Stupid vga tricks...

In the process of doing an install from VN's CD on an IBM Aptiva I get the
following:

aux/vga: vgactlw: