home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Amiga Plus 2000 #5
/
Amiga Plus CD - 2000 - No. 5.iso
/
Tools
/
Dev
/
Barfly
/
Release-Notes
< prev
next >
Wrap
Text File
|
2000-01-01
|
21KB
|
779 lines
o BASM Todo
- Fix GlobalOptimize problem that happens with a certain source
for Markus Schmall but though I have the same source I am unable
to reproduce it....sigh
- Fixed PIR(was 1013) to 1023...what a fuckup.
- Add PPC mtspr/mfspr Machine/Supervisor check
o BASM Changes
V2.16
- Fixed:
o move.l #$ffxx????,d0 optimize bug (Bert Jahn)
o dc.d,dc.s stopped the parser afterwards because the
A0 register was trashed.
V2.15
- Fixed:
o The mffs opcode was sort of random.
Urgsss...
V2.14
- Changed:
o Removed the barfly.library usage
V2.13
- Added:
o Added a lot 750 special registers
o Added G4 special registers
o Added Altivec support.
V2.12
- Fixed:
o subic was assembled as subi
V2.11
- Fixed:
o bCClr caused a parser errobut
- Added:
o 16bit xref support
V2.10
- Fixed:
o Really dumb btst #AFF_68020 bug instead #AFB.
Nobody has reported this in the last 5 years...oh well:-)
o lwarx->lwa...could also happened with other instructions
- Added:
o fpecr,vrsave registers added
- Changed:
o QUAD handles 16 byte alignment now(though for hunk format it doesn`t make sense)
* cnop max is now 32 byte alignment
V2.09
- I forgot:-)
V2.08
- Fixed:
o PPC relative16 size check
o FNOP instruction size increase was wrong which caused wrong addresses
calculation(B. Nost)
V2.08
- Fixed:
o xref_hunk long->word access in the elf object module
V2.07
- Fixed:
o Global Optimize xref problem
- Changed:
o absref16,8 -> dext16,8 (SLINK)
V2.06
- Added:
o ELF object state
o (xdef symbol - normal symbol) & x didn`t work because of eor l_xdef.
Now this bit is removed because it has no valuable information in that stage.
V2.05
- Fixed:
o Deadlock problem
o IDNT was very *dangerous*. It overwrote memory areas
o Fixed mttbl,mttbu spr number
o mullw
- Added:
o Release version disables the PPC because the ELF object output
stage isn`t done yet and the PPC part is custom for my needs.
It`s not completely done.
o mmcr0 for 604e
o mfmmcr0
o mtmmcr0
o pir
o mfpir
o mtpir
o eascii eor-value,"string":-)
o Added PPC supervisor checks
- Changed:
o Now a >=68020 is needed
V2.04
- Fixed:
o "cmpw" parsing bug fixed
o added "+,-" to branch parser
o added 603e tlbli and tlbld instructions
V2.03
- Added:
o "extlwi" PPC alias
o "extrwi" PPC alias
o "rotlwi" PPC alias
o "rotrwi" PPC alias
o "slwi" PPC alias
o "srwi" PPC alias
o "clrlwi" PPC alias
o "clrrwi" PPC alias
o "clrlslwi" PPC alias
o "rotlw" PPC alias
- Fixed:
o Random write bug introduced by the mf/mt aliases.
Used A6 instead of A5
o cmp(lk)i encoding in certain cases
o currentline set symbol output cyberguard hit in listingmode bug
V2.02
- Added:
o "mr" PPC alias
o "not" PPC alias
o mf????? PPC alias commands
o mt????? PPC alias commands
o fixed mffs,mtfs register parsing to use fpuregister parsing
(PPC is sick)
- Fixed:
o "cmp" 3 operand format
V2.01
- Fixed:
o Fixed a deadloop parser problem when used something like equr? ? != p
o Fixed a problem with mffs only accepting fpu registers
o Fixed problems with fpu load and store commands with symbol fpu
registers.
o Fixed cmpi encoding problems
V2.0
- Fixed:
o Global Optimize Problem with mungwall and xdef label + doubledefine
(Borge)
o bfxxx bug with a Width of 32...
o fmovem.x <ea>,SINGLEREG bug
o fmovem.x SINGLEREG,<ea> bug
o fixed a float #x sideeffect which may cause bad code..
o OD00 was checked by #0,an optimization instead of OD01
o align was broken...stupid me. Has no effect to cnop
o in some cases (xn,yn) created yn.w instead of yn.l
o fixed a problem with fmovem.l
o fixed dxx.b missing problem(PPC)
o fixed indirect register limitation(PPC)
o fixed wrong argument order for andi and ori commands(PPC)
o bclr* codes wrong..fixed
o bclr ;comment parsing bug fixed
o fixed wrong mfspr,mtspr spr ordering(IBM's engineers are fucked in their brains)
o hid0,hid1,dmiss,dcmp,hash1,hash2,imiss,icmp,rpa
spr registers added
o Listing problem with overlong lines..
o Fixed a lot broken PPC codes...too lazy to list them
- Added:
o PowerPC assembler mnemonics
o ppc601,ppc602,ppc603,ppc604,ppc620 switches into the different
machines
- Changed:
o cmp.x #y,dn now creates cmpi.x #y,dn instead of cmp.x #y,dn
V1.131
- Fixed:
o add.x #x,dn is now compiled to addi.x #x,dn...really weird.
