Return to g-rex.mdwn CVS log

Up to [NetBSD Developer Wiki] / wikisrc / users / rkujawa

Bump version.

    1: [[!meta title="G-REX"]]
    2: 
    3: Programming the G-REX PCI bridge
    4: 
    5: document version 0.4
    6: 
    7: # 0. Introduction
    8: 
    9: This document describes software/hardware interface of the G-REX PCI bridge 
   10: for Amiga computers. What you're reading is a result of reverse engineering, 
   11: which was long and difficult process. 
   12: 
   13: Next time when you're going to buy a hardware product for your Amiga, don't 
   14: forget to ask the vendor to make the programming documentation publicly 
   15: available! Remeber that hardware without software is just a piece of junk...
   16: and you can't write software without hardware documentation.
   17: 
   18: In case you've noticed an error in this document please let me know. 
   19: 
   20: # 1. Theory of operation
   21: 
   22: # 1a. Hardware 
   23: 
   24: Three versions of G-REX exist:
   25: 
   26: * G-REX 1200 (for Amiga 1200 equipped with BlizzardPPC)
   27: * G-REX 4000D (for Amiga 4000 equipped with CyberStormPPC)
   28: * G-REX 4000T (for Amiga 4000T equipped with CyberStormPPC)
   29: 
   30: There were at least two different revisions of G-REX 1200. Later revision 
   31: (marked "Neue Version") probably does support DMA in first two slots. I'm not 
   32: sure if it is possible to detect revision of the G-REX in software.
   33: 
   34: Blizzard PPC hardware revision 0 is not compatible with G-REX
   35: (revision 2 is certainly compatible, not sure about revision 1). 
   36: 
   37: There's a rumor that most G-REX 4000T were recalled due to hardware problem.
   38: 
   39: G-REX is connected to local expansion slot present on CyberStorm PPC and 
   40: Blizzard PPC. These slots have different physical connectors but signals seem 
   41: to be mostly the same.
   42: 
   43: The bridge itself is an evolution of PCI bridge used previously on 
   44: CyberVisionPPC and BlizzardVisionPPC cards. These products share a lot of 
   45: similiarities (at least when it comes to PCI interface). In fact CVPPC/BVPPC 
   46: can be treated as a special one-slot version of G-REX. Maybe actually it's the 
   47: other way around ;-). 
   48: 
   49: All memory spaces of G-REX are directly visible and addressable in Amiga memory
   50: space, unlike in Mediator. Firmware allocates memory space as needed, depending
   51: on what cards are installed.
   52: 
   53: # 1b. Firmware
   54: 
   55: G-REX firmware is a part of Flash ROM present on Blizzard PPC and CyberStorm
   56: PPC boards. Known CSPPC firmware revisions supporting G-REX include 44.69 and 
   57: 44.71.
   58: 
   59: It does the dirty job of assigning PCI resources (BARs, interrupt lines, 
   60: etc.) before the OS is running. Therefore G-REX does not need any special 
   61: initialization.
   62: 
   63: # 2. Memory map
   64: 
   65: G-REX is configured as multipie AutoConf boards. Confusingly, they all have the 
   66: same vendor (8512) and product (101).
   67: 
   68: 0xFFFA0000 - PCI I/O register space, 64KB.
   69: 
   70: 0xFFFC0000 - PCI configuration space, 128KB.
   71: 
   72: 0xFFFE0000 - Bridge configuration registers, 4kB.
   73: 
   74: 0x80000000 - PCI memory space, variable size and number of boards, depending on cards installed. 
   75: 
   76: # 2a. PCI configuration space (0xFFFC0000)
   77: 
   78: Access to configuration space is a bit tricky. Be warned that access to 
   79: addresses not used by G-REX generates bus error (esp. to configuration 
   80: locations which are unused because there is no card in the slot). Depending on 
   81: how these errors are supported in your OS, it may be important to trap them and
   82: handle correctly. 
   83: 
   84: Configuration data for first slot (device 0) is accessible at offset +0x1000, 
   85: location for the next slots can be obtained by shifting the bit:
   86: 
   87: Configuration space address + (offset << device number)
   88: 
   89: For example to obtain configuration space for the second slot (device 1): 
   90: 
   91: 0xFFFC0000 + (0x1000 << 1) = 0xFFFC2000
   92: 
   93: For the third slot (device 2): 
   94: 
   95: 0xFFFC0000 + (0x1000 << 2) = 0xFFFC4000
   96: 
   97: and so on.
   98: 
   99: How to access device functions is not well analyzed, however funtion 0 is
  100: always available at address computed by the above equation. Function 1 is
  101: available at offset +0x100. One could assume that accessing the next
  102: device functions is possible by shifting the bit (as with device access),
  103: but that was not tested, becasue cards with more than two functions are not
  104: common. 
  105: 
  106: On CVPPC/BVPPC configuration space is accessible at offset +0x0 (but there
  107: are also mirrors through whole configuration space).
  108: 
