This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: mips16 float and complex support
- From: Sandra Loosemore <sandra at codesourcery dot com>
- To: GCC Patches <gcc-patches at gcc dot gnu dot org>, richard at codesourcery dot com
- Date: Wed, 23 May 2007 16:06:37 -0400
- Subject: Re: PATCH: mips16 float and complex support
- References: <46538D21.5020007@codesourcery.com> <87d50s3r90.fsf@firetop.home>
Richard Sandiford wrote:
OK with those changes. Thanks for putting so much effort into this.
OK, the patch has been committed with the suggested tweaks.
-Sandra
2007-05-23 Sandra Loosemore <sandra@codesourcery.com>
Nigel Stephens <nigel@mips.com>
Richard Sandiford <richard@codesourcery.com>
gcc/
Fix up MIPS16 hard float and add support for complex.
* config/mips/mips.h (TARGET_HARD_FLOAT_ABI): New.
(TARGET_SOFT_FLOAT_ABI): New.
(TARGET_CPU_CPP_BUILTINS): Define __mips_hard_float and
__mips_soft_float to reflect the ABI in use, not whether the
FPU is directly accessible (e.g., in MIPS16 mode).
(UNITS_PER_HWFPVALUE): Use TARGET_SOFT_FLOAT_ABI.
(UNITS_PER_FPVALUE): Likewise.
* config/mips/mips.c (mips_expand_call): Remove redundant
TARGET_MIPS16 check.
(mips_arg_regno): New.
(function_arg_advance): When setting bits in cum->fp_code for
MIPS16, don't subtract 1 from cum->arg_number, since it is now
zero-based.
(function_arg): Use mips_arg_regno.
(mips_return_mode_in_fpr_p): New.
(mips16_call_stub_mode_suffix): New.
(mips16_cfun_returns_in_fpr_p): New.
(mips_save_reg_p): Use mips16_cfun_returns_in_fpr_p.
(mips_output_function_prologue): Test mips16_hard_float, not
!TARGET_SOFT_FLOAT, to decide when a function stub is required.
(mips_expand_epilogue): Call MIPS16 helper routines to copy
return value into a floating-point register.
(mips_can_use_return_insn): Use mips16_cfun_returns_in_fpr_p.
(mips_function_value): Rewrite to use mips_return_mode_in_fpr_p.
(mips16_fp_args): Handle MIPS32r2 ISA which supports
TARGET_FLOAT64, and use mfhc1/mthc1 to copy the most significant
word of double arguments from or to the high bits of 64-bit
floating point registers.
(build_mips16_function_stub): Fill in DECL_RESULT for stubdecl.
(mips16_fpret_double): New helper function.
(build_mips16_call_stub): Use mips16_return_mode_in_fpr_p. Add
support for complex modes. Fill in DECL_RESULT for stubdecl.
(mips_init_libfuncs): Remove redundant TARGET_MIPS16 check.
* config/mips/mips16.S
(RET, ARG1, ARG2): New.
(MERGE_GPRf, MERGE_GPRt): New.
(DELAYt, DELAYf): New.
(MOVE_SF_BYTE0, MOVE_SI_BYTE0): New.
(MOVE_SF_BYTE4, MOVE_SF_BYTE8): New.
(MOVE_DF_BYTE0, MOVE_DF_BYTE8): New.
(MOVE_SF_RET, MOVE_SC_RET, MOVE_DF_RET, MOVE_DC_RET, MOVE_SI_RET): New.
(SFOP): Renamed to...
(OPSF3): This, and macro-ified. Updated all uses.
(SFOP2): Renamed to...
(OPSF2): This, and macro-ified. Updated all uses.
(SFCMP): Renamed to...
(CMPSF): This, and macro-ified. Updated all uses.
(SFREVCMP): Renamed to...
(REVCMPSF): This, and macro-ified. Updated all uses.
(__mips16_floatsisf, __mips16_fix_truncsfsi): Macro-ified.
(LDDBL1, LDDBL2, RETDBL): Deleted.
(DFOP): Renamed to...
(OPDF3): This, and macro-ified. Updated all uses.
(DFOP2): Renamed to...
(OPDF2): This, and macro-ified. Updated all uses.
(__mips16_extendsfdf2, __mips16_truncdfsf2): Macro-ified.
(DFCMP): Renamed to...
(CMPDF): This, and macro-ified. Updated all uses.
(DFREVCMP): Renamed to...
(REVCMPDF): This, and macro-ified. Updated all uses.
(__mips16_floatsidf, __mips16_fix_truncdfsi): Macro-ified.
(RET_FUNCTION): New.
(__mips16_ret_sf, __mips16_ret_df): Macro-ified.
(__mips16_ret_sc, __mips16_ret_dc): New.
(STUB_ARGS_0, STUB_ARGS_1, STUB_ARGS_5, STUB_ARGS_9, STUB_ARGS_2,
STUB_ARGS_6, STUB_ARGS_10): New.
(CALL_STUB_NO_RET): New.
(__mips16_call_stub_1): Macro-ified.
(__mips16_call_stub_5): Macro-ified.
(__mips16_call_stub_2): Macro-ified.
(__mips16_call_stub_6): Macro-ified.
(__mips16_call_stub_9): Macro-ified.
(__mips16_call_stub_10): Macro-ified.
(CALL_STUB_RET): New.
(__mips16_call_stub_sf_0): Macro-ified.
(__mips16_call_stub_sf_1): Macro-ified.
(__mips16_call_stub_sf_5): Macro-ified.
(__mips16_call_stub_sf_2): Macro-ified.
(__mips16_call_stub_sf_6): Macro-ified.
(__mips16_call_stub_sf_9): Macro-ified.
(__mips16_call_stub_sf_10): Macro-ified.
(__mips16_call_stub_df_0): Macro-ified.
(__mips16_call_stub_df_1): Macro-ified.
(__mips16_call_stub_df_5): Macro-ified.
(__mips16_call_stub_df_2): Macro-ified.
(__mips16_call_stub_df_6): Macro-ified.
(__mips16_call_stub_df_9): Macro-ified.
(__mips16_call_stub_df_10): Macro-ified.
(__mips16_call_stub_sc_0): New.
(__mips16_call_stub_sc_1): New.
(__mips16_call_stub_sc_5): New.
(__mips16_call_stub_sc_2): New.
(__mips16_call_stub_sc_6): New.
(__mips16_call_stub_sc_9): New.
(__mips16_call_stub_sc_10): New.
(__mips16_call_stub_dc_0): New.
(__mips16_call_stub_dc_1): New.
(__mips16_call_stub_dc_5): New.
(__mips16_call_stub_dc_2): New.
(__mips16_call_stub_dc_6): New.
(__mips16_call_stub_dc_9): New.
(__mips16_call_stub_dc_10): New.
* config/mips/t-elf (LIB1ASMFUNCS): Add MIPS16 floating-point stubs.
* config/mips/t-isa3264 (LIB1ASMFUNCS): Likewise.
* config/mips/t-r2900 (LIB1ASMFUNCS): Likewise.
gcc/testsuite/
* gcc.target/mips/inter/mips16_stubs_1_main.c: New.
* gcc.target/mips/inter/mips16_stubs_1_x.c: New.
* gcc.target/mips/inter/mips16_stubs_1_y.c: New.
* gcc.target/mips/inter/mips16-inter.exp: New.