LevGdIZWbTvv+SXA";   
   er.org";	posting-account="U2FsdGVkX185ZFM17Lc3bR7+j7oXlZL1"   
   comp.os.os2.utilities:147   
   From: Ilya Zakharevich    
      
   On 2011-03-31, Jonathan de Boyne Pollard  wrote:   
   >  BTW, does 'check' including fixing FAT32 ? The current FAT32 check    
   >> often leads to a msg like: Please run u$ Scandisk to fix your    
   >> partition :-/   
   >>   
   > Allan now knows that the answer to this is "yes", he being the first    
   > person other than me to have successfully checked a FAT32 partition with    
   > CHKVOL.  He's also seen CHKVOL tell him about some oddities in his FATs    
   > that not even Microsoft's CHKDSK (from Windows NT) was telling him    
   > about.  (-:   
   >   
   > Others have not been so lucky.  It turns out that there's a    
   > not-so-well-known problem with the FAT32 IFS driver and access to    
   > *large* partitions (certainly larger than Allan's ~4GiB at any rate)    
   > that affects any program that tries to read the partition directly, as    
   > of course CHKVOL does.  One cannot even seek to block #0 and read it.     
   > I'm trying a few workaround strategies for this.   
      
   You might be seeing the same problem as what I reported about a year   
   ago: reading by anything larger than 512 B blocks is "unreliable" -   
   gives junk.  But, IIRC, I have seen it even with smallish partitions   
   (do not remember the details - it should be on Usenet).   
      
   E.g., my Perl module for post-mortems on FAT reads by 512 B when run   
   on OS/2 (of course, this is painfully slow).  But at least I did not   
   see any OTHER problem with my test cases...   
      
   Ilya   
      
   --- Internet Rex 2.31   
    * Origin: U.C. Berkeley Math. Department. (1:261/20.999)