49036 - Segmentation Fault during mozilla startup in OpenBSD (aparently in libcaps)

Status: VERIFIED DUPLICATE of bug 44426

(Core :: Security: CAPS, defect, P3)

Description

[Zach Garner]

From Bugzilla Helper:
User-Agent: Mozilla/4.7 [en] (X11; U; OpenBSD 2.7 i386; Nav)
BuildID:    

As can be seen from the output below, mozilla core dumps with a segmentation
fault from a call to strlen from libcaps.

Reproducible: Always
Steps to Reproduce:
1. Build Mozilla on OpenBSD with standard configure options.
2. Try to run it.

Actual Results:  
trotsky$ ./mozilla-bin 
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: autoregistering succeeded
Segmentation fault (core dumped)
trotsky$ 

Expected Results:  							
It Should run, successfully  :)

From running Mozilla in GDB:
(gdb) r
Starting program: /usr/home/zach/mozilla/dist/bin/mozilla-bin 
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: autoregistering succeeded

Program received signal SIGSEGV, Segmentation fault.
0x405de601 in strlen ()
(gdb) bt
#0  0x405de601 in strlen ()
#1  0x40da5484 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libcaps.so.1.0
#2  0x4079f188 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libxpconnect.so.1.0
#3  0x407f8dc5 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libxpconnect.so.1.0
#4  0x4080024f in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libxpconnect.so.1.0
#5  0x40801a3d in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libxpconnect.so.1.0
#6  0x402edd8f in js_ValueToString (cx=0x2b800, v=0) at jsstr.c:2286
#7  0x4029ead3 in js_ReportUncaughtException (cx=0x2b800) at jsexn.c:645
#8  0x4027ca7a in JS_ExecuteScript (cx=0x2b800, obj=0xb39f0, script=0xfb800, 
    rval=0xdfbfcf60) at jsapi.c:2668
#9  0x40d43ba8 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#10 0x40d43198 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#11 0x40d42786 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#12 0x40d423d7 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#13 0x40d41a09 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#14 0x40d41764 in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libjsloader.so.1.0
#15 0x40171c30 in AutoRegister_enumerate (key=0xdbc00, aData=0x117580, 
    aClosure=0xdfbfd6b8) at nsComponentManager.cpp:1961
#16 0x400c4319 in _hashEnumerate (he=0xf9790, i=0, arg=0xdfbfd4d4)
    at nsHashtable.cpp:99
#17 0x4032431c in PL_HashTableEnumerateEntries (ht=0x293c0, 
    f=0x400c42e0 <_hashEnumerate(PLHashEntry *, int, void *)>, arg=0xdfbfd4d4)
    at plhash.c:413
#18 0x400c494b in nsHashtable::Enumerate (this=0x2c3d0, 
    aEnumFunc=0x40171ba0 <AutoRegister_enumerate(nsHashKey *, void *, void *)>,
closure=0xdfbfd6b8) at nsHashtable.cpp:237
#19 0x400c811d in nsSupportsHashtable::Enumerate (this=0x2c3d0, 
    aEnumFunc=0x40171ba0 <AutoRegister_enumerate(nsHashKey *, void *, void *)>,
closure=0xdfbfd6b8) at nsHashtable.h:135
#20 0x40172a59 in nsComponentManagerImpl::AutoRegister (this=0x2e880, when=0, 
    inDirSpec=0x0) at nsComponentManager.cpp:2081
#21 0x40195f60 in nsComponentManager::AutoRegister (when=0, directory=0x0)
    at nsRepository.cpp:200
#22 0x40c8691c in ?? ()
   from /usr/home/zach/mozilla/dist/bin/components/libxpinstall.so.1.0
#23 0x506a in main1 (argc=1, argv=0xdfbfdb80, nativeApp=0x0)
    at nsAppRunner.cpp:786
#24 0x6a42 in main (argc=1, argv=0xdfbfdb80) at nsAppRunner.cpp:1094

------------------------------------------------------------

Also, I'm running OpenBSD 2.7 on Pentium II with 160meg of ram. I've heard
similar complaints about running mozilla on the openbsd-misc mailling list, but
couldnt find a bug report. I've tried, and received the same results, with
Milestone 17. This report is based off of a CVS checkout on Aug 14, 2000.

Comment 1

[Zach Garner]

If i break just before line 792 where this is declared:
  PRBool needAutoreg = PR_TRUE;
I assume that I should be able to do a "print PR_TRUE" in gdb and get its value,
but instead I get "No symbol "PR_TRUE" in current context."

This is around the area (within nsAppRunner, at least) that ends up segfaulting,
so maybe that has something to do with it (and there is a call to
su->StartupTasks(&needAutoreg); right after needAutoreg is defined). 

Comment 2

[Zach Garner]

Sorry, that was line 792 of nsAppRunner.cpp.

Comment 3

[Zach Garner]

Ok, me again. I do not know how helpfull it is but here is some more output from
GDB:
(gdb) break nsAppRunner.cpp:786
Breakpoint 3 at 0x5000: file nsAppRunner.cpp, line 786.
(gdb) r
Starting program: /usr/home/zach/mozilla/dist/bin/mozilla-bin 

Breakpoint 3, main1 (argc=1, argv=0xdfbfdb00, nativeApp=0x0) at
nsAppRunner.cpp:786
(gdb) step
nsCOMPtr<nsISoftwareUpdate>::operator-> (this=0xdfbfd9bc) at
../../dist/include/nsCOMPtr.h:649
(gdb) step
(gdb) step
nsCOMPtr<nsISoftwareUpdate>::get (this=0xdfbfd9bc) at
../../dist/include/nsCOMPtr.h:631
(gdb) step
(gdb) step
nsCOMPtr<nsISoftwareUpdate>::operator-> (this=0xdfbfd9bc) at
../../dist/include/nsCOMPtr.h:650
(gdb) step
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: autoregistering succeeded

Program received signal SIGSEGV, Segmentation fault.
0x405de601 in strlen ()
(gdb) 

Comment 4

[Zach Garner]

Ok, i've finally figured out that this is a duplicate of 44426.

Comment 5

[Zach Garner]

*** This bug has been marked as a duplicate of 44426 ***

Comment 6

[Scott Brodmerkle]

Verified dupe.

Comment 7

[David Krause]

Sorry, for the spam.  Setting OS to the OpenBSD field.