8000 Releases · loongson/luajit2 · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Releases: loongson/luajit2

v2.1-20250529-loongarch64

17 Jun 08:50
Compare
Choose a tag to compare

Changelog in loongarch64

  • LOONGARCH64: Fixed data loss in number type caused by 32-bit sign extension in BC_UNM
  • LOONGARCH64: Fixed the old PC being overwritten by the new PC in BC_ISNEXT
  • LOONGARCH64: Fixed ffi_callback to ensure global_State is relocated to GL
  • LOONGARCH64: Optimize register allocation to use RID_TMP instead of ra_scratch
  • LOONGARCH64: Fix tmp register restored after guard execution
  • LOONGARCH64: Optimize the implementation mechanism of guard exit
  • LOONGARCH64: Fix offset calculation between current PC and target in exit stub
  • LOONGARCH64: Fix the bug in generating machine instructions in asm_mulov
  • LOONGARCH64: Fix "li traceno" range check and instruction in exit stub
  • LOONGARCH64: Adjust assembly instructions to improve code readability
  • LOONGARCH64: Replace "addi.w rd, r0, si12" with "l12i rd, si12" to improve code readability
  • LOONGARCH64: Optimize the comparison method of certain conditional branches
  • LOONGARCH64: Fix the bug of register allocation overlap in asm_hrefk
  • LOONGARCH64: Replace "addi.d rd, r0, si12" with "l12i rd, si12" to improve code readability
  • LOONGARCH64: Replace "jirl r0, ra, 0" with "ret" to improve code readability
  • LOONGARCH64: Replace "or CARG1, L, r0" with "mv CARG1, L" to improve code readability
  • LOONGARCH64: Switch from DISPATCH base to JGL base to minimize instruction count
  • LOONGARCH64: Simplify function headers by eliminating redundant logic
  • LOONGARCH64: Eliminate unnecessary register moves and optimize index storage timing
  • LOONGARCH64: Optimize return ops
  • LOONGARCH64: Optimize target jump calculation by streamlining base offset subtraction
  • LOONGARCH64: Optimize calls and vararg handling
  • LOONGARCH64: Fix BC_TSETS_Z overwriting the mark register during the write barrier
  • LOONGARCH64: Optimize the table, upvalue, and function ops
  • LOONGARCH64: Optimize the BC_KNUM
  • LOONGARCH64: Optimize binary arith ops
  • LOONGARCH64: Fix the BC_UNM int error
  • LOONGARCH64: Optimize unary test and copy ops
  • LOONGARCH64: Fix min/max return register error
  • LOONGARCH64: Fix vm_next register dirty read bug
  • LOONGARCH64: Add the definition and usage of the CFR
  • LOONGARCH64: Fix FP to integer conversions
  • LOONGARCH64: Fix pcall() error case
  • LOONGARCH64: Bump copyright date
  • LOONGARCH64: Optimizing the vm_next function
  • LOONGARCH64: Fixed the bugs of movfcsr2gr_2, movgr2fcsr_2, movgr2cf_2 decode in dynasm
  • LOONGARCH64: Fix the bug in Hard-float round to integer function
  • LOONGARCH64: Optimized the use of some conditional branch instructions
  • LOONGARCH64: Fixed the bug in pcall and xpcall function
  • LOONGARCH64: Fixed the bug in math_minmax function

v2.1-20240815-loongarch64

14 Nov 01:05
48e56bd
Compare
Choose a tag to compare

补丁对应分支: v2.1-agentzh-loongarch64
补丁对应tag: v2.1-20240815-loongarch64

0