[luatex] luatex segfault on mips64el hardware

luigi scarso luigi.scarso at gmail.com
Tue Feb 21 13:24:16 CET 2012

On Tue, Feb 21, 2012 at 4:40 AM, Mojca Miklavec
<mojca.miklavec.lists at gmail.com> wrote:
> On Mon, Feb 20, 2012 at 20:59, Kip Warner wrote:
>> On Mon, 2012-02-20 at 20:10 -0600, Mojca Miklavec wrote:
>>> You could have also asked me. I already have all the necessary
>>> binaries (natively compiled on Kip's machine). It might be that Karl
>>> will eventually put them online on TUG's ftp server as well (or maybe
>>> he did that already).
>> Hey Luigi and Mojca. It might actually be a good idea if Mojca takes a
>> stab at it since she already has a shell account on the machine in
>> question and has experience with it.
ok for me
> I don't know how to reproduce the error though.
> When I said "you could have asked me" I mostly meant providing set of
> binaries, also because Luigi might be a lot better in bug hunting than
> I am.
uh it depends to the bug...

Anyway the overall idea is to setup a virtual machine that is as much
as possible similar to the target one,
if the last one is not at hand.

For example the qemu 0.9.1 on my Ubuntu LTS  doesn't support Loongson,
while the latest 1.0.1 has some kind of support -- it doesn't work on
my machine, anyway; the daily images from
frozen the cpu on 100% .
Usually Debian offers a quick way to setup a vm: either download the
netinst from
~180MByte or if the connection is ok use a netinst
Anyway, with a modern adsl updating a vm is not a problem.

The problem for luatex is to build the executable for the target machine.
The preferred way is of course use gcc on the target machine, but the
slowdown can be 10 times.
It is preferable to build a cross-toolchain on host, e.g
or check if the binaries from
are ok
or whatever else (I e.g

Another example is gub
useful to cross-compile luatex for MacOs and Windows
Puredarwin offers a vm for macos
but until now I've had no time to test it.


More information about the luatex mailing list