-----------------------------------------------------------------------------
Changes since release 2.0:
- Fixed defects in the list management of PCI boards.
- Forgot to map the legacy IO window through via the MMU.
- Added a preferences file ("ENVARC:PCI-Configuration").
- Fixed code if run on an MMU-less system.
- Tested successfully on a Firestorm and A4000 Mediator,
more tests welcome.
- Apparently, the PCIInit tool that came with the previous
version was broken. This got fixed as well.
-----------------------------------------------------------------------------
Changes for release 2.2:
- Added functions for allocating, releasing and announcing PCI
memory for DMA bounce buffers.
- Added an emulation of the prometheus.library.
- Added commands to reserve IO space and memory space.
-----------------------------------------------------------------------------
Changes for release 3.2:
- The functions for bounce buffer allocation have been debugged.
- Added two new tags for FindBoardTagList() to find a board by
a memory range. This allows P96 to identify an OpenPCI handled
board from the frame buffer address.
- Added development files such as the .fd file and the includes.
-----------------------------------------------------------------------------
Changes for release 4.2:
- This release adds support for PCI-2-PCI or PCI-2-PCIe bridges.
However, such bridges require additional hardware support to
allow accessing devices behind such bridges. At the time of
writing, only the Firestorm PCI bridge is able to generate
such extended PCI configuration cycles and thus, this is the
only bridge that supports bridges.
Support is currently limited to bridges that offer a single
PCI bus behind the bridge. This should apply to the majority
of PCI plug-in cards.
- Added functionality to remove the cybpci.library from the
system to stop interfering with the openpci.library.
-----------------------------------------------------------------------------
Changes for release 4.3:
- The PCI scanning functions were broken and could not continue
correctly from a middle of a scan, i.e. if the previous device
passed in was non-NULL.
- The Grex detection algorithm was a bit uncareful can could have
hung machines. It now tests for the presence of the ID tags of
the corresponding boards.
- The Grex interrupt server did not test all flags necessary to
ensure that an interrupt was triggered.
- Grex boars board detection now also includes the CVisionPPC if
it is present, and the early PCI bridges.
- Grex detection now also includes the PCI reset sequence that
might have been missing if the PCI firmware was not initialized.
- PCIInit no longer fails with an Alert if no PCI bridge boards
were detected.
-----------------------------------------------------------------------------
Changes for release 4.7:
- A new command SetCacheLineSize was added to the
PCI-Configuration file. This command sets the
cache line size field in the PCI config area. Typically,
it should be set to 0.
- Support for G-Rex was finalized. The G-Rex board is now
fully initialized and PCI configuration cycles now also
generate parity correctly.
- The G-Rex PCI configuration code now also detects the
legacy PCI bridge which connects to the CVisionPPC. This
potentially requires an update of the CVisionPPC graphics
card driver as it may now use the openpci.library as
an optional component.
- The G-Rex interrupt handler is now fully debugged and
working.
- Mediator A1200 device detection has been verified on
real hardware. Still missing confirmation that the MMU
window handling works not only under emulation.
-----------------------------------------------------------------------------
Changes for release 5.2:
- A couple of new tags to GetBoardAttrs() allow to
retrieve the interrupt line, the legacy IO space or
the offset from the PCI address space to 68K virtual
addresses.
- The G-Rex support functions did not fill in the device
number correctly.
- Includes for assembler developers were included.
- Added sources and binaries for an lspci example binary
that lists the devices on the PCI bus(es).
-----------------------------------------------------------------------------
Changes for release 5.3:
- FindBoardTagList() had a defect and crashed if it was
called. This got fixed.
- The cybpci.library is now removed immediately after
detecting a G-Rex, avoiding a conflict when the PCI
configuration is changed under the feed of cybpci.
- The include files still misdocumented one field
("master") incorrectly as 32-bit whereas it is just
16 bit, also under the version 1 of openpci.
- Forgot to include an assembler lvo file defining the
library offsets.
