61970 - OpenBSD non-debug build fails to open a window.

Status: VERIFIED DUPLICATE of bug 124958

(SeaMonkey :: Build Config, defect, P3)

Description

[David Krause]

OpenBSD 2.8-current (DAVID) #2: Mon Nov 27 20:34:12 CST 2000

hermes:david {131} uname -a
OpenBSD hermes 2.8 DAVID#2 i386

hermes:david {127} gcc -v
Reading specs from /usr/lib/gcc-lib/i386-unknown-openbsd2.8/2.95.3/specs
gcc version 2.95.3 19991030 (prerelease)

I'm trying to get the latest mozilla cvs code working on OpenBSD-Current. 
Previously it would crash on startup (bug 49575).  The last few days it has just
been printing some debug info and exits gracefully.  I'm using an unpatched
source tree except for the couple of changes needed in bug 60730.

hermes:bin {125} ./mozilla
./run-mozilla.sh ./mozilla-bin
MOZILLA_FIVE_HOME=.
  LD_LIBRARY_PATH=.
     LIBRARY_PATH=.:./components
       SHLIB_PATH=.
          LIBPATH=.
       ADDON_PATH=.
      MOZ_PROGRAM=./mozilla-bin
      MOZ_TOOLKIT=
        moz_debug=0
     moz_debugger=
hermes:bin {126} ls -al ~/.mozilla
total 4
drwx------   2 david  david   512 Dec  5 00:36 .
drwxr-xr-x  18 david  david  1024 Dec  5 00:52 ..
hermes:bin {127} ls -al component.reg
-rw-r--r--  1 david  david  89591 Dec  5 00:37 component.reg
hermes:bin {128}

As you can see it looks like it modified the component.reg, but nothing was
created in the profile dir.  I'm CCing David Baron in hope that he will know
what is wrong.  He's been really helpful so far. Thanks to everybody that helped
so far.  I will attach a ktrace and kdump output (similiar to strace on linux).

Comment 1

[David Krause]

Here is the ktrace/kdump output.  This file is really huge so I put it up on a
website:  http://test.netcentral.com/daviddrop/kdump.out  I'm looking through it
now trying to find an answer.  Any help is appreciated.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Since OpenBSD has serious problems with static constructors (does it ever run
them?), I'd have to think that they're a likely cause (see bug 49575).  They're
against the C++ portability guidelines, but people still use them occasionally.
 I've been trying to get rid of them slowly, but it can be rather painful at
times to figure out how to fix them.  Considering how the number still in the
code, I'd be surprised if you were able to run on OpenBSD right now, although
it's not all *that* hard to fix all of the ones encountered while starting up
(perhaps a day or two of work to fix them in a reasonable way, although it would
be faster to just fix all of them by leaking things right and left).

It would also be useful to see the debug info that it prints before failing.
Also, aren't you still running with patches from, e.g., bug 49575?

Comment 3

[David Krause]

I was running with the patches on bug 49575, but then it stopped crashing and
just exited on startup.  So, I removed all the patches applied from bug 49575 to
see if that had anything to do with it, and it still does the same thing.

The only debug info that it prints before failing is what I have below.
Basically just a listing of the environment variables.  Then it runs for about
30 sec printing nothing and then exits.

Comment 4

[cls]

Setting milestones to Future.

Comment 5

[hacker formerly known as [e-mail]]

*** This bug has been marked as a duplicate of 124958 ***

Comment 6

[Jon Granrose]

verified dup.