Date: Thu, 19 May 2005 18:06:35 -0400 From: Alex Ionescu Subject: Re: [ros-dev] Re: [ros-svn] [hbirr] 15381: - Reference the file object in IopSecurityFile. In-reply-to: <428CEFA9.4040404@gmx.de> To: ReactOS Development List Message-id: <428D0DEB.9020906@videotron.ca> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7BIT X-Accept-Language: en-us, en User-Agent: Mozilla Thunderbird 1.0 (X11/20041207) References: <000601c55b12$a8a7b7f0$6b01a8c0@PENELOPE> <428B76CB.9040401@videotron.ca> <428CEFA9.4040404@gmx.de> Hartmut Birr wrote: >Alex Ionescu wrote: > > > >>hbirr@svn.reactos.com wrote: >> >> >> >>>- Reference the file object in IopSecurityFile. - Don't set >>>FO_DIRECT_DEVICE_OPEN in IoCreateStreamFileObject. - Disabled the >>>setting of IRP_NOCACHE, because vfat cannot handle cached and non >>>cached requests for the same file. - Set the correct device object >>>in NtWriteFile. >>> >>> >>Hi, >> >>This change breaks usetup. As I said, IoCreateStreamFileObject must be >>left with the current hack intact. >> >>Best regards, >>Alex Ionescu >> >> >> >Without this changes, vfatfs does crash very often while parsing a >directory and if the directory is larger than one cluster. I see this >while compiling ros on ros. After a crash, chkdsk has many to do. Please >remove this hack immediately. > >- Hartmut > > This hack has been in our tree forever! I did not add it. The only way to get rid of it are to rewrite IoCreateFile, and to do that, we need a proper parse routine, and to do that, we need a proper ObFindObject (i have a patch that's been looked at and will commit tonight), and to do that we need to fix our obcreate/insert (i've done this) and to do that we need the right structures (ive done this) and to do that we need this and that and this (ive done them). So yes I"m trying my best to fix the parse routines to the registry/file code can work properly, but until then this hack must stay. But once again, i did NOT add it. It was ALWAYS there. So if it's causing problems, then ROS always had that problem. Removing the hack (Which I agree with and is correct) breaks usetup MBR writing. So I don't know what we can do at this point. Maybe the bug is in vfatfs. Best regards, Alex Ionescu