ffox_A3509 trouble

[Torsten.]

Did you get the "latest" version from here?

http://www.hp.com/go/firefox

==> https://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=HPUXJAVAFFTB

[Alex Smirnoff]

May 2010, yes. I downloaded it just a few weeks ago, so the version is definitely latest one. Should i provide a list of system patches applied?

[Dennis Handly]

>I inserted "ldd $prog" just before executing binary in run-mozilla.sh

There were no ldd errors, did this get one of your signal 10 or 11?

>Anything I can do to diagnose it further?

Use gdb and your core file to get a stack trace.

gdb /opt/firefox/firefox-bin core
(gdb) bt
(gdb) info reg
(gdb) info shared
(gdb) disas $pc-4*16 $pc+4*4
(gdb) q

[Alex Smirnoff]

Here it is. Backtrace looks similar if i start it via "two-line" method recommended above or just via /opt/firefox/firefox shell script -- the crash is still in the same module and offset.

[Dennis Handly]

#3
#4 0xc6dc0130 in FcStrListCreate+0x58

Unfortunately firefox had a signal handler so you need to redo some steps using the registers and PC from frame 4:
(gdb) frame 4
(gdb) info reg
(gdb) disas 0xc6dc0130-4*16 0xc6dc0130+4*4
(gdb) q

[Alex Smirnoff]

(gdb) frame 4
#4 0xc6bf0130 in FcStrListCreate+0x58 () from /opt/firefox/depend/lib//libfontconfig.sl
(gdb) info reg
flags: 2f000041
r1: 7afb9930 rp/r2: c6bf0123
r3: 40887e78 r4: c6bec7d8
r5: 405a92a0 r6: f
r7: 7f7f3230 r8: 405a9328
r9: 40000000 r10: 13b0
r11: 0 r12: c
r13: 1 r14: 100000
r15: 0 r16: 8001084
r17: 40015bd8 r18: 40025598
r19: 7afba130 r20: 7afbc0fc
r21: c6be5600 r22: 0
arg3/r23: 8e794d18 arg2/r24: f0fff00
arg1/r25: 8 arg0/r26: e
dp/gp/r27: 40001248 ret0/r28: 40886c90
ret1/ap/r29: 0 sp/r30: 7f7f3470
mrp/r31: 40886c90 sar/cr11: 19
pcoqh: c6bf0133 pcsqh: b4ad800
pcoqt: c6bf0137 pcsqt: b4ad800
eiem/cr15: ffffffff iir/cr19: 4af80000
isr/cr20: b4ad800 ior/cr21: 8e794d18
ipsw/cr22: 4ff0f goto: 1
sr4: 9a8c000 sr0: b4ad800
sr1: b4ad800 sr2: 0
sr3: 0 sr5: a9f3000
sr6: b4ad800 sr7: b4ad800
rctr/cr0: 24 pidr1/cr8: ffffffff
pidr2/cr9: 801 ccr/cr10: 381d0
pidr3/cr12: 1 pidr4/cr13: 35720
cr24: 8 cr25: 0
cr26: ad35f0 mpsfu_high: 7aa25048
mpsfu_low: c5cab8 mpsfu_ovflo: 0
pad: bcf050 fpsr: 88340000
fpe1: 0 fpe2: 0
fpe3: 0 fpe4: 0
fpe5: 0 fpe6: 0
fpe7: 0
(gdb) disas 0xc6bf0130-4*16 0xc6bf0130+4*4
Dump of assembler code from 0xc6bf00f0 to 0xc6bf0140:
0xc6bf00f0 : ldw -0x20(%sp),%r19
0xc6bf00f4 :
cmpb,<> %r0,%ret0,0xc6bf0110
0xc6bf00f8 : copy %ret0,%r31
0xc6bf00fc : ldi 0,%r31
0xc6bf0100 : ldw -0x54(%sp),%rp
0xc6bf0104 : copy %r31,%ret0
0xc6bf0108 : ret
0xc6bf010c : ldo -0x40(%sp),%sp
0xc6bf0110 : stw %r31,-0x40(%sp)
0xc6bf0114 : ldi 0xe,%r26
0xc6bf0118 : call 0xc6bd6740
0xc6bf011c : ldi 8,%r25
0xc6bf0120 : ldw -0x20(%sp),%r19
0xc6bf0124 : ldw -0x40(%sp),%r31
0xc6bf0128 : ldw -0x38(%sp),%r23
0xc6bf012c : stw %r23,0(%r31)
0xc6bf0130 : ldw 0(%r23),%r24
0xc6bf0134 : ldo 1(%r24),%r25
0xc6bf0138 : stw %r25,0(%r23)
0xc6bf013c : b 0xc6bf0100
End of assembler dump.
(gdb)

[Dennis Handly · Accepted solution]

r23: 8e794d18
0xc6bf0130 : ldw 0(%r23),%r24

It looks like r23 has a bad pointer.

In talking with the lab, they requested the following:
1) the profile could be corrupted. We could validate this by creating a temporary home and try it out.
$ mkdir /tmp/firefox_home
$ export HOME=/tmp/firefox_home
$ opt/firefox/firefox

2) Complete patch details on the system.

3) Make sure firefox is installed properly on the system. Requires the following output (/tmp/fire_dump) to validate this.
$ for i in $(find /opt/firefox); do
$ what $i >> /tmp/file_dump
$ done

[Alex Smirnoff]

1) Excellent! It really worked (had to copy X authority file as well, though).

So we nailed it down. I tried to mv out .mozilla/firefox directory before, but it seems to be not enough as it did not help -- unlike creating new home. How do i find other files that may cause FF to crash?

[Alex Smirnoff]

..found it. clearing ~/.fontconfig resolved the issue completely. Quite strange, isn't it?

[Dennis Handly]

>found it. Quite strange, isn't it?

I'm glad we were able to help.