linux-mips-fnet
[Top] [All Lists]

Re: Dead function optimisation Call for Testers

To: ralf@uni-koblenz.de (Ralf Baechle)
Subject: Re: Dead function optimisation Call for Testers
From: greyham@research.canon.com.au (Graham Stoney)
Date: Tue, 9 May 2000 10:58:36 +1000 (EST)
Cc: greyham@research.canon.com.au (Graham Stoney), linux@engr.sgi.com, linux-mips@fnet.fr, linux-mips@vger.rutgers.edu
In-reply-to: <20000509003437.A19318@uni-koblenz.de> from "Ralf Baechle" at May 09, 2000 12:34:37 AM
Hi Ralf,

Ralf Baechle writes:
> does you optimization still guarantee that the order of functions in the
> C source is the order of functions in the final executable?  Linux/MIPS
> relies on that.

Yes and No. It trips up a bug in the linker which reverses the order of the
functions in each file. This caused grief at first because the exception table
used to trap bad user space accesses got out of order. I had to put the
exception table entries in seperate sections which also get reordered
accordingly to work around it.

The latest binutils snapshots have the bug fixed, and will preserve the order
of the functions. Could you please elaborate on where Linux/MIPS relies on
this?

Thanks,
Graham
-- 
Graham Stoney
Principal Hardware/Software Engineer
Canon Information Systems Research Australia
Ph: +61 2 9805 2909  Fax: +61 2 9805 2929

<Prev in Thread] Current Thread [Next in Thread>