This is the mail archive of the
mailing list for the GCC project.
Re: Conditional call instruction
- From: Geoff Keating <geoffk at geoffk dot org>
- To: dje at watson dot ibm dot com
- Cc: dalej at apple dot com, gcc at gcc dot gnu dot org
- Date: Wed, 13 Feb 2002 15:02:38 -0800
- Subject: Re: Conditional call instruction
- References: <200202132256.RAA23166@makai.watson.ibm.com>
- Reply-to: Geoff Keating <geoffk at redhat dot com>
> cc: Dale Johannesen <email@example.com>, firstname.lastname@example.org
> Date: Wed, 13 Feb 2002 17:56:04 -0500
> From: David Edelsohn <email@example.com>
> X-OriginalArrivalTime: 13 Feb 2002 22:56:06.0531 (UTC) FILETIME=[9EA48D30:01C1B4E1]
> >>>>> Geoff Keating writes:
> Geoff> It is unusable because the range of the call is only 32k.
> That limitation only affects the branch instructions with a target
> address. What is more useful on PowerPC are the branch condition to count
> register (bcctr) and branch conditional to link register (bclr)
> instructions which allow conditional call (at the expense of placing the
> target address in a register) and conditional return.
Conditional return we do now, so that's OK.
I'm not sure about conditional indirect calls. If they started to be
used for direct calls, they might turn out to be slower than a call
with a branch around it, because of lack of prefetching.
- Geoffrey Keating <firstname.lastname@example.org> <email@example.com>