Re: Creating a hardfloat package
- To: Konstantinos Margaritis <markos@genesi-usa.com>
- Cc: "Jeremiah C. Foster" <jeremiah@jeremiahfoster.com>, debian-arm@lists.debian.org
- Subject: Re: Creating a hardfloat package
- From: Matt Sealey <matt@genesi-usa.com>
- Date: Sun, 1 May 2011 18:08:32 -0500
- Message-id: <BANLkTinbFZX9H6DbFbTQmz81j4CfehZe=g@mail.gmail.com>
- In-reply-to: <BANLkTinjQyprPzJg5p8pYt9_+nn46e+o2w@mail.gmail.com>
- References: <20110413223918.GA18457@localhost> <BANLkTinjQyprPzJg5p8pYt9_+nn46e+o2w@mail.gmail.com>
On Thu, Apr 14, 2011 at 2:32 AM, Konstantinos Margaritis
<markos@genesi-usa.com> wrote:
> On 14 April 2011 01:39, Jeremiah C. Foster <jeremiah@jeremiahfoster.com> wrote:
>> Hello!
>
> Hi,
>
> 3. For packages that fail on armhf but succeed on armel, it's highly likely that
> this is due to some thumb2 bug. Ubuntu is great help there, many packages
> have already been fixed on ubuntu-armel for thumb2 problems and most of
> the time it's just a case of backporting the thumb2 patches (as was the case
> with libmad and others), and now I'm working on backporting fixes for
> coq, blcr, openmsx. Dave Martin from ARM has done a great job in finding and
> fixing those.
Isn't there a quick fix here by building them with -marm instead of -mthumb?
While being *totally* thumb2 is a noble goal (and a required one IMO),
at the moment
none of the supported processors are thumb2-only. Yes, running Debian
on a Cortex-M4
would be wicked awesome, but not today, is basically my view on it
If it gets the rest of the system building with maybe a small
percentage -marm'd up (I
am sure a few Ubuntu packages still do this) that is a more noble goal
than blocking.
All we need is a way of marking the package as -marm'd and still needs
-mthumb'ing.
--
Matt Sealey <matt@genesi-usa.com>
Product Development Analyst, Genesi USA, Inc.
Reply to: