vfat bug (hartmut?)

[Gunnar Dalsnes]

This problem prevents ros from deleting a file/directory (observed while 
running wine kernel32 reg. tests with param "path")

around line  295 in vfat\finfo.c: (svn blame says Hartmut made the last 
changes around here)

------------------------------------------------
   if (DispositionInfo->DeleteFile)
     {
       if (MmFlushImageSection (FileObject->SectionObjectPointer, 
MmFlushForDelete))
         {
           if (FCB->OpenHandleCount > 1)
             {
          DPRINT1("%d %x\n", FCB->OpenHandleCount, 
CcGetFileObjectFromSectionPtrs(FileObject->SectionObjectPointer));
               Status = STATUS_ACCESS_DENIED;
             }
           else
             {
               FCB->Flags |= FCB_DELETE_PENDING;
               FileObject->DeletePending = TRUE;
             }
         }
       else
         {
  ------------------------------------------------
Why deny deletion with STATUS_ACCESS_DENIED if FCB->OpenHandleCount > 1 
??????

Gunnar

[Hartmut Birr]

Gunnar Dalsnes schrieb:
> This problem prevents ros from deleting a file/directory (observed
> while running wine kernel32 reg. tests with param "path")
>
> around line  295 in vfat\finfo.c: (svn blame says Hartmut made the
> last changes around here)
>
> if (DispositionInfo->DeleteFile)
>     {
>       if (MmFlushImageSection (FileObject->SectionObjectPointer,
> MmFlushForDelete))
>         {
>           if (FCB->OpenHandleCount > 1)
>             {
>           DPRINT1("%d %x\n", FCB->OpenHandleCount,
> CcGetFileObjectFromSectionPtrs(FileObject->SectionObjectPointer));
>               Status = STATUS_ACCESS_DENIED;
>             }
>           else
>             {
>               FCB->Flags |= FCB_DELETE_PENDING;
>               FileObject->DeletePending = TRUE;
>             }
>         }
>       else
>         {
>
> Why deny deletion with STATUS_ACCESS_DENIED if FCB->OpenHandleCount >
> 1 ??????

It is only possible to delete a file if the file is not mapped (as image
or data section) and if nobody else has open this file. The second
condition is the reason for the handle check. There exist one situation
for more than one handle. The file can be open with FILE_SHARE_DELETE.
Currently it is not implemented.

- Hartmut

[James Tabor]

Hi!
Hartmut Birr wrote:
> Gunnar Dalsnes schrieb:
> It is only possible to delete a file if the file is not mapped (as image
> or data section) and if nobody else has open this file. The second
> condition is the reason for the handle check. There exist one situation
> for more than one handle. The file can be open with FILE_SHARE_DELETE.
> Currently it is not implemented.
> Hartmut

Just to give a heads up on vfat! There is a Free Space bug. I have to run
win98 scandisk on the ros drive to fix it.
James

[Hartmut Birr]

James Tabor schrieb:
> Just to give a heads up on vfat! There is a Free Space bug. I have to run
> win98 scandisk on the ros drive to fix it.
> James

Did you found a real error or only a message about a mismatched free
size? Vfat does not update the available cluster count on shutdown.

- Hartmut

[James Tabor]

Hartmut Birr wrote:
> James Tabor schrieb:
> > Just to give a heads up on vfat! There is a Free Space bug. I have to run
> > win98 scandisk on the ros drive to fix it.
> > James
>
> Did you found a real error or only a message about a mismatched free
> size? Vfat does not update the available cluster count on shutdown.
> Hartmut

That's right! Mismatch free size every time.
James

[Magnus Olsen]

Which fat version does you use in windows 98?
 vfat32a or vfat32b ?

vfat32a does have some bugs
1. Mismatch free size  betwin vfat32a and vfat32b
2. can not store biger that 2GB files on one parition.

that only two bugs I know exists in vfat32a
only way check this is to see how large file you can store. that does
not be corrupt.

----- Original Message ----- 
From: "James Tabor" [e-mail]
To: "ReactOS Development List" [e-mail]
Sent: Wednesday, March 16, 2005 8:43 PM
Subject: Re: [ros-dev] vfat bug (hartmut?)
> Hartmut Birr wrote:
> > James Tabor schrieb:
> > > Just to give a heads up on vfat! There is a Free Space bug. I have to

run
> > > win98 scandisk on the ros drive to fix it.
> > > James
> >
> > Did you found a real error or only a message about a mismatched free
> > size? Vfat does not update the available cluster count on shutdown.
> > Hartmut
>
> That's right! Mismatch free size every time.
> James
> _______________________________________________
> Ros-dev mailing list
> [e-mail]
> http://reactos.com:8080/mailman/listinfo/ros-dev

[Jakob Eriksson]

Magnus Olsen wrote:
> Which fat version does you use in windows 98?
> vfat32a or vfat32b ?
>
> vfat32a does have some bugs
> Mismatch free size  betwin vfat32a and vfat32b
> can not store biger that 2GB files on one parition.
>
> that only two bugs I know exists in vfat32a
> only way check this is to see how large file you can store. that does
> not be corrupt.

Is there no ext2 driver or something for NT we can use?

regards,
Jakob

[Gunnar Dalsnes]

> It is only possible to delete a file if the file is not mapped (as image
> or data section) and if nobody else has open this file. The second
> condition is the reason for the handle check. There exist one situation
> for more than one handle. The file can be open with FILE_SHARE_DELETE.
> Currently it is not implemented.

I have checked some other fs' (a couple of ext fs') and they do no such 
check. Io(Set/Check)ShareAccess should already have taken care of 
denying access in the right cases (when create/open file). Removing the 
handle check fixes my problem:-)

Gunnar
> Hartmut
>
> Ros-dev mailing list
> [e-mail]
> http://reactos.com:8080/mailman/listinfo/ros-dev