Re: msdos/int21 CREAT special cases

[Andreas Mohr]

On Mon, May 26, 2003 at 01:04:49PM +0200, Ferenc Wagner wrote:
> Volume label creation should fail, not create a regular file.
> Similarly for file names ending in a slash (this is actually
> needed for PKSFX Version 2.04g to unpack directory trees).
>                                             Feri.
>
> Index: msdos/int21.c
>
> RCS file: /home/wine/wine/msdos/int21.c,v
> retrieving revision 1.93
> diff -u -r1.93 int21.c
> --- msdos/int21.c	19 May 2003 21:40:05 -0000	1.93
> +++ msdos/int21.c	26 May 2003 10:48:21 -0000
> @@ -474,8 +474,18 @@
>  }
>  static BOOL INT21_CreateFile( CONTEXT86 *context )
>  {
> - SET_AX( context, _lcreat16( CTX_SEG_OFF_TO_LIN(context, context->SegDs,
> -
>                                               context->Edx ), CX_reg(context) ) );
> - LPCSTR path=CTX_SEG_OFF_TO_LIN(context, context->SegDs, context->Edx);
> -
> - if (CX_reg(context) == 0x0008) {
> -
>    SetLastError( ERROR_ACCESS_DENIED );
> -
>    return TRUE;
> - }
> - if (path[strlen(path)-1] == '/')
> - {
> -
>    SetLastError( ERROR_FILE_NOT_FOUND );
> -
>    return TRUE;
> - }
> - SET_AX( context, _lcreat16( path, CX_reg(context) ) );
> return (AX_reg(context) == (WORD)HFILE_ERROR16);
>
> }

Hmm, and is there a reason to not do it in _lcreat16() instead?
Or is _lcreat16() really "clever" enough to handle volume labels
"properly" if called directly?
Somehow I doubt it... ;)
Or in fact you could walk further up the chain and maybe find that even
_lcreat() should be the function to have that check instead...
(or, for that matter, even CreateFileA and thus CreateFileW...)

So my rough guess is that this should probably be prevented in CreateFileW
instead even.

msdos/int21 CREAT special cases

[Ferenc Wagner]

Andreas Mohr  writes:
> On Mon, May 26, 2003 at 01:04:49PM +0200, Ferenc Wagner wrote:
> > Volume label creation should fail, not create a regular file.
> > Similarly for file names ending in a slash.
>
> Hmm, and is there a reason to not do it in _lcreat16() instead?
> Or is _lcreat16() really "clever" enough to handle volume labels
> "properly" if called directly?
> Somehow I doubt it... ;)
> Or in fact you could walk further up the chain and maybe find that even
> _lcreat() should be the function to have that check instead...
> (or, for that matter, even CreateFileA and thus CreateFileW...)

Yes, this would be reasonable.  This was Duane's original
idea, actually (part of the code is from him).  However, the
tests I submitted some days ago into the kernel32 file tests
show that _lcreat() actually behaves the same in Wine and
Windows: masks off the not allowed flag bits and does not
care about trailing slashes.  On NT4 at least.  Since I have
no access to recent conformance testing results on different
versions, I went with what I had.  I did not test CreateFile
or _lcreat16(), though.
                                                Feri.