Gecko/20110303 Thunderbird/3.1.9   
   rogrammer.misc,comp.os.os2.misc   
   UTC)   
   os2.utilities:169 comp.os.os2.programmer.misc:2087 comp.os.os2.misc:3132   
   From: Jonathan de Boyne Pollard    
      
   > (I suggest that those with JFS source code access check what SFFSI    
   > structure is passed to an "L" function.)   
   >   
      
   Thanks to Dave Yeo, I've been able to check this out.  The "L" functions    
   take the same sffsi structure, except that it has two 64-bit fields    
   tacked on to the end: sfi_sizel and sfi_positionl.  Obviously, by their    
   names, these replace the 32-bit sfi_size and sfi_position fields earlier    
   in the structure.  Quite how the kernel tells which set of fields, the    
   32-bit ones or the 64-bit ones, an IFS driver is maintaining is    
   unclear.  I see no sign of JFS setting a flag to tell the kernel which    
   fields it is maintaining, and it maintains the 64-bit fields, and *only*    
   the 64-bit fields, even in the non-"L" entrypoints.  (The non-"L"    
   entrypoints simply call the "L" ones.  Perhaps this is a JFS bug.  Have    
   there been reports of the 2003 JFS.IFS not working on OS/2 prior to    
   version 4.5?  The 4.0 and earlier kernels will be expecting the 32-bit    
   fields to be maintained.  They probably won't even be allocating the    
   extra space for the 64-bit ones.)   
      
      
   --- Internet Rex 2.31   
    * Origin: virginmedia.com (1:261/20.999)