Indesign CS2 get Corrupted file error.

[GoogleCodeExporter]

What steps will reproduce the problem?
1. mirror a folder on C:\test to a drive k:
2. open a file with indesign CS2 and try to save it to k:\
3. indesign says the file is corrupted

What is the expected output? What do you see instead?
Indesign should save the file, instead it says it's corrupted. FileMon report 
that the first Write 
after two SetEndofFile (with 2 diff size) recevie an error INVALID_PARAMETER. 

What version of the product are you using? On what operating system?
I m using the version 0.42 and windows XP SP2.

Please provide any additional information below.
I d like to try FileMon with the last snapshot on the SVN as it seems some 
error have been 
corrected for that issue.  If you could send me it already compiled it would 
help me a lot. I ll try 
to compile it myself.

Original issue reported on code.google.com by dek...@gmail.com on 10 Feb 2009 at 1:20

[GoogleCodeExporter]

Please find a trace of filemon while I was trying to save a file on the volume.

Original comment by dek...@gmail.com on 10 Feb 2009 at 1:23

Attachments:

• [Picture 28.png](https://storage.googleapis.com/google-code-attachments/dokan/issue-64/comment-1/Picture 28.png)

[GoogleCodeExporter]

The issue is still present in the current svn trunk. The trace is also similar 
to what i have put in the png up there.

Original comment by dek...@gmail.com on 11 Feb 2009 at 9:38

[GoogleCodeExporter]

From the filespy trace it seems it s linked to the lack of FASTIO_* support. 
From what I read there :

http://msdn.microsoft.com/en-us/library/ms790753.aspx

It seems to confirm that it s important as well. This could probably fix some 
other issues.

Original comment by dek...@gmail.com on 12 Feb 2009 at 5:36

[GoogleCodeExporter]

Thank you very match for your information. I tried implementing fast io 
dispatch 
routines, but they seems not to be related the issue.
I also found a bug in write dispatch routine when Fileman is monitoring IO.
I updated the svn repository. Similar problems are happened when saving file on 
offiece2007. I'm still investing the issue.

Original comment by asa...@gmail.com on 12 Feb 2009 at 4:06

[GoogleCodeExporter]

Hi,

I think I pinpoined where the error coming from. Using straceNT on Indesign i 
collected a trace application 
wise :

[T2464] CreateFileW(3f0b4e0, c0000000, 0, 0, ...) = 49c
[T2464] GetFileSize(49c, 12ed04, 4cd03b8, 3f0b440, ...) = 0
[T2464] SetFilePointer(49c, 6000, 12ecf8, 0, ...) = 6000
[T2464] SetEndOfFile(49c, 1, 3, 4f482e74, ...) = 1
[T2464] SetFilePointer(49c, 16000, 12eca4, 0, ...) = 16000
[T2464] SetEndOfFile(49c, 6, 10, 4f481c2e, ...) = 1
[T2464] SetFilePointer(49c, 0, 12ed08, 0, ...) = 0
[T2464] WriteFile(49c, 3ed0890, 1000, 12ed20, ...) = 1
[T2464] FlushFileBuffers(49c, 3f0b580, 3f0b440, 4f4891ad, ...) = 1
[T2464] WriteFile(49c, 3ed0890, 1000, 12ed18, ...) = 0
[T2464] GetLastError(3f0b580, 12ed44, 4f4879e2, 1, ...) = 57
Exception = e06d7363, Address = 7c81eb33 (first-chance!)
[T2464] GetLastError(0, 12ed38, 4edc8e18, 4f49db98, ...) = 57
[T2464] GetLastError(0, 12ed38, 4edc8eb6, 4edc8e98, ...) = 57

I have reproduced the error in a C program (Compile it with cl writetest.c - 
sorry change the filename i was 
lazy to get it through argv[1]). 

If you remove the FlushFileBuffer the 2nd write works. If you keep it the 2nd 
one get a 0x57 which is ERROR_INVALID_PARAMETER.

Original comment by dek...@gmail.com on 13 Feb 2009 at 8:40

Attachments:

• writetest.c

[GoogleCodeExporter]

From what i read in fatsys source :

    //
    //  CcFlushCache is always synchronous, so if we can't wait enqueue
    //  the irp to the Fsp.
    //

    if ( !FlagOn(IrpContext->Flags, IRP_CONTEXT_FLAG_WAIT) ) {

Would it be possible that the way dokan handle a flush by sending a 
STATUS_PENDING being the cause of the 
error for the next write ? (pure speculation from me).

Original comment by dek...@gmail.com on 13 Feb 2009 at 8:49

[GoogleCodeExporter]

Hello,
I could finally fix the issue. Your test program was really useful.
The problem was in flush dispatch routine. I updated the repository.
Thank you very match.

Original comment by asa...@gmail.com on 13 Feb 2009 at 3:05

• Changed state: Fixed