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;
MIME-Version: 1.0
References: <20210316050801.2446401-1-jim.cromie@gmail.com> <87ft0rvl1r.fsf@linux.intel.com>
In-Reply-To: <87ft0rvl1r.fsf@linux.intel.com>
From:   jim.cromie@gmail.com
Date:   Sat, 1 May 2021 20:16:59 -0600
Message-ID: <CAJfuBxyJHu3JdZ++0RU3DEDUS-i-4iT990P+zihFG_nKfyjVFA@mail.gmail.com>
Subject: Re: [RFC PATCH v3 00/18] dynamic debug diet plan
To:     Andi Kleen <ak@linux.intel.com>
Cc:     Jason Baron <jbaron@akamai.com>,
        Greg KH <gregkh@linuxfoundation.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Mar 18, 2021 at 10:12 PM Andi Kleen <ak@linux.intel.com> wrote:
>
> Jim Cromie <jim.cromie@gmail.com> writes:
>
> > CONFIG_DYNAMIC_DEBUG creates a struct _ddebug (56 bytes) for each
> > callsite, which includes 3 pointers to: module, filename, function.
> > These are repetetive, and compressible, this patch series goes about
> > doing that, it:
>
> So how much memory does it actually save?
> -Andi

sorry for late reply, html mode got switched on and I didnt see kickback

on my laptop/build, master has 165kb, about 70k is the compressible data,
RLE column-wize could get close to 60% on my data. so ~40kb ?

3 things to do to get the savings:
figure the compression,
figure the hash holding enabled/used/expanded pr_debug decorations
(maybe optional, depending on indexed/seek decompress time)
drop the site pointer, with some anonymous union struct combo to blend
header with callsites cleanly