*** 9.1 (15.02.1999) - new: WHDLTAG_TIME_GET to get actual date and time (MrLarmer) - fix: problems with some 68030 systems fixed (WHDLoad uses some special hardware features in conjunction with the MMU related operations, some hardware does not support this (Apollo A630 and Magnum MK2 are known for this), on such systems you can't enable the MMU/S option - otherwise the machine will hang) (reported by Chris Thomas, Maxim Ilyn and Jess Sosnoski) *** 9.0 (17.01.1999) - chg: performance of resload_LoadFileDecrunch improved - chg: access fault handler for 68030 rewritten: - better performance - cia snooper implemented - icm and alarm will be saved and dumped - setting overlay bit (ciaa.pra) catched - setting motor/step (ciab.prb) catched - bad writes to ddra/ddrb catched - blitsize check fixed, supports all except line-mode now (ascending/descending mode, negative modulos, odd pointers and odd modulos) - a callback function can be setup (resload_Control) which will called by WHDLoad if an Access Fault occurs, the function can modify the bad access and decide if the program should be continued or terminated, see autodoc for further infos - fix: potgo set to input mode now, fixes right mouse button problem with ArcherMcLeanPool and JimmyWhiteSnooker (MrLarmer) - new: WHDLTAG_VERSION_GET, WHDLTAG_REVISION_GET, WHDLTAG_BUILD_GET for usage with resload_Control to get the major and minor version number and the build number of WHDLoad currently running, the build number is granted to be uniq and can be used to distinguish between different sub versions (MrLarmer) - new: pressing Ctrl-Amiga-Amiga will cause WHDLoad to quit, depending on machine speed this may avoid cold reboots due a not restored execbase (Graham) - fix: Snoop works again on 68060, was broken in 8.2 - chg: the vector table is not write protected if Snoop on a 68060 is enabled - new: option ReadDelay/N to cause WHDLoad to wait some time between reading files from disk and switching the OS off (Jeff) - new: flag WHDLF_EmulTrapV to forward exceptions to the zero page handler if the vbr is moved (Graham) *** 8.2 (19.12.1998) - chg: the vector table (VBR) is write protected now, to catch invalid to accesses to it (MrLarmer), requires that the MMU is used by WHDLoad, NoVBRMove is not set and no freezer is detected by WHDLoad in memory - fix: problems with PAL and NTSC in conjunction with Picasso 96 monitor drivers fixed (Tim Kovack) *** 8.1 (13.12.1998) - chg: detection of HrtMon 2.24 implemented - fix: Access Fault in resload_Relocate removed - new: some changes and enhancements for HrtMon 2.25 - fix: Snoop works again for 68030 machines (was broken in v8.0) - chg: all unsaved registers after executing a resload function (D0-D1/A0-A1 as far as they don't contain a return code) will be destroyed explicit now, for compatibility reasons *** 8.0 (22.11.1998) not public released! - chg: if the write offset for a resload_SaveFileOffset operation lies outside the current length of the file and therefore the file must be enlarged before writing, WHDLoad tries now to fill this area with $ee, if there is not enough memory available for this, WHDLoad falls back to the old style by using dos.SetFileSize(), with the result that the enlarged area will contain random data - fix: deadlock in switch routine fixed which occured when lightpen has been activated (e.g. "Die Hard 2") - fix: cia's will be completly initialized by WHDLoad now, and also better handled during switch - chg: also disk images will be cached dynamically now - fix: small bug in DIC removed (MrLarmer) - fix: resload_LoadFileOffset accepts also parameters upto the BaseMem bound now (DarkAngel) - new: flag WHDLF_NoKbd to allow NoVBRMove also for programs which doesn't check the keyboard from the PORTS ($68) interrupt - new: function resload_Relocate to relocate an standard AmigaDOS executable in memory, beware: uses a lot of stack and is not tested with all hunk types, see autodoc for more details - new: function resload_Delay to wait a specified time or a button to be pressed (time independed from PAL/NTSC display) - new: slave can request expansion memory (fast memory) now ATTENTION: slave structure has been enlarged - new: TDREASON_MUSTREG (DarkAngel) - new: WHDLTAG_IOERR_GET for usage with resload_Control to check if all names have been fitted in resload_ListFiles and to distinguish a file with a size of 0 from an nonexistent file in resload_GetFileSize (Harry) - new: option D/S for debuging, if enabled the software freezer (Hrt/TK) will be activated at the first instruction of the Slave - new: flag WHDLF_EmulLineA to emulate Line-A opcodes if the vbr is moved - new: function resload_DeleteFile to delete files (MrLarmer) - new: program SP which can save iff pictures from the new coredump files created by WHDLoad, works for most standard picture formats (Snoop* is required of course) - new: sources for all programs except WHDLoad included, please note that most of them are copyrighted - chg: many internal changes, also the problems with BloodMoney and CelticLegends seems to have been solved - fix: Restart should work reliable now - fix: small bug in ITD removed (Harry) *** 7.0 (01.07.1998) - chg: new revision scheme, version number of WHDLoad is now equal to highest possible version number for a Slave, separate build number - new: new tag for resload_Control to setup a function to restore destroyed custom registers during switch between OS and