On Thu, Jun 03, 2004 at 06:14:58PM -0400, Christopher Faylor wrote: >On Thu, Jun 03, 2004 at 04:35:00PM -0400, Christopher Faylor wrote: >>On Thu, Jun 03, 2004 at 03:53:40PM -0400, David Fritz wrote: >>>I don't how much you want to rely on undocumented features, but ntdll.dll >>>exports a function called RtlIsDosDeviceName_U(). The WINE implementation >>>has the following to say about it: >>> >>> >>>/*********************************************************************** >>> * RtlIsDosDeviceName_U (NTDLL.@) >>> * >>> * Check if the given DOS path contains a DOS device name. >>> * >>> * Returns the length of the device name in the low word and its >>> * position in the high word (both in bytes, not WCHARs), or 0 if no >>> * device name is found. >>> */ >>>ULONG WINAPI RtlIsDosDeviceName_U( PCWSTR dos_name ) >> >>THANK YOU! This is what I was vaguely remembering. > >Interestingly enough, I just added some checking to fhandler_base::open which >used RtlIsDosDeviceName_U. It caused a reboot of my XP system every time >I tried it. That's a first for XP.
Oops. No, that was the result of passing garbage to InitializeObjectAttributes apparently. Seems like a pretty serious XP bug regardless. I've checked in a minimal change which protects against calling NtCreateFile on a dos device. Now what was the method that was used to get rid of those pesky NULs and COM1s on my disk again? cgf
