source: branches/x8664-call

Revision Log Mode:


Legend:

Added
Modified
Copied or renamed
Diff Rev Age Author Log Message
(edit) @6406   13 years gb Define a .SPnmkunwind for PPC; needs to be bootstrapped.
(edit) @6405   13 years gb Arrange to pass old_valence through exception handlers, so we're less …
(edit) @6404   13 years gb Squelch "foo is deprecated" warnings.
(edit) @6403   13 years gb Arrange to pass old_valence through exception handlers, so we're less …
(edit) @6402   13 years gb PATHNAME-DEVICE is NIL for physical-pathnames, :UNSPECIFIC for …
(edit) @6401   13 years gb LOGICAL-PATHNAME: signal a type-error if there's no logical-host. …
(edit) @6400   13 years gb :IF-EXISTS defaults to :NEW-VERSION if the pathname's version is :NEWEST.
(edit) @6399   13 years gb In SET-GF-ARG-INFO, clear the dt cache if APO changes.
(edit) @6398   13 years gb Use a CMOVE in EQL.
(edit) @6397   13 years gb FLOAT: arrrgh.
(edit) @6396   13 years gb Remove the extraneous "6"; don't want to know how it got there ... …
(edit) @6395   13 years gb Need to revive the old MKUNWIND, to keep PROGV from losing.
(edit) @6385   13 years gb Bump.
(edit) @6384   13 years gb Reassign registers: %rcx = %imm2, %r10 = %temp2. This basically means …
(edit) @6383   13 years gb Don't think that %ra0 and %temp2 are disjoint. (They aren't, and we …
(edit) @6382   13 years gb Use %imm2 - not %temp2 - when building a catch frame.
(edit) @6381   13 years gb Don't use %temp2 (which is aliased to %ra0, still used occasionally.) …
(edit) @6380   13 years gb x8664-imm-regs now include %imm2.
(edit) @6379   13 years gb Things which do shifts - including %FIXNUM-GCD - should use %imm2, …
(edit) @6378   13 years gb Things which do shifts - including %FIXNUM-GCD - should use %imm2, …
(edit) @6377   13 years gb When unwinding the stack and returning values, don't let %imm2 be used …
(edit) @6376   13 years gb Don't need to worry about keeping %rcx from confusing the GC when %cl …
(edit) @6375   13 years gb RCX is an immediate register (and also %nargs.q, and still used for …
(edit) @6374   13 years gb Define and use repret instruction.
(edit) @6373   13 years gb x8664-setup-ftd: deal with cross-compilation issues.
(edit) @6372   13 years gb Gak. .SPmkcatch1v wants return address on stack.
(edit) @6371   13 years gb LAP changes to support new x86-64 calling sequence.
(edit) @6370   13 years gb LAP changes to support new x86-64 calling sequence (multiple-values, …
(edit) @6369   13 years gb don't DECLARE IGNORE something that probably hasn't existed in years.
(edit) @6368   13 years gb Printing a single float might work; need to check that …
(edit) @6367   13 years gb If we interrupt late in the write-barrier, force the thread to return …
(edit) @6366   13 years gb The one and only call to _SPfuncall in the world pushes a return address.
(edit) @6365   13 years gb Lose most of the "new" call/ret-based subprim variants; most things …
(edit) @6364   13 years gb Lose most of the "new" call/ret-based subprim variants; most things …
(edit) @6363   13 years gb Add gtra, just about when we're about to stop keeping return addresses …
(edit) @6362   13 years gb Don't set foreign record types when cross-compiling; too much …
(edit) @6361   13 years gb Use an ecase to initialize ftd interface dir; #. won't work if the ftd …
(edit) @6360   13 years gb Spell arg names correctly in ppc64 callback. Struct misalignment …
(edit) @6359   13 years gb Functions (and most subprims) called with return address on top of …
(edit) @6358   13 years gb Functions (and most subprims) called with return address on top of …
(edit) @6357   13 years gb Functions (and most subprims) called with return address on top of …
(edit) @6356   13 years gb Spell FreeBSD package name correctly, for cross-compile.
(edit) @6355   13 years gb Lose most of the "new" subprims added a few weeks ago. Most things …
(edit) @6354   13 years gb Handle other label operands in (e.g.) CALL instructions.
(edit) @6353   13 years gb handle_fault: check for valid cmain before calling out to it.
(edit) @6352   13 years gb makes64 is an xcall.
(edit) @6351   13 years gb new unwind: call via lea/jmp, but do interrupt-level binding out-of-line.