Though i can't find that documentation anymore it was illegal
in the new documentation....but it works.
o Yet another time...fixed ds...now finally:-)
(Markus Schmall and Hans Raaf)
- Changed:
o <> handling "IN" a macro parameter...not the behaviour at the begin/end
V1.130
- Fixed:
o Yet another time...changed something and broke something else.
(dx,blb and incbin)..happens when you don't have time working on it.
V1.129
- Fixed:
o Serious Memory-Crash problem with very large block allocations
with dx,blb and incbin.(Borge)
o some string problems with dc/db(Resource:-)
o reloc32short align bug...forgot a 1 somewhere:-)
- Improved:
o Now allow section "text" without additional parameters for
SAS asm code.
V1.128
- Fixed:
o IFD/IFND didn`t check the equr symbol list(Bert Jahn)
o treg equr a0
move.l ([1234,treg],5678),d0
test = 10
move.l ([1234,treg],test),d0
bug fixed.(Bert Jahn)
V1.127
- Fixed:
o Sorry..rs symbols were hosed by some mistake
V1.126
- Fixed:
o printx "\n" problems in macros
o errorstring \0 end problems
o FINALLY....tarahhh(party) found this -iinclude: bug that sometimes
caused that includes weren't found. The reason was a not
initialized flag in certain circumstances and it was pretty
hard to reproduce.(matthias,mjl)
o RS/FS/SO problem caused by a using a new 32bit mul routine.
o Smalldata,optimizing,xref problem(matthias)
o Global Optimize Problem with constructs like this.
start: jsr a
rts
blk.l 50000,0
a beq start ;random offset
(Markus Schmall)
o pea a(pc),pea a.l->a.w optimizing possible again and without
64kwarning.(chrisr)
o Fixed -s1 debug hunk source filename problems...hopefully(Denny)
- Added:
o Now there are new optimize options to allow a finer control
over the -OD0 optimizing.
V1.125
- Fixed:
o Missing check in InitError/Warning
V1.124
- Fixed:
o Cleared some tempflags though these are needed closely afterwards.
Ooops:-)
V1.123
- Fixed:
o RS.b has a bug with numbers >$ffff
o dsb didn't work because of some simple syntax bug..i'm sure it
worked some time. Mhhhh
o Fixed several problems with the Arexxmode
- Added:
o BINITWARNING,BGETWARNING,BNEXTWARNING Arexx commands
V1.122
- Fixed:
o Showing internal labels when using -h
o RS,FO,SO changing the last label when there's no label in
the current line.
o String Offset in the error output when using the -Z option
o Mungwall hit in connection with the barfly.library and a flush.
(Sorry...)
- Improved:
o Movem didn't handle register symbols.
o Multi-Assign support and the new option -F that switches
back into the older way to store paths as locks instead of
strings.
- Added:
o the command:filecom "string"
o Optimize move.w -2(a0),a0
V1.121
- Fixed:
o a bsr+rts optimizing problem when bsr was a bcc and this bcc was
optimized to a nop
o incdir problems
V1.120
- Fixed:
o CALLM#? macro problem
o fmovem.l #$xxxxxxxx bug with more than 1 register
V1.119
- Fixed:
o Serious optimize bug
for example cmp.l #0,10(a0)->tst.l 10(a0)+dc.w 10
- Improved:
o Cursor position for Warnings
o Align now accepts values up to 16...
o Documentation
- Added:
o Warning for instructions that access different sections by
pc-relative addressmodes.
V1.118
- Fixed:
o Forget to unlock -i<includepath> paths.
o -l,lisfile problems...sorry
o several errors in the manual
(More to follow...)
V1.117
- Fixed:
o Possible problems with writing of xdef,xref symbols
that can lead to write file sizes that have nothing
to do with the real file size. Used word instead of long.
V1.116 (Stupid)
- Fixed:
o Possible enforcer hit in certain circumstances
V1.115
- Fixed:
o Silly bug in verbose output. The optimize passes
were accidently output when multiple pass wasn't
even activated.
