Name: krT82822 Date: 05/06/99


(note: report prompted by problem in big RMI app, but also seen in smaller GUI app now available as test case --kevin.ryan@eng)

I have a dual processor NT box running JDK 1.2. I periodically get
crashes into NT's Dr. Watson when running a Java application.
I cannot reproduce the cause, but when it happens, the NT stack
trace is almost always the same, occurring in the function
RtlpWaitForCriticalSection. I have NEVER seen this crash when
running the same Java code with one of the CPUs turned off. And
I've never seen it when running the same code on a single
processor NT. But I have seen this same crash with more than one
Java application, so I don't think it's my code. An example NT
stack trace from Dr. Watson is shown below.

function: RtlpWaitForCriticalSection
        77f6cdc2 e920010000 jmp RtlpWaitForCriticalSection+0x196 (77f6cee7)
        77f6cdc7 803d8859fa7701 ds:77fa5988=01
                                  cmp byte ptr [fltused+0xca8 (77fa5988)],0x1
        77f6cdce 1bc0 sbb eax,eax
        77f6cdd0 25d05afa77 and eax,0x77fa5ad0
        77f6cdd5 837e1000 cmp dword ptr [esi+0x10],0x0 ds:024b43da=????????
        77f6cdd9 8945fc mov [ebp-0x4],eax ss:0720e02e=????????
        77f6cddc 7506 jnz RtlpWaitForCriticalSection+0x93 (77f6cde4)
        77f6cdde 56 push esi
        77f6cddf e8b6fdffff call RtlSetCriticalSectionSpinCount+0x70 (77f6cb9a)
        77f6cde4 8b06 mov eax,[esi] ds:009559d4=00000000
FAULT ->77f6cde6 ff4010 inc dword ptr [eax+0x10] ds:01b5ea06=ce79ce79
        77f6cde9 8b06 mov eax,[esi] ds:009559d4=00000000
        77f6cdeb ff75fc push dword ptr [ebp-0x4] ss:0720e02e=????????
        77f6cdee 6a00 push 0x0
        77f6cdf0 ff4014 inc dword ptr [eax+0x14] ds:01b5ea06=ce79ce79
        77f6cdf3 ff7610 push dword ptr [esi+0x10] ds:024b43da=????????
        77f6cdf6 e855b4ffff call NtWaitForSingleObject (77f68250)
        77f6cdfb 3d02010000 cmp eax,0x102
        77f6ce00 0f85b2000000 jne RtlpWaitForCriticalSection+0x167 (77f6ceb8)
        77f6ce06 57 push edi
        77f6ce07 68bcccf677 push 0x77f6ccbc
        77f6ce0c e873940000 call DbgPrint (77f76284)

java -version says
Classic VM (build JDK-1.2-V, native threads)

java -fullversion says
java full version "JDK-1.2-V"

Note. I call this a "consistent" crash because when it crashes,
the stack trace is almost always the same. But it is not
repeatable. I.e., it happens without warning and in no discernible
pattern at completely different points in the Java code and even
with completely different Java programs. I cannot CAUSE it to
happen, though I am certain it will happen again if I wait long
enough. "Long enough" can be minutes or days.
(Review ID: 56532)
======================================================================