source: trunk/source/compiler

Revision Log Mode:


Legend:

Added
Modified
Copied or renamed
Diff Rev Age Author Log Message
(edit) @14727   9 years gb Move ACODE-CONSTANT-P from acode-rewrite.lisp to nx2.lisp, since the …
(edit) @14726   9 years gb Acode optimizers for 2-arg +,-,*,/. Generalize acode constant-folding.
(edit) @14725   9 years gb Constant-fold calls to SQRT. Need a bit somewhere.
(edit) @14707   9 years gb MISC-REF-C-DOUBLE-FLOAT: need to scale the index.
(edit) @14705   9 years gb Still a work in progress, but ... in progress, again.
(edit) @14703   9 years rme In nx-declared-inline-p and nx-declared-notinline-p, stop searching …
(edit) @14696   9 years gb Remove stray "(and nil ..." conditionalization from ACODE-OPTIMIZE-LOGIOR2.
(edit) @14695   9 years gb In x862-%ilog[ior|and|xor]2 (and in some subset of the analogous …
(edit) @14622   9 years rme Remove some left-over debugging stuff.
(edit) @14621   9 years rme x8632::tcr-bias is supposed to be 0 on non-Windows platforms.
(edit) @14619   9 years rme Merge shrink-tcr branch. This enables the 32-bit Windows lisp to run …
(edit) @14576   9 years gb Handle SETF function names in AUGMENT-ENVIRONMENT. Fixes ticket:807.
(edit) @14559   9 years gb In LDB compiler macro, just do a LOGAND if the bytespec's position is …
(edit) @14558   9 years gb Change subprims origin to #x04002000 on Android.
(edit) @14557   9 years gb %fixnum-to-double: make the single-float temp an arg, as a a …
(edit) @14556   9 years gb Let's try that again …
(edit) @14555   9 years gb Try to introduce the concept of "backend-subprims-bias", where a …
(edit) @14552   9 years gb Package name change.
(edit) @14548   9 years gb Don't try to share a single FFI interface file between LinuxARM and …
(edit) @14543   9 years gb Make things a little easier to cross-compile.
(edit) @14518   9 years gb No more conditional UUOs on ARM.
(edit) @14507   9 years gb Different.
(edit) @14476   9 years gz Fix use of nx-t/nx-null in decomp-form. Tweak some debug code.
(edit) @14472   9 years gb lisp-kernel/linuxarm/Makefile: just use -mfpu=vfp (not vfpv2) in …
(edit) @14463   9 years rme In NX1-FUNCTION, ensure that we have a lambda expression before …
(edit) @14459   9 years gb In DEFNXDECL FTYPE, record the canonical function name (not the list …
(edit) @14457   9 years rme Explicitly list x8664::allocptr (aka temp0) as a temporary in the cons …
(edit) @14439   9 years rme Change left over *nx-t* to (make-nx-t) in arm2-fixnum-overflow.
(edit) @14421   9 years gb systems.lisp, compile-ccl.lisp, l1-boot-2.lisp: Compile and load …
(edit) @14418   9 years gb Un-botch MAKE-NX-T.
(edit) @14417   9 years gb No more *NX-NIL*, *NX-T*.
(edit) @14406   9 years gb Start to bootstrap a change which uses "normal" acode operators to …
(edit) @14402   9 years gb Define ISB, DSB, DMB instructions. (A prerequisite to actually using …
(edit) @14382   9 years rme Update insert-ppc-label for pickier defstruct. …
(edit) @14375   9 years gb In the frontend: binary boolean operations assert their result's …
(edit) @14374   9 years gb Try to generate better code in some cases in the 64-bit LRI/LRIU …
(edit) @14372   9 years gb (concatenate 'string ,@sequences) => (concat-to-string ,@sequences)
(edit) @14369   9 years gz Make :unknown-type-in-declaration only be used for unknown types, no …
(edit) @14357   9 years gb Acode optimization for logior/logxor/logand in PPC backend, too.
(edit) @14355   9 years gb Use new stuff for logior/logxor/logand in ARM backend.
(edit) @14351   9 years rme Remove conditionalizations on ccl-qres.
(edit) @14348   9 years gb Define (obvious) compiler-macros on LOG{AND,OR}C{1,2}. Define …
(edit) @14343   9 years gb In DO-LEXICAL-REFERENCE: if we push a register, do so without updating …
(edit) @14342   9 years gb In x862-do-lexical-reference: look in the register map before pushing …
(edit) @14340   9 years gb %ILOGNOT, ASH backend stuff for ARM.
(edit) @14339   9 years gb %ILOGNOT, ASH for PPC. %ILSL vinsns: do logical (unsigned) compare of …
(edit) @14338   9 years gb Implement %ILOGNOT vinsns. Improve %ILSL visnsn (knock wood.)
(edit) @14337   9 years gb Implement %ILOGNOT, ASH in the x86 backend. (The macroexpansion of …
(edit) @14336   9 years gb BACKEND-USE-OPERATOR ("portable" way to do X862-USE-OPERATOR, …
