GitLab

Detail:
  TOGPSRR does the same as TOGPSR except that the bits to toggle
    are held in a register.
  Hdr:Sprite contains declarations that are mistranslated by Hdr2H 1.1.
    Hdr2H has been updated to handle these declarations and is first
    present in RiscOS/Library 0.45.
Admin:
  Required by DualSerial 0.24 and later.
  Requires Library 0.45 to export a correct C version of Hdr:Sprite

Version 0.88. Tagged as 'HdrSrc-0_88'

Version 0.87. Tagged as 'HdrSrc-0_87'

Hdr:NdrDebug made fully 32-bit compatible (only change is to the code switched
    using the debug_nullterminatedstrings variable).

Version 0.86. Tagged as 'HdrSrc-0_86'

Version 0.85. Tagged as 'HdrSrc-0_85'

* Added $oldpsr extension to SETPSR, CLRPSR and TOGPSR.
* Removed reliance on ObjAsm default macro parameters in $oldpsr extension.
* Replaced SavePSR macro calls with literal mrs instructions in 32-bit macros.

Version 0.84. Tagged as 'HdrSrc-0_84'

Detail:
  Improved some of the PSR wiggling macros with an optional extra
    register into which to save the old PSR.  Saves an extra MRS
    instruction in some cases.
  Some macros changed to invoke SavePSR macro instead of having a copy
    of the macro body.
Admin:
  Required by ScrBlank 2.30.

Version 0.83. Tagged as 'HdrSrc-0_83'

LDR/STR instead of single-register LDM/STM.

Version 0.82. Tagged as 'HdrSrc-0_82'

Detail:

  * Error "Ldr26" added to list of Podule manager errors: "Loader is
    not 32-bit compatible"

Admin:

  * Verified OK by Pineapple resource allocations; next release of
    HdrSrc files from Pineapple should contain this message.

Version 0.81. Tagged as 'HdrSrc-0_81'

Detail:
  Regressing the version of HdrSrc in the STB-400 build didn't
  work because some of the other components are 32-bit clean
  and assume that the EXITS macro can take a flags parameter,
  which it doesn't in older HdrSrc version (eg: PS2Driver).
Admin:
  It'll be tested before most people read this log.

Version 0.80. Tagged as 'HdrSrc-0_80'

More 32-bit tweaks. Some assembler macros extended to accept condition codes and PSR flag subset specifiers.

CPU.FPA header file added.

Version 0.79. Tagged as 'HdrSrc-0_79'

Admin:
  Not tested.  Version 0.77 was broken.

Version 0.78. Tagged as 'HdrSrc-0_78'

Detail:
  New SWI chunks
  New filetypes
  New error blocks
  New printer dumper
  New service calls

Admin:
  E-mail received Marth 24th 2000

Version 0.77. Tagged as 'HdrSrc-0_77'

Detail:
  Safe area was 8,8,8,8. Is now 8,6,8,6
Admin:
  Untested (but Kevin's maths indicates it will work!)

Version 0.76. Tagged as 'HdrSrc-0_76'

Detail:
  There is a duplicate allocation at 114/115 - one has been commented out
  to give subsequent filing systems the correct number.
Admin:
  Pineapple has been informed - next update should incorporate a fix.

Version 0.75. Tagged as 'HdrSrc-0_75'

Detail:
  Use cvs diff to see the changes.
Admin:
  E-mail received 27 Feb 2000.

Version 0.74. Tagged as 'HdrSrc-0_74'

Detail:
  MACFROMNVRAM was intended to force the Kernel to fetch the MAC
  address from NVRAM for Customer F 5. In practice it is not needed
  as its main purpose was to avoid getting the MAC address from
  the RawMachineID but this will not happen in Customer F 5 as
  RawMachineID will not be set.

Admin:
  Not tested.
  You must use Kernel 5.17 or later with this HdrSrc.

