D. Known bugs

If your think you found a bug in Future Drive Accelerator, please send a description see section D.1 New bugr to the author see section 10. Where to send bug reports, comments and orders?.

Here is a list of known bugs and problems:

Bugs:

• ABackUp has a problem with the RAM-Disk patch of Future Drive Accelerator V1.4.
  => If a file is packed with XPK and there is no space in the RamDisk, ABackup thinks the file is 2GigaByte large.
  Will be fixed in the near future, when the author has completly rewritten the memory-routines.
• AvailMemory(largest) gives a value that is too small.
  => There is a larger continous block; so "not used FastMemory" can be set tof 1%.
  The real value can be determined if you start Avail from the shell, then quit FDA, and again start Avail.
  Will be fixed in the near future, when the author has completly rewritten the memory-routines.
• If the "Size of Track" is different from disk to disk, the memory will be fragmented.
  Will be fixed in the near future, when the author has completly rewritten the memory-routines.

Not real bugs:

• If Not used ChipMemory, see section 7.1 The Mainwindow is set to 100%, no track can be moved to make room.
  => The track is removed from the cache.
  Will be improved soon, when the author has rewritten the memory-routines.
• Doens't support AbortIO,CloseDevice yet.
• Supports, but not optimal, AllocMem(reverse).
  => AllocMem reserves a memory area which is the last free. Normally this is in an early region, because the cache is behind. See section 6. How to install and configure Future Drive Accelerator.
  Will be fixed in the near future, when the author has completly rewritten the memory-routines.
• ViNCEd V3.63 doesn't use AllocEntry or AllocPooled from exec library, to maintain an own poolmemory, to bundle small memory allocations. (It replaces AllocPooled with an own routine.)
  => The cache is very often flushed unneccesary. That's a pity, because for instance the horizontal scrolling is a nice feature.
  KingCON V1.3 makes no problems.
  Will be fixed in the near future, when the author has completly rewritten the memory-routines.
• vdisk.device 2.7 (C) 1994-1997 by Etienne Vogt (like ramdrive.device)
  Doesn't work together, because vdisk does not use OS-memory functions. Is not that tragic, because it doesn't support FastFileSystem V44.5, too.

Future Drive Accelerator has been tested intensively with Enforcer/MungWall and PatchWork.

Patchwork has helped to figure out two very resistant bugs.

• CopyMem(Quick)
  size must not only be LongWordAlligned (size MOD 8=0), but also Destination & Source. The bug appeared only after several 1000 times and caused chaos in memory.
• CopyMem(Quick)
  "memory areas are overlapping (incremental)" seemed to work, but sooner or later, there will be chaos in memory, too, at least with CopyMemQuickerV2.8. This doesn't mean that CopyMemQuicker is not a good program. See section A. How to speed up Future Drive Accelerator even more.

Problems of other cache-programs

• PowerCache
  
  • Has caused checksum error while writing, because Read/Write request are not put in a queue. Especially because the settings are saved to late before a reboot.
  • Cache usage is not flexible.
  • Doesn't re-use the cache when reinserting a disk.
• DynamiCache
  • Does not support Write Async.
  • Does not support SCSI-direct-FileSystems.
  • Does not support harddisk with more than 4 GigaBytes.
  • Doesn't re-use the cache when re-inseritng a disk.
  • Fragments the memory extremely.
  • Does not support AllocAbs and Allocate to free the cache for other programs.
  • If flush is executed (memory full), the cache is deleted, although the memory is not too short anyway.
  • Does not cache accesses from ArtStudioPro 3 and Imagine.