(edit) @14335   9 years gb Optionally (mostly optional for bootstrapping reasons) expect the …
(edit) @14331   9 years gb Bootstrapping: replace unused operators with new ones for ASH, LOGNOT. …
(edit) @14323   9 years gz Back out of r14313
(edit) @14313   9 years gb Map :UNKNOWN-TYPE-IN-DECLARATION to UNDEFINED-TYPE-REFERENCE, so that …
(edit) @14309   9 years rme A somewhat inelegant change to nx1-typespec-for-typep: when called …
(edit) @14306   9 years rme Add :whine keyword argument to nx1-typespec-for-typep. Don't call …
(edit) @14304   9 years rme New function acode-xxx-form-p. Use it in functions …
(edit) @14303   9 years rme Indentation in x862-%logand2.
(edit) @14302   9 years gb Avoid register conflicts in ARM2-TWO-TARGETED-REG-FORMS, especially …
(edit) @14301   9 years rme Print "big" (> 100) numbers using #x notation instead of in decimal.
(edit) @14299   9 years gz Encode code coverage acode strings to save space
(edit) @14297   9 years gb Some small changes to ACODE-FORM-TYPE: try to intersect, don't return …
(edit) @14286   9 years gb Propagate r14113 from ARM branch to trunk.
(edit) @14258   9 years gz Defstruct changes: Get rid of *defstruct-share-accessor-functions*. …
(edit) @14240   9 years rme Remove leftovers (remnants of operator %fixnum-mask-to-natural) from …
(edit) @14235   9 years rme Abandon my attempts to do more low-level tricks with logand and logior …
(edit) @14231   9 years rme Open-code ASH in cases where the integer is a fixnum, and the shift …
(edit) @14229   9 years rme Re-enable some changes that try to do a better job on LOGAND and …
(edit) @14222   9 years gb Don't "naturally" align a record field if the field type's natural …
(edit) @14200   9 years gb Apply rme's fix to ARM2-UNWIND-SET.
(edit) @14196   9 years rme Remove remnants of *x862-tos-reg* hack (from r7938), which is …
(edit) @14195   9 years rme x862-stack-to-register: use regmap to avoid reading from the stack …
(edit) @14194   9 years rme Start at keeping track of which registers contain which stack …
(edit) @14193   9 years rme Add :call, :subprim-call, and :jumplr attributes to some vinsns.
(edit) @14187   9 years gz Fixes in acode code coverage (r14085, r14110, r14164, r14165, r14172
(edit) @14183   9 years gb 64-bit alignment in ff-call.
(edit) @14176   9 years rme Fix dopey typo.
(edit) @14175   9 years rme Arrange for the disassembler on Darwin and Windows to show r11 as …
(edit) @14171   9 years gb Darwin/ARM changes.
(edit) @14159   9 years gb Unscramble variable-c-frame stuff. Use it to implement #'%FF-CALL.
(edit) @14156   9 years gb FF-CALL on win64 is sufficiently different from other x8664 platforms …
(edit) @14147   9 years gb kernel-import for WAIT-FOR-SIGNAL.
(edit) @14121   9 years gb FTYPE declaration for %DEFINE-ARM-LAP-FUNCTION.
(edit) @14119   9 years gb Changes from ARM branch. Need testing …
(edit) @14086   9 years gz in decomp-afunc, handle circular refs
(edit) @14047   9 years rme Undo, for the time being, the effects of r13887. Those changes are …
(edit) @14044   9 years gz support for reporting code coverage of acode, needs more testing
(edit) @13971   9 years gb Define kernel import for WAIT-FOR-SIGNAL.
(edit) @13966   9 years gz Record code coverage acode string and positions
(edit) @13891   9 years gz Add DECOMP-ACODE, decompile acode into something more readable.
(edit) @13890   9 years rme Ugly additions to nx-logand-2-op to truncate a large integer operand …
(edit) @13887   9 years rme Make the compiler work a little harder to identify cases where LOGAND …
(edit) @13814   9 years rme Delete .cvsignore files since we no longer use CVS.
(edit) @13813   9 years rme nx1-combination: indentation
(edit) @13792   9 years gb New pre-GC notification mechanism.
(edit) @13782   9 years rme Remove some cruft.
(edit) @13675   9 years gz Rename feature ccl-0711 to ccl-qres
(edit) @13561   10 years gb In the vinsns and subprims for special variable reference, try to use …
(edit) @13538   10 years rme Push things around a bit in x862-vset1 to localize use of …
(edit) @13488   10 years gz Be more thorough in converting declaration types for typechecking.
(edit) @13469   10 years gb PPC2-DIGEST-SYMBOLS, X862-DIGEST-SYMBOLS: make the addresses vector in …
Note: See TracRevisionLog for help on using the revision log.