This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: TARGET_STRING problem on ppc
- From: Kumar Gala <kumar dot gala at motorola dot com>
- To: Aldy Hernandez <aldyh at redhat dot com>
- Cc: David Edelsohn <dje at watson dot ibm dot com>, GCC Mailinglist <gcc at gcc dot gnu dot org>, Geoff Keating <geoffk at geoffk dot org>
- Date: Tue, 11 Mar 2003 15:58:18 -0600 (CST)
- Subject: Re: TARGET_STRING problem on ppc
David's correct that the architecture spec's having string instructions,
however it does not require the actually implement them in hardware. It
is perfectly acceptable for them the be emulated in software.
However, I think Aldy that your kludge will still be required because who
really wants to trap to the OS and emulate.
As a note, you can expect future processors from Motorola to most likely
NOT implement the string ops in hardware (so make it a generic kludge, not
just an e500 specific one).
- kumar
On Tue, 11 Mar 2003, Aldy Hernandez wrote:
>
> On Tuesday, March 11, 2003, at 12:49 PM, David Edelsohn wrote:
>
> >>>>>> Aldy Hernandez writes:
> >
> > Aldy> The e500 doesn't have string load/store instructions (though it
> > does
> > Aldy> have load/store multiple instructions). Before I kludge a
> > special case
> > Aldy> for the e500, is there a reason why MASK_STRING is set
> > regardless of if
> > Aldy> the CPU has that capability?
> >
> > Because the string instructions are not optional on PowerPC. The
> > e500 is looking less and less like a PowerPC processor. If you want to
> > disable setting it automatically it for e500, go ahead.
>
> Crap crap. Ok, my apologies to both you and Geoff, but a kludge is
> coming up...
>
> Aldy (who didn't design the chip)
>