Doug Shields wrote in a message to Thomas Waldman:

DS> 11-12-1995  07:41:07  SYS3175  PID 0044
DS> R:\VACPPCD\IBMCPP\BIN\DDE4VB30.EXE
DS> c0000005
DS> 1704a4bf
DS> P1=00000000  P2=ffffffff  P3=XXXXXXXX  P4=XXXXXXXX  
DS> EAX=ab741b68  EBX=ab741b68  ECX=0022af84  EDX=00241ec8
DS> ESI=002e0000  EDI=00000004  
DS> DS=0053  DSACC=d0f3  DSLIM=1bffffff  
DS> ES=0053  ESACC=d0f3  ESLIM=1bffffff  
DS> FS=150b  FSACC=00f3  FSLIM=00000030
DS> GS=0000  GSACC=****  GSLIM=********
DS> CS:EIP=005b:1704a4bf  CSACC=d0df  CSLIM=1bffffff
DS> SS:ESP=0053:00241e50  SSACC=d0f3  SSLIM=1bffffff
DS> EBP=00241e68  FLG=00012206

DS> ESVM13.DLL 0002:0002a4bf

That's a thunk error.  Something tried to call 16-bit code from 32-bit code
and either forgot to adjust the stack or failed in the attempt.  However,
the failure occurred inside OS/2 itself, and was only triggered by your
application. The 0xab741b68 address is a linear memory pointer into the
area used by OS/2 for Ring 0 data space.  The 0x002????? addresses are
pointers to Ring 3 process space, with 0x0022???? pointing to static and
0x0024???? pointing to automatic (stack) variables.  The actual error code
(0xc0000005) is useless: it just means fatal access violation.  The
selectors are very unusual, especially the limits. This almost looks like a
pipe problem?

-- Mike


---