11.23 (PA-RISC) (64bit) : Need help Hitting with some bug

[wipro_1]

Dear All,

Our Oracle Team is hitting with a bug
ORA-07445: exception encountered: core dump [11] [4286564440] [240] [0] [] []
The corresponding trace file will include the following in the stack:
_doprnt
_fprintf
U_get_previous_frame
skdstgframe
Error from U_get_previous_frame_x is 1
Stack is not Windable
Stack has no Unwind_descriptor
+++++

Oracle support says, not getting the call stack is HP UX problem. Involve your System Administrator. They reported similar bug/problem which was encountered in earlier release of HP UX 11.10 version and was fixed by HP patch number : PHSS_16849: s700_800 11.0 LIBCL,

But currently we've equivalent patch in hp-ux 11.23.Patch number is PHSS_37958.See the patch details below.
Patch Name: PHSS_37958
Patch Description: s700_800 11.23 LIBCL patch
Creation Date: 08/08/04
Post Date: 08/08/13
Hardware Platforms - OS Releases:
s700: 11.23
s800: 11.23
Products: N/A
Filesets:
OS-Core.CORE-SHLIBS,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,v=HP
ProgSupport.LANG-64ALIB,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP
OS-Core.CORE-64SLIB,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,v=HP
ProgSupport.LANG-MIN,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP
+++++++++++++++++++++++++++++++++++++++++++

Please suggest a solution in this regards.

Thanks in Advance.

[Dennis Handly]

What program is running? Is it with threads?
If so, you could have a thread stack overflow.
You shouldn't see U_get_previous_frame in a stack trace unless there was a recursive signal. You could try getting a stack trace from the core file with gdb.

In any case, Oracle must debug their own problem.

PHSS_37958 is the latest libcl patch. It should fix the problem Oracle found with sigaltstack(2).

[wipro_1]

Thanks Denny, here is the stack which I got from our Oracle Team

----- Call Stack Trace -----
calling call entry argument values in hex
location type point (? means dubious value)
-------------------- -------- -------------------- ----------------------------
Error from U_get_previous_frame_x is 1
Stack is not Windable
Stack has no Unwind_descriptor
sdsnm: dlmodinfo failed for address : 0xc000000069f4eaa8
ksedmp()+184 ? ksedst() 4000000001978CFF ?
000000002 ? 000000000 ?
C000000070FEC660 ?
ksfdmp()+32 ? ksedmp() 40000000016B3857 ?
00000000F ? 000000000 ?
000000000 ?
kgeriv()+152 ? ksfdmp() 000000000 ? 000000000 ?
000000000 ?
8000000100008328 ?
kgeasi()+172 ? kgeriv() 8000000100008328 ?
C000000065DA3C90 ?
8000000100149380 ?
40000000016070BF ?
kspgsp()+388 ? kgeasi() 800003FFC0000E98 ?
00013CFA8 ?
80000001001497B8 ?
000000020 ?
kjcvcgc()+880 ? kspgsp() 000000000 ?
C0000000A1A51160 ?
80000001001493C0 ?
C000000064E6FD90 ?
0xC000000069F4EAA8 ? kjcvcgc() C000000065DA3D10 ?
4100000000000000 ?
800000010000B510 ?
000000020 ?

--------------------- Binary Stack Dump -----

[Steven E. Protter]

Shalom,

Oracle's greatest hits.

Error from U_get_previous_frame_x is 1
Stack is not Windable
Stack has no Unwind_descriptor

It is a HP-UX Os problem.

Though Oracle support is obligated to provide you the OS patch ID necessary to fix it.

I would start with the documentation at http://metalink.oracle.com

Gather the patch list and dependencies from the itrc.hp.com patch database.

Then go through the metalink notes on this problem. Again gather the patches and dependencies from itrc.hp.com patch database.

I would then build and install a patch depot.

Then relink oracle.

Then try and reproduce the error.

SEP

[Dennis Handly]

>here is the stack which I got from our Oracle Team

This is NIH. I want a stack trace from the tried and true U_STACK_TRACE or gdb. Accept no substitutes.

[jgobbur]

any update on this isssue?

