summaryrefslogtreecommitdiff
path: root/src/lj_asm.c (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* Fix handling of floats in x86/x64 backend.Mike Pall2011-01-051-13/+17
|
* Add support for non-constant integer IR_MUL to backend.Mike Pall2011-01-031-27/+13
|
* Add support for integer IR_NEG to backend.Mike Pall2011-01-031-4/+9
|
* Improve uint64_t <-> FP conversions in x64 backend.Mike Pall2011-01-021-13/+17
|
* Add support for uint64_t <-> FP conversions to x64 backend.Mike Pall2011-01-021-15/+34
|
* Add 8/16 to 32 bit sign/zero-extension variants to CONV.Mike Pall2010-12-311-1/+24
|
* Fix register allocation for FP <- int conversion.Mike Pall2010-12-311-1/+1
|
* Refactoring of conversion ops, part 5: drop TOINT/TOI64/TONUM.Mike Pall2010-12-311-44/+2
|
* Copy destination type for CONV from ir->t to op2, too.Mike Pall2010-12-311-1/+1
|
* Add support for IRT_FLOAT to XLOAD/XSTORE.Mike Pall2010-12-301-2/+5
|
* Refactoring of conversion ops, part 1: add IR_CONV.Mike Pall2010-12-301-1/+95
|
* Improve register allocation for loops with variable shifts.Mike Pall2010-12-231-1/+4
|
* Fix XSTORE for IRT_NUM constants.Mike Pall2010-12-201-1/+2
|
* FFI: Drop IR_CNEWI. Add IR_CNEWP only for pointers/refs.Mike Pall2010-12-171-58/+32
|
* FFI: Add IR_CNEW/IR_CNEWI to allocate/init cdata objects.Mike Pall2010-12-111-1/+90
|
* x86/x64 backend: keep invariants on the right rather than fusing loads.Mike Pall2010-12-091-3/+6
|
* Avoid fusing loads if there are multiple references.Mike Pall2010-12-081-8/+13
|
* Fix IRT_NUM support for XLOAD/XSTORE.Mike Pall2010-12-081-2/+3
|
* FFI: Turn cdata indexing into x86/x64 [base+idx*sz+ofs] addressing.Mike Pall2010-12-081-4/+36
|
* Fix x64 code generation for A16+REX prefixed instructions.Mike Pall2010-12-081-0/+1
|
* Fix register allocation for 8 bit stores in x86 backend.Mike Pall2010-12-071-2/+10
|
* Add IRT_NUM support to XLOAD/XSTORE.Mike Pall2010-12-061-2/+4
|
* Add IR_TOI64.Mike Pall2010-12-061-0/+19
|
* Fix 64 bit shifts in backend. Fix shift by 0.Mike Pall2010-12-061-2/+2
|
* Fix XLOAD/XSTORE references.Mike Pall2010-12-061-2/+3
|
* Add support for integer IR_MUL.Mike Pall2010-12-061-1/+31
|
* Support all kinds of XLOAD/XSTORE references in backend.Mike Pall2010-12-061-6/+13
| | | | Fuse pointer arithmetic, too.
* Avoid warning.Mike Pall2010-12-061-0/+1
|
* Add IR_XSTORE.Mike Pall2010-12-061-4/+9
|
* Add support for 64 bit integer arithmetic to x64 backend.Mike Pall2010-12-061-33/+51
|
* Fix xmm spill/restore broken by b1fb71fb.Mike Pall2010-12-051-7/+16
|
* Add IR_KINT64.Mike Pall2010-12-051-21/+44
|
* Cleanup 64 bit IR type handling.Mike Pall2010-12-051-21/+23
|
* Make sure to use irt_toitype() macro everywhere.Mike Pall2010-12-051-8/+8
|
* Cleanup architecture, ABI and OS definitions.Mike Pall2010-11-161-6/+2
|
* Fix conflict between loop branch inversion and HREF+NE/EQ merging.Mike Pall2010-11-051-5/+7
|
* Decouple SLOAD type and optional conversion.Mike Pall2010-10-111-6/+7
|
* x64: Fix type check for numbers in compiled code.Mike Pall2010-10-111-5/+21
|
* Avoid fusing potentially negative indexes into AREF on x64.Mike Pall2010-09-181-2/+3
|
* Add IR_VLOAD for vararg loads.Mike Pall2010-09-141-3/+8
| | | | Also fixes the broken AA improvement in the last commit.
* Record vararg expressions with varargs defined off-trace.Mike Pall2010-09-131-18/+29
| | | | Add SLOAD variant to access the frame type/size.
* Untangle some target dependencies.Mike Pall2010-08-031-1/+1
|
* Abstract out pointer hash to hashrot(). Tune hash constants.Mike Pall2010-07-211-10/+7
|
* Fix passing of constant args in FPRs on x64.Mike Pall2010-07-141-1/+1
|
* Add weak guards. Emit TNEW/TDUP with a guard bit.Mike Pall2010-05-081-21/+18
|
* Simplify management of current trace. Drop lazy save.Mike Pall2010-04-251-2/+2
|
* Turn traces into true GC objects (GCtrace).Mike Pall2010-04-251-8/+8
|
* Simplify GC step calls from on-trace code.Mike Pall2010-04-191-39/+28
|
* Replace on-trace GC frame syncing with interpreter exit.Mike Pall2010-04-191-49/+17
| | | | | | | Need to sync GC objects to stack only during atomic GC phase. Need to setup a proper frame structure only for calling finalizers. Force an exit to the interpreter and let it handle the uncommon cases. Finally solves the "NYI: gcstep sync with frames" issue.
* Fix setup of RD when dispatching to function headers after exit.Mike Pall2010-04-091-5/+6
|
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-19 22:39:44 +0000