Magnetic Tape Data Storage. Part 1: Tape Drive - IO Commands

• Download source files - 3.2 KB

Introduction

This article describes the simple way to implement Read/Write operation on tape device. Attached zip file contains TapeOperator class that exposes Load, Read, Write and Close methods and BlockSize property. All these can be used as part of backup utility.

Step 1 - Configuration

Congratulations, you've bought a tape device and connected it to your PC. Now you can implement simple backup utility in C#. Actually there are three ways to work with your tape:

1. You can operate it via driver
2. You can work with it via some management software (if it's provided by the device manufacturer)
3. And the last, simplest way to do it is to operate it via OS handle. This is our case; I'll show to you how you can operate it via OS handle

It is well known that each external device is interpreted by OS as a file, it's true for tape devices as well. The only thing that you have to find out before we start is the appropriate device file name. Open the device manager Window of your PC and you'll see your tape device under the Tape drives node (see picture 1 below).

Perform mouse right click on your tape device node and choose Tape Symbolic Name tab. In this tab you can see tape symbolic name, Tape0 in our example (see Picture 2). This name can be used to create a handle for your device.

So the appropriate file name for your device is @file://./Tape0.

Step 2 - Implementation

TapeOperator methods description:

1. Load: gets as parameter name of your tape device ( "\\.\Tape0"). It creates a handle to the device by CreateFile and performs PrepareTape which executes certain procedures to prepare the given tape for I/O, both methods are part of Win32 API and therefore INTEROP is used to invoke it. Now you have the OS handle for your tape device, and it can be interpreted like a regular file handle, I mean you can open System.IO.FileStream. The only difference is that you can't seek opened FileStream, instead use SetTapePosition Win32 API (see read and write operations).
   C#

   /// <summary>
   /// Loads tape with given name. 
   /// </summary>
   
   public void Load( string tapeName )
   {
       // Try to open the file.
       
       m_handleValue = CreateFile(
           tapeName,
           GENERIC_READ | GENERIC_WRITE,
           0,
           IntPtr.Zero,
           OPEN_EXISTING,
           FILE_ATTRIBUTE_ARCHIVE | FILE_FLAG_BACKUP_SEMANTICS,
           IntPtr.Zero
           );
       if ( m_handleValue.IsInvalid )
       {
           throw new TapeOperatorWin32Exception(
               "CreateFile", Marshal.GetLastWin32Error() );
       }
       // Load the tape
       
       int result = PrepareTape(
           m_handleValue,
           TAPE_LOAD,
           TRUE
           );
       if ( result != NO_ERROR )
       {
           throw new TapeOperatorWin32Exception(
                "PrepareTape", Marshal.GetLastWin32Error() );
       }
       m_stream = new FileStream(
           m_handleValue,
           FileAccess.ReadWrite,
           65536,
           false
           );
   }

2. Write: gets the start position and byte array to write. As it was shown, now you can access your tape device via the opened FileStream. Just invoke FileStrem.Write and FileSream.Flush methods. One thing you have to remember, all I/O operations on tape must be done in multiplies of block size. Each device has min, default and max block sizes (in my example I've used default block size - 65536), these values can be fetched by GetTapeParameters method (Win32 API), for more information, see BlockSize property implementation.
   C#

   /// <summary>
   /// Writes to the tape given stream starting from given position
   /// </summary>
   /// <param name="startPos"></param>
   /// <param name="stream"></param>
   
   public void Write( long startPos, byte[] stream )
   {
       // Get number of blocks that will be needed to perform write
       
       uint numberOfBlocks = GetBlocksNumber( stream.Length );
       // Updates tape's current position
       
       SetTapePosition( startPos );
       
       byte[] arrayToWrite = new byte[ numberOfBlocks * BlockSize ];
       Array.Copy( stream, arrayToWrite, stream.Length );
       // Write data to the device
       
       m_stream.Write( stream, 0, stream.Length );
       m_stream.Flush();
   }

3. Read: gets the start positions and number of bytes to read.

   *By the way, pay attention on SetTapePosition (Win32 API) last parameter, it's type is BOOL in WIN32 definition. Don't pass .NET Boolean type for BOOL and BOOLEAN of WIN32. The size of .NET Boolean is 2 bytes, the size of BOOL is 4 bytes and size of BOOLEAN is 1 byte, therefore each Win32 API method with BOOL or BO<code>OLEAN parameter will return error for .NET Boolean pass attempt.

   C#

   /// <summary>
   /// Read one logical block from tape 
   /// starting on the given position
   /// </summary>
   /// <returns></returns>
   
   public byte[] Read( long startPosition )
   {
       byte[] buffer = new byte[ BlockSize ];
       SetTapePosition( startPosition );
       
       m_stream.Read( buffer, 0, buffer.Length );
       m_stream.Flush();
       return buffer;
   }

4. Close: closes the stream and releases unmanaged resources. You can add to this code call to LoadTape with TAPE_UNLOAD parameter to eject tape from drive.
   C#

   /// <summary>
   /// Closes handler of the current tape
   /// </summary>
   
   public void Close()
   {
       if ( m_handleValue != null &&
           !m_handleValue.IsInvalid &&
           !m_handleValue.IsClosed )
       {
           m_handleValue.Close();
       }
   }

5. BlockSize: This property returns the tape's default block size by invocation of GetTapeParameters method. One of the GetTapeParameters is a reference to the structure that is filled by Win32 method. So Marshal class is used to allocate unmanaged memory and copy between managed and unmanaged structures. See the source code below:
   C#

   /// <summary>
   /// Returns default block size for current
   /// device
   /// </summary>
   public uint BlockSize
   {
       get
       {
           IntPtr ptr = IntPtr.Zero;
           try
           {
               if ( !m_driveInfo.HasValue )
               {
                   m_driveInfo = new DriveInfo();
                   // Allocate unmanaged memory
                   
                   int size = Marshal.SizeOf( m_driveInfo );
                   ptr = Marshal.AllocHGlobal( size );
                   Marshal.StructureToPtr(
                       m_driveInfo,
                       ptr,
                       false
                   );
                   
                   int result = 0;
                   if ( ( result = GetTapeParameters(
                       m_handleValue,
                       DRIVE_PARAMS,
                       ref size,
                       ptr ) ) != NO_ERROR )
                   {
                       throw new TapeOperatorWin32Exception(
                           "GetTapeParameters", 
                           Marshal.GetLastWin32Error() );    
                   }
                   // Get managed media Info
                   
                   m_driveInfo = ( DriveInfo )
                       Marshal.PtrToStructure
               ( ptr, typeof( DriveInfo ) );
               }
               
               return m_driveInfo.Value.DefaultBlockSize;
           }
           finally
           {
               if ( ptr != IntPtr.Zero )
               {
                   Marshal.FreeHGlobal( ptr );
               }
           }
       }

6. See attached files for more information about defined constants, private variables, private methods and PINVOKE declarations.
7. Good luck!