regs.o
 perf-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
tools/perf/arch/arm64/util/machine.c new file +27 -0
new file mode 100644
index 0000000..d41b27e
@@ -0,0 +1,27 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include <string.h>
+#include "debug.h"
+#include "symbol.h"
+
+/* On arm64, kernel text segment start at high memory address,
+ * for example 0xffff 0000 8xxx xxxx. Modules start at a low memory
+ * address, like 0xffff 0000 00ax xxxx. When only samll amount of
+ * memory is used by modules, gap between end of module's text segment
+ * and start of kernel text segment may be reach 2G.
+ * Therefore do not fill this gap and do not assign it to the kernel dso map.
+ */
+
+#define SYMBOL_LIMIT (1 << 12) /* 4K */
+
+void arch__symbols__fixup_end(struct symbol *p, struct symbol *c)
+{
+	if ((strchr(p->name, '[') && strchr(c->name, '[') == NULL) ||
+			(strchr(p->name, '[') == NULL && strchr(c->name, '[')))
+		/* Limit range of last symbol in module and kernel */
+		p->end += SYMBOL_LIMIT;
+	else
+		p->end = c->start;
+	pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
+}