WHDLoadCD32

These are the differences to the normal WHDLoad:

• it does not require nor use a WHDLoad keyfile, therefore CDs with WHDLoadCD32 on it can be distributed freely
• only the 68020 CPU is supported, which means also no MMU support
• fast memory is not supported
• there is absolut no writing support, neither savegames, highscores nor coredumps (a CD is usually read-only)
• many options have been removed which are not useful on a CD³² to make the executable smaller and save memory, check Usage and Options to see which options are omitted
• the following resload functions are not implemented and using them will terminate WHDLoad with a return value TDREASON_ILLEGALARGS:
  • resload_DiskLoadDev
  • resload_ProtectRead
  • resload_ProtectReadWrite
  • resload_ProtectWrite
  • resload_ProtectRemove
  • resload_ProtectSMC
  • resload_Log
• the following resload functions are dummies, they will do nothing but return always success:
  • resload_SaveFile
  • resload_SaveFileOffset
  • resload_DeleteFile
• there is no freezer/monitor support built in
• there is no internal check for sufficient stackspace as in the normal WHDLoad, to my tests the default of 4096 bytes should enough for WHDLoadCD32

Hints for using WHDLoad on a CD³²

• If install gives the choice between diskimages and files use files. Because many smaller files can be better cached than less big files.
• If an install provides the ability to compress files using ProPack/FImp use it and compress all files (RNC1 gives best compression ratio). Compressed files need less memory for Preload and cause nearly no delay in loading speed.
• Use AllocMemReverse before running WHDLoadCD32 (or an adequate replacement). AllocMemReverse is a small program included in the dev-package of WHDLoad. It patches the exec function AllocMem to always use the memory attribute MEM_REVERSE for the allocations. That will result in memory getting allocated from the top of the free memory to the bottom and not from bottom to top as normal. This will hold the lower memory regions free. The background is, that WHDLoad needs memory starting at address 0, the start of chipmem. All memory which is used in this lower region needs to be swapped out. To save this swapped memory additional memory is required. The more memory is used in the low memory regions the less is available for the installed program and for preloading data files.
• Always use the Data option. Preload does not know which files will be loaded by the installed program, therefore it loads all files it finds. To avoid that unnecessary files are loaded create a directory which only contains the required files. Delete all obsolete stuff like #?.info, #?.Slave, ReadMe, Manual, .... Then call WHDLoad with Slave and Data directory as argument, e.g. WHDLoadCD32 :Slaves/XYZ.Slave Data=:Data/XYZ Preload. There is no need to change the current directory to anywhere.
• Don't use XPK compressed data files. XPK decompression needs much memory because XPK libraries must be in memory, decompression itself needs memory and decompression will cause memory fragmentation.