Closed Bug 13160 Opened 25 years ago Closed 25 years ago

Solaris crashes on startup

Categories

(SeaMonkey :: Build Config, defect, P1)

Sun
Solaris

Tracking

(Not tracked)

VERIFIED FIXED

People

(Reporter: drl, Assigned: mcafee)

References

()

Details

For the last few weeks, I've been unable to invoke apprunner
successfully, it always segfaults on startup.  Environment
is Solaris 7, Ultra 5 (UltraSPARC).  I built GTK+ v1.2.4 to
see if that would help but it does not.  I'm using the nightly
builds in mozilla/nightly/latest builds from the FTP server.

Below is some runtime output;  What else can I provide that will
be of use in resolving this problem?


(drl@jethro)145% pwd
/tmp/package
(drl@jethro)146% ls
apprunner              libplc3.so             libzlib.so
chrome                 libplds3.so            mozilla-apprunner.sh
component.reg          libpng.so              mozilla-viewer.sh
components             libpref.so             netscape.cfg
defaults               libraptorgfx.so        plugins
libappcores.so         libraptorhtmlpars.so   regExport
libgfx_gtk.so          libraptorplugin.so     res
libjpeg.so             libraptorview.so       run-mozilla.sh
libjsdom.so            libraptorwebwidget.so  viewer_gtk
libmozjs.so            libsecfree.so          vreg
libmsgbaseutil.so      libwidget_gtk.so       xpidl
libnsappshell.so       libxlibrgb.so          xpt_dump
libnspr3.so            libxpcom.so            xpt_link
(drl@jethro)147% ./m*ap*
MOZILLA_FIVE_HOME=/tmp/package

LD_LIBRARY_PATH=/tmp/package:/tmp/package:/opt/pkgs/GTKglib/v1.2.4/lib:/opt/pkgs/GTKgtk/v1.2.4/lib=
      MOZ_PROGRAM=./apprunner
      MOZ_TOOLKIT=
        moz_debug=0
     moz_debugger=
Segmentation Fault - core dumped
(drl@jethro)148% truss apprunner
...
open("/tmp/package/libgfx_gtk.so", O_RDONLY)	= 11
fstat(11, 0xFFBEE4CC)				= 0
mmap(0x00000000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE, 11, 0) = 0xFEF40000
mmap(0x00000000, 1810432, PROT_READ|PROT_EXEC, MAP_PRIVATE, 11, 0) = 0xFDB00000
munmap(0xFDCA6000, 57344)			= 0
mmap(0xFDCB4000, 22004, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_FIXED,
11, 1720320) = 0xFDCB4000
close(11)					= 0
open("/dev/zero", O_RDONLY)			= 11
mmap(0x00000000, 8192, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE, 11, 0) =
0xFDF70000
mprotect(0xFDB00000, 1724786, PROT_READ|PROT_WRITE|PROT_EXEC) = 0
mprotect(0xFDB00000, 1724786, PROT_READ|PROT_EXEC) = 0
close(11)					= 0
munmap(0xFEF40000, 8192)			= 0
    Incurred fault #6, FLTBOUNDS  %pc = 0xFDC65670
      siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
    Received signal #11, SIGSEGV [caught]
      siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
sysconfig(_CONFIG_SIGRT_MIN)			= 38
sigprocmask(SIG_SETMASK, 0xFEE6D4E8, 0x00000000) = 0
sigaction(SIGSEGV, 0xFFBEE680, 0x00000000)	= 0
sysconfig(_CONFIG_SIGRT_MIN)			= 38
setcontext(0xFFBEE840)
    Incurred fault #6, FLTBOUNDS  %pc = 0xFDC65670
      siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
    Received signal #11, SIGSEGV [default]
      siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
	*** process killed ***
(drl@jethro)149%

I note that the last mmapped seems to have no symbol table info;
perhaps something is running strip inappropriately?
Assignee: don → cyeh
Severity: blocker → critical
Component: Browser-General → Build Config
Chris,

Are there known problems with the Solaris builds?  Does anything here look
suspect to you?
Yep, mcafee sez Solaris has been horked for awhile ...
*** Bug 13916 has been marked as a duplicate of this bug. ***
Severity: critical → major
Priority: P3 → P1
Summary: unable to start apprunner for at least 3 weeks on S7-64bit → Solaris crashes on startup
Target Milestone: M11
I have a dup of this bug; resummarizing, marking other bug dup.
Blocks: 8413
Blocks: 6253
Blocks: 10699
Assignee: cyeh → mcafee
viewer runs Ok, apprunner does not:

ld.so.1: ./apprunner: fatal: relocation error: file
/usr/openwin/lib/locale/iso8859-1/xomEuro.so.2: symbol _XlcCompileResourceList:
referenced symbol not found
I've just run into the _XlcCompileResourceList crash on Solaris 2.6. A quick
search of SunSolve turned up bug report #4239074, from which:

Work around:
Edit the file :
/usr/openwin/lib/locale/iso8859-1/XI18N_OBJS
Remove the line
< XOM     xomEuro         _XomGenericOpenOM       # XOM_open
Add the line
> XOM     common/xomLTRTTB        _XomGenericOpenOM # XOM_open

which fixes it for me (i.e. I can bring apprunner up once I've made this
change).

The bad news seems to be that there's currently no official fix and the file you
need to edit is owned by root.
Depends on: 17220
There seems to be multiple problems with Solaris apprunner starting up,
please file separate bugs, this bug will depend on those.
Depends on: 17221
Blocks: 5890
No longer blocks: 8413
Depends on: 8413
Looks like this fixes waterson's build, we should release-note this
for Solaris 2.6.
Target Milestone: M11 → M12
m12.
*** Bug 13202 has been marked as a duplicate of this bug. ***
QA Contact: leger → granrose
*** Bug 13202 has been marked as a duplicate of this bug. ***
Updating QA contact to a release person.  Internal Core QA does not test
Solaris.
*** Bug 16167 has been marked as a duplicate of this bug. ***
Status: NEW → RESOLVED
Closed: 25 years ago
Resolution: --- → FIXED
Fixed!
I realize that this bug has been marked closed but I still can't run the nightly
builds from ftp.mozilla.org.  I've slowly been accumulating the dependencies of
Mozilla so that I could build it myself locally and finally have everything
needed.  Using the 1999-11-28 source tarball, GTK v1.2.6, GLIB v1.2.6 and libIDL
v0.6.8 with gmake v3.76.1 and SunPro Workshop v4.2, I was able to build a
pre-M12 Mozilla locally (still under sparc64-sun-solaris7) that runs fine, if
somewhat slowly, until application exit (when there are a number of failed
assertions).  The problem with the nightly builds may simply be in the packaging
or it may be a difference between Solaris 2.5.1 and Solaris (2.)7.  Unforunately
I have no sparc64 machines that run something older than Solaris 7
Here's docs that got me going on Solaris 2.6:

  http://www.mozilla.org/unix/solaris.html

This seems to unblock at least 2.6/32-bit people.
I have built & run with WS4.2 build, I believe gcc-2.7.2.3
runs Ok too.
Status: RESOLVED → VERIFIED
verifying.
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.