Exception in OpenMCDF 2.4.0 (Worked in 2.3.0)

[aunverdorben-dap]

When I run this code with OpenMCDF 2.4.0 it will crash when writing the 9th stream.

{
    string tempFile = @"d:\OpenMCDFCrash.dat";

    // create big MCDF file
    CFSVersion version = CFSVersion.Ver_3;
    CFSConfiguration configuration = CFSConfiguration.Default;
    CFSUpdateMode updateMode = CFSUpdateMode.Update;

    using (CompoundFile compoundFile = new CompoundFile(version, configuration))
    {
        compoundFile.SaveAs(tempFile);
        compoundFile.Close();
    }

    using (var compoundFile = new CompoundFile(tempFile, updateMode, configuration))
    {
        byte[] data = new byte[6617123];

        for (int i = 0; i < 100; i++)
        {
            var cfStream = compoundFile.RootStorage.AddStream($"Stream {i}.dat");
            cfStream.SetData(data);
        }

        compoundFile.Commit();
    }
}

I get:

System.IO.IOException: Seek before origin
at OpenMcdf.StreamView.Seek(Int64 offset, SeekOrigin origin)
at OpenMcdf.StreamRW.Seek(Int64 count, SeekOrigin origin)
at OpenMcdf.CompoundFile.AllocateDIFATSectorChain(List1 FATsectorChain) at OpenMcdf.CompoundFile.AllocateFATSectorChain(List1 sectorChain)
at OpenMcdf.CompoundFile.AllocateSectorChain(List1 sectorChain) at OpenMcdf.CompoundFile.SetSectorChain(List1 sectorChain)
at OpenMcdf.CompoundFile.SetStreamLength(CFItem cfItem, Int64 length)
at OpenMcdf.CompoundFile.WriteData(CFItem cfItem, Byte[] buffer, Int64 position, Int32 offset, Int32 count)
at OpenMcdf.CompoundFile.WriteData(CFItem cfItem, Int64 position, Byte[] buffer)
at OpenMcdf.CompoundFile.WriteData(CFItem cfItem, Byte[] buffer)
at OpenMcdf.CFStream.SetData(Byte[] data)
at OpenMCDF2._4._0CrashRepro.Program.Main(String[] args) in

[jeremy-visionaid]

While I can verify that this is new to v2.4.0, I'm not sure if I'd be able to fix it... While I made a lot of changes on 2.4, most of them were for code quality improvements and validation rather than functional changes (I put in the check which meant that the seek arguments were validated, so it crashes instead of corrupting the file).

So, I suspect that this is probably a regression from fixing a different crash that was happening on v2.3 (#184). However, that issue was the thing that really made me think that a rewrite for v3 was probably a better course of action than trying to continue to improve/fix v2.

So, my advice would be to try 3.0.0-preview.4 and see if that resolves the problem for you. If there's any problems with 3.0.0 then I'll gladly fix it, otherwise I think this one would be down to @ironfede since he knows the v2 code better than I do.

[ironfede]

Fixed in 2.4.1 There was a wrong calculation for DFAT sectors numbers -> regression from 2.3 branch.
Thank you @aunverdorben-dap for having found this issue.

[jeremy-visionaid]

Awesome, thanks for handling that one @ironfede!

[aunverdorben-dap]

Fixed in 2.4.1 There was a wrong calculation for DFAT sectors numbers -> regression from 2.3 branch. Thank you

Thank you for the quick fix!