-----------------------------------------------------------------------------
Changes for release 5.4:
- The Grex board installs now its own DMA memory provider
which allocates RAM from the board RAM. This should work
as PCI devices on the GRex can do DMA into the CPU RAM,
unlike other boards.
- Fixed a wrong pointer reference when removing cybpci.
-----------------------------------------------------------------------------
Changes for release 5.5:
- The code now also removes the P5 PCI ConfigDevs from autoconf as they
should not be used anymore. Instead, PCI access should be routed through
the openpci.library and its expansion nodes.
-----------------------------------------------------------------------------
Changes for release 6.2:
- Due to a defect in a header file, creating MMU tables for the
A1200 mediator boards did not work.
- In case the MMU table setup did not work, the code did not clear
device flags requesting MMU window processing.
- Creating expansion library ConfigDevs did not check whether some
of the BARs have not been initialized.
-----------------------------------------------------------------------------
Changes for release 6.3:
- During preferences parsing, a structure to hold the parsed arguments
was not released, causing a mild memory loss.
- A new command was added to the PCI-Configuration file, namely
"EMULATE". This command allows to enable emulation of third party
libraries, for example "Emulate Prometheus" will enable the prometheus
emulation layer, including a "fake" configDev structure pointing to the
legacy IO region. This works regardless of the actual PCI board
installed. Note that while previous releases always enabled the
Prometheus emulation, this emulation now requires explicit user
interaction by editing "ENVARC:PCI-Configuration".
- When initializing PCI devices for one of the 1200 Mediators,
the code overwrote parts of the device structure, implicitly
allocating the device. This got fixed.
- The SetBoardAttrs() function now expects a "struct Node *" as
argument to PRM_BoardOwner, and GetBoardAttrs() returns one.
Unfortunately, the prometheus library documentation was not very
clear on what the expected argument should be. Similarly, the
library stores now its own base address (and thus a struct Node *)
as owner on pci_obtain_card().
-----------------------------------------------------------------------------
Changes for release 6.5:
- This version now fixes the interrupt routing through PCI
bridges and also configures the InterruptLine PCI configuration
register to include the interrupt the device is actually
generating.
-----------------------------------------------------------------------------
Changes for release 6.6:
- Interrupt enabling and disabling now not only happens at
bridge level, but also at PCI device level through the
device command register.
-----------------------------------------------------------------------------
Changes for release 6.7:
- Interrupt management has been improved further by disabling
the interrupt at bridge level only if there is no additional
interrupt server queued to listen to the interrupt. This
prevents problems if multiple interrupt servers for multiple
devices are installed, but one of the servers is removed.
- Fixed the include files, the arguments should be TagItem *,
not TagList *, and the FindBoard() function follows now the
convention of the remaining tag based calls by using an A
as appendix for the regular function, and the function name
without the appendix for the varadic function.
-----------------------------------------------------------------------------
Changes for release 7.2:
- Memory and IO base and limit generation for PCI-to-PCI
bridges was wrong because it set the upper limit to the
exclusive, not the inclusive largest address, thus requesting
the bridge to map more than the absolute minimum address range
through the bridge.
- Bridge memory and IO limits were not rounded correctly to
the granularity of the bridge registers.
- The bridge subordinate bus configuration byte was set
incorrectly in case more than a single bus was behind the
bridge; in fact, bridges could not detect more than a single
bus behind them at all. This was fixed.
- Installation of the interrupt line on the Mediator A1200
closed the PCI configuration bank and thus prevented
configuration of any but the first PCI device.
- In case a VGA device is found behind the bridge, the code
now ensures that access to the legacy VGA registers is
mapped through the bridge.
- This release disables memory prefetching at the bridge
level. There is probably little to be gained as the 68K
and the Zorro bus dominates the access time, and
prefetching may interact badly with memory mapped IO
registers.
-----------------------------------------------------------------------------
Changes for release 7.2.1:
- The library remained unchanged, but the archive got
reorganized a bit. This version includes an installer
script and also a guide.
-----------------------------------------------------------------------------
Changes for release 7.2.2:
- This release fixed the default source directory of the
installer.