(edit) @6350   13 years gb gtra for linuxx8664.
(edit) @6349   13 years gb New catch, unwind-protect stuff.
(edit) @6348   13 years gb New catch, unwind-protect stuff. define xcall-subprim, start to use it.
(edit) @6347   13 years gb (Finally) don't need to set %fn to jump to it, can jump to …
(edit) @6346   13 years gb New mkcatch, mkunwind.
(edit) @6345   13 years gb New mkcatch, mkunwind.
(edit) @6344   13 years gb Back out of the "delta" stuff: probably more confusing that way.
(edit) @6343   13 years gb Use init-nclosure to make closures. LABELS needs to account for extra …
(edit) @6342   13 years gb Add INIT-NCLOSURE: closures now start with an instruction that sets …
(edit) @6341   13 years gb .SPncall-closure.
(edit) @6340   13 years gb In newfangled .SPnbuiltin_ash, return via ret (not jmp ra0) if shift …
(edit) @6339   13 years gb Correct an assertion made in a comment. Should really throw most of …
(edit) @6338   13 years gb Indentation …
(edit) @6337   13 years gb Start to handle new "builtin" subprims.
(edit) @6336   13 years gb Add delta to label gensyms, commented instruction addresses. Use delta …
(edit) @6335   13 years gb ESTABLISH-FN, XCALL-SUBPRIM.
(edit) @6334   13 years gb In mark_xp(), mark rip as function if it looks like rip points to the …
(edit) @6333   13 years gb aref1/aset1: fix array dims check: want a list of length 1.
(edit) @6332   13 years gb Cant use (:array :uint 2) in the cold load without making a fuss about it.
(edit) @6331   13 years gb ref-global.l
(edit) @6330   13 years gb Bump.
(edit) @6329   13 years gb Remove some stuff leftover from 68k->ppc MCL transition.
(edit) @6328   13 years gb Indentation/comment changes only.
(edit) @6327   13 years gb Fix tra checking.
(edit) @6326   13 years gb FreeBSD CPU-COUNT sysctl call looked bad.
(edit) @6325   13 years gb Tried to use shorter insns; losing battle.
(edit) @6324   13 years gb Can use :talign, call instructions in some cases.
(edit) @6323   13 years gb Use xcall-subprim-2 (uses x86 CALL to call subprims) for a few numeric …
(edit) @6322   13 years gb Prepend an rip-relative LEA that sets %fn; this will mean (with GC …
(edit) @6321   13 years gb Increase gf-code-size. Might shrink again soon.
(edit) @6318   13 years gb Support :talign pseudo-op. Don't bother emitting SAVE-CLEANUP-CONTEXT …
(edit) @6317   13 years gb Add variants of many subprims which take their return addresses on the …
(edit) @6316   13 years gb Add variants of many subprims which take their return addresses on the …
(edit) @6315   13 years gb mark_xp: if rip is unmarked and tagged as a return address, treat it …
(edit) @6314   13 years gb Handle SIGSYS, 'cause Leopard can't handle the truth.
(edit) @6313   13 years gb Support for :talign pseudo-op: aligns the -end- of the next …
(edit) @6312   13 years gb xcall-subprim (for testing).
(edit) @6311   13 years gb Recognize subprim addresses in CALL operands.
(edit) @6310   13 years gb Allow use of :talign pseudo-op in vinsns.
(edit) @6309   13 years gb Define "new" subprims (where "new" means "will be called via x8664 …
(edit) @6306   13 years gb Use new scheme for mapping from tra to containing fn. Needs testing on …
(edit) @6305   13 years gb Use new scheme to map from tra to function/offset.
(edit) @6304   13 years gb Constants related to recognizing the RIP-relative instruction after …
(edit) @6303   13 years gb recover-fn-from-rip.
(edit) @6302   13 years gb recover-fn-from-rip.
(edit) @6301   13 years gb New scheme - mandatory recovery of %FN relative to %RIP - causes …
(edit) @6300   13 years gb When recovering %fn from %rip, use a temporary to reference the entry …
(edit) @6299   13 years gb Don't define RECOVER-FN-FROM-RA0. Do define RECOVER-FN-FROM-RIP. Use …
(edit) @6298   13 years gb In MATCH-TEMPLATE-VREGS, handle locals/temps which refer to global labels.
(edit) @6297   13 years gb Accept (name (:label n)) as a vinsn temporary.
(edit) @6296   13 years gb Recognize memory addresses of the form (@ (: label) (% rip)) as if …
(edit) @6295   13 years gb Do the right/obvious thing for memory addresses of the form (@ (:
(edit) @6294   13 years gb Some changes for new tagged-return-address scheme.
Note: See TracRevisionLog for help on using the revision log.