1. 20 Jun, 2004 1 commit
  2. 26 Jan, 2004 1 commit
    • Kevin Bracey's avatar
      * Denormalised numbers now held consistently in their nominal precision. · ce61b80e
      Kevin Bracey authored
        MVF/MNF/ABS of a denormalised number without a change of precision
        treated specially, much like signalling NaNs. This prevents spurious
        underflow traps if they're enabled. Previously
      
             ; Underflow traps on
             LDFS    F0,DenormalisedSingle
             MVFS    F0,F0
      
        or
      
             ; Underflow traps off
             LDFD    F0,SmallestNormalDouble
             MUFD    F0,F0,#0.5
             ; Turn underflow traps on
             STFD    F0,xxx
      
        would have generated traps.
      
      * LDFP/STFP rearrangement of NaN significands reinstated, as DDT
        can now make use of it.
      
      * Typos in over/underflow code for LDFP corrected.
      
      * Makefile changed to build in rm.${SYSTEM} instead of rm.${MACHINE}.
      
      * Experimental dynamic rounding controls added to FPSR (for emulator
        only). Currently disabled; does not affect LDFP or STFP yet.
      
      Version 4.21. Tagged as 'FPASC-4_21'
      ce61b80e
  3. 21 Feb, 2003 1 commit
  4. 12 Apr, 2002 1 commit
    • Steve Revill's avatar
      * Fixed install phase; · 928be59e
      Steve Revill authored
        * Moved stripdepnd out of Makefile;
        * Variants passed on amu command line.
      Detail:
        * You can now specify the precise build variant of the FPEmulator by passing
          System=... and FPE_APCS=... on the amu command line. There are now more
          files in riscos.s to set the options for various types (as per Wimp).
          The default FPE as was is now built with: System="RO380" FPE_APCS="3/32bit"
      Admin:
        Builds.
      Version 4.19. Tagged as 'FPASC-4_19'
      928be59e
  5. 04 Mar, 2002 1 commit
    • Kevin Bracey's avatar
      Two changes to ARM's version 1.17 core: · 8f2c3a3e
      Kevin Bracey authored
      * Generated NaNs were always negative - they were supposed to have a sign
        based on the operands. As a side effect, some internal routines were not
        obeying their advertised API - this has been corrected.
      * Generated NaNs no longer contained a reason code - this was a deliberate
        ARM change, but I'm going to have a philosophical disagreement and reverse
        it.
      
      Version 4.18. Tagged as 'FPASC-4_18'
      8f2c3a3e
  6. 31 Aug, 2001 2 commits
  7. 25 Jul, 2001 1 commit
  8. 31 May, 2001 1 commit
  9. 02 Oct, 2000 1 commit
  10. 24 Mar, 2000 2 commits
    • Kevin Bracey's avatar
      Fixed error in multiply routine reported by RISC OS Ltd. · aa5adf86
      Kevin Bracey authored
      Detail:
        When the long multiply option was added on the Ursula branch, an inexplicable
        change to a single instruction in the original non-long multiply code
        appeared.
        Result would have been slightly inaccurate multiplies when both operands
        had more than 32 bits of mantissa.
        Original code restored.
      
      Version 4.13. Tagged as 'FPASC-4_13'
      aa5adf86
    • Kevin Bracey's avatar
      Multiple context version of FP emulator and FPA support code - FP in callbacks... · a2049d5a
      Kevin Bracey authored
      Multiple context version of FP emulator and FPA support code - FP in callbacks and interrupts can now be yours.
      
        Exception dumps now available (integer and FP).
      
      Detail:
      * Changed to multiple-context version of the core. This will allow floating-
        point to be used safely in callbacks and interrupt routines (as long as
        you call the appropriate context changing SWIs). This activates the
        dormant SWIs FPEmulator_ChangeContext, FPEmulator_ContextLength and
        FPEmulator_InitContext. FPEmulator_DeactivateContext and
        FPEmulator_ActivateContext now have full functionality.
      
      * Added new SWI interfaces to core: FPEmulator_Abort, FPEmulator_LoadContext
        and FPEmulator_SaveContext.
      
      * On FP exceptions, integer registers are copied into the exception dump area.
        (TO DO: registers are USR mode ones - should take pains to figure out
        SVC or IRQ ones as required).
      
      * The FPEmulator now has an exception dump area for floating point registers -
        a suitable version of Debugger will show these with *ShowFPRegs.
      
      * FPEmulator_ChangeContext and FPEmulator_DeactivateContext now return -1
        for the default context.
      
      * Run time detection of SVC stack position added.
      
      * No longer creates a MessageTrans descriptor on the stack.
      
      * Removed broken STM^ fudge from the FPA veneer, as the only StrongARMs that
        need it can't use the FPA.
      
      * Fixed a typo in core_abort.
      
      * Stopped continuous emulation runs crossing 4K page boundaries, with a view
        to improving abort handling.
      
      * Use of DREG debugging macros fixed (line endings were being inserted in the
        wrong place).
      
      * Prototype code to catch aborts inside the emulator and change the registers
        to make it look like the emulated instruction aborted. This is cute, but
        causes problems with anything like Ursula's lazy task-swapping. Really this
        needs to be done only on fatal aborts. Compiled out for now.
      
      Version 4.12. Tagged as 'FPASC-4_12'
      a2049d5a
  11. 26 Nov, 1999 1 commit
  12. 03 Aug, 1999 1 commit
    • Kevin Bracey's avatar
      Ursula branch merged. This contains: · 5b362524
      Kevin Bracey authored
        Use of UMULL (compiled out).
        Service call table.
        Runs in UND32 mode (requires latest Kernel), giving a speed boost.
      
      Version 4.10. Tagged as 'FPASC-4_10'
      5b362524
  13. 21 Sep, 1998 2 commits
  14. 02 May, 1997 1 commit
  15. 21 Jan, 1997 1 commit
  16. 21 Nov, 1996 1 commit
  17. 05 Nov, 1996 1 commit