-----------------------------------------------------------------------------
Changes for release 8.2:
- The emulate command in PCI-Configuration was extended.
-----------------------------------------------------------------------------
Changes for release 9.2:
- The built-in DMA memory provider for the G-Rex
host bridge now also allocates cache-disabled
memory by going through the PCI library, and
resets the caching mode as soon as the memory
is released.
- This version adds experimental support for PCI
to PCI bridges on the Mediator bridges. This is
experimental and to some degree based on guesswork,
so feedback is required.
-----------------------------------------------------------------------------
Changes for release 9.3:
- The PCI device scanner for the mediator bridges
bypassed PCI bridges early and thus could have
never supported bridges in first place.
Nevertheless, most mediator boards are not
capable to support bridges, unfortunately.
- The IO and memory regions for PCI devices
are now only enabled after BAR registers have
been loaded with the final addresses.
-----------------------------------------------------------------------------
Changes for release 10.2:
- Preferences are now parsed before BAR registers
are sized. This has the effect that the IGNORE
command for BARs will also prevent that the
ignored BARs are even written to.
- Matching a PCI by class code in PCI-configuration
could have trashed memory.
- The SetInterrupt command in PCI-Configuration failed
to reroute interrupts and thus was mostly useless.
- SetInterrupt and SetPri may have failed to configure
PCI devices correctly on the A1200 mediator.
- In PCI-Configuration, the device function is now
separated by a dot from the device ID to mirror
the syntax of lspci.
- Added a new SLOT keyword to most commands for
PCI-Configuration to identify a device by its bus
number, slot number and (optionally) function number.
-----------------------------------------------------------------------------
Changes for release 10.3:
- Bridge configuration was refactored again. Bridge
and parent device are now passed already into the
constructor of a PCI device, and not only when it
is added.
- Fixed a trashed register in the Mediator board
scan.
- The Mediator scan did not pass the parent device
correctly into the device creator, thus giving
devices behind bridges wrong bus IDs.
- For some strange reason, the Mediator does not
seem to decode the address bits for selecting a
device behind a PCI bridge correctly; it seems
to ignore them. Therefore, the Mediator currently
supports only a single device behind a bridge.
- Some bridges take apparently several milliseconds
to accept a configuration. Now the library inserts
a 50ms delay after having installed the target
busses after setting up a PCI bridge.
- The time delay function was broken and passed in
the wrong command to the timer.device.
-----------------------------------------------------------------------------
Changes for release 10.4:
- The algorithm for finding an address space window
for the A1200 mediators has been changed. It no
longer looks behind the last memory segment, but
uses the largest hole from the 24bit limit up.
- The installer script has been updated. It now
checks whether the archive is corrupt, and also
backs up the PCI-Configuration file if it
existed.
- PCIInit no longer checks whether the version
of openpci is larger or equal than its own
version. Version 3 of openpci is now sufficient
since that version introduced the MMU initialization
API.
-----------------------------------------------------------------------------
Changes for release 10.5:
- When tearing down the library due to an Expunge,
the code trashed memory due to an incorrectly
loaded memory.
- When scanning for devices, the scan no longer
stops after the first device of a multi-function
device that claims not be of multi-function type.
-----------------------------------------------------------------------------
Changes for release 11.2:
- GetBoardAttrs() and SetBoardAttrs() could loop forever
on unknown tags, the end marker for the known tags was
incorrect.
- The PCI BAR register parser did not compute the parity
correctly for some boards.
- Two new tags were added for GetBoardAttr(), namely
PRM_PCIMemWindowLow and PRM_PCIMemWindowHigh, which
provide the PCI address range available for a device
at config time, or the PCI address range a device is
mapped into later. These are PCI addresses, not 68K
addresses.
- SetBoardAttr() can now write PRM_MemorySizeX and
PRM_MemoryFlagsX at config time to dynamically size
PCI devices. This is useful for tools called from
within the PCI-Configuration file.
- Fixed a missing register initialization when unloading
tools from LIBS:PCI.