  109: See [[p5pb_pci_conf_read()|http://nxr.netbsd.org/xref/src/sys/arch/amiga/pci/p5pb.c#p5pb_pci_conf_read]] and [[p5pb_pci_conf_write()|http://nxr.netbsd.org/xref/src/sys/arch/amiga/pci/p5pb.c#p5pb_pci_conf_read]] functions. 
  110: 
  111: # 2b. PCI I/O registers space (0xFFFA0000)
  112: 
  113: This space offers access to I/O registers of all PCI cards.
  114: 
  115: On G-REX BAR addresses in this space are treated as absolute.
  116: 
  117: On CVPPC/BVPPC BAR addresses in this space are treated as relative to 
  118: 0xFFFA0000. Card with I/O BAR set to 0x100 will actually be available 
  119: at 0xFFFA0100. 
  120: 
  121: # 2c. PCI memory space (0x80000000)
  122: 
  123: This space offers access to memory (and memory-mapped registers) of PCI cards. 
  124: Each PCI memory BAR is assigned a separate AutoConf board during firmware 
  125: initialization.
  126: 
  127: For example Voodoo 3, which has two 32MB memory BARs, will be visible as 
  128: two 8512/101 boards somewhere at 0x80000000 (or later).
  129: 
  130: Addresses in this space are treated as absolute. Memory BAR register set to 
  131: 0x80000000 means it is configured at this address.
  132: 
  133: On CVPPC/BVPPC this space is present at different address - 0xE0000000.
  134: 
  135: # 2d. Bridge configuration registers (0xFFFE0000)
  136: 
  137: Offset - meaning
  138: 
  139: 0x0000 - Endianness swapper mode, write 0x02 to switch bridge into 
  140: big endian mode
  141: 
  142: 0x0010 - Interrupt enable, write 0x01 to enable interrupts (INT2 on Amiga side)
  143: 
  144: No need to fiddle with these registers, as they've been already configured 
  145: properly by the firmware.
  146: 
  147: # 3. Detecting the G-REX
  148: 
  149: Since AutoConf entries are created by the firmware, it is not possible to
  150: detect G-REX easily if the correct firmware is not installed.
  151: 
  152: Detecting the G-REX is done by looking for Phase5 vendor ID (8512) and product
  153: ID 101. Keep in mind that there will be more than one such board present, as
  154: expained above.
  155: 
  156: It is possible to misdetect CVPPC/BVPPC as G-REX, since it uses the same vendor 
  157: and product ID if G-REX firmware is installed. With older firmware versions 
  158: these cards have no associated AutoConf entries.
  159: 
  160: Differentiating between CVPPC/BVPPC and G-REX in this situation is possible
  161: by looking for Texas Instruments TVP4020 vendor and product ID at the beginning
  162: of PCI configuration space. Configuration data for Permedia 2 chip will be
  163: available at offset 0x0 on CVPPC/BVPPC, but on G-REX first slot is located
  164: at offset 0x1000. See [[p5pb_identify_bridge()|http://nxr.netbsd.org/xref/src/sys/arch/amiga/pci/p5pb.c#p5pb_identify_bridge]] 
  165: and [[p5pb_cvppc_probe()|http://nxr.netbsd.org/xref/src/sys/arch/amiga/pci/p5pb.c#p5pb_cvppc_probe]] functions
  166: in the NetBSD driver.
  167: 
  168: # 4. Reconfiguring the bus
  169: 
  170: If needed, it's possible to reconfigure bus just by writing new values into 
  171: configuration space. Keep in mind that any previously initialized chips will 
  172: need to be reset and initialized again (for example 3Dfx Voodoo 3, which is
  173: initialized by the firmware so it can display early startup menu). 
  174: 
  175: # 5. Interrupts
  176: 
  177: All interrupts are converted into Amiga INT2 interrupt. There's no such thing 
  178: as interrupt acknowledge register. However, there seems to be an interrupt 
  179: enable register (see "Bridge configuration registers" above).
  180: 
  181: # 6. DMA
  182: 
  183: The bridge is certainly capable of real busmaster DMA, but it needs further 
  184: reverse engineering.
  185: 
  186: [TO BE COMPLETED]
  187: 
  188: 
  189: G-REX 4000D probably has busmaster DMA capability in all slots. G-REX 1200 has
  190: busmaster DMA in first or two first slots depending on hardware revision.
  191: 
  192: # 7. Sample PCI bridge driver implementation
  193: 
  194: The NetBSD [[p5pb|http://netbsd.gw.com/cgi-bin/man-cgi?p5pb+4.amiga+NetBSD-current]] 
  195: driver serves as an example driver implementation. It was written using the 
  196: same knowledge that went into this document.
  197: 
  198: The driver consists of several files in [[src/sys/arch/amiga/pci|http://nxr.netbsd.org/xref/src/sys/arch/amiga/pci/]] directory.
  199: 
  200: * p5membar.c - Dummy driver handling AutoConf resources.
  201: * p5membarvar.h - Structures used by the p5membar.
  202: * p5pb.c - Main driver code.
  203: * p5pbreg.h - Inlcude file containing register locations.
  204: * p5pbvar.h - Structures used by the p5pb.
  205: 
  206: The p5pb does attach on top of p5bus, however p5membar drivers attach on top of zbus (since 8512/101 entries are seen as Zorro boards).
  207: 
  208: # 8. Thanks
  209: 
  210: [[AmiBay|http://www.amibay.com]] users d0pefish and ramborolf helped testing 
  211: early versions of p5pb driver. Without their help this document would not 
  212: exist.
  213: