Re: cause for armel nss (i.e., firefox) build segfault
On 2007-05-06 03:40 +0200, Lennert Buytenhek wrote:
> The reason that building nss (i.e., firefox) would segfault on ARM EABI
> systems is an assumption about the layout of the jmp_buf structure in
> the nspr library (which nss depends on) that does hold on old-ABI but
> no longer holds on EABI. The attached patch fixes this assumption,
> and fixes the shlibsign segfault during the building of nss (it also
> fixes a floating point byte order assumption.)
That's a fine late-night hack.
Only this week I met some ARM Corp people who very very pleased with
your efforts but wondering when they might get a browser (better than
dillo) to demo to people. I did point out that the code was there and
they could fix it themselves, rather than just waiting for someone
else to get round to it :-)
They will now be even more impressed.
> Looks sane?
> It's of course a stupid idea to depend on a particular layout of the
> jmp_buf structure in application programs, but oh well. People write
> ugly buggy code, film at 11.
Especially mozilla people. They had code that depended on the float
format too IIRC.
Principal hats: Balloonz - Toby Churchill - Aleph One - Debian