V1.114
- Fixed:
o Option OD5 had no function
cmp.l #0,ea wasn't optimized by accident
V1.113
- Fixed:
o some faults in the manual
o Arexx crashes/stops
- Changed:
o BGETERROR Output format..instead of : it's using | as a
seperator.
V1.112
- Fixed:
o Basm writes an incorrect SAS D1 Hunk when the filename is
long word aligned....this leads to a serious BDebug crash
while loading.
o Verbose output bug with the global optimize counter when
Optimize wasn't activated.
V1.111
- Fixed:
o Basm forgets in the .while Macro the conditional branch.
Parser problem...could be visible in other things.
o Several serious code bugs i added close before the final
release into the fpu command handling(Not by intention:-b)
o RTM,RTD error output improved
o cachefile.library crash in a certain situation(read error)
o enabled cpush for 68060
o bfextu parsing problem...i know it worked sometime:-B
o divul.l/mulx.l 68060 error detection was suboptimal:-B
o macros starting with the substring FMOVECR fooled the parser
o double symbols in symbol hunk
o bkpt machine detection for 68010/20 broken
o cas2.b bug
o reworked processor/instruction table in the manual.
It's worth it now:-B
o FMOVE.S #0.01,FP0 error...includes a lot weird float #x problems
o Crash while listing symbols that include floating point
- Added:
o cargs <#offset,>symbol1,symbol2,...
o equ.x equ.d equ.s equ.p
- Improved:
o Docs about optimizing and the miserable Instruction Table.
V1.110
- Fixed:
o Under certain circumstances read problems with precompiled includes.
Better recreate your precompiled includes.
- Improved
o Precompiled Hash error reason.
V1.109
- Fixed:
o Possible Enforcer Hit in BFINS
V1.108
- Fixed:
o ([Base,an],OD) and optimize on results in buggy code
V1.107
- Fixed:
o (H.MSG).l parsing problems
o link a5,#-4<TAB><TAB>..bug
o Problem with SET and Labels..(Neil Blackett)
o BASM crash with Snoopdos 1.7 when started from the WB...
Wondering that i haven't noticed it earlier.
o $01 Bug
o BASM Format problem
- Improved:
o Error location output
o Now it doesn't ignore >> in macro parameters
o Now accepts (bd,dn) if BD Width = 0..in this case Width=.w is assumed.
V1.106
- Fixed:
o dc.b " " Bug
V1.105
- Fixed:
o Now it outputs relocatable also if XREF symbols are accessed.(Alex)
o Problems with the position of Bytexrefs,moveqxrefs and damaging data
- Changed:
o dc.b|w|l "" behaviour(now prolly more compatible to this broken standard)
o y(pc,xxx) now allowed with y=value..
o movem now accepts d0-7 and a0-7 terms
- Improved:
o Multipass error condition detection(northway)
- Added:
o move.l -4(an),an->move.l -(an),an optimizing(Alex)
o Option -J that creates the file env:BDebugProgram
that BDebug checks when it is started with no parameter.
o command BDEBUGARG that can be used to define an argument
in env:BDebugProgram
V1.104
- Initial Release
##########################################################################
##########################################################################
##########################################################################
##########################################################################
o BDEBUG Todo
- DissWindow on lvo function pop up an autodocs window if requested
- divul.l (a0,d0.l*4),d3:d2 EA calculation problem..basicly for all 4 byte commands(should be fixed now too)
- Fix GCC multi source problem
- Fix cybergfx penmode problem in edit mode
- PC-Segtracker Support
- Rewrite everything to C....well mostly.
o BDEBUG Changes
V1.692
- Changed:
o More sophisticated FPU usage/detection
- Added:
o TRAP_LIB_RESTORE
V1.691
- Added:
o TRAP_LIBD0_D1
V1.690
- Changed:
o Removed the barfly.library usage
o Added segtracker support into the regwindow(quick hack)
o Added TRAP_FUNC emulation extensions...not satisfying yet..but well
V1.689
- Fixed:
o Listview problems on truecolour screens
(I finally had to fix that because of the
CVisionPPC WB 16bit screen..fast enough now:-)
V1.688
- Fixed:
o CRASHEDTASK config cmd enforcer hit
V1.687
- Fixed:
o Mungwall in the Alias management(Bjorge)
V1.686
- Fixed:
o bfxxx bug with a Width of 32...
o FD Reader cut off names with digits
o Allocvec in the tasklist generator had a broken
TRUE test
o Fixed divu->sbcd disassemble bug
o EA calculation with move and if the 1st operand needs an extra word/long
o Barfly.Include.S changes
o some comments(bad english)
o STRUCTURE name,offset where offset != 0
for example asl.i/FileRequester used this and
the StructWindow showed the wrong values/offsets
- Improved:
o Log Output also shows registers now
(But the log output style sucks in general)
o I always wondered why i couldn`t set an illegal into a lib to crash in LibInit(RamLib)
task but i was always too lazy to find the reason until now:-)
ramlib task has a tc_TrapCode of NULL which is illegal and because it was an unknown
routine i skipped it in my patch task trapcode function.
