This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [arm] Use EABI unwind library
- From: Richard Earnshaw <rearnsha at arm dot com>
- To: Richard Henderson <rth at redhat dot com>
- Cc: Paul Brook <paul at codesourcery dot com>, gcc-patches <gcc-patches at gcc dot gnu dot org>, libstdc++ at gcc dot gnu dot org, Mark Mitchell <mark at codesourcery dot com>
- Date: Tue, 12 Oct 2004 14:57:05 +0100
- Subject: Re: [arm] Use EABI unwind library
- References: <200409220346.19550.paul@codesourcery.com> <20040923063203.GA3721@redhat.com> <200409231320.40588.paul@codesourcery.com> <20040923203731.GE8766@redhat.com>
On Thu, 2004-09-23 at 21:37, Richard Henderson wrote:
> On Thu, Sep 23, 2004 at 01:20:40PM +0100, Paul Brook wrote:
> > Because __cxa_call_unexpected (as defined by the EABI) requires the function
> > exception specification as an array of pointers[1] to type_into objects.
>
> I'm thinking less and less of this ARM EABI. Take an existing ABI and
> make all sorts of arbitrary changes to it. Someone should be smacked.
>
> Can you work around this by fiddling with ASM_PREFERRED_EH_DATA_FORMAT,
> such that you don't need changes to generic code?
>
> As for the rest, I think adding 15 different hooks just to satisfy one
> new ABI is dumb. We might as well admit it and just test is_arm_eabi
> all over the place instead.[1]
>
>
> r~
>
> [1] Or rip it all out and wait for the ARM folk to come to their senses.
> But of course, I'm not being paid to do the work, so I can say this. ;-)
The changes aren't arbitrary. They're there because the existing
mechanisms are too expensive for our customers.
People using IA64 may not care about code density and encoding
efficiency (I assume that they do, but it's probably not as high on
there list of concerns as other issues), but I can assure you that ARM's
customers do.
R.