Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
        id S1751458AbdIOR0a (ORCPT <rfc822;w@1wt.eu>);
        Fri, 15 Sep 2017 13:26:30 -0400
Received: from mail.kernel.org ([198.145.29.99]:45266 "EHLO mail.kernel.org"
        rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
        id S1751197AbdIOR03 (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
        Fri, 15 Sep 2017 13:26:29 -0400
DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 60D2821BCE
Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=kernel.org
Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=acme@kernel.org
Date: Fri, 15 Sep 2017 14:26:26 -0300
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: "Liang, Kan" <kan.liang@intel.com>
Cc: "peterz@infradead.org" <peterz@infradead.org>,
        "mingo@redhat.com" <mingo@redhat.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "jolsa@kernel.org" <jolsa@kernel.org>,
        "namhyung@kernel.org" <namhyung@kernel.org>,
        "Hunter, Adrian" <adrian.hunter@intel.com>,
        "Odzioba, Lukasz" <lukasz.odzioba@intel.com>,
        "ak@linux.intel.com" <ak@linux.intel.com>
Subject: Re: [PATCH RFC V2 00/10] perf top optimization
Message-ID: <20170915172625.GA14469@kernel.org>
References: <1505096603-215017-1-git-send-email-kan.liang@intel.com>
 <20170913152506.GK5866@kernel.org>
 <37D7C6CF3E00A74B8858931C1DB2F077537AF31A@SHSMSX103.ccr.corp.intel.com>
 <20170913153819.GL5866@kernel.org>
 <20170914211946.GB10371@kernel.org>
 <37D7C6CF3E00A74B8858931C1DB2F077537BF0A8@SHSMSX103.ccr.corp.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <37D7C6CF3E00A74B8858931C1DB2F077537BF0A8@SHSMSX103.ccr.corp.intel.com>
X-Url: http://acmel.wordpress.com
User-Agent: Mutt/1.8.3 (2017-05-23)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Content-Length: 2982
Lines: 78

Em Fri, Sep 15, 2017 at 03:11:51PM +0000, Liang, Kan escreveu:
> > Em Wed, Sep 13, 2017 at 12:38:19PM -0300, Arnaldo Carvalho de Melo
> > escreveu:
> > > Em Wed, Sep 13, 2017 at 03:29:44PM +0000, Liang, Kan escreveu:
> > > > >
> > > > > Em Sun, Sep 10, 2017 at 07:23:13PM -0700, kan.liang@intel.com
> > escreveu:
> > > > >
> > > > > So I got the first two patches already merged, and made some
> > > > > comments about the other patches, please check those,
> > > > >
> > > >
> > > > Thanks for the review Arnaldo.
> > > >
> > > > I will take a close look for the comments.
> > > > For the next version, I only need to include patch 3-10, correct?
> > >
> > > Right, and go from my perf/core branch. The hashtable patch is still
> > > not there as I am running tests before pushing out, but it should be
> > > there later today.
> > 
> > So, its at my repo, branch tmp.perf/threads_hashtable
> > 
> > But 'perf trace' is broken, please take a look below:
> > 
> > [root@jouet ~]# gdb -c core
> > GNU gdb (GDB) Fedora 8.0-20.fc26
> > <SNIP>
> > Core was generated by `perf trace -e block:block_bio_queue'.
> > Program terminated with signal SIGSEGV, Segmentation fault.
> > #0  0x000000000051089a in ?? ()
> > (gdb) file perf
> > Reading symbols from perf...done.
> > (gdb) bt
> > #0  0x000000000051089a in ____machine__findnew_thread
> > (machine=0x3dfcab0, threads=0x3dfca78, pid=-1, tid=-1, create=false) at
> > util/machine.c:429
> 
> I think the root cause is tid==-1. So the index of hashtable will be -1.
> The patch as below should fix it.
> 
> diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h
> index e6d5381..3c564b8 100644
> --- a/tools/perf/util/machine.h
> +++ b/tools/perf/util/machine.h
> @@ -57,7 +57,7 @@ struct machine {
>  
>  static inline struct threads *machine__threads(struct machine *machine, pid_t tid)
>  {
> -	return &machine->threads[tid % THREADS__TABLE_SIZE];
> +	return &machine->threads[(unsigned int)tid % THREADS__TABLE_SIZE];
>  }
>  
>  static inline
> 
> 
> There should be another issue which was introduced by  
> 33013b9a5607 ("perf machine: Optimize a bit the machine__findnew_thread() methods")
> It should use tid not pid to get the threads.
> 
> diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
> index 90ae9c7..ddeea05 100644
> --- a/tools/perf/util/machine.c
> +++ b/tools/perf/util/machine.c
> @@ -473,7 +473,7 @@ static struct thread *____machine__findnew_thread(struct machine *machine,
>  
>  struct thread *__machine__findnew_thread(struct machine *machine, pid_t pid, pid_t tid)
>  {
> -	return ____machine__findnew_thread(machine, machine__threads(machine, pid), pid, tid, true);
> +	return ____machine__findnew_thread(machine, machine__threads(machine, tid), pid, tid, true);
>  }
> 
> They are small fixes. I think it's better to merge them with the old patches.
> Should I include the modified hashtable patches in V3?

I'll add these now and test, then push another branch, ok?

- Arnaldo