Commit Graph

29 Commits

Author SHA1 Message Date
Jonas Maebe
a9b5e27bc1 * finally got around making GetNextInstruction return false when
the current pai object is a AsmBlockStart marker
  * changed a loop in aopt386 which was incompatible with this change
1999-10-23 14:44:24 +00:00
Jonas Maebe
4590ff9524 + some comments 1999-05-08 20:39:02 +00:00
Jonas Maebe
05acbe4077 * fix for crash when the first instruction of a sequence that gets
optimized is removed (this situation can't occur aymore now)
1999-04-18 17:57:17 +00:00
peter
2bfdf91c5c * assembler inlining working for ag386bin 1999-03-31 13:55:03 +00:00
Jonas Maebe
64c509adb8 * fixed crash when there are two asm blocks right after each other 1998-12-29 19:58:27 +00:00
Jonas Maebe
8cf4da6d5e + optimize pascal code surrounding assembler blocks 1998-12-29 18:48:15 +00:00
peter
1b369bb4bb + globtype,tokens,version unit splitted from globals 1998-12-11 00:02:37 +00:00
Jonas Maebe
f2da32cd91 * changed optimizer switches + cleanup of DestroyRefs in daopt386.pas 1998-08-19 16:07:36 +00:00
Jonas Maebe
a3b5867cb8 * removed $ before and after Log in comment 1998-08-06 19:40:25 +00:00
florian
76ee202b37 * some fixes for ansi strings
* $log$ to $Log$ changed
1998-08-05 16:00:08 +00:00
Jonas Maebe
2c0d277164 + disposal of labeltable after optimizations 1998-08-04 16:27:18 +00:00
Jonas Maebe
892964eaf2 * split the optimizer 1998-07-23 19:31:18 +00:00
Jonas Maebe
879b7d0288 * change "pop reg/push reg to "mov (%esp), reg" with -Ox, remove otherwise 1998-07-15 16:06:00 +00:00
peter
1bffd4e290 * released NEWINPUT 1998-07-14 14:46:36 +00:00
peter
c2d5abdfed + targetcpu
* cleaner pmodules for newppu
1998-06-16 08:56:17 +00:00
peter
4fc1780fe4 + $ASMMODE 1998-05-30 14:31:02 +00:00
Jonas Maebe
837fa04627 * final bugfilx for mov optimizes, remake3 with optimizations works again! 1998-05-24 18:42:37 +00:00
Jonas Maebe
9f0c2e3a26 * 2 bugs fixed in mov peepholeoptimizes 1998-05-24 15:20:59 +00:00
peter
b3865ae063 + aktasmmode, aktoptprocessor, aktoutputformat
+ smartlink per module $SMARTLINK-/+ (like MMX) and moved to aktswitches
  + $LIBNAME to set the library name where the unit will be put in
  * splitted cgi386 a bit (codeseg to large for bp7)
  * nasm, tasm works again. nasm moved to ag386nsm.pas
1998-05-23 01:20:53 +00:00
Jonas Maebe
6c86e16f46 * bugfix in a_mov optimizations; completed bugfix of "sub $2,esp;...;movw reg, y(%esp)" 1998-05-10 12:06:30 +00:00
pierre
6fc80b783f * better position info with UseTokenInfo
UseTokenInfo greatly simplified
  + added check for changed tree after first time firstpass
    (if we could remove all the cases were it happen
    we could skip all firstpass if firstpasscount > 1)
    Only with ExtDebug
1998-05-06 08:38:32 +00:00
pierre
5cdd60cac8 + added some code for ansistring (not complete nor working yet)
* corrected operator overloading
  * corrected nasm output
  + started inline procedures
  + added starstarn : use ** for exponentiation (^ gave problems)
  + started UseTokenInfo cond to get accurate positions
1998-04-29 10:33:40 +00:00
florian
bef30e693c * fixes of Jonas applied 1998-04-23 21:52:08 +00:00
peter
6af82802c2 * fixed $ifdef regalloc 1998-04-21 11:30:14 +00:00
Jonas Maebe
2f9ba77432 *** empty log message *** 1998-04-16 16:53:56 +00:00
Jonas Maebe
2938e014d2 * fixed bug where "imul 12,reg" was replaced with a wrong lea sequence 1998-04-08 19:12:28 +00:00
florian
8cb9f5ea5b * aopt386 compiles with TP
* correct line number is displayed, if a #0 is in the input
1998-03-29 17:27:58 +00:00
florian
2d790e0158 * secondin bugfix (m68k and i386)
* overflow checking bugfix (m68k and i386) -- pretty useless in
    secondadd, since everything is done using 32-bit
  * loading pointer to routines hopefully fixed (m68k)
  * flags problem with calls to RTL internal routines fixed (still strcmp
    to fix) (m68k)
  * #ELSE was still incorrect (didn't take care of the previous level)
  * problem with filenames in the command line solved
  * problem with mangledname solved
  * linking name problem solved (was case insensitive)
  * double id problem and potential crash solved
  * stop after first error
  * and=>test problem removed
  * correct read for all float types
  * 2 sigsegv fixes and a cosmetic fix for Internal Error
  * push/pop is now correct optimized (=> mov (%esp),reg)
1998-03-28 23:09:53 +00:00
root
d1a1f501c8 Initial revision 1998-03-25 11:18:12 +00:00