Hi folks,

wine-941017 still doesn't work for me.

Katsuragawa Naoki san, you said in a posting to
comp.emulators.ms-windows.wine that you had it working just fine under
NetBSD-1.0 and XFree86-3.1 by #define'ing sc_efl to sc_eflags and
whacking the #include statements in dos_fs.c.  I have at NetBSD-1.0
(supped yesterday), but wine-941017 doesn't work for me.  Are you
running the final 1.0 bits?  (it sounds like you are, but I want to be
sure).  My kernel says:

# uname -a
NetBSD kolvir 1.0 NetBSD 1.0 (KOLVIR) #53: Tue Oct 25 00:18:50 EDT 1994     jtk@kolvir:/u1/NetBSD-1.0/src/sys/arch/i386/compile/KOLVIR i386

What type of processor do you have?  Genuine Intel, or a clone?  386,
486, P5?  what speed?  [I don't think any of these should make a
difference, but there must be *something* different between our machines
that's relevant].  I have a real Intel 486/33.

I've built with 'gmake MAKE=gmake', as the README sort-of suggests.
[Otherwise I get problems with WINESTAT being defined or not defined--I
don't understand yet why the BSD pmake builds are broken but GNU make
works OK?]  I tried building as 'gmake CDEBUGFLAGS= MAKE=gmake', but
that didn't alter the brokenness.

The lossage for sol.exe is:
...
InitializeLoadedDLLs 00126AC0
InitializeLoadedDLLs 00000000
Initializing DLLs
Initializing sysres, cs:ip 00c7:0000, ds 00cf
rv = 1
Segmentation fault in Wine program (1f:1016c541).  Please debug
In win_fault 1f:1016c541
In 16 bit mode.
Reading symbols from file wine.sym
Removing BPs
0xffffff82(_end+0xffee2332):  Address 0xffffff82 is out of bounds.

Wine-dbg>x/i $pc
0x009fff82(_end+0x8e2332):  orw    (%bx,%si),%ax
Wine-dbg>info reg
Register dump:
 CS:ffffffff SS:0000 DS:320000 ES:6f5a4 GS:0027 FS:0027
 EIP:0000ff82 ESP:00000027 EBP:00000027 EFLAGS:f7bfd85e
 EAX:000018e0 EBX:00000246 ECX:00000027 EDX:0000d832
 EDI:0000c541 ESI:0000001f

If I run it under gdb, I get the same old "strlen" bug:

(gdb) handle 10 nostop
(gdb) handle 10 noprint
Signal          Stop    Print   Pass to program Description
SIGBUS (10)     No      No      Yes             Bus error
(gdb) run
Starting program: /u1/src/wine941017/wine -relaydbg  /msdos/windows/sol.exe
...
Program received signal SIGSEGV (11), Segmentation fault
0x1016c541 in strlen ()
(gdb) where
#0  0x1016c541 in strlen ()
#1  0xd706b8 in end ()
#2  0xccf5 in DLLRelay ()
#3  0xbf79 in CallTo32 ()
#4  0x3b327 in StartNEprogram ()
#5  0x3a063 in _WinMain ()
#6  0x4c5b1 in main ()
#7  0x100b0060 in end ()
#8  0xf7bfdcc5 in end ()
#9  0x2f6c6961 in end ()
Error accessing memory address 0x6d2f7261: Operation not permitted.
(gdb) x/i $pc
0x1016c541 <strlen+13>: repnz scasb %ds:(%esi),%al

When I put in the 940912 patch from Christos Zoulas to adjust the stack
offsets for certain calls in relay.c, sol.exe works enough to start
spewing infinite loop garbage:
...
InitializeLoadedDLLs 00000000
Initializing DLLs
Initializing sysres, cs:ip 00c7:0000, ds 00cf
rv = 1
MessageBox(0714, 009FFEA4='Not enough memory to display card faces when cards move. Outline dragging has been enabled.', 009F04DA='Solitaire', 0030)
MessageBox // before RegisterClass, 'Not enough memory to display card faces when cards move. Outline dragging has been enabled.' 'Solitaire' !
MessageBox WM_CREATE hWnd=0E64 !
MessageBox WM_CREATE title='Solitaire' str='Not enough memory to display card faces when cards move. Outline dragging has been enabled.' !
MsgBox LoadIcon Exclamation !
MessageBox WM_SHOWWINDOW hWnd=0E64 !
MessageBox // before Msg Loop !
MessageBox WM_PAINT hWnd=0E64 !
MessageBox WM_PAINT // BeginPaint returned BAD hDC !
BeginPaint not called on WM_PAINT for hwnd 3684!
MessageBox WM_PAINT hWnd=0E64 !
MessageBox WM_PAINT // BeginPaint returned BAD hDC !
BeginPaint not called on WM_PAINT for hwnd 3684!
... and on and on ...

With that patch, winmine.exe works enough to paint its initial window, and
it then croaks im memcpy().

I can't ever get a backtrace from the built-in debugger (it seems to
loop forever just after printing "0 " for the first frame).

So, how did you get things working?

==John