******************************************************************************* * * powerpc.library V7 * based on the ppc.library * ******************************************************************************* 3. 6.97 - first internal release - V7.0 21. 6.97 - New library functions 'AllocVecPPC' and 'FreeVecPPC' for memory allocation, including user-defined alignment. - V7.01 21. 7.97 - New 68K library functions 'AllocVec32' and 'FreeVec32' allow to allocate memory which can be shared with PPC tasks - New 68K library function 'SPrintF68K' allows to print out debug information to serial/sushi. - V7.02 30. 7.97 - General adaption to the new ppc.library due to major changes of ppc.library (i.e. dynamic binding). The file dummy.elf has been replaced by a new one - Added private Cacheflush-routine to every internal 68K-Call because the automatic cache flushing of the actual ppc.library (actually 44.17) doesn't work anymore. - V7.03 1. 8.97 - Changed the assembler macros RUNPOWERPC and RUNPOWERPC_XL (powerpc.i) to support direct PPC-Library-Calls from 68K- Functions. Note: the parameters 2-4 were shifted to the right by one. New parameter 2: The Library-Offset (used in the same way like in the macros RUN68K and RUN68K_XL). 11. 8.97 - Small bug fixed causing problems with debug output enabled. - V7.041 - IMPORTANT! Special Note: The powerpc.library V7 does NOT work with ppc.library 44.20. Reason: The signaling-mechanism of the ppc.library is definitively broken! Probably more reasons... 16. 9.97 - V7 works again (special hack implemented to bypass ppc.library bugs). - Removed private Cacheflush (seems to be obsolete now). - V7.05 18. 9.97 - New library functions GetSysTimePPC, AddTimePPC, SubTimePPC and CmpTimePPC for time measurement - New small demo program 'pixelOmania' to measure the context switch performance (i.e. to compare V7 and V8+) - Reinstalled private Cacheflush (isn't obsolete at all...) - V7.06 29. 9.97 - New env variable 'earlyterm' implemented. If the variable is not 0, then the mirror task of a 68k task is removed at the last CloseLibrary of the 68k task, instead at removal of the task. Hope, this will fix the problems with WShell. - V7.07 8.10.97 - RunPPC now interprets the PP_OFFSET field in the same way as Run68K does. If PP_OFFSET is zero, than it works as before, if it is not zero, a PPC library call is performed with the base in PP_CODE and the library vector offset in PP_OFFSET. - V7.08 22.11.97 - Oh, DAMN. Exec tasks couldn't use RunPPC because of DOS calls. These calls are now handled by a new support process called 'Babylon5'. - V7.1 25. 3.98 - Stackswapping implemented to eliminate RAMLIB stack problems V7.2