execute.c File Reference

---

Detailed Description

Author:

Raviraj Joshi (raviraj@student.unsw.edu.au)

Saeid Nooshabadi (saeid@unsw.edu.au)

Date:

1 August 2004

Version:

1.1

The 'execute.c' file contains functions that execute the following floating point instructions:

• Data Processing:
  • floating point addition
  • fused multiplication and addition
  • subtraction
  • multiplication
  • division
  • copy
  • absolute
  • negation
  • square root
  • compare
  • conversions

• Register Transfer:
  • Sn = Rd
  • Rd = Sn
  • Dn[31:0] -> Rd
  • Rd = Dn[31:0]
  • Dn[63:32] -> Rd
  • Rd = Dn[63:32]
  • SystemReg(fn) = Rd
  • Rd = SystemReg(fn)

• Load Store:
  • store a number of single precision float registers to memory
  • load a number of single precision float registers from memory
  • store a number of double precision float registers to memory
  • load a number of double precision float registers from memory
  • store a number of single precision float registers to memory with increment Rn
  • load a number of single precision float registers from memory with increment Rn
  • store a number of double precision float registers to memory with increment Rn
  • load a number of double precision float registers from memory with increment Rn
  • store a single precision float register to memory with negative offset
  • load a single precision float register from memory with negative offset
  • store a double precision float register to memory with negative offset
  • load a double precision float register from memory with negative offset
  • store a number of single precision float registers to memory with decrement Rn
  • load a number of single precision float registers from memory with decrement Rn
  • store a number of double precision float registers to memory with decrement Rn
  • load a number of double precision float registers from memory with decrement Rn
  • store a single precision float registers to memory with positive offset
  • load a single precision float registers from memory with positive offset
  • store a double precision float registers to memory with positive offset
  • load a double precision float registers from memory with positive offset

See the 'CHANGES.txt' file for list of modifications made to this file.

Definition in file execute.c.

#include "armvfp.h"

Include dependency graph for execute.c:

Go to the source code of this file.

Functions
void	execute_dp (void)
void	execute_regt (unsigned long int *BankedReg)
void	execute_ldst (unsigned long int *BankedReg)

---

Function Documentation

void execute_dp (  void   )

The 'execute_dp' function executes the following decoded ARM VFP data processing instructions: floating point addition FADDS or FADDD: Fd = Fn + Fm fused multiplication and addition FMACS or FMACD: Fd = Fd + (Fn * Fm) FNMACS or FNMACD: Fd = Fd - (Fn * Fm) FMSCS or FMSCD: Fd = -Fd + (Fn * Fm) FNMSCS or FNMSCD: Fd = -Fd - (Fn * Fm) subtraction FSUBS or FSUBD: Fd = Fn - Fm multiplication FMULS or FMULD: Fd = Fn * Fm FNMULS or FNMULD: Fd = -(Fn * Fm) division FDIVS or FDIVD: Fd = Fn / Fm copy FCPYS or FCPYD: Fd = Fm absolute FABSS or FABSD: Fd = abs(Fm) negation FNEGS or FNEGD: Fd = -Fm square root FSQRTS or FSQRTD: Fd = sqrt(Fm) compare FCMPS or FCMPD: Compare Fd with Fm, no exceptions on quiet NaNs FCMPES or FCMPED: Compare Fd with Fm, with exceptions on quiet NaNs FCMPZS or FCMPZD: Compare Fd with 0, no exceptions on quiet NaNs FCMPEZS or FCMPEZD: Compare Fd with 0, with exceptions on quiet NaNs conversions FUITOS or FUITOD: Unsigned integer -> floating point conversions FSITOS or FSITOD: Signed integer -> floating point conversions FCVTDS or FCVTSD: Single <-> Double precision conversions FTUIOS or FTUIOD: Floating point -> unsigned integer conversions FTOUIZS or FTOUIZD: Floating point -> unsigned integer conversions, RZ mode FTOSIS or FTOSID: Floating point -> signed integer conversions FTOSIZS or FTOSIZD: Floating point -> signed integer conversions, RZ mode Definition at line 109 of file execute.c. References check_flag(), Instruction::cp_num, Instruction::ext_opcode, fabsolute(), faddition(), fcompare(), fcompare_exp(), fcompare_zero(), fcompare_zero_exp(), fcopy(), fcvt_float_signedint(), fcvt_float_unsignedint(), fcvt_float_unsignedint(), fcvt_signedint_float(), fcvt_single_double(), fcvt_unsignedint_float(), Instruction::fd, fdivision(), fetch_data(), Instruction::fm, fmultiplication(), Instruction::fn, fnegation(), fsqrt(), fsubtraction(), fused_mul_add(), inst, Instruction::opcode, return_data(), and Float::sign. Referenced by vfp_data_proc(). Here is the call graph for this function:

void execute_ldst (  unsigned long int *  BankedReg  )

Parameters: BankedReg  The banked register. The 'execute_ldst' function executes the following decoded ARM VFP load store instructions: store a number of single precision float registers to memory load a number of single precision float registers from memory store a number of double precision float registers to memory load a number of double precision float registers from memory store a number of single precision float registers to memory with increment Rn load a number of single precision float registers from memory with increment Rn store a number of double precision float registers to memory with increment Rn load a number of double precision float registers from memory with increment Rn store a single precision float register to memory with negative offset load a single precision float register from memory with negative offset store a double precision float register to memory with negative offset load a double precision float register from memory with negative offset store a number of single precision float registers to memory with decrement Rn load a number of single precision float registers from memory with decrement Rn store a number of double precision float registers to memory with decrement Rn load a number of double precision float registers from memory with decrement Rn store a single precision float registers to memory with positive offset load a single precision float registers from memory with positive offset store a double precision float registers to memory with positive offset load a double precision float registers from memory with positive offset Definition at line 416 of file execute.c. References Instruction::cp_num, ENDIAN, Instruction::fd, Instruction::fn, inst, Instruction::instruction, Instruction::opcode, and s. Referenced by vfp_ld_st().

void execute_regt (  unsigned long int *  BankedReg  )

Parameters: BankedReg  The banked register. The 'execute_regt' function executes the following decoded ARM VFP register transfer instructions: FMSR: Sn = Rd FMRS: Rd = Sn FMDLR: Dn[31:0] -> Rd FMRDL: Rd = Dn[31:0] FMDHR: Dn[63:32] -> Rd FMRDH: Rd = Dn[63:32] FMXR: SystemReg(fn) = Rd FMRX: Rd = SystemReg(fn) Definition at line 296 of file execute.c. References Instruction::cp_num, ENDIAN, Instruction::fd, Instruction::fn, fpexc, fpscr, fpsid, inst, Instruction::opcode, and s. Referenced by vfp_reg_tran().

---