Version 0.73. Tagged as 'HdrSrc-0_73'

Needed by STB-400 MPEGDriver.

Version 0.72. Tagged as 'HdrSrc-0_72'

Lazarus.
Minor optimisation to RemovePSRFromReg macro.

Version 0.71. Tagged as 'HdrSrc-0_71'

Detail:
  New Ethernet allocation: SWI chunk, error chunk.
  Lots of other new SWI chunks.
  Three new filetypes.
  Eleven new error chunks (including SVBGeneric)
  One new pointer device
Admin:
  Synchronisation of all master header files.
  E-mail received 22 Jan 2000

Version 0.70. Tagged as 'HdrSrc-0_70'

Detail:
  Used GBLA instead of GBLL for MACFROMNVRAM - sorry.
Admin:
  Tested.

Version 0.69. Tagged as 'HdrSrc-0_69'

Detail:
  MACFROMNVRAM and MACNVRAM2COPIES are TRUE for Customer F 5 Ethernet machines.

  If TRUE then OS_ReadSysInfo 4 will read the Ethernet MAC Address from the from bytes
  0-13 of the NVRAM area, instead of trying to read it from anywhere else.
  This is intended for use with a 24LC128 device (which has no OTP area and no clock).

  The NVRAM data is:
  	0-5	MAC Address, first copy
  	6	Checksum (LSB of sum of bytes 0-5, XOR'd with &FF)
  	7-12	MAC Address, second copy
  	13	Checksum, second copy

  Note that these are physical addresses within the device. Logical addresses &F0-&FF map
  to physical addresses &00-&0F.

  A SetMac utility is provided in Customer F 5 !Boot.Utils to enable the MAC address to be
  programmed for prototypes and development machines.

  Since it is dangerous to rely on the NVRAM to hold our MAC address when it is not
  protected in any way, we store two copies of it, each with their own checksum.
  If either copy becomes damaged we can repair it using the other copy.
Admin:
  Not tested.

  If used for Customer F 3 or 4, then you must use the F3UI v0.51 or
  later as it defines the NVRAM tags that are used.

Version 0.68. Tagged as 'HdrSrc-0_68'

Version 0.67. Tagged as 'HdrSrc-0_67'

Detail:
  This stops the errors occurring in 26-bit builds when the EntryS
    macro is invoked without quotes around the register list.
Admin:
  Fixes problems with the latest ARM module not building.


Version 0.66. Tagged as 'HdrSrc-0_66'

Detail:
  The 32-bit version of WritePSRc takes an optional third parameter
    which is the condition code to embed in the instruction.
    The 26-bit mode version now takes the same parameters, thus
    allowing Sound2 (SoundScheduler) to build.
Admin:
  Tested in Phoebe and 32 machine builds of Sound2.
  Required by SoundScheduler 1.25 and later (Sound2-1_25)

Version 0.65. Tagged as 'HdrSrc-0_65'

Detail:

  * The combo cip on Lazarus has, until now, been clocked at 14.318MHz.
    Issue 2 of the hardware will use a 24MHz clocking frequency instead;
    all existing issue 1 boards will be modified appropriately.
    "hdr.Machine.Machine" file changed to set ComboClock14MHz flag to
    false.

Admin:

  * Tested on modified Lazarus hardware.

Version 0.64. Tagged as 'HdrSrc-0_64'

Detail:
  Pointless jumping over the generated code or inserting conditional
    instructions which will never be executed.

