40950 - Need xptcall port for linux/ia64

Status: RESOLVED FIXED

(Core :: XPCOM, defect, P3)

Description

[jim_nance]

I am attempting to get mozilla going on linux/ia64.  So far its been fairly
simple to get everything working except the xptcall stuff.  This requires
someone who knows both ia64 assembler and the internals of g++.  There are
probably about 10 people in the world who fit that description.  I figure
half of them work at cygnus.  I am hoping that Chris knows who they are
and can convince them they want to port this code, which is why I assigned
this bug to him.

Comment 1

[Christopher Blizzard (:blizzard)]

The rumor is that there is a lot left to be desired with C++ support on ia64. 
I'll ask around though.  Maybe someone at Intel already has one done.

Comment 2

[Christopher Blizzard (:blizzard)]

wtc seems to be on this.

Comment 3

[Wan-Teh Chang]

Reassigned to [e-mail].

Comment 4

[Wan-Teh Chang]

Sorry, I forgot that it is possible to assign
a bug to the owner of the selected component.

Comment 5

[John Bandhauer]

Assigning this to me is pretty much pointless. We rely on volunteers who are 
interested and capable of porting to a give platform to provide these platform 
ports. I'll add this platform to the list on the xptcall status page - 
http://lxr.mozilla.org/seamonkey/source/xpcom/reflect/xptcall/status.html - 
but other than that I don't have a plan for action on this. Anyone who wants to 
see this port happen should spread the word and find the person to do it.

Comment 6

[jim_nance]

Yes, please add it to the status page.  Having it up there may help us
find someone to do this port.

Comment 7

[John Bandhauer]

I continue my quest to concentrate 'nix xptcall bugs in the shaver camp.

Comment 8

[Mike Shaver (:shaver emeritus)]

I have a "volunteer" for this, and we might get it in time for 0.8.1.

Comment 9

[Mike Shaver (:shaver emeritus)]

My volunteer bailed.  Sometime I'll try and use the simulator environment to
hack it up myself.

Comment 10

[Bill Nottingham]

David Mosberger got something working using libffi. Once I actually get
a build that starts, I'll post the patch.

Comment 11

[Mike Shaver (:shaver emeritus)]

Hey Bill, did you get it working?  We could use the libffi port to ease our gcc3
woes on other platforms as well...

Comment 12

[David Bradley]

Might be interested in bug 82286 I'll cross post this to that bug. May encounter
similar issues.

Comment 13

[Bill Nottingham]

Not yet. I can dig up the patch and post it. Basically, the behavior I'm seeing is
that mozilla on startup grinds for about 20 seconds or so, and then silently
exits 1.

Comment 14

[Martin Lawyer]

http://bugzilla.mozilla.org/show_bug.cgi?id=146722

Bug 146722 addresses this issue for HPUX on ia64. 

Comment 15

[Christopher Blizzard (:blizzard)]

Attachment: patch for linux

This is a patch that Ulrich sent me and seems to work on Linux.  It's derived
from the HPUX code, so the copyrights are intact in the file.  I already have
an rs=shaver on this, so I'll check it in shortly.

Comment 16

[Brendan Eich [:brendan]]

Comment on attachment 125960 [details] [diff] [review]
patch for linux

Just for the record.

/be

Comment 17

[Christopher Blizzard (:blizzard)]

Checked in on the tip and the 1.4 branch.  Yay!

Comment 18

[Olli Pettay [:smaug][[e-mail]]]

Someone should have updated http://lxr.mozilla.org/seamonkey/source/xpcom/reflect/xptcall/status.html

Comment 19

[Wan-Teh Chang]

Attachment: Patch for status.html

Update status.html with the Linux ia64 status.

I determined the contributors of the Linux ia64 xptcall
patch based on comment 15 and comment 14.

Comment 20

[Mike Shaver (:shaver emeritus)]

Comment on attachment 207430 [details] [diff] [review]
Patch for status.html

Approved (and implicitly in the future, Wan-Teh, should you want to make further updates).

Comment 21

[Wan-Teh Chang]

Comment on attachment 207430 [details] [diff] [review]
Patch for status.html

I updated status.html on the Mozilla trunk (1.9 alpha).

Checking in status.html;
/cvsroot/mozilla/xpcom/reflect/xptcall/status.html,v  <--  status.html
new revision: 1.29; previous revision: 1.28
done