V1.685
- Fixed:
o Memory Window search/next requester was broken
V1.684
- Fixed:
o mulu.l ,dn n always 0 bug
o SnoopMemory Window didn`t work at all.
Wondering why nobody noticed:-)
Was caused by a source cleanup months
ago...don't even know when i did that.
But it worked one time...seriously:-)
o Multiselect crashes with certain menus
o Fixed crash with Memory Fill
V1.683
- Fixed:
o "OpenScreen\n" config file parse error.
V1.682
- Fixed:
o potential enforcer hit in the config-loader
V1.681
- Fixed:
o Mungwall hit in connection with the barfly.library and a flush.
(Sorry...)
- Improved:
o I never really liked the Task listview and the code was already
that old it started to smell....so i rewrote it.
Now you can detect a process at the [] after the Task name.
The [] extension shows the process number and the cli name if
these exist.
o Now the information windows that haven't reacted on the cursor
buttons react now
V1.680
- Fixed:
o Ugghh.accidently i fucked up the Breakpoint List mouseclick
o Select Screen Trouble
o CyberGfx Screen problems
o misc stuff
- Added:
o Progdir: support
V1.679
- Fixed:
o BDebug crashes on a 060 when it activates Cyberguard's
hit catching.
V1.678
- Fixed:
o fmovem.l #$xxxxxxxx bug with more than 1 register
- Added:
o SysiHack support
V1.677
- Fixed:
o Debug Task problem with 68060...(i must have programmed that
in delirium)
V1.676
- Fixed:
o Mouseclick bug in a big regwindow with the (EA) field.
- Added:
o Shift-Tab goes through the windows backwards
V1.675
- Fixed:
o Broken Screenselector.
o RegWindow ID bug
V1.674
- Fixed:
o Wrong 1st address in the Resident window
o Config Error Linenumber bug
o 68040LC/68060LC crash
- Improved:
o SourceWindow status output improved for PCs with no LINE
information.
o Significant Speed Up in the Sourcewindow offset/line search.
Visible with bigger programs
o Smartrefresh for Warning msgs in the Sourcewindow
- Added:
o ADDSTRUCTFILE= config command...allows the addition of
custom struct files into CUSTOM/
o NEWWINDOW menu option in the Structwindow so that if you
use links(like in Node) it should open a new window instead
of using the current.
o NEWWINDOW new STRUCTFLAGS= config option
V1.673
- Fixed:
o Uggghh...long time bug..disassembler outputs fmovecr #x,d0
instead of fmovecr #x,fpn
- Added:
o Cyberstorm 060 Support added
V1.672
- Fixed:
o Crash when configurated that Breakpoint/Watchpoint
windows are opened by the Barfly.Debugconfig file.
Really fixed ? I don't even know why it crashes...
It crashes when i get the Supervisor register and i
have not the slightest clue why it happens only when
these 2 windows are opened. I removed the Supervisor
routine because it's not really necessary to load
these registers at that time.
- Improved:
o SnoopWindow full dimension parsing.
V1.671
- Fixed:
o Watchpoint Argument Check Hit
o A1200+88x,BDebug crashes when it loads a file.
o Show CLI problems with Processes that don't have a CLI-Ptr
V1.670
- Fixed:
o ScreenFont config parameter parsing bug i added saturday
o GCC Sourcelevel problem with the LINE Cursor position.
o Source File Pathrequester Hit bug fixed..forgot to save a register
o Next Task Debug crash...sigh..was caused by the key code i added:-(
o Memory Loss
o BASM Debug Loading Problem
- Added:
o Multi-Select now supported
V1.669
- Fixed:
o Crashed Task Hit and Crash Bug fixed..sorry added it this Saturday.
Never change working code.
V1.668
- Fixed:
o The Bug that Enforcer Catch is accidently not activated.
V1.667
- Fixed:
o The Bug that the Quit Menu isn't disabled but the Quit Button is
disabled....i added that last week.
o Bug with Corrupt Window Ptr(WShell) when BDebug is opened on
a PubScreen
o CloseGadget could break Quit Button/Menu disable
i had added 2 weeks ago
- Added:
o Select Screen Menu to the control window that allows to select
screens and PubScreens.
o If BDebug is started with no parameter it checks if there's
a env:BDebugProgram and starts that program.
V1.666
- Initial Release