[prev in list] [next in list] [prev in thread] [next in thread] 

List:       haskell-jhc
Subject:    Re: [jhc] JHC's internal format conflicting with ARM instruction
From:       Simon Broadhead <sbroadhead () gmail ! com>
Date:       2011-01-07 6:09:10
Message-ID: AANLkTin-AMYXmqpMFGGx7NpKMAkCEFwayV6WCh5r=LBe () mail ! gmail ! com
[Download RAW message or body]

As far as I can tell there is no such attribute for GCC. However,
there is good news! I just managed to get it to work by adding '-marm'
to the compiler flags, which I suppose prevents it from using Thumb at
all (I had already turned off 'Compile for Thumb' in Xcode but I guess
that doesn't do the same thing). It isn't necessarily an ideal
solution but it works, and that's good enough for now. If there is a
preprocessor flag that indicates whether Thumb or ARM is being used
(which I didn't see in my cursory glance at Google), it would be easy
to have eval set the proper bits. Even if not, it would be easy enough
to add one to the build script. It would just require a bit more
diligence to make sure the flag reflects reality at all times.

Thanks,

Simon

On Fri, Jan 7, 2011 at 12:59 AM, John Meacham <john@repetae.net> wrote:
> Actually, The only functions that need tag bits are the evaluation functions
> and they are only called from the eval function, so all we need to do is
> force them all to use the same instruction set then we can just set the
> correct lower bits in eval.  If gcc can't be persuaded to do this, then a
> simple assembly stub would work.
>
> On Jan 6, 2011 9:12 PM, "John Meacham" <john@repetae.net> wrote:
>

_______________________________________________
jhc mailing list
jhc@haskell.org
http://www.haskell.org/mailman/listinfo/jhc

[prev in list] [next in list] [prev in thread] [next in thread]