D.1 New bugr

1. Which version do you use?
2. What was set in FDAPref, see section 7.2 The Device Window? (Removeable Disk?, ...)
3. What says FDAPref, see section 7.3 Informations about the drive, about the non-functioning partition?
4. If other programs do not work as they should, do they without FDA?
5. If DOS-checksum (Read/Write) error occured, please activate FDAPref Protect tracks in memory with checksummes, see section 7.1 The Mainwindow. If FDA reports checksum errors, it is not a FDA BUG. Or try to install another FileSystem, if the harddisk has more than 4 GigaByte capacity. See section 11.1 How to use harddisks bigger than 4 GigaBytes..
   Or remove static dust from your RAM-modules. See section B. How to use bad memory.
6. How did the bug appear? How can it be reproduced?
7. Which hardware/software (Mem, OS, FileSystem, Device-Software) was used?
8. Was there a Guru-alert (error messages)?
   In FDA are Guru-number (error messages) built-in. These start with $0FDA00##. If there was another number, please start SegTracker from the Enforcer-package. Start Tnt or if possible Enforcer or replacement programs like CyberGuard (MC68060). Wait for the bug to show up again.
   Write down the error message.
   SegTracker is important because the author would like to know when the BUG occours. See section D.1.1 SegTracker, Enforcer und Co..
9. If something like this appears, please start Segtracker before.

   Guru #0005: Division durch 0
   Dx 00007FFF 00000000 08C6FFFF 00000000  7FFFFFFF 00000000 081AAE54 00000000
   Ax 081AADEC 08C5F9D6 08C5F9DA 08C5F88A  081AAE54 08C5FC34 08000AB0 08C5F820
   segment FDAPref
   hunk       0
   offset  0000365E
   sr 0000
   pc 08C553E6
   <RETURN>
   

D.1.1 SegTracker, Enforcer und Co.

Please read on if you have a processor with MMU, or want to know whether you have.

Needed files

Enforcer supports MC68060,

CyberGuard supports MC68060 DisASM,

MungWall to detect Bad written Programs.

needed to display output (Virtua Terminal) (Work also if the MultTasking is switched off), a replacement for Sushi.

How to run Debugging-Tools?

• SegTracker (Needed for locate (Hunk-offset) the bug)
  RUN Devel:Debug/Enforcer/SegTracker
• Enforcer
  MultiCX/RebootOff (if MultiCX is running)
  RUN Enforcer SHOWPC VERBOSE DEADLY STACKCHECK AREGCHECK DREGCHECK RAWIO (need MungFriend)
  RUN Enforcer SHOWPC VERBOSE DEADLY STACKCHECK AREGCHECK DREGCHECK FILE=CON:0/0/640/100/Enforcer/Auto/Close
  LawBreaker (Any Output? Yes the MMU is working.)
• CyberGuard
  RUN CyberGuard SHOWPC VERBOSE DEADLY STACKCHECK AREGCHECK DREGCHECK FILE=CON:0/0/640/100/CyberGuard/Auto/Close PCLINES=4 SHOWDISS
  LawBreaker (Any Output? Yes the MMU is working.)
• MungWall (Detect bad written Programms (CheckSumme-Error))
  RUN MungWall INFO SHOWFAIL SHOWPC SHOWHUNK SHOWSTACK NAMETAG
• Start Output Display Buffer (Virtual Terminal) (MungFriend ON RawSer->CON:)
  RUN Devel:Debug/MungFriend INSTALL SIZE 40000 trace flash nobells noserial
• Convert, only, GuruNumber in Hunk-offset, doesn`t need an MMU.
  Devel:Debug/Tnt FORCE

How to show the output?

• MungFriend Type to Window
  Devel:Debug/MungFriend >CON:0/0/696/200/MungFriendSerOutput/Auto/Close/Wait/SHELL TYPE
• MungFriend Write to File
  Devel:Debug/MungFriend Write `RequestFile DRAWER=SYS:T/ FILE=SerialOutput.txt TITLE="OutputFile"`
• MungFriend Clear
  Devel:Debug/MungFriend CLEAR
• MungFriend Info (Is any inside?)
  Devel:Debug/MungFriend INFO
• MungFriend ON RawSer->CON After Reset/Dead-GURU: (OutPut is still alife)
  Devel:Debug/MungFriend update trace
• MungFriend OFF
  Devel:Debug/MungFriend REMOVE

Go to the first, previous, next, last section, table of contents.