installed program (Harry) - chg: functions resload_LoadFileOffset, resload_SaveFile and resload_SaveFileOffset can be used with a filesize of zero now (Harry) - chg: the functions resload_Protect#? are checking now if the protected area matches the current SSP, if it does, WHDLoad will stop and show a requester "Unacceptible Arguments", this has been build in to reduce the likelihood of a Double Bus Fault, thanks to Mr Larmer for reporting this problem - fix: check of workbench arguments a bit more secure now - new: new Slave flags WHDLF_Req68020 and WHDLF_ReqAGA, new TDREASON_REQ68020 and TDREASON_REQAGA, new tag WHDLTAG_CHIPREVBITS_GET for resload_Control, to handle programs which require a MC68020 or the AGA Chip Set (Mr Larmer) - new: TDREASON_MUSTNTSC and TDREASON_MUSTPAL to terminate if the installed program requires a specific display which isn't enabled (Mr Larmer) - chg: PreLoad now collects also files from sub directories - chg: example sources (slaves/imager) reworked to assemble with Asm-Pro v1.12 (still some problems with labels in macros, but this is a problem of AsmPro) - new: support for additional NewIcon in Install Templates (MrLarmer) the following problems are known and still unsolved: - no exit possible from BloodMoney/CelticLegends - Nitro does hang with PRELOAD enabled, use WHDLoad v0.608 or disable PRELOAD to go around this *** 1.92 (08.06.1998) - fix: problem with Clown'o'mania (©1989 Starbyte) does not work with WHDLoad v1.1 has been fixed, the reason was a different initialized ciaa.icr.tb bit, to fix this and future problems WHDLoad does init the CIA interrupt mask explicit now, all interrupts are disabled except the CIAA SP interrupt (used for keyboard) which is enabled. I have tried some installs and it seems to work ok, anyway it is possible that some installs may no longer work because they may expect that some interrupts are already enabled on startup - these installs must be fixed for this new WHDLoad - new: TDREASON_OSEMUFAIL to support the OS emulation module - chg: the message for instruction stream access faults has been corrected (Harry) - new: option Expert/S, this option must be enabled to make core dumps using the DebugKey/K/N now, the option has been added to avoid that normal users press the wrong key by mistake (Harry) - new: options ExecuteStartup and ExecuteCleanup to execute a command at start and at end of WHDLoad *** 1.1 (25.05.1998) - fix: stack usage on resload_Decrunch and resload_LoadFileDecrunch decreased, speed of RNC1 decrunching improved a lot, its a really a good idea to patch the decrunch routine in games with resload_Decrunch if it uses RNC1 ! - new: DIC can create partial images now (option SIZE/K) - new: WHDLoad performs now a "SetClock LOAD" at exit - fix: some minor fixes in the gfx degrading routine - fix: disk-dma check in os-switch modified, it switches dma off now instead of showing the white-color-cycle-bug-screen (because it's nowhere documented else: red - copper cannot switched off, blue - ints cannot switched off, green - blitter is busy but dma is off; after pressing LMB WHDLoad continues but it's undefined what will happen) - new: new program "ITD" (ImageToDisk) to write disk images back to floppy or RAD: can format the disk, can write partial images, no verify, see documentation for more info - fix: problems with second copperlist on reactivating os graphics - new: blitter wait check if SnoopOCS/ECS/AGA enabled - new: if available whdload will use reqtools.library instead intuitions requesters now, because they are screen centered and have better keyboard support - new: with SnoopOCS/ECS/AGA all cpu used custom registers will be dumped now - chg: tooltypes NTSC/PAL will now affect the display init routine (Intuition.OpenScreen()) therefore you need the appropriate monitor files installed - new: functions resload_Protect#? to snoop memory accesses in various ways, very useful for developer... (requested by MrLarmer) - new: function resload_LoadFileOffset which is similar to resload_DiskLoad but for files - new: flag WHDLF_NoDivZero for ws_Flags to prevent WHDLoad from exit if a "Division by Zero" exception occurs - chg: all resload functions has been reworked to no longer modify registers a5 and a6, this solves problems with demos/games which expect that the contens of these registers are valid also in an interrupt, for example SWIV expects a5=global-data and a6=custom-base, the only exception from this rule are the functions resload_Decrunch and resload_LoadFileDecrunch when working on rnc packed files, in this case a5 is temporarly used (it is a bit hard to rewrite the whole decrunch routines and I'm low on time of course...) in general it is a good idea if the program you like to install does'nt work with WHDLoad to check the interrupt routines, if the interrupts make assuptions similar the above (e.g a4=anything) you have to disable the interrupts before calling any resload function and reenable after - chg: WHDLoad checks now the current stackpointer each time the Slave calls a resload function, if the SP is lower than $180 or outside the BaseMem area the program will be terminated and WHDLoad shows a requester about, this should avoid problems like destroyed vector tables or deadly double Access Faults in conjunction with an enabled Memory Protection - chg: it is no longer required that SetPatch is executed before WHDLoad on 68040/68060 equipped machines because WHDLoad