I have a patch. It builds. It doesn't run. Previously, it ran for about thirty seconds, then silently exited 1. Now it just hangs indefinitely when started - it doesn't appear to be doing much. I'm currently trying to rebuild with gcc3 to see if it makes any difference - I doubt it will. Here as almost a placeholder. If we get this working before we put out the ia64 release, it doesn't make sense to not put it in. But it's not a 'must-fix' in the normal sense.
Is the patch a real patch or is it that libffi monstrocity?
Monstrosity, even.
It's the patch that uses libffi.
What version of Mozilla do you have it ported to? Also, what are the licensing concerns here?
I'm currently trying it on 0.9.2.1. 'ported' is a strong word. 'Made it apply' is better. :) What licensing concerns would there be?
Hm, building with gcc3 does not seem to help.
If I strace the hanging process, it's looping on: poll([{fd=11, events=POLLIN}, {fd=15, events=POLLIN}, {fd=13, events=POLLIN}], 3, 7 = 0 gettimeofday() (twice) ioctl(1,0x541b,[0]) = 0 ...
Got me. Is gdb usable?
g_main_poll, from g_main_iterate.
Sounds like it's waiting for X events.
Thats entirely possible. However, there's no window, so I'm not sure how to give it any X events. :)
We have working builds for ia64.