Date:   Fri, 17 Mar 2023 19:53:05 +0300
From:   Alexey Dobriyan <adobriyan@gmail.com>
To:     Bagas Sanjaya <bagasdotme@gmail.com>
Cc:     Randy Dunlap <rdunlap@infradead.org>, akpm@linux-foundation.org,
        linux-kernel@vger.kernel.org,
        Linux Doc Mailing List <linux-doc@vger.kernel.org>,
        Jonathan Corbet <corbet@lwn.net>
Subject: Re: [PATCH] ELF: document some de-facto PT_* ABI quirks
Message-ID: <cc449f31-071c-48b8-a6a8-7b91f205c3c6@p183>
References: <2acb586c-08a9-42d9-a41e-7986cc1383ea@p183>
 <e262ea00-a027-9073-812e-7e034d75e718@infradead.org>
 <ZBQjRLiXOwfmoIs+@debian.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <ZBQjRLiXOwfmoIs+@debian.me>
Precedence: bulk

On Fri, Mar 17, 2023 at 03:22:28PM +0700, Bagas Sanjaya wrote:
> On Tue, Mar 14, 2023 at 07:34:11PM -0700, Randy Dunlap wrote:
> > Hi,
> > 
> > [adding linux-doc for other interested parties]
> 
> Unfortunately akpm had already applied this doc as 60b38b7cbb295d ("ELF:
> document some de-facto PT_* ABI quirks") while it being reviewed and
> doesn't have any consensus yet.
> 
> > And could the document have a title, like:
> > 
> > =========================
> > ELF header usage in Linux
> > =========================
> 
> The current doc path is Documentation/ELF/ELF.rst, which means that
> readers expect to find general info about the executable format, not
> some sort of trivia/niche like this.

General info is in ELF spec. This document is intended to be Linux
specific stuff you won't find anywhere but source.

I'll write down overmapping rules as well.

> > (I just made that up. Feel free to change it. :)
> > 
> > Also, the .rst file should be added to some chapter in the current
> > documentation tree, such as under "Other documentation", so add this file name
> > to Documentation/staging/index.rst. In fact this file could live in
> > Documentation/staging instead of in Documentation/ELF/ (IMO of course).
> 
> If there are more ELF docs there then a separate directory may be
> warranted.

This is codification of what Linux has been doing for years:
e.g pre 2.4.11 executables with multiple PT_INTERP segments were
rejected.