STATUS_ACCESS_VIOLATION in egcs-2.91.57 compiled program
Tue Aug 17 06:44:00 GMT 1999

Dear cygwin gurus,

for porting C programs from UNIX to NT, I  downloaded the complete Cygwin B20.1
on 1999-07-29 and installed it under NT 4.0 SP4.

The first program compiled by the included egcs-2.91.57 and run as "sqbproc <
sqbinput" aborts for most but not all instances of sqbinput, outputting
STATUS_ACCESS_VIOLATION on stderr and in sqbproc.exe.core (a stack trace).

Due to its unresolved redirection problem (see mailing list archives
.../1999-04/msg00355.html) gdb could not be used for investigation. Analyzing
the generated stack traces ("core dumps") by the aid of the linker map shows
different origines of the aborts (see below), whereas in all cases pc's in the
uppermost four stack frames don't differ and look similar to those reported in
.../1999-06/msg00637.html. Sorry I could not localize addresses above 0x61000000
and 0x77F00000.

Compilation is done with options -g -O0 -Wall; no warnings are seen besides 4
"defined but not used" (from lex and yacc). Options -O1 to O2 as suggested
anywhere have no effect. Linking for increased heap size (-Wl,heap,... , to 10
times the default of 0x00100000) is reflected in the map's or objdump's
"size_of_heap_reserve" but does not affect aborting. The binary's size is appox.
490 Kb.

The behavior found is hard to understand. For many years the program works fine
under SunOS 4.1 to 5.6 (gcc up to as well as other compilers) and
recently under NT when compiled by MSC. Using [f]lex and [a]yacc (up to now
under SunOS), this C program implements a special description language and
processes a data set, written in that language and taken from stdin, into a set
of tables output as C files consisting of arrays. For doing so it opens about 40
files for writing and reading (or vice versa, all in binary mode) during an init
phase and then calls malloc very often.

Four data sets were tested. Due to the principle they have the same structure
though differing in size (approx. 4/47/88/22 Kb). They cause the program to work
differently. The smallest one passes, the second induces an abort by occasion of
the first fopen call, but the largest one passes all these fopens producing the
exception from inside an user function much later on, probably when calling
malloc. The fourth aborts similarly, but without malloc being involved, as far
as I can see (address > 0x61...).

Diagnostic output from running data set 2 is enclosed here:
(Note: On my NT machine Cygwin and MKS Toolkit must coexist.)

Output from stderr:
[main] c:\users\alker\bin\sqbproc.exe 1000 (0) handle_exceptions:
[main] sqbproc 1000 (0) handle_exceptions:
     Dumping stack trace to sqbproc.exe.core

sqbproc.exe.core (stack trace):
[main] sqbproc 1000 (0) exception: trapped!
[main] sqbproc 1000 (0) exception: code 0xC0000005 at 0x6100D89B
[main] sqbproc 1000 (0) exception: ax 0x42A910 bx 0x0 cx 0x61061104 dx 0x10
[main] sqbproc 1000 (0) exception: si 0x401044 di 0x0 bp 0x247CD50 sp 0x247CD4C
[main] sqbproc 1000 (0) exception: exception is: STATUS_ACCESS_VIOLATION
[main] sqbproc 1000 (0) stack: Stack trace:
[main] sqbproc 1000 (0) stack: frame 0: sp = 0x247CB60, pc = 0x6100A2C3
[main] sqbproc 1000 (0) stack: frame 1: sp = 0x247CB9C, pc = 0x77FA4846
[main] sqbproc 1000 (0) stack: frame 2: sp = 0x247CBC0, pc = 0x77F99013
[main] sqbproc 1000 (0) stack: frame 3: sp = 0x247CC4C, pc = 0x77F86392
[main] sqbproc 1000 (0) stack: frame 4: sp = 0x247CD50, pc = 0x6100D95D
[main] sqbproc 1000 (1) stack: frame 5: sp = 0x247CD6C, pc = 0x610194E0
[main] sqbproc 1000 (0) stack: frame 6: sp = 0x247CD88, pc = 0x61019345
[main] sqbproc 1000 (0) stack: frame 7: sp = 0x247CE10, pc = 0x6103098E
[main] sqbproc 1000 (0) stack: frame 8: sp = 0x247CE44, pc = 0x6105D7E2
[main] sqbproc 1000 (0) stack: frame 9: sp = 0x247CE58, pc = 0x610527CF
[main] sqbproc 1000 (0) stack: frame 10: sp = 0x247CE88, pc = 0x6105283A
[main] sqbproc 1000 (0) stack: frame 11: sp = 0x247CE9C, pc = 0x409143
[main] sqbproc 1000 (0) stack: frame 12: sp = 0x247CEBC, pc = 0x4011A4
[main] sqbproc 1000 (0) stack: frame 13: sp = 0x247CED0, pc = 0x405752
[main] sqbproc 1000 (0) stack: frame 14: sp = 0x247FE14, pc = 0x410DAD
[main] sqbproc 1000 (0) stack: frame 15: sp = 0x247FE1C, pc = 0x61004402
[main] sqbproc 1000 (0) stack:
     End of stack trace (more stack frames may be present)

bash-2.02$ cygcheck -s

Cygnus Win95/NT Configuration Diagnostics
Current System Time: Tue Aug 17 10:43:46 1999