Detail:
  The 26-bit and 32-bit capable macros have been added.
  Header files choose which set of macros to use based on <Machine>.
  If you want the new macros, include "Hdr:APCS.<APCS>" in addition
    to the usual inclusions at the top of assembler files.
  Choices are based on the settings of macros No32bitCode and
    No26bitCode.  By default, all existing machines define these
    as {TRUE} and {FALSE} respectively.  This yields the same macros
    as before (tested in Customer F 5 build) and should yield code which
    functions on RISC OS 3.1.  Any other combination of settings is
    untested at this time.
  New assembler code should use the macros FunctionEntry, Return et al
    to ensure that they build correctly.
  New assembler code should be written to not require flag preservation
    across internal function calls and to not assume it on external
    function calls where it is possible to do so.  DO NOT simply replace
    "MOVS pc, lr" with "MOV pc, lr" as this doesn't work properly.
    Replace "MOVS pc, lr" with "Return ,LinkNotStacked"
  Do not write code which simply masks out &FC000003 from pc or some kind
    or pc originated value (eg. on r14 to attempt to remove the PSR flags
    from it.  Use the RemovePSRFromReg macro instead.
Admin:
  Tested with Customer F 5 build.
  Requires FPEmulator 4.11
  Requires Portable 0.65 (or Portable75 0.04)
  Required by any new or modified assembler code in future, pretty much.
    Such components may not necessarily specify a dependency on this
    or any other later version of HdrSrc.

Version 0.63. Tagged as 'HdrSrc-0_63'

Detail:
  New external declarations: error base, service call.
  New country declarations.
Admin:
  E-mail received 18 Nov 1999

Version 0.62. Tagged as 'HdrSrc-0_62'

  Added Window module service calls.
Detail:
  The Window module was using service calls &82880-4 (matching its SWI
    chunk).  This is now registered in the Services file.
  SWI chunk name corrections:
    Toolkit -> Toolbox
    Windows -> Window
    Menus -> Menu
    IconBar -> Iconbar
  The SWI chunk name changes affect the symbols generated in Hdr:SWIs, and
    by the newly exported header files in RiscOS/Sources/Toolbox/Common
    which use the symbols defined in ISV-SWIs as the SWI symbol prefix.
Admin:
  Pineapple have OK'ed these changes.
  New files e-mailed to Pineapple as current masters.

Version 0.61. Tagged as 'HdrSrc-0_61'

Detail:
  Start off a new range for encoding alphabets that'll only appear in fonts,
  never as territory information
Admin:
  E-mail received 13 Nov 1999.

Version 0.60. Tagged as 'HdrSrc-0_60'

Detail:
  MediaNet and RTSP SWI chunks no longer contain underscores.
  Fixed a missing allocation.
Admin:
  E-mail received on 9 Nov 1999.

Version 0.59. Tagged as 'HdrSrc-0_59'

Detail:
  All mkdir commands are preceded by do to ensure system variables
    are expanded before mkdir sees them.
  Relies on the header directory variables instead of <Build$Dir>
    since the export directories may not be <Build$Dir>.Export any
    more.
Admin:
  This change is required if <CExport$Dir> or <Hdr$Dir> do not exist
    when HdrSrc's export phase is executed.

Version 0.58. Tagged as 'HdrSrc-0_58'

Version 0.57. Tagged as 'HdrSrc-0_57'

Detail:
  ErrorBase_MediaCtrl added for Pace (arodger)
Admin:
  E-mail received 2 Nov 1999.

Version 0.56. Tagged as 'HdrSrc-0_56'

Detail:
  New filetype allocation for 3rd party.
Admin:
  E-mail received 31 October 1999.  Scary.

Version 0.55. Tagged as 'HdrSrc-0_55'

Changed Lazarus ROM timing to 8 cycles = 125ns (was 10 cycles = 156ns).

Version 0.54. Tagged as 'HdrSrc-0_54'

Detail:
  New filetype allocations for 3rd parties.
Admin:
  E-mail received 25 October 1999.

Version 0.53. Tagged as 'HdrSrc-0_53'

Removed conditional compilation from reassigned IOMD interrupt stuff. Both sets of allocations are now included, prefixed by IOMD_ and IOMDr_.

This complicates the Kernel, where it is fixed at compile time, but it should
simplify everthing else, which will handle both at run time.

Version 0.52. Tagged as 'HdrSrc-0_52'