GitLab

Fixed properly in DeviceFS-0_72.

Version 0.79. Tagged as 'Parallel-0_79'

ioeb_busy: saved a whole instruction in the NE case
ioeb_irqwrite: we are entered with R3=IO controller, which is not useful, so reload HardwareBase each time
ioeb_strobe: the call to the HAL corrupts the byte we're printing, so stack it either side of the call

Tested printing some geometrical shapes from a PrintOut file from Draw.
Used *COPY MyTest parallel: and
     *COPY MyTest fastparallel:
and also checked with !PICSuite that the Parallel_ SWIs allow low level access to the control register (used for bit bashing).

Version 0.78. Tagged as 'Parallel-0_78'

Common.s: Misplaced comment moved
IOEB.s: Make sure pp_ctrl is relative to HardwareBase
        Preserve C bit over the HAL call
Parallel.s: Misplaced ASSERT moved
This module still stiffs when trying to print, but it at least doesn't data abort too!

Version 0.77. Tagged as 'Parallel-0_77'

Now uses OS_Memory throughout to determine controller addresses rather than calling the private HAL_ControllerAddress API.
Be devious in all the register pokes by making HardwareBase be roughly half way between the two sets of standard parallel registers and ECP mode ones, so they're always in range for an immediate offset of LDRB/STRB so the code doesn't thrash the literal pool so much.
Where possible, eliminate HAL conditional code by initialising HardwareBase in the non HAL case thus simplifying code paths a bit.
Be a bit less ugly in the 0.5us delays, since the combo chip is using Type B cycles a dummy read of a SuperIO chip will waste 0.5us rather than needing to back calculate the IOEB registers.
Expand tabs, use OSBytes header file.

Version 0.76. Tagged as 'Parallel-0_76'

Detail:
  s/IOEB - Now uses OS_Memory 9 to look for controller #34 instead of HAL_ControllerAddress to look for controller #6. Makes it a bit more portable/future-proof and fixes the HAL_ControllerAddress allocation clash.
Admin:
  Tested in ROM softload on RiscPC


Version 0.75. Tagged as 'Parallel-0_75'

Detail:
  s/Parallel - Use HAL_ControllerAddress to get the address of the 82C710. IOMD IRQ numbers are assumed.
  s/Common - Use HAL_ControllerAddress to get the address of IOMD when measuring the CPU speed
  s/FastCent, s/IOEB - Use HardwareBase for hardware access and HAL calls for IRQ handling instead of fixed IOC/IOMD address & direct memory poking
  s/IOEB, s/Parallel - HAL version of UpdateIOEBCtrl gets copied into the module workspace to allow it to be used with the old calling standard (since we need to look up the C710 address)
  s/IOC - Add code that'll warn if this file gets built for a HAL target
Admin:
  Tested on RiscPC via IOMD ROM softload.
  Code is rather rough and could do with cleaning up in places (e.g. direct poking of timer registers in s/Common, and less-than-optimal register save/restore in ClearPINTR macro)


Version 0.74. Tagged as 'Parallel-0_74'

Detail:
  Uses suffixed file extensions for compatiblity with both objasm and asasm.
Admin:
  Supplied by Peter Naulls, tested at ROOL

Version 0.73. Not tagged

Detail:
  The tag in the Messages file for OnLine was incorrectly saved as Online.
Admin:
  Local STB


Version 0.73. Tagged as 'Parallel-0_73'

Detail:
  Resources.Finland.Messages file translated
Admin:
  Local STB


Version 0.72. Tagged as 'Parallel-0_72'

Detail:
  Added Resources.Taiwan and Resources.Taiwan.Messages
Admin:
  Local STB


Version 0.71. Tagged as 'Parallel-0_71'

Detail:
  Messages file translated into Korean
Admin:
  Local STB


Version 0.70. Tagged as 'Parallel-0_70'