- Added example source code for an init tool for sizing
PCI devices dynamically.
- Fixed the description of the autoconf vendor IDs in the
autodoc file.
-----------------------------------------------------------------------------
Changes for release 12.2:
- The "lspci" program accepts now an additional argument,
namely "NUMERIC". If set, then it does not attempt to
resolve the PCI vendor and device ID to human-readable
numbers but rather prints their hex values.
- Dynamic PCI initialization through external segments
was augmented. Init functions receive now one additional
argument in a1, and may return a (non-NULL) pointer.
Initially, Init functions are called with this argument
set to NULL, and receive in register a0 a pointer to a
RDArgs structure for command line parsing. If they return
a non-NULL pointer and not a small number as error code,
they are called *once again* the PCI environment is
completely setup. They then receive NULL as RDArgs in a0,
but its own (previous) return code in a1 to complete
a potential initialization of a device.
-----------------------------------------------------------------------------
Changes for release 12.3:
- lspci sources included the wrong header,
<strings.h> is BSD-only, should be <string.h> which is POSIX.
- Fixed a defect in the mediator emulation setup.
-----------------------------------------------------------------------------
Changes for release 12.4:
- This version fixes a race condition in initializing the
emulated library bases.
- As side effect, emulated libraries are now already showing
within exec once openpci has been loaded. MMU initialization
is now also performed by opening one of the emulated pci
libraries.
-----------------------------------------------------------------------------
Changes for release 12.5:
- The GetPhysicalAddress() and GetVirtualAddress() of the
prometheus emulation library expected the input address
in register a0 instead of d0. This got fixed.
-----------------------------------------------------------------------------
Changes for release 12.6:
- When PCI BAR sizes are defined through SetBoardAttrs(),
the library no longer attempts to read back the
corresponding BAR as it assumes that the corresponding
BAR is non-standard and does not allow read-access.
Instead, the defined BAR size is assumed to be correct.
-----------------------------------------------------------------------------
Changes for release 12.7:
- The ReserveMemSpace function in PCI-Configuration could
hang forever if a bridge does not support access to the
low memory PCI area and reservation failed.
- In case address space allocation for a BAR fails, the
corresponding device is disabled, excluded from further
reservations and removed from the listed device list now.
- Translation between 68K addresses and PCI addresses is now
aware of the memory region reserved by ReserveMemSpace
and maps addresses below the reservation limit through,
regardless whether any BAR actually uses the region.
-----------------------------------------------------------------------------
Changes for release 12.8:
- The legacy VGA window, made accessible by ReserveMemSpace
was not properly mapped by the MMU, and neither received
an Expansion ConfigDev node. This has now been fixed. The
product ID for this region is now 7 (see also openpci.doc).
- The PCI address space algorithm was updated to make better
use of a non-aligned upper PCI memory window. Now, smaller
PCI BARs are placed there first to make best use of the
available address space.
- The guide was updated to describe the legacy VGA window.
-----------------------------------------------------------------------------
Changes for release 12.9:
- The pci_logic_to_physic_addr() and related functions
were actually broken because one of the used instructions
did not (as expected) generate condition codes for
determine error conditions. This has been fixed.
-----------------------------------------------------------------------------
Changes for release 13.2:
- Added the "VirtualMapping" command for PCI-Configuration.
If this is enabled - which is the default - the PCI
address space of the A1200 mediator boards is enlarged by
dynamically (virtually) mapping addresses of the 68K
to the small PCI window. Can be turned off to always have
physically mapped PCI addresses, though only few devices
then fit into the relatively small address space.
- Flushing the library now restores the original ConfigDev
structures.
- If the emulation code was flushed, the code forgot to
remove the emulated expansion structures and thus
could have crashed the machines.
- If the PCI bus was rescanned after flushing the library,
the MMU blocked access to non-existing devices and thus
caused the code to trigger MuForce hits - for devices that
are not present. The library now bypasses the MMU for
scanning. It already did for Firestorm and G-Rex.