determinates the current CPU itself now - chg: for 68060 only (68040 also??): the newer versions of 68060.library (starting v41.1) does remap the memory $0-$1000 (the first page) to a fast mem location using the MMU, this will make major problems if the option NoMMU is activated: in this case WHDLoad will not change any MMU related registers and this address translation will remain intact !!! if the installed program tries to use this area for any dma actions (e.g. a copperlist in this area) the results will be unforeseeable, because the cpu will read/write the fast-mem location and dma's the real chip-mem which contains random data !!! I don't want to disable the NoMMU option for 68040/060 machines, therefore the following has been changed: the longword at address $1000 does now contain the value -2 which is the copper instruction cend (was on address $300 in earlier WHDLoad versions), the minimum BaseMemSize has been increased to $2000. If NoMMU is selected and the above mentioned address translation is active an requester will inform the user about the existing problem. - new: options ButtonWait/S and Custom1/K/N ... Custom5/K/N for slave specific things (requested by Harry) - chg: WriteDelay/K/N does also affect the FileLog/S feature now, this means that after each write to the logfile WHDLoad will wait the time specified via WriteDelay/K/N to minimize the danger of filesystem corupption - chg: install scripts has been reworked, PLEASE check them ! - chg: sources has been changed and new added, PLEASE check them ! - chg: WHDLoad is shareware now, read documention for more info *** 0.608 (03.09.1997) - new: all configuration stuff rewritten, global config file see guide for more informations - chg: SURVIVERESET removed (obsolete I think) - new: support for HrtMon 2.1 and ThrillKill 1.04 implemented (auto detect and NMI forwarding) - new: function resload_Control to get the AttnFlags - fix: enforcer hit in resload_SaveFile removed - new: function resload_SaveFileOffset for partial writes - chg: ws_DontCache works now also for preload of disk images file preload does no longer loads "(ReadMe|#?.(info|slave))" ("Disk.#[0-9]" also not as before) - chg: better support for NTSC amigas (Raymond Earle) - new: option PAL/S to force PAL display on NTSC amigas - new: options Cache/S,DCache/S,NoDCache/S to allow full user control over the caches - new: using MMU for memory protection and cache managment (!!!) - new: Quit/Debug-key managment trough new options QuitKey/K/N and DebugKey/K/N, every new slave should support this via the slave version 4 extensions ws_keydebug and ws_keyexit, see one of the example slave sources for an example - chg: on whdload is writing data (resload_SaveFile) it will do a special check for RAM: now, so it is possible to use RAM: as installation destination now - chg: macros patch and patchs optimized - new: more sources added to the package *** 0.322 (25.06.1997) - fix: sometimes whdload has quit on random keystrokes if the vbr was moved - fix: deadlock problems in OS-Swap routine - new: option COREDUMP to make a dump even on TDREASON_OK *** 0.288 (16.06.1997) - fix: sprite dma switching off - chg: ws_ExecInstall can be set to zero now (disables SURVIVERESET) - new: ws_keys, whdload manages quit key self (see autodoc) - new: whdload handles all MMU stuff, result is that whdload is now Enforcer/Cyberguard and VMM proof - new: ** vbr shadowing ** (see autodoc) tooltype NOVBRMOVE added to disable that - new: slave flag EmulTrap for ws_Flags - chg: execbase changed from $ffff0000 to $f0000001 (any use of this should create an "Address Error" exception) - chg: Data Cache is always off and cannot be enabled via resload_SetCACR - fix: resload_FlushCache fixed for 040/060 - fix: register dump changed a lot - new: exitcode TDREASON_WRONGVER indicating a bad program version - chg: all registers have now a defined value at startup - fix: filelog output fixed for data writes - new: Patcher, FindAccess, Cmp added to the package autodoc, include and guide updated Install templates added *** 0.163 (02.03.1997) - fix: better handling of trackdisk.device, headpos is flashed always now - new: trackdisk errors as text instead numbers - new: whdload option NOAUTOVEC (see whdload.guide) *** 0.147 (31.01.1997) - fix: on slave load failed, we were in an endless loop - new: function resload_CRC16 implemented *** 0.143 (07.01.1997) - fix: restart point changed, that slave will be reloaded (static vars) - new: status of cia timers added to register dump - new: cia timer handling improved - fix: ISP/MSP highlighting on user-mode - chg: file cache changed, resload_GetFileSize from cache, resload_SaveFile to cache, now ws_DontCache is nearly obsolete - chg: PRELOAD changed, caches now automatically diskimages and files, independent from WHDLF_Disk - fix: resload_SaveFile checks now for disk validation and free space to avoid deadlocks - fix: file loading from cache has always copied one byte to much - fix: WRITEDELAY parsing on CLI-start - new: resload_DiskLoadDev for install slaves (see autodoc) - chg: better error detection on slave-whdload interface - chg: Macros in whdload.i improved *** 0.88 (24.11.1996) - file logging implemented - register dump improved - dump file names changed (now PROGDIR:.whdl_#?) - on TDREASON_DEBUG sr and pc now taken from primary/secondary result - InstallBB added to the package *** 0.41 (05.09.1996) - first public release