GVP Binddrivers Causes Black Screen

[thebajaguy]

Enable your ROM
Connect your device, leave the power off.
Boot from a disk with a stripped down OS of the ROM you are using.  
On the disk, Include a GVPSCSICTRL tool from the GVP utilities disk.  
Have a partition tool handy.  If not FastPrep/ExpertPrep, then have HDToolBox handy.  Use HDToolBox gvpscsi.device

After boot up*, add power to your device.

To rescan the bus for missed devices, you can use gvpscsictrl -r

Use your partitioning tool.

Note that I think there is either an incompatible file system in the boot block, or more likely corrupted boot block information on that media.  The above rescan should be more tolerant of the bad boot block.

some gvpscsictrl info - http://www.verycomputer.com/2_a4679a1bc606cb53_1.htm

gvpscsi.device is tolerant of LUNs on devices, but some devices that make LUNs are stupid and can sometimes mirror devices at higher LUNs.  Note that there is a LastLUN feature in the RDB (seen in ExpertPrep screen) if you encounter this kind of thing.

*Binddrivers also works at this point if the ROM is disabled and driver in Expansion folder of SYS: - test your v4.15 Expansion drawer version with no ROM enabled this way.

Disabling/removing the ROM causes the RAM not to be software-mapped into the memory list.  AddMem 01000000-01FFFFFF (for 16MB, less for less installed). if you need it in a pinch.  ROM mapping is the better way.

This kind of thing was an issue on the A3001 AT interface if the old prep tools corrupted the boot block or the FFS installed in the boot block.  The binddrivers way was how we cleared it.

Robert Miranda - Fmr GVP Tech Support.

[thebajaguy]

For historical purposes, v3.07 was the first release of gvpscsi.device.  Longer bus scan on ID 0 and other IDs to make sure slower devices were found.  First Series I support.  Much better removable media support, and more efficient Polled I/O performance on the Series I.  Many bugs of the 33C93/33C93A were worked around.  Driver seeks out FastRAM or 32-bit FastRAM RAM to copy itself into.  Good LUN support.  SCSI-driect compliant and RDB compliant programming interface.

3.12 was a change to the scan process, faster boot was the result (shorter scan period).  GVPSCSICtrl added to rescan (insert in startup) as needed, or run manually if a device gets powered later.  First Combo030 support appeared here.

Versions from 3.12 to 3.15 were mainly additions for the first group of 68030 + RAM + SCSI cards.  One was a bug fix for a DMA to onboard 32-bit RAM bug found on the Rev3 surface-mount version.  Another was to recognize the Rev 4 card that replaced it (these were the Combo 030 cards, with surface-mount 68030's), and another was to incorporate a few hack fixes for HC8/A500HD8 Series II like the AdSpeed 68000 cache accelerator that fails to hook the OS cache control routines (though illegally tried to sniff the Zorro expansion bus, but doesn't cache ChipRAM because of the same issue - you are not allowed), the serial.device fixes to limit heavy DMA transfers when time-sensitive serial.deivce is open at higher baud rates, and maybe some others I am not recalling.  3.15 is the last version to support the Series I cards.  SCSI-level code was not modified as far as I know after 3.12.  I have a v3.9 beta in my stock, and it's much like the v3.7/3.07 release.

Ignoring any labels on ROMs, most of the 3.x series have the AT interface code for the AT interface on the A30xx sandwich accelerator boards burned higher in the 27C256 ROM/EPROM.  Some FastROM 4.0 labels are actually v3.3 AT and v3.15 SCSI code on them.  The first gvpscsi.device release was 4.5 on the newer G-Force PGA-chip 68030 accelerators and the 68040 board for the A2000.  Lost from the code was Series I controller support, and the AT code in a seperate area of the ROM was dropped from the image.  SCSI-level code was not modified from the later 3.x series.  I know the Supra 28Mhz accelerator had the same issue as the AdSpeed, so they revamped the kludge control tool option as 68KCache, and the DMA-tranfer size limiter tool can be pointed at any .device in the system, and activates when the device is in an open state.  I think there are v4.5, 4.12, 4.14, and 4.15 (read Ralph's explanation when it was released - though never on a produced card).