X-Authentication-Warning: ieva01.lanet.lv: pavenis owned process doing -bs
Date: Tue, 6 Jul 1999 21:58:01 +0300 (WET)
From: Andris Pavenis <pavenis AT lanet DOT lv>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: salvador <salvador AT inti DOT gov DOT ar>, djgpp-workers AT delorie DOT com,
        Robert Hoehne <robert DOT hoehne AT gmx DOT net>
Subject: Re: .align directives in libc.a
In-Reply-To: <Pine.SUN.3.91.990706160240.28849A-100000@is>
Message-ID: <Pine.A41.4.05.9907062151210.57466-100000@ieva01.lanet.lv>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk



On Tue, 6 Jul 1999, Eli Zaretskii wrote:

> 
> On Tue, 6 Jul 1999, salvador wrote:
> 
> > > The alignment pseudo-ops allow to define the value of the byte with
> > > which to pad.  Thus, ".p2align 3,0xcc" should pad to 8-byte boundary
> > > with Int 3 instructions.
> > >
> > > But why is this a good idea?
> > 
> > To trap jumps in this space.
> 
> Can GCC be configured or told at run time which filler to use for 
> alignment?
> 
> For that matter, can it be told to emit a smarter align directive, like 
> ".balign 16,0xcc,7"?
> 

It's not so simple. Normally gcc (cc1, cc1plus) does not specify filler
character (eg. .p2align 2) but we had to use .p2align 2,0 for 
.ehframe (or otherwise exception suppoer would break). That should
be normally fixed in binutils but that can be done for DJGPP with
development snapshots only which is not acceptable for ordinary user.

If we specify filler in target related config. files we should make
sure filler will be used only where needed. That would require much
grepping gcc sources, and more likely modifications to cc1, cc1plus etc

And perhaps also fixing newly introduced bugs. I would like to stay off
these things

Andris