-----------------------------------------------------------------------------
Changes for release 13.3:
- The mediator emulation code did not pass the device/function
offset correctly to the lower level config space accessor
functions and therefore failed to read or write configuration
registers at all.
- DMA memory handling for mediator emulation was unfortunately
broken. While this is fixed, DMA memory is only available through
the emulation if the underlying bridge maps 68K addresses
directly to PCI addresses, without any offset.
-----------------------------------------------------------------------------
Changes for release 13.4:
- The alloc_dma_mem() function returned nonsense instead of NULL
if no DMA memory provider is available.
- The emulation functions now call (as much as possible) through
the openpci API.
-----------------------------------------------------------------------------
Changes for release 14.2:
- The PCI address space/BAR allocation algorithm was revisited
again. It is now some kind of "buddy" allocation scheme that
should be (nearly) optimal and should be able to squeeze in
BARs even if top and bottom end of the window are not
allocated.
- The lower PCI memory area end for the Prometheus was not properly
indicated, thus potentially causing BARs to leak into the I/O or
config areas.
- The upper end of the PCI memory area for the firestorm was
extended a little bit to make more room. This requires testing
whether real hardware behaves as documented.
- Added two new functions to the API, ObtainPCIRegion() and
ReleasePCIRegion() which temporarily can provide access to
any area in the PCI config space. The typical use case for
these functions is to write to legacy VGA space. Not all PCI
host adapters will support such accesses. Included are currently
the firestorm and the Mediators (potentially). Example source
is included (ObtainExample.c) to demonstrate the API.
- As the API is extended, pragmas and headers have been regenerated.
-----------------------------------------------------------------------------
Changes for release 14.3:
- Fixed bridge allocation (again, hopefully).
-----------------------------------------------------------------------------
Changes for release 14.4:
- The granularity of which bridge limits are allocated now also depends
on the devices behind the bridge (and, recursively, behind bridges
behind the bridge). This should also make bridge limit allocation almost
optimal.
- Enlarged the delay to move out of reset to 100ms, and the delay to
start scanning behind a bridge to 100ms.
-----------------------------------------------------------------------------
Changes for release 14.5:
- openpci now installs a patch into AllocMem if one of the emulated
libraries is opened and by that provides DMA memory to drivers that
ask for DMA memory this way. The allocation then goes directly down
to P96 without requiring a memory pool in exec. This method has the
advantage that DMA memory cannot be allocated "by accident" if the
system runs low on memory because such memory might be incompatible
to forms of Zorro DMA as it may be virtually mapped. The patch
is removed again once the emulation is closed again.
-----------------------------------------------------------------------------
Changes for release 14.6:
- This release fixes a crash when attempting to remove one of the
emulation libraries.
-----------------------------------------------------------------------------
Changes for release 15.2:
- The PCI-Configuration file offers a new command, namely "Hint".
With this command, openpci can be hinted which BARs of a PCI
device contain memory-mapped registers and which contain video
RAM. The purpose of this command is that it allows openpci to
support VGA graphics cards even in restricted configurations with
only two small PCI windows accessible from the 68K side without
using a virtual MMU mapping window. If such a hint is found,
openpci makes the first 4MB of the IO window and another 4MB
window into vdeo RAM accessible to the 68K side. This makes
some graphics cards such as the Voodoo workable without MMU tricks
in such limited environments, but also restricts the usable video
RAM to 4MB. Driver developers need to take care that they clamp
the video RAM to the size delivered from the pci_dev structure
and do not attemp to read them from hardware. Furthermore,
MMIO registers are in such setups not reachable by DMA - this should
hopefully be not a restriction.
- lspci received a new command line option "PCIADDRS" to print
the addresses of the PCI bar registers as seen from the PCI space.
The default is to print the addresses as used by the 68K side.
- The DMA memory provider is now more careful and checks whether
PCI and 68K addresses differ for memory allocated for some emulation
modules that cannot handle such translations.
