GitLab

Detail:
  In the previous version, Portable_Idle was called each time that the callback
  chain was examined. This is only desirable in cases where we're polling the
  keyboard, waiting before scrolling the text window; in the normal case, it's
  an unnecessary slow-down. Now fixed to only call Portable_Idle if we're
  waiting for a key up/down/timeout.
Admin:
  Not tested.

Version 5.28. Tagged as 'Kernel-5_28'

  In particular, they can go off when Scroll Lock etc have frozen the text
  output.
Detail:
  The callback chain is examined at least once per line scroll. Additionally,
  it is examined repeatedly while Scroll Lock or Ctrl-Shift are acting, while
  idling between lines if Ctrl is held down, and while waiting for the Shift
  key to be pressed at the bottom of a screen when in page mode.
  The opportunity is also taken to call Portable_Idle after each time that the
  callback chain has been examined.
Admin:
  Not tested.

Version 5.27. Tagged as 'Kernel-5_27'

Detail:
	It is often the case that modules fail on receipt of a PostInit
cervice call. The Kernel already has an option to display debugging on module
initialisation (DebugROMInit), but this does not help if a module crashes
during the PostInit stage.
	To aid debugging of the PostInit stage of module initialisation, a
new flag (DebugROMPostInit) has been added to the vanilla service call
handler. This flag displays the name of each module that the PostInit is
being dispatched to, and then displays whether control has passed back to the
kernel. Hence crashes of a module during PostInit can be detected.

Admin:
	Note that this debug option only works in the vanilla service call
handler. If your build uses the chocolate handler and you wish to debug
PostInit of modules, then set it temporarilly to use vanilla handlers.
	Tested in Lazarus builds both with and without the option switched.

Version 5.26. Not tagged

  Fixed multiple internationalisation of error message.
Detail:
  Somebody had decided to write a comment that R0-R2 needed to be
    unstacked before exiting the dynamic area remove routine, but
    then forgot to do so.  The crash is the same each time, as the
    address of MessageTrans error buffer it was trying to return
    is the value mistakenly loaded in R15.
  Kernel doesn't try to translate error messages multiple times.
    Not only does this give rise to errors about not being able
    to translate error messages, but causes MessageTrans to have to
    search repeatedly for non-existant tokens, slowing things down
    enormously.
Admin:
  Tested in Ursula build.
  Not a complete fix - FileCore now leaks map & buffer dynamic areas
    whenever the RAM disc size is altered, but that needs to be fixed
    in FileCore (it needs to accept that OS_ChangeDynamicArea is not
    re-entrant and delete the DAs on a callback)

Version 5.26. Tagged as 'Kernel-5_26'

Detail:

  * Lazarus builds of the Kernel failed with the following error:

      Immediate value out of range at line 458 in file "s.NewReset"
              ADR    R2, IRQ_Test_CTRL_or_R_Pressed

    This is due to the ADR going out of range.  Changed to use ADDR
    macro instead to rectify this.

Admin:

  * Untested at time of check-in; to be verified in next Lazarus development
    build.

Version 5.25. Tagged as 'Kernel-5_25'

Some POST changes which got lost recommitted.

Version 5.24. Not tagged

Had one of those weekend brainstorms - managed to speed up SWI despatcher _and_ add Thumb support to it.

Fixed OS_BreakPt - was confused by PC/PSR split.

Version 5.24. Not tagged

* IOMD register listing removed from POST (for Tom).

This version seems to be pretty stable in a 26-bit desktop build. Apart from
silly modules relying on the manner in which SWIs corrupt R14_svc :(

Version 5.24. Tagged as 'Kernel-5_24'

  RPCEm update.
* Register allocation in default ErrorV handler fixed - problems occured when
  callbacks were triggered on way out.
* OS_Byte 19 didn't manipulate interrupt disable flag correctly in 26-bit
  builds.
* Stray bit of debugging left in sprite code many years ago removed.

Version 5.23. Not tagged

Detail:
  "Podule" number now displayed again in *ROMModule output - flag
     preservation issue caused it to disappear in 5.23.
  *Eval output no longer misses its trailing space, neither do "Podule" or
    "Extn ROM" in *ROMModules output.
  Heap manager now works again in non-SVC modes.
  Exception dump now contains faked up 26-bit PC+PSR lookalike.
Admin:
  Assembled.

Details:
  The Kernel will now compile to produce a pure 32-bit system if No26bitCode is
  set to TRUE.
  If No26bitCode is FALSE, then the Kernel will be a standard 26-bit Kernel,
  although some internal changes have taken place to minimise compile
  switches between the two cases. See Docs.32bit for more technical info.

  The hardest part was the flood-fill...

Other changes:
  Pointer shape changes now take place on the next VSync, rather than actually
  WAITING for the VSync. Turning the Hourglass on shouldn't slow your machine
  down by 5% now :)

  Lots of really crusty pre-IOMD code removed.

Admin:
  Tested in 32 and 26-bit forms in a limited desktop build. Basically, this
  will need to see a lot of use to iron out difficulties. I'd like anyone who
  has a non-frozen project to at least attempt using this Kernel.

Version 5.23. Tagged as 'Kernel-5_23'

Detail:
  Most of the centisecond timers were incremented very early in the Timer0
  interrupt routine, but MetroGnome was incremented after we had called
  TickerV. Routines on TickerV are allowed to enable interrupts, so any
  interrupt routines that use OS_ReadMonotonicTime and IRQRQA are unable to
  accurately determine if the monotonic time is one tick out-of-date or not.
  MetroGnome is now incremented with the other timers.
Admin:
  Tested with the timer code in STB-400 MPEGDriver.

Version 5.22. Tagged as 'Kernel-5_22'

Version 5.21. Tagged as 'Kernel-5_21'

Detail:
  The code for *ROMModules uses OS_PrettyPrint, which seems unconcerned
    by such trivial things as leading spaces, which accounts for the
    misaligned output.  The code now starts by using ASCII 31 (hard
    space for OS_PrettyPrint's purposes) as the leading indent character
    to avoid confusion.
Admin:
  Tested on desktop build in both command-line and TaskWindows.

Version 5.20. Tagged as 'Kernel-5_20'

Detail:
  When Machine=Phoebe the Kernel failed to build due to use of OSVersionID.
  Commented out this line.  Other machine settings unaffected.
Admin:
  Tested on Ursula build: Products/Ursula/modules (tag Ursula010a).

Version 5.19. Tagged as 'Kernel-5_19'

Detail:
  Three new OS SWIs for RISC OS Ltd.
Admin:
  E-mail received 27 Feb 2000
  No conflicts with Pace definitions.

Version 5.18. Tagged as 'Kernel-5_18'

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.
  MACFROMNVRAM will be removed from HdrSrc.

Version 5.17. Tagged as 'Kernel-5_17'

STB and NC builds. Remerged correctly.

Version 5.16. Tagged as 'Kernel-5_16'

  Ensured that M_Phoebe builds set UtilityModule version to 4.00
Detail:
  The softload utility relies on the existence of the extra reason codes
    to OS_ReadSysInfo introduced in Ursula.  The main kernel now supports
    these too (they are simply interfaces to read kernel capabilities and
    configuration - eg. addresses and sizes of UND and SVC mode stacks)
  Avoid OS_ReadSysInfo 9 - ROL have used it for reading the ROM personality
    information (and it's not in our kernel)
  Added some of the new macros into Copro15ops required by the ABT dump
    area code (returned by OS_ReadSysInfo 7) and added the code into ARM600
    to store abort information there.
Admin:
  Required by softload utility for Ursula builds.
  Tested on Risc PC.

