[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 7 Jul 2017 11:51:11 -0300
From: Arnaldo Carvalho de Melo <acme@...nel.org>
To: Jin Yao <yao.jin@...ux.intel.com>
Cc: Jiri Olsa <jolsa@...nel.org>, peterz@...radead.org,
Ingo Molnar <mingo@...nel.org>,
alexander.shishkin@...ux.intel.com, Linux-kernel@...r.kernel.org,
ak@...ux.intel.com, kan.liang@...el.com, yao.jin@...el.com
Subject: Re: [PATCH v4 2/2] perf report: Implement visual marker for macro
fusion in annotate
Em Fri, Jul 07, 2017 at 01:06:35PM +0800, Jin Yao escreveu:
> For marking the fused instructions clearly, This patch adds a line
> before the first instruction of pair and joins it with the arrow of the
> jump.
>
> For example, when je is selected in annotate view, the line before cmpl
> is displayed and joins the arrow of je.
>
> │ ┌──cmpl $0x0,argp_program_version_hook
> 81.93 │ ├──je 20
> │ │ lock cmpxchg %esi,0x38a9a4(%rip)
> │ │↓ jne 29
> │ │↓ jmp 43
> 11.47 │20:└─→cmpxch %esi,0x38a999(%rip)
>
> That means the cmpl+je is fused instruction pair and they should be
> considered together.
I applied this one, no unnecessary parsing of cpuid done at each
jump->target arrow rendering, much better, thanks!
One thing for a follow up patch:
We have this when the cursor is at a jump instruction:
│ ┌──test %ecx,%ecx
-> │ ├──je 714cf
│ │ mov LINES+0xb40,%edx
│ │ test %edx,%edx
│ │↓ je 71580
│714cf:└─→mov LINES+0x10c8,%eax
But if we go up a line, to that "test" instruction, we get:
-> │ test %ecx,%ecx
│ ↓ je 714cf
│ mov LINES+0xb40,%edx
│ test %edx,%edx
│ ↓ je 71580
│714cf: mov LINES+0x10c8,%eax
I suggest that this be changed to:
-> │ ┌─test %ecx,%ecx
│ ↓ je 714cf
│ mov LINES+0xb40,%edx
│ test %edx,%edx
│ ↓ je 71580
│714cf: mov LINES+0x10c8,%eax
I.e. even before going to the jump instruction line with the cursor, we
would see the fused instructions.
To do that perhaps we should improve annotate_browser__draw_current_jump
to improve that part that looks for is_valid_jump() to consider
instructions that could be fused with jumps for the machine where the
perf data came from, etc.
But the current situation is better already, thanks for your work,
applied!
- Arnaldo
Powered by blists - more mailing lists