Date: Tue, 18 Jan 2005 18:12:54 -0200 From: =?ISO-8859-1?Q?Jo=E3o_Carlos_Mendes_Lu=EDs?= <jonny@jonny.eng.br> To: Astrodog <astrodog@gmail.com> Cc: Andreas Klemm <andreas@freebsd.org> Subject: Re: [RFC] what to name linux 32-bit compat Message-ID: <41ED6DC6.1000604@jonny.eng.br> In-Reply-To: <2fd864e050118063747f5caa3@mail.gmail.com> References: <20050117203818.GA29131@dragon.nuxi.com> <200501172146.17965.jhb@FreeBSD.org> <20050118073612.GA10427@titan.klemm.apsfilter.org> <2fd864e050118063747f5caa3@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Astrodog wrote:
> On Tue, 18 Jan 2005 08:36:13 +0100, Andreas Klemm <andreas@freebsd.org> wrote:
>
>>On Mon, Jan 17, 2005 at 09:46:17PM -0500, John Baldwin wrote:
>>
>>>Personally, I think /compat/linux32 and /compat/linux (for linux64) would be
>>>the best way to go. The idea being that /compat/linux runs native binaries
>>>on any given arch, and if there's more than one arch supported, the
>>>non-native ones get the funky names.
>>
>>Am not 100% sure but it might be a win to re-think this for port
>>builders/designers.
>>
>>It might be beneficial not to use such "implicit" rules for naming
>>like your suggestion for taking /compat/linux as native arch.
>>
>>I would perhaps name /compat/linux32 and /compat/linux64 explicitely,
>>which might be a win and is IMHO not too expensive.
>>
>>Uname should IMHO get a new switch to print out default architecture
>>of being 32 or 64 bit.
>>
>>So together with uname and the above naming scheme you have all you need
>>and is compareable to what we already have (concerning uname) for
>>cpu architecture.
>>
>>Do perhaps other BSD teams have added an mechanism like that or
>>do they get 32/4 bit out of /proc ?
>>
>>Best regards
>>
>> Andreas ///
>
>
> Not sure if it got lost, so I'm gonna summarize what I said earlier,
> Why not have /compat/linux32, and /compat/linux64 (For things that
> require one or the other), then just have /compat/linux linked to the
> native setup for the machine? That gives the ease of /compat/linux for
> the native stuff, without causing the problems Andreas pointed out.
Thats exactly what I was going to write now. Symlink is the solution!
> Also allows people with "clean" 64-bit friendly code to just use
> /compat/linux on AMD64 or i386, and have things work in whatever way
> is native to the machine.
>
> This could also be extended to other archs, if that ever becomes an
> issue. (linuxppc, linuxsparc64, whatever), with linux still pointing
> to native.
--
João Carlos Mendes Luís
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41ED6DC6.1000604>