Version 5.15. Tagged as 'Kernel-5_15'

Detail:
  If E2ROMSupport is {FALSE} then the kernel fails to build due to the
    use a symbols that are only defined if E2ROMSupport is {TRUE}.  This
    is now fixed by stopping the symbols being used where they should not
    be (NVRAM and RTC related symbols).
Admin:
  Built.

Version 5.14. Tagged as 'Kernel-5_14'

Version 5.13. Tagged as 'Kernel-5_13'

POST now uses option from Hdr:Machine.<Machine> to determine whether or not to flash the front panel LEDs (Lazarus doesn't want this function).

Version 5.12. Tagged as 'Kernel-5_12'

Changed default PaletteV handler so that read palette reason code returns all four supremacy bits in the palette entries (previously it cleared all except bit 7). This brings it into line with the bulk read reason code.

Version 5.11. Tagged as 'Kernel-5_11'

Kernel now always assumes that 4K and 8K ATMEL EEPROM devices have their top quarter write protected, even if IOMD_C_EEPROMProtect isn't set.

Version 5.10. Tagged as 'Kernel-5_10'

  MACFROMNVRAM is set. This is set in HdrSrc for Customer F machines.

Detail:
  If MACFROMNVRAM is TRUE then OS_ReadSysInfo 4 will read the Ethernet MAC
  Address from bytes 0-13 of the NVRAM area, instead of trying to read it
  from anywhere else. (14 bytes are ysed to store 2 copies and 2
  checksums).

  This is intended for use with a 24LC128 device (which has no OTP area
  and no clock).

  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. At present only the first copy is used, but the
  plan is to check both copies and both checksums.

Admin:
  This version of the Kernel needs Programmer/HdrSrc v0.68 or later.

  If MACFROMNVRAM is enabled, ensure that the MACADDRESS tag in the
  NVRAM file is set to point at &F0 and that MACADDRESSCHECKSUM
  is set to point at &F6. Both should have options L and P to ensure
  they are protected from NVRAM resets and other writes.

Version 5.09. Tagged as 'Kernel-5_09'

*Set Sys$Year would scramble the time. It would also fail to internationalise the error generated by *Set Year 20000.

Tested on a 32-bit system - changes grafted back on to 26-bit Kernel.

Version 5.08. Tagged as 'Kernel-5_08'

OS_ReadSysInfo 2 now reports whether the IIC bus is fast (in bit 24 of R2), and whether I/O clocks should be stopped when idling the processor (bit 25).

OS_Byte 19 is a bit more careful in its use of Portable_Idle - an edge case
where the vsync interrupt was already pending now returns immediately.

Version 5.07. Tagged as 'Kernel-5_07'

Version 5.06. Tagged as 'Kernel-5_06'

Power down A to D convertors on reset.

Version 5.05. Tagged as 'Kernel-5_05'

Version 5.04. Tagged as 'Kernel-5_04'

Version 5.03. Tagged as 'Kernel-5_03'

Version 5.02. Tagged as 'Kernel-5_02'

If reassigned, ReadSysInfo3 returns &00x22200 in R0, to indicate that the
serial and parallel ports aren't compatible, and IDE and floppy are
unavailable.

Version 5.01. Tagged as 'Kernel-5_01'

This has been fixed. In addition, SMC669 and UMC669 chips are reported as
a different chip configuration by OS_ReadSysInfo 3 (values 4 and 5
respectively).
A few assertions added to catch the remaining cases where the RCMM stuff
won't work - those cases will involve a bit more reordering of hardware
initialisation.

Version 5.00. Tagged as 'Kernel-5_00'

Version 4.99. Tagged as 'Kernel-4_99'

Right arrow on a remote also acts as Delete.

Version 4.98. Tagged as 'Kernel-4_98'

Version 4.97. Tagged as 'Kernel-4_97'

When screen is blanked, DACs are turned off (60mA saving).
If DPMS state 3 comes on, sync lines are set low.

Version 4.96. Tagged as 'Kernel-4_96'

Version 4.95. Tagged as 'Kernel-4_95'