Received: by 2002:a05:6a10:9848:0:0:0:0 with SMTP id x8csp914577pxf; Wed, 7 Apr 2021 14:57:51 -0700 (PDT) X-Google-Smtp-Source: ABdhPJzEJI26gMTj535pEPJTc44rUyGyZH1Z19Ec2LCpu3gPXjCsBadLkNxQ9Fs65Lccwc14D7Eh X-Received: by 2002:a17:90b:3cd:: with SMTP id go13mr5223462pjb.148.1617832671526; Wed, 07 Apr 2021 14:57:51 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1617832671; cv=none; d=google.com; s=arc-20160816; b=PtRhoiX0Tu9+1usVpaMIBPvoq5aK1ZFa4KApTO/2dTM/DrNvx3Ij4VDX5+eB3MOxuq QKXtXenzv4NtSElfMEnkiM6v1r2wrgy9CG7suyn4dV0HMZLru+4g1qAspiESLV9Uvgry DIQku5Q5Fic2+MozZr7fcS9OFPhGvVZVLUNdU1+X6eTWOBP1lHs8r3rYbASwt7Lnl/w7 BAxBZFNNvXhegB/G1WCOEgi31LAta8m8XZGf0jZMvpAj9MT3ZMxIEBsQWkbI5kpYkfsh TOGXbeZAdUR/qzTb1KJ6mnby2Aw097ejWYEu2pr+0gVguKTZwg/jkrYBl+k60mK/v6Cp OFwg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:in-reply-to:content-transfer-encoding :content-disposition:mime-version:references:message-id:subject:cc :to:from:date:dkim-signature; bh=7ihQnz4BslxJwYoWPxOA2kVwy/tYRgTQuTSCpsWIblc=; b=vXoR5Pt1/s+kL5EVcNq91/NZuZcndlFfYv/ve11popAWYXeGlQRpaa62QYwZ2TQyc8 6cXGL5Vf1pIUDcRQXsrmUENs2sz+cjMVAoOirpXpJyerBHHeDAYhyMsMdcMD4pcYbh/Y TMg1vNJvB/m/jHE/Peg0VIxTBIwrMMn/Lvo16qAGXVwwvWD7M4/NkvojyVo6oY6yxUQb q6wwC6rA9rZ4oc/HgnPfo9OoAfyQyLdrIIh7/sfpzbE3O96QYO42E9EJNE7OdBchPa5M EfwQi9mWzDO0EXhdCH6UNz/Z3bjSnFAsWw55mltYfUoeQNe2mFHy2kzcSDR+2n+KsInx Abvg== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@kernel.org header.s=k20201202 header.b=VsTJ3V1o; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=kernel.org Return-Path: Received: from vger.kernel.org (vger.kernel.org. [23.128.96.18]) by mx.google.com with ESMTP id e2si9379423plt.255.2021.04.07.14.57.39; Wed, 07 Apr 2021 14:57:51 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18; Authentication-Results: mx.google.com; dkim=pass header.i=@kernel.org header.s=k20201202 header.b=VsTJ3V1o; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229512AbhDGT3R (ORCPT + 99 others); Wed, 7 Apr 2021 15:29:17 -0400 Received: from mail.kernel.org ([198.145.29.99]:43226 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1345181AbhDGT3I (ORCPT ); Wed, 7 Apr 2021 15:29:08 -0400 Received: by mail.kernel.org (Postfix) with ESMTPSA id EE3BB610E7; Wed, 7 Apr 2021 19:28:57 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1617823738; bh=d8BhiruXdLuB2TuWDxIGECMg94c6t0ZVddn66ylrU0Y=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=VsTJ3V1oNq5ffOuOM2vaARGSsW9Ie5p6+HqwcCZhiB/czUuAX6/qDA03fgbUCxvLD fqqseuJ1DPtjrqE3QqPBvLsURfgKqd/C2QVRy3M19mFI/tktfvpcNt8SHkLwwG4zRG XaT2jfZPoH3/B2KiADDl1dsrsB17BWvJ76RfE0GL6dkuKUg49wg6FXR6f8WjGTQomG JbVCn/6AtkM6hKJL8E7LDjGNaSkBGCIw9DXoj2PUqugVQizgliBtjf9DulRD903akL Miap6cgLVjY4pDGFhNtKnaUmdLA8/3X9h0VEYqfZ4a/uzPvHgB8ZX6ja2Z6VO++mPy Umc0RoIZNegRA== Received: by quaco.ghostprotocols.net (Postfix, from userid 1000) id 286B040647; Wed, 7 Apr 2021 16:28:55 -0300 (-03) Date: Wed, 7 Apr 2021 16:28:55 -0300 From: Arnaldo Carvalho de Melo To: Martin =?utf-8?B?TGnFoWth?= Cc: linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org Subject: Re: [PATCH] perf annotate: improve --stdio mode Message-ID: References: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: X-Url: http://acmel.wordpress.com Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Em Fri, Mar 19, 2021 at 10:38:26AM +0100, Martin Liška escreveu: > PING Can you please try to refresh the patch? It isn't applying, probably my bad for not having processed it already :-\ - Arnaldo > On 2/26/21 10:24 AM, Martin Liška wrote: > > On 2/23/21 8:47 PM, Arnaldo Carvalho de Melo wrote: > > > Em Sun, Feb 21, 2021 at 01:46:36PM +0100, Martin Liška escreveu: > > > > The patch changes the output format in 2 ways: > > > > - line number is displayed for all source lines (matching TUI mode) > > > > > > Are you aware of 'perf annotate --stdio2' ? If the goal is to make the > > > stdio mode better, doing it in that mode would be best, as it was done > > > to share as much code as possible, not just the looks, with the TUI > > > mode. > > > > Yes, I'm aware of it. My motivation is to generate a HTML perf annotate report > > and I see the following parts of --stdio2 not ideal: > > > > - coloring is not available (--stdio-color=always does not work) > > - 'Sorted summary for file ' is missing so one can't easily search for > >   hot spots in browser > > - source line number are displayed, but not the source files > > - there's a missing option for 'Toggle disassembler output/simplified view' which > >   is available in TUI mode > > > > That said, the stdio2 annotation report is quite different and so handy for my use case. > > > > > > > > I kept --stdio around because changing the output in that way could > > > annoy people used to that format. > > > > Sure. But I think the current format provides quite broken visual layout: > > > >       0.00 :   405ef1: inc    %r15 > >       0.01 :   405ef4: vfmadd213sd 0x2b9b3(%rip),%xmm0,%xmm3        # 4318b0 <_IO_stdin_used+0x8b0> > >    eff.c:1811    0.67 :   405efd: vfmadd213sd 0x2b9b2(%rip),%xmm0,%xmm3        # 4318b8 <_IO_stdin_used+0x8b8> > >           :            TA + tmpsd * (TB + > > > > vs. > > > >       0.00 :   405ef1: inc    %r15 > >       0.01 :   405ef4: vfmadd213sd 0x2b9b3(%rip),%xmm0,%xmm3        # 4318b0 <_IO_stdin_used+0x8b0> > >       0.67 :   405efd: vfmadd213sd 0x2b9b2(%rip),%xmm0,%xmm3        # 4318b8 <_IO_stdin_used+0x8b8> // eff.c:1811 > >            : 1810   TA + tmpsd * (TB + > > > > I bet also the current users of --stdio mode would benefit from it. > > What do you think? > > > > Thanks, > > Martin > > > > > > > > Please take a look at 'man perf-config' and see what can be configured > > > for both 'perf annotate --tui' and 'perf annotate --stdio2'. > > > > > > Perhaps we can do something like: > > > > > > perf config annotate.stdio=tui_like > > > > > > And, for completeness have: > > > > > > perf config annotate.stdio=classical > > > > > > wdyt? > > > > > > Looking at the other patches now. > > > > > > - Arnaldo > > > > > > > - source locations for the hottest lines are printed > > > >    at the line end in order to preserve layout > > > > > > > > Before: > > > > > > > >      0.00 :   405ef1: inc    %r15 > > > >           :            tmpsd * (TD + tmpsd * TDD))); > > > >      0.01 :   405ef4: vfmadd213sd 0x2b9b3(%rip),%xmm0,%xmm3        # 4318b0 <_IO_stdin_used+0x8b0> > > > >           :            tmpsd * (TC + > > > >   eff.c:1811    0.67 :   405efd: vfmadd213sd 0x2b9b2(%rip),%xmm0,%xmm3        # 4318b8 <_IO_stdin_used+0x8b8> > > > >           :            TA + tmpsd * (TB + > > > >      0.35 :   405f06: vfmadd213sd 0x2b9b1(%rip),%xmm0,%xmm3        # 4318c0 <_IO_stdin_used+0x8c0> > > > >           :            dumbo = > > > >   eff.c:1809    1.41 :   405f0f: vfmadd213sd 0x2b9b0(%rip),%xmm0,%xmm3        # 4318c8 <_IO_stdin_used+0x8c8> > > > >           :            sumi -= sj * tmpsd * dij2i * dumbo; > > > >   eff.c:1813    2.58 :   405f18: vmulsd %xmm3,%xmm0,%xmm0 > > > >      2.81 :   405f1c: vfnmadd213sd 0x30(%rsp),%xmm1,%xmm0 > > > >      3.78 :   405f23: vmovsd %xmm0,0x30(%rsp) > > > >           :            for (k = 0; k < lpears[i] + upears[i]; k++) { > > > >   eff.c:1761    0.90 :   405f29: cmp    %r15d,%r12d > > > > > > > > After: > > > > > > > >      0.00 :   405ef1: inc    %r15 > > > >           : 1812   tmpsd * (TD + tmpsd * TDD))); > > > >      0.01 :   405ef4: vfmadd213sd 0x2b9b3(%rip),%xmm0,%xmm3        # 4318b0 <_IO_stdin_used+0x8b0> > > > >           : 1811   tmpsd * (TC + > > > >      0.67 :   405efd: vfmadd213sd 0x2b9b2(%rip),%xmm0,%xmm3        # 4318b8 <_IO_stdin_used+0x8b8> // eff.c:1811 > > > >           : 1810   TA + tmpsd * (TB + > > > >      0.35 :   405f06: vfmadd213sd 0x2b9b1(%rip),%xmm0,%xmm3        # 4318c0 <_IO_stdin_used+0x8c0> > > > >           : 1809   dumbo = > > > >      1.41 :   405f0f: vfmadd213sd 0x2b9b0(%rip),%xmm0,%xmm3        # 4318c8 <_IO_stdin_used+0x8c8> // eff.c:1809 > > > >           : 1813   sumi -= sj * tmpsd * dij2i * dumbo; > > > >      2.58 :   405f18: vmulsd %xmm3,%xmm0,%xmm0 // eff.c:1813 > > > >      2.81 :   405f1c: vfnmadd213sd 0x30(%rsp),%xmm1,%xmm0 > > > >      3.78 :   405f23: vmovsd %xmm0,0x30(%rsp) > > > >           : 1761   for (k = 0; k < lpears[i] + upears[i]; k++) { > > > > > > > > Where e.g. '// eff.c:1811' shares the same color as the percentantage > > > > at the line beginning. > > > > > > > > Signed-off-by: Martin Liška > > > > --- > > > >   tools/perf/util/annotate.c | 30 ++++++++++++++---------------- > > > >   1 file changed, 14 insertions(+), 16 deletions(-) > > > > > > > > diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c > > > > index e60841b86d27..80542012ec1b 100644 > > > > --- a/tools/perf/util/annotate.c > > > > +++ b/tools/perf/util/annotate.c > > > > @@ -1366,7 +1366,6 @@ annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start > > > >   { > > > >       struct disasm_line *dl = container_of(al, struct disasm_line, al); > > > >       static const char *prev_line; > > > > -    static const char *prev_color; > > > >       if (al->offset != -1) { > > > >           double max_percent = 0.0; > > > > @@ -1405,20 +1404,6 @@ annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start > > > >           color = get_percent_color(max_percent); > > > > -        /* > > > > -         * Also color the filename and line if needed, with > > > > -         * the same color than the percentage. Don't print it > > > > -         * twice for close colored addr with the same filename:line > > > > -         */ > > > > -        if (al->path) { > > > > -            if (!prev_line || strcmp(prev_line, al->path) > > > > -                       || color != prev_color) { > > > > -                color_fprintf(stdout, color, " %s", al->path); > > > > -                prev_line = al->path; > > > > -                prev_color = color; > > > > -            } > > > > -        } > > > > - > > > >           for (i = 0; i < nr_percent; i++) { > > > >               struct annotation_data *data = &al->data[i]; > > > >               double percent; > > > > @@ -1439,6 +1424,19 @@ annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start > > > >           printf(" : "); > > > >           disasm_line__print(dl, start, addr_fmt_width); > > > > + > > > > +        /* > > > > +         * Also color the filename and line if needed, with > > > > +         * the same color than the percentage. Don't print it > > > > +         * twice for close colored addr with the same filename:line > > > > +         */ > > > > +        if (al->path) { > > > > +            if (!prev_line || strcmp(prev_line, al->path)) { > > > > +                color_fprintf(stdout, color, " // %s", al->path); > > > > +                prev_line = al->path; > > > > +            } > > > > +        } > > > > + > > > >           printf("\n"); > > > >       } else if (max_lines && printed >= max_lines) > > > >           return 1; > > > > @@ -1454,7 +1452,7 @@ annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start > > > >           if (!*al->line) > > > >               printf(" %*s:\n", width, " "); > > > >           else > > > > -            printf(" %*s:     %*s %s\n", width, " ", addr_fmt_width, " ", al->line); > > > > +            printf(" %*s: %-*d %s\n", width, " ", addr_fmt_width, al->line_nr, al->line); > > > >       } > > > >       return 0; > > > > -- > > > > 2.30.1 > > > > > > > > > > -- - Arnaldo