Commit History

作者 SHA1 備註 提交日期
  sergei f80ce76a69 + MIPS: emulate "flags", i.e. support LOC_FLAGS location. This allows to generate differently optimized code for branching and for conversion to register, typically saving a register and instruction per compare. 12 年之前
  sergei 9a6edd0fb8 * MIPS: handle restoring GP after calls without GAS macro processing, removes ugly workaround for GAS bug. 12 年之前
  sergei 9e4cc57768 * MIPS: handle 8 and 16-bit arithmetic shifts internally, by shifting argument left by 24/16 bits, followed with 32-bit arithmetic shift right by appropriately adjusted amount. 12 年之前
  sergei 59d6df4fca * MIPS: replaced opcode mapping functions with array, much shorter that way. Separate "overflow" mapping is also no longer needed. 12 年之前
  sergei 7a28815182 * r24895 used wrong expression for swapping sides of comparison, and it went undetected by tests. Fixed. 12 年之前
  sergei c855868a3d * MIPS: get rid of macros in comparison operations, use immediate operands for comparison with constants when possible. 12 年之前
  sergei 89c9cdf6c4 + MIPS: implemented parameter location reusing, eliminating second copy of (potentially large) records passed by value. When parameter is passed both in registers and stack, let it have a single LOC_REFERENCE location on callee side, and store relevant registers on stack (into 16-byte area reserved by ABI) early in prologue. 12 年之前
  sergei 8b8553991a + MIPS: prevent coalescing written-to registers with $sp,$fp,$zero and $at. 12 年之前
  sergei 6a8e4f0381 * MIPS: generate real instructions, not macros, for comparisons with $zero. 12 年之前
  sergei fb88cc4257 * TCGMIPS.a_load_reg_reg: reduce code duplication, and don't generate same register move for OS_32->OS_S32 and vice versa. Such moves explode into at least 4 instructions if register needs spilling, after which they are no longer recognized and cannot be removed by reg.allocator. So it's much better not to generate them in first place. 12 年之前
  sergei 7e0ae2e984 * MIPS: fixed cgsize2subreg to return correct result for float registers. 12 年之前
  sergei 562714129f * MIPS: get completely rid of LI macro, generate equivalent CPU instructions instead. 12 年之前
  sergei c31321c2fe * TCGMIPS.handle_reg_const_reg(): fixed to generate 'real' CPU instructions, so macro processing by assembler is no longer needed. 12 年之前
  sergei d367148f75 - Removed obsolete comments (copypasted from other CPU code, most likely). 12 年之前
  sergei 300289dd89 * MIPS: reworked 64-bit code generation, implemented overflow checking and optimized operations with constants. 12 年之前
  sergei bfd7401541 * MIPS: overflow checking added in r24445 works only when source and destination of operation are different registers. Fixed cases of operations on same register. 12 年之前
  sergei 7cfc737866 * MIPS: rewrote 32-bit code generation methods, reducing code duplication. 12 年之前
  sergei 504b6754b7 * MIPS small improvements: 12 年之前
  pierre 702effaad0 Force use of PIC compatible calling for register variable calls as procvars might be internal or external and thus require PIC 12 年之前
  sergei 6f1997f5e5 + MIPS: added profiling support 12 年之前
  sergei 1c652eb8f9 MIPS procinfo improvements: 12 年之前
  sergei 823e3ea398 MIPS: reworked PIC/call code: 12 年之前
  sergei a1b50f643e - removed never used function. 12 年之前
  sergei d190d4676b - MIPS: removed comment generation in function prologues, it clutters listing more than it helps. Also removed some unused variables. 12 年之前
  sergei 354ebb822a * MIPS: fixed layout of stack frame in case procedure does only low-level calls (via a_call_name, etc.). A 16-byte outgoing parameter area must be allocated regardless of actual parameter count. 12 年之前
  sergei 69c8f6cf7d * Typo in previous commit 12 年之前
  sergei d7c96dea8b * MIPS: refactoring: removed handle_load_store and handle_load_store_fpu. 12 年之前
  sergei 30247d8961 * MIPS: fixed parameter handling for big-endian targets, and removed a lot of garbage. 12 年之前
  pierre 2916235cfe Fix make_simple_ref for pic address with big offsets 12 年之前
  sergei d82387ff72 * MIPS: fixed parameter management to properly align records smaller than 32 bits on big-endian targets. 12 年之前