-----------------------------------------------------------------------------
Changes for release 16.2:
- When allocating DMA memory from a G-Rex, and no suitable memory
was found, the previous release crashed with a bad memory alert.
The fixed code just does not deliver any memory.
- The emulation code now also accepts G-Rex memory as DMA memory
and provides it as such under emulation.
- The memory allocation patch for the emulation will now also
request G-Rex memory if DMA memory is desired through AllocMem().
- A new command was added to PCI-Configuration, namely DMAMemSource.
This command allows to select the board from which DMA memory is
taken if requested through the (old) openpci interface that does
not take a target board for which the memory is desired. For
example "DMAMemSource G-Rex" designates the G-Rex board in a multi-
PCI setup as source board for DMA memory, asuming that DMA-requesting
devices are plugged into this board.
-----------------------------------------------------------------------------
Changes for release 17.2:
- Added a command to disable the second PCI window on some
boards as it is not always working, in some configurations.
Alternatively, the board should be jumpered to 4MB only.
-----------------------------------------------------------------------------
Changes for release 17.3:
- Due to a left-over debugging switch, the 17.2 version
always handled mediator bridge boards as if they had an
8MB window.
-----------------------------------------------------------------------------
Changes for release 17.4:
- Interrupt handling on the mediator bridges improved and
should now be able to handle multiple simultaneous
interrupts at once.
-----------------------------------------------------------------------------
Changes for release 17.5:
- At most 512MB are reserved for the MMU window of the
small mediators as the window management of the hardware
cannot handle larger windows.
-----------------------------------------------------------------------------
Changes for release 17.6:
- When reading BAR registers, the scanner now disregards
those registers that are not negatives of powers of 2
and thus likely corrupt.
- One mediator emulation function had to keep CPU
registers intact.
- Fixed computation of the announced Mediator window mask.
-----------------------------------------------------------------------------
Changes for release 17.7:
- The algorithm for determining whether devices of a bridge
fit into an MMU window was a bit too conservative and
did not attempt to place BARs ideally, i.e. failed to
mirror the actual generic allocation algorithm of the
bridge. This could have caused failure to allocate devices
within an MMU window at all.
-----------------------------------------------------------------------------
Changes for release 17.8:
- Due to an oversight, the G-Rex emulation was always
enabled, even if not requested. This has been fixed.
-----------------------------------------------------------------------------
Changes for release 17.9:
- Interrupt handling was reworked. Even if an interrupt
handler signaled that it was responsible for an
interrupt, the interrupt is passed on as the server
cannot exclude that another device triggered an interrupt
simultaenously.
- Fixed the algorithm to determine invalid BARs.
-----------------------------------------------------------------------------
Changes for release 17.10:
- Interrupt handling for mediator was reworked again.
Interrupts are now acknowledged by writing into the
interrupt enable register once the interrupt source
has been detected.
- The "find PCI device by class" function in the mediator
emulation was broken and got fixed.
-----------------------------------------------------------------------------
Changes for release 17.11:
- The mediator interrupt mask computation was still not
correct.
-----------------------------------------------------------------------------
Changes for release 17.12:
- The logic to refuse invalid BAR registers unfortunately
also refused to accept 16 bit I/O registers. This was
fixed.
-----------------------------------------------------------------------------
Changes for release 17.13:
- Window allocation changed; a supervisor allocated window takes
now priority and is not mapped out unless an overlapping window
access requires so.
-----------------------------------------------------------------------------
Changes for release 17.14:
- Added support for emulation of interrupt probing.
-----------------------------------------------------------------------------
Changes for release 40.2:
- Added a workaround for a defect in PatchControl. This
program is, however, not recommended due to defects
when patching the exec memory allocation functions. It
is recommended to replace it by TRSaferPtch.lha.
-----------------------------------------------------------------------------
This archive contains an implementation of the openpci.library following
the API defined by Benjamin Vernoux, ported to AmigaOs 4 by St?phane Guillard.