Detail:
  A CallAfter was occasionally trying to write a byte whilst the reverse nibble transfer was in progress. Pending CallAfters are removed before the reverse nibble and reinstated afterwards.
  A number of errors in the reverse nibble negotiate have been fixed.
Admin:
  10000 trials of alternately reading device ID and sending 250 zero bytes gives no errors with HP930C and HP K80

Version 0.69. Tagged as 'Parallel-0_69'

Detail:
  Explan have added error elucidation from a printer
  They have also fixed a number of bugs in ioctl read device id
  I have done an interim fix of another bug in the reverse nibble code
Admin:
  Tested on RPC. No errors in several thousand reads of device ID and no errors on several thousand reads of device ID interleaved with sending data

Version 0.68. Tagged as 'Parallel-0_68'

Detail:
  these relate to reading the device id only.
  In ioctl terms;
  when reading size the data word is an integer word.
  when reading data the size is the first two nibbles as in the IEE1286 spec.
  This appears to only work when the file is open for writing only.  The
  reasons for this as yet uninvestigated.
Admin:
  This has been tested but not during a print job etc. only as a stand alone
  thing.

Version 0.67. Tagged as 'Parallel-0_67'

  Updated to build using objasm instead of aasm.
  Sources changed to be objasm-compatible.
Admin:
  Requires Library 0.71 or later.
  Requires BuildSys 3.06 or later.
  Requires Env 0.65 or later.

Version 0.66. Tagged as 'Parallel-0_66'

Detail:
  This allows applications other than the parallel device driver to determine
  what the setting of output lines means.  The Parallel Device driver will
  still raise errors based on its idea of what is the cause of the printer error
  is perhaps this behavior can be configured.
Admin:
   This has been loosly tested and the code reviewed.  In essence it is a simple
   change.

Version 0.65. Tagged as 'Parallel-0_65'

Detail:
  This is the result of the 32-bit branch merge.
Admin:
  This module has received modest testing.  It has not been exhaustively
    tested.  Projects taking this version should be careful with it as it
    may contain bugs.

Version 0.64. Tagged as 'Parallel-0_64'

Detail:
  this allows the host to query the peripheral as to what it is.  As per
  IEEE1284-1994 spec.
Admin:
  This has not been excessivly tested as it does not interfere with the
  usual workings of the parallel port.  It has been tested with the
  lexmark Z12 and a HP 950

Version 0.63. Tagged as 'Parallel-0_63'

    (the translators)

Detail:

  * Import of (minor) corrections in the Turkish translations.

Admin:

  * Verified by Alpnet.

Version 0.62. Tagged as 'Parallel-0_62'

Detail:

  * Translated resources added; note that not all strings are translated,
    only those that are used on NC products.

Admin:

  * Untested.

  * France, Germany, Italy and Spain use the Latin1 alphabet; Turkey uses
    Latin5.

Version 0.61. Tagged as 'Parallel-0_61'

Version 0.60. Tagged as 'Parallel-0_60'

Added support for "version 2" parallel: if ReadSysInfo 3 says parallel type 2, then it behaves as before, but claims device 2 rather than 0, as the interface is on a different interrupt pin.

Version 0.59. Tagged as 'Parallel-0_59'

Version 0.58. Tagged as 'Parallel-0_58'

(i.e., Element 14 owned).

Version 0.57. Tagged as 'Parallel-0_57'

Version 0.56. Tagged as 'Parallel-0_56'

Version 0.55. Tagged as 'Parallel-0_55'

Version 0.54. Tagged as 'Parallel-0_54'

Version 0.53. Not tagged

Version 0.53. Not tagged

FastParallel: device no longer faults monitor RX/TX calls.

Version 0.53. Tagged as 'Parallel-0_53'

Exchanged the "out of paper" message for "printer not ready".  This is because it is actually impossible to tell when an "out of paper" error occurs on all supported printers; instead we now use the generic "printer not ready" message.

This change applied to all resource sets with the exception of Japan.

Version 0.52. Tagged as 'Parallel-0_52'

Spinner branch merged.
Inclusion of Hdr:CMOS removed.

Version 0.51. Tagged as 'Parallel-0_51'