Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux

perf script: Add --insn-trace for instruction decoding

Add a --insn-trace short hand option for decoding and disassembling
instruction streams for intel_pt. This automatically pipes the output
into the xed disassembler to generate disassembled instructions. This
just makes this use model much nicer to use.

Before

% perf record -e intel_pt// ...
% perf script --itrace=i0ns --ns -F +insn,-event,-period | xed -F insn: -A -64
swapper 0 [000] 17276.429606186: ffffffff81010486 pt_config ([kernel.kallsyms]) nopl %eax, (%rax,%rax,1)
swapper 0 [000] 17276.429606186: ffffffff8101048b pt_config ([kernel.kallsyms]) add $0x10, %rsp
swapper 0 [000] 17276.429606186: ffffffff8101048f pt_config ([kernel.kallsyms]) popq %rbx
swapper 0 [000] 17276.429606186: ffffffff81010490 pt_config ([kernel.kallsyms]) popq %rbp
swapper 0 [000] 17276.429606186: ffffffff81010491 pt_config ([kernel.kallsyms]) popq %r12
swapper 0 [000] 17276.429606186: ffffffff81010493 pt_config ([kernel.kallsyms]) popq %r13
swapper 0 [000] 17276.429606186: ffffffff81010495 pt_config ([kernel.kallsyms]) popq %r14
swapper 0 [000] 17276.429606186: ffffffff81010497 pt_config ([kernel.kallsyms]) popq %r15
swapper 0 [000] 17276.429606186: ffffffff81010499 pt_config ([kernel.kallsyms]) retq
swapper 0 [000] 17276.429606186: ffffffff8101063e pt_event_add ([kernel.kallsyms]) cmpl $0x1, 0x1b0(%rbx)
swapper 0 [000] 17276.429606186: ffffffff81010645 pt_event_add ([kernel.kallsyms]) mov $0xffffffea, %eax
swapper 0 [000] 17276.429606186: ffffffff8101064a pt_event_add ([kernel.kallsyms]) mov $0x0, %edx
swapper 0 [000] 17276.429606186: ffffffff8101064f pt_event_add ([kernel.kallsyms]) popq %rbx
swapper 0 [000] 17276.429606186: ffffffff81010650 pt_event_add ([kernel.kallsyms]) cmovnz %edx, %eax
swapper 0 [000] 17276.429606186: ffffffff81010653 pt_event_add ([kernel.kallsyms]) jmp 0xffffffff81010635
swapper 0 [000] 17276.429606186: ffffffff81010635 pt_event_add ([kernel.kallsyms]) retq
swapper 0 [000] 17276.429606186: ffffffff8115e687 event_sched_in.isra.107 ([kernel.kallsyms]) test %eax, %eax

Now:

% perf record -e intel_pt// ...
% perf script --insn-trace --xed
... same output ...

XED needs to be installed with:

$ git clone https://github.com/intelxed/mbuild.git mbuild
$ git clone https://github.com/intelxed/xed
$ cd xed
$ ./mfile.py
$ ./mfile.py examples
$ sudo ./mfile.py --prefix=/usr/local install
$ sudo cp obj/examples/xed /usr/local/bin
$ xed | head -3
ERROR: required argument(s) were missing
Copyright (C) 2017, Intel Corporation. All rights reserved.
XED version: [v10.0-328-g7d62c8c49b7b]
$

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: http://lkml.kernel.org/r/20180920180540.14039-2-andi@firstfloor.org
[ Fixed up whitespace damage, added the 'mfile.py examples + cp obj/examples/xed ... ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

authored by

Andi Kleen and committed by
Arnaldo Carvalho de Melo
b585ebdb 76099f98

+49
+19
tools/perf/Documentation/build-xed.txt
··· 1 + 2 + For --xed the xed tool is needed. Here is how to install it: 3 + 4 + $ git clone https://github.com/intelxed/mbuild.git mbuild 5 + $ git clone https://github.com/intelxed/xed 6 + $ cd xed 7 + $ ./mfile.py --share 8 + $ ./mfile.py examples 9 + $ sudo ./mfile.py --prefix=/usr/local install 10 + $ sudo ldconfig 11 + $ sudo cp obj/examples/xed /usr/local/bin 12 + 13 + Basic xed testing: 14 + 15 + $ xed | head -3 16 + ERROR: required argument(s) were missing 17 + Copyright (C) 2017, Intel Corporation. All rights reserved. 18 + XED version: [v10.0-328-g7d62c8c49b7b] 19 + $
+7
tools/perf/Documentation/perf-script.txt
··· 383 383 will be printed. Each entry has function name and file/line. Enabled by 384 384 default, disable with --no-inline. 385 385 386 + --insn-trace:: 387 + Show instruction stream for intel_pt traces. Combine with --xed to 388 + show disassembly. 389 + 390 + --xed:: 391 + Run xed disassembler on output. Requires installing the xed disassembler. 392 + 386 393 SEE ALSO 387 394 -------- 388 395 linkperf:perf-record[1], linkperf:perf-script-perl[1],
+23
tools/perf/builtin-script.c
··· 44 44 #include <sys/stat.h> 45 45 #include <fcntl.h> 46 46 #include <unistd.h> 47 + #include <subcmd/pager.h> 47 48 48 49 #include "sane_ctype.h" 49 50 ··· 3104 3103 #define perf_script__process_auxtrace_info 0 3105 3104 #endif 3106 3105 3106 + static int parse_insn_trace(const struct option *opt __maybe_unused, 3107 + const char *str __maybe_unused, 3108 + int unset __maybe_unused) 3109 + { 3110 + parse_output_fields(NULL, "+insn,-event,-period", 0); 3111 + itrace_parse_synth_opts(opt, "i0ns", 0); 3112 + nanosecs = true; 3113 + return 0; 3114 + } 3115 + 3116 + static int parse_xed(const struct option *opt __maybe_unused, 3117 + const char *str __maybe_unused, 3118 + int unset __maybe_unused) 3119 + { 3120 + force_pager("xed -F insn: -A -64 | less"); 3121 + return 0; 3122 + } 3123 + 3107 3124 int cmd_script(int argc, const char **argv) 3108 3125 { 3109 3126 bool show_full_info = false; ··· 3206 3187 "system-wide collection from all CPUs"), 3207 3188 OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]", 3208 3189 "only consider these symbols"), 3190 + OPT_CALLBACK_OPTARG(0, "insn-trace", &itrace_synth_opts, NULL, NULL, 3191 + "Decode instructions from itrace", parse_insn_trace), 3192 + OPT_CALLBACK_OPTARG(0, "xed", NULL, NULL, NULL, 3193 + "Run xed disassembler on output", parse_xed), 3209 3194 OPT_STRING(0, "stop-bt", &symbol_conf.bt_stop_list_str, "symbol[,symbol...]", 3210 3195 "Stop display of callgraph at these symbols"), 3211 3196 OPT_STRING('C', "cpu", &cpu_list, "cpu", "list of cpus to profile"),