WinNT Ver 4.0 build 1381 Service Pack 4

Path:   /c/users/alker/bin

SysDir: C:\WINNT\System32
WinDir: C:\WINNT

HOME = `/c/users/alker'
PWD = `/c/users/alker/bin'

Use `-r' to scan registry

a:\ fd           N/A    N/A
c:\ hd  NTFS    4353Mb  24% CP CS UN PA FC     NTFS
g:\ net NTFS    2060Mb  58% CP CS UN PA FC
h:\ net SUNWNFS  1090Mb  67% CP                 magnus
k:\ net SUNWNFS  4781Mb 100%                    donau
m:\ net SUNWNFS  4959Mb  96% CP                 magnus
r:\ cd           N/A    N/A
s:\ net SUNWNFS   193Mb  43% CP                 magnus

c:\cygnus\root                            /        native  text=binary
h:\swarchiv\src\gpib                      /asg     native  text=binary
c:\cygnus\cygwin-b20\H-i586-cygwin32\bin  /bin     native  text=binary
c:                                        /c       native  text=binary
h:                                        /h       native  text=binary
c:\temp                                   /tmp     native  text=binary

Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\bash.exe
Found: c:\mks\mksnt\cat.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\cat.exe
Found: .\cpp.exe
Found: c:\users\alker\bin\cpp.exe
Found: c:\mks\mksnt\find.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\find.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\gcc.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\gdb.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\ld.exe
Found: c:\mks\mksnt\ls.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\ls.exe
Found: c:\mks\mksnt\make.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\make.exe
Found: c:\mks\mksnt\sh.exe
Found: c:\cygnus\CYGWIN~1\H-I586~1\bin\sh.exe

  371k 1998/12/01 c:\cygnus\CYGWIN~1\H-I586~1\bin\cygtcl80.dll
    5k 1998/12/01 c:\cygnus\CYGWIN~1\H-I586~1\bin\cygtclpip80.dll
   10k 1998/12/01 c:\cygnus\CYGWIN~1\H-I586~1\bin\cygtclreg80.dll
  600k 1998/12/01 c:\cygnus\CYGWIN~1\H-I586~1\bin\cygtk80.dll
  446k 1998/12/03 c:\cygnus\CYGWIN~1\H-I586~1\bin\cygwin1.dll

bash-2.02$ cygcheck -v sqbproc.exe
Found: .\sqbproc.exe
Found: c:\users\alker\bin\sqbproc.exe
Warning: .\sqbproc.exe hides c:\users\alker\bin\sqbproc.exe
.\sqbproc.exe - os=4.0 img=1.0 sys=4.0
  c:\cygnus\CYGWIN~1\H-I586~1\bin\cygwin1.dll - os=4.0 img=1.0 sys=4.0
    "cygwin1.dll" v0.0 ts=1998/12/4 5:39
    C:\WINNT\System32\advapi32.dll - os=4.0 img=4.0 sys=4.0
      "ADVAPI32.dll" v0.0 ts=1998/9/29 22:44
      C:\WINNT\System32\ntdll.dll - os=4.0 img=4.0 sys=4.0
        "ntdll.dll" v0.0 ts=1998/10/8 21:30
      C:\WINNT\System32\KERNEL32.dll - os=4.0 img=4.0 sys=4.0
        "KERNEL32.dll" v0.0 ts=1998/10/13 10:38
        C:\WINNT\System32\ntdll.dll (already done)
      C:\WINNT\System32\USER32.dll - os=4.0 img=4.0 sys=4.0
        "USER32.dll" v0.0 ts=1998/10/13 10:39
        C:\WINNT\System32\ntdll.dll (already done)
        C:\WINNT\System32\KERNEL32.dll (already done)
        C:\WINNT\System32\GDI32.dll - os=4.0 img=4.0 sys=4.0
          "GDI32.dll" v0.0 ts=1998/9/23 8:27
          C:\WINNT\System32\ntdll.dll (already done)
          C:\WINNT\System32\KERNEL32.dll (already done)
          C:\WINNT\System32\USER32.dll (recursive)
          C:\WINNT\System32\ADVAPI32.dll (recursive)
        C:\WINNT\System32\ADVAPI32.dll (recursive)
      C:\WINNT\System32\RPCRT4.dll - os=4.0 img=4.0 sys=4.0
        "RPCRT4.dll" v0.0 ts=1998/10/9 21:19
        C:\WINNT\System32\ntdll.dll (already done)
        C:\WINNT\System32\KERNEL32.dll (already done)
        C:\WINNT\System32\ADVAPI32.dll (recursive)
    C:\WINNT\System32\kernel32.dll (already done)
    C:\WINNT\System32\user32.dll (already done)
  C:\WINNT\System32\kernel32.dll (already done)

(End of diagnostic output)

Any hint would be highly appreciated.


Dietrich Alker

E-mail: or

c/o Wavetek Wandel Goltermann, Dept. 1ET
POB 1262, 72795 Eningen, Germany

Want to unsubscribe from this list?
Send a message to

More information about the Cygwin mailing list