forums.silverfrost.com

[sparge]

I know there is a FAQ associated with this phenomenon; I just reacquainted myself with it. In my case the unknown state is somewhere in ntdll.dll. My reason for posting is to to report an associated runime error. The code I'm working on only starting pausing in ntdll.dll this morning. In a bid to find out what bit of my code was suddenly ending up paused in ntdll.dll, in a loop with a count of 5, I started single-stepping and managed to provoke the following runtime error. Does it indicate that in this specific instance there is a real issue to be tracked down and addressed, or does it just indicate that there may be unpredictable consequences of trying too closely to scrutinize sdbg.exe and ntdl.dll slugging it out at assembler level ?

Runtime error from program:c:\program files\salford software\ftn95\sdbg.exe
Access Violation
The instruction at address 036fcef6 attempted to read from location 00000000

036fb502 decode_instruction [+19f4]
036fb4c5 decode_remote_instruction [+001b]
10018d7d text_asm::calculate(<ptr>void) [+00c2]
10018c97 text_asm::set_namespace(<ptr>structÄnamespace) [+0046]
100082e6 debugit(void) [+0458]
100094fb sdbg_main [+1591]
00401000 main [+0076]
004010ba SALFStart [+06ff]

eax=00000020 ebx=00000073 ecx=00000006
edx=7c90e514 esi=00000000 edi=0361f1d9
ebp=0361f180 esp=0361f09c IOPL=2
ds=0023 es=0023 fs=003b
gs=0000 cs=001b ss=0023
flgs=00010206 [NC EP NZ SN DN NV]

036fcef6 lodsb
036fcef7 lodsb
036fcef8 lodsb

[sparge]

For the record, I've tracked down the cause of this now, though it took quite a lot of code rearrangement so I could mix DEBUG-compiled and CECKMATE-compiled code before I was able to do so. Turned out I was trying to "DEALLOCATE a dangling FORTRAN pointer", which I would never have realised without CHECKMATE.

So SDBG pausing in unknown assembler can also be indicative of a bug in user code. It might be worth updating:

http://forums.silverfrost.com/viewtopic.php?t=171

to say so.