We've been having intermittent problems with one of our server apps (the "server" is a subclass of GdkServerInterface) where it will terminate with a vmtrap whose walkback doesn't point (usefully) to any particular error - the stack seems to show a process happily chugging along, only to fall off a cliff. It had previously only happened on systems whose logs I can't post, but it seems to have shown up on a development box.
We're currently using VA 7.5; we've had this error (that I know of) on VMWare running WinServer2K3 32-bit and on (non-VMWAre) WinServer2K3 64-bit.
The vmtrap.log is as follows:
-Platform Information-------------------
VM Timestamp: 4.0,(NC) 1/22/2007 (72)
CPU Architecture: i386
OS Type: WIN32_NT
OS Version: 5.2 build 3790 Service Pack 2
-State Information----------------------
Handler=0x10058770
vmState=0x00000000
vmError=0x00000035
ExceptionCode=0xc0000005
ExceptionAddress=0x1003f0a2
ContextFlags=0x0001003f
Module=C:\Program
Files\JAS\bin\ESVM40.dll
Module_base_address=0x10000000
Offset_in_DLL=0x0003f0a2
EDI=0x0014c3f0
ESI=0x0000003b
EAX=0x00000001
EBX=0x00000014
ECX=0x0014c440
EDX=0x00bfb2d8
EBP=0x00149990
ESP=0x00a0fcd4
EIP=0x1003f0a2
-----------------------------------------
Image snapshot: vmtrap.img
-Walkback--------------------------------
Collection>>detect:ifNone:
[] in ProcessorScheduler>>okayToTerminateRun
AbtStackDumpStartUp class>>run
EmSystemConfiguration>>startUp
[] in <optimized block> EmSystemConfiguration>>xdImageStartup
From the log comes:
VA Smalltalk, Version 7.5
VM Timestamp: 4.0,(NC) 1/22/2007 (72)
(C) Copyright Instantiations 1994, 2007. All rights reserved.
(C) Copyright International Business Machines Corp. 1994, 2006. All rights reserved.
Virtual Machine PID: 1712
Debug: Dump stack trace due to error 53
Collection>>detect:ifNone:
[] in ProcessorScheduler>>okayToTerminateRun
AbtStackDumpStartUp class>>run
EmSystemConfiguration>>startUp
[] in <optimized block> EmSystemConfiguration>>xdImageStartup
End Debug: Stack Trace End
We've also seen the same general error (the vmtrap.log is identical down until the EDI/ESI/etc section), but with the top of the walkback being Behavior>>sstBasicLoadFrom:register:
We've added various work-arounds (including automatically restarting the service when it dies), but it would be nice to know what is actually happening so we can fix it, accept it, or throw the whole app away and build something new. Anyone have any clues as to what might be going on?
Thanks in advance!
Brian
PS. What's the deal with the vmtrap.img? Any way I can dig anything human-readable out of that file? Would I even understand it if I could?