Unlike the former implementation by Benjamin and St?phane, this version does
not depend on an underlying bridge-board specific library but operates on top
of the hardware directly. It is also compatible to the mmu.library and uses
its services for banking PCI windows. This version also avoids unnecessary
long MMU table setup times implied by some PCI bridge boards that drill
unnecessarily large wholes into the 68K address space.
This version supports the following PCI bridge boards:
- The original Prometheus board
- The Prometheus Firestorm board
- The Grex bridge board
- The A4000 Mediator boards (all variants)
- The A1200 Mediator boards (all variants)
Restrictions:
- The A4000 Mediator board does not support MMU mapping and is restricted
to a 512MB PCI window (see below for the rationale).
- The mechanism to allocate memory from RTG graphic boards as buffer
for PCI DMA memory transactions requires at least version 3.5.0 of
the P96 software. At the time of writing, this version is not yet
available. It is likely going to be published in approximately
December 2024.
Some of these restrictions will be lifted in the future. The future of
this library depends on you - please help its autor by providing test
results.
------------------------------------------------------------------------------
Installation:
- Run the installer included in the archive. Alternatively:
- Make a backup of your current openpci.library, if you use it.
- Copy the openpci.library to LIBS:, replacing other versions.
- Third-party PCI bridge libraries are no longer needed. They can be
kept or removed.
- Install "PCIInit" to LIBS:mmu.
- Edit the "ENVARC:MMU-Configuration" file and add the following line
at its end:
PCIInit
This MMU tool loads the openpci.library into RAM, locks it and restricts
the MMU table building to the areas actually occupied by PCI components.
This will speed up MMU table build-up dramatically.
Now reboot. PCI implementations based on openpci based PCI implementations
will continue to work.
------------------------------------------------------------------------------
The THOR-Software Licence (v3, January 2nd 2021)
This License applies to the computer programs known as the "openpci.library"
and its sources. The "Program", below, refers to such program. The
"Archive" refers to the package of distribution, as prepared by the author
of the Program, Thomas Richter. Each licensee is addressed as "you".
The Program and the data in the archive are freely distributable
under the restrictions stated below, but are also Copyright (c)
Thomas Richter.
Distribution of the Program, the Archive and the data in the Archive by a
commercial organization without written permission from the author to any
third party is prohibited if any payment is made in connection with such
distribution, whether directly (as in payment for a copy of the Program) or
indirectly (as in payment for some service related to the Program, or
payment for some product or service that includes a copy of the Program
"without charge"; these are only examples, and not an exhaustive
enumeration of prohibited activities).
However, the following methods of distribution involving payment shall not
in and of themselves be a violation of this restriction:
(i) Distributing the Program on a physical data carrier (e.g. CD-ROM,
DVD, USB-Stick, Disk...) provided that:
a) the Archive is reproduced entirely and verbatim on such data carrier,
including especially this licence agreement;
b) the data carrier is made available to the public for a nominal
fee only, i.e. for a fee that covers the costs of the data carrier,
and shipment of the data carrier;
c) a data carrier with the Program installed is made available to the
author for free except for shipment costs, and
d) provided further that all information on said data carrier is
redistributable for non-commercial purposes without charge.
Redistribution of a modified version of the Archive, the Program or the
contents of the Archive is prohibited in any way, by any organization,
regardless whether commercial or non-commercial. Everything must be kept
together, in original and unmodified form.
Limitations.
THE PROGRAM IS PROVIDED TO YOU "AS IS", WITHOUT WARRANTY. THERE IS NO
WARRANTY FOR THE PROGRAM, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY RIGHTS. THE ENTIRE
RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD
THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY
SERVICING, REPAIR OR CORRECTION.
IF YOU DO NOT ACCEPT THIS LICENCE, YOU MUST DELETE THE PROGRAM, THE ARCHIVE
AND ALL DATA OF THIS ARCHIVE FROM YOUR STORAGE SYSTEM. YOU ACCEPT THIS
LICENCE BY USING OR REDISTRIBUTING THE PROGRAM.
Thomas Richter
-----------------------------------------------------------------------------
So long,
Thomas (March 2025)
|
aminet net>