40915d76c6
Name change : tail_em.vend -> end_em.s
ceriel1985-09-19 14:03:13 +00:00
8668e313f8
bug fixed: same problem as in r1.3, now with csb. Solution: only do block fusion if first block end on bra.
bal1985-09-19 11:14:55 +00:00
a18c5dd9c4
bug fixed: last instruction of first block in block fusion optimization can be a conditional branch. (bug detected by Ceriel).
bal1985-09-18 16:27:29 +00:00
53f6b51cde
sprintf does not return pointer to string so the assert(s==string) does not make any sense.
bal1985-09-06 11:39:06 +00:00
6a4d9703cc
exit(0) added (trouble with make)
bal1985-09-06 11:38:00 +00:00
0d9459bdd6
MAXSTAB set to 2500 (problem with argument list of ROM for new C front-end).
bal1985-09-06 11:06:32 +00:00
5d86ead6c0
There was a bug in the "str $1==0" instruction. It said: str $1==0 | source4 | remove(ALL) "jsb\t.strlb" | | | It should of course have been str $1==0 | STACK | jsb\t.strlb" | | |
ceriel1985-07-21 11:34:33 +00:00
7b4b53af21
bugs for ngf 8 and fif 4 fixed (asp 16 --> asp 8).
bal1985-07-19 18:02:25 +00:00
2e5bf801a0
patterns for "loc loc cii/cui/ciu/cuu $1==$2" added.
bal1985-07-19 17:58:34 +00:00
8749fb1da8
Using new magic number for 'aal' archives.
keie1985-06-10 12:41:51 +00:00
a90254c045
Added AALMAG, for making distiction between 'led' archives with SYMDEF and archives made/altered with 'arch'.
keie1985-06-10 11:28:06 +00:00
0db1db10b8
Bug fixed: cg generated instructions like "and.l #2,a4", which are not allowed on the 68000. The condition "inreg($...) < 2" is added to every EM-pattern that accesses local non-register variables. So the ordering of the patterns for regvar/non-regvar is no longer important.
bal1985-05-01 12:53:33 +00:00
eb025dae5c
Added a message to the effect that 'local commons' can not be handled.
keie1985-04-29 11:55:30 +00:00
fa912e6f37
This version produces local commons in its symbol table. It cannot be used because 'led' con't handle that.
keie1985-04-29 11:50:24 +00:00
cec503a1b4
Changed trap number into bit shifted by number for ignore mask. The whole example is, at leat, unclear. The trap for EFOVL never occurs!
keie1985-04-25 13:37:59 +00:00
d90761dd33
fixup commit for branch 'unlabeled-2.2.1'
cvs2hg1985-04-23 12:55:02 +00:00
5472fafa56
Add hoc solution of .align problem.
keie1985-04-23 12:55:01 +00:00
a0a7a48c3b
The reading of the input is now machine independent.
em1985-04-22 15:13:24 +00:00