[jgobbur]

I am getting similar kind of isssue during test run.
i.e.
68600000002 ?
_sigreturn() call sspuser() 800003FF00000001 ?
000000002 ?
800000010028C16C ?
80000001002748B0 ?
000000048 ?
1000000000000000 ?
800000000000000 ? 07FFFFFFC ?
_memmove()+408 call _sigreturn() 13B2AEA0E7706 ?
40000000027B0210 ?
3058000808A1252B ?
800003FFF7F0A698 ?
80000001002748B0 ?
4000000000B80B38 ?
8000000100017318 ?
8000000100016B30 ?
kco_copy_preimage() call 4000000006D67810 800003FFF800ECB0 ?
+56 400000000179E870 ?
800003FFF7EA8388 ?
10000000B ? 10025E8B0 ?
00025F0B0 ? 01482DAC0 ?
A7A319684B ?
NOTICE: Signal generated by another process. A call stack
dump does not mean that a problem exists in this process

*** 2010-12-24 10:28:00.437
Exception [type: SIGSEGV, Unknown Error] [ADDR:0xFFFFFFFFFFFFFAD0] [PC:0xC000000
000751E54, U_get_previous_frame_x()+1140] [exception issued by pid: -1, uid: -13
28] [flags: 0x2, count: 4]
Not recording an ORA-7445 (previous attmpet failed)
Registers:
r1: 7 r22: 800003fff8012950 sr4: 9c4000
rp: 71bd943 arg3: 0 sr0: 279e400
r3: 800000010038f478 arg2: 40000000071bd943 sr1: 6142800
r4: 8000000100388550 arg1: 800000010038fa60 sr2: 0
r5: c0000000007519e0 arg0: 520 sr3: 0
r6: 0 dp: 800003fff7f59ed0 sr5: 8d27800
r7: 17238 ret0: 0 sr6: a3e4800
r8: 10 ret1: 0 sr7: 279e400
r9: fffffffffffffae0 sp: 800000010038f8b0 cr0: c85238
r10: 0 r31: 800003fff8012950 cr8: 8000000000000000
r11: 0 sar: 2a cr9: 10bfaa0
r12: 71bd943 pcoqh: c000000000751e57 ccr: c000000000000000
r13: 800000010036fad8 pcsqh: 279e400 cr12: 101a1a0
r14: 8 pcoqt: c000000000751e5b cr13: 3941c
r15: 40000000024e6400 pcsqt: 279e400 cr24: 5ffe00000
r16: 800000010036fa38 eiem: ffffffffffffffff cr25: 80
r17: 72 iir: 51373fe1 cr26: 547659000
r18: 30 isr: 0 mpsfu_hi: 800003fff7ee5490
r19: 800003fff8012e70 ior: fffffffffffffad0 mpsfu_lo: 0
r20: 520 ipsw: 804010f mpsfu_ov: 0
r21: 800003fff8012e70 goto: ffffffff pad: 8000009
ssexhd: crashing the process...
Background_Core_Dump = partial


Pls let me know if any patches are missing here. In this OS, LIBCL latest patch is used(PHSS:37959)

[Dennis Handly]

>jgobbur: any update on this issue?

No, the OP never provided the requested info.
Your new thread:
http://h30499.www3.hp.com/t5/General/U-get-previous-frame-x-in-hpux-11-31-platform/m-p/4734506#M147297

[Rita C Workman]

Is your shared memory large enough?

Quiet day so I did a little research on this and found this:

Error from U_get_previous_frame_x is 1
Stack is not Windable
Stack has no Unwind_descriptor

What now?
This happens because you are running an older version on HP-UX on the newer PA8000 hardware architecture. You can work around it by getting the stack trace from the core file using the t command in xdb. Change to your core_dump_dest directory and find the subdirectory for the right core file, then start up xdb as follows.
xdb $ORACLE_HOME/bin/oracle core

-----
So, as has been asked by the experts above, get a proper stack trace - send to Oracle - have them review and let you know what patch to load to fix this specific bug of theirs.

Rgrds,
Rita

[Dennis Handly]

>Rita: You can work around it by getting the stack trace from the core file using the t command in xdb.

(This won't work too well now since xdb hasn't been there since 10.20 and won't work in 64 bit mode. Instead use gdb and bt.)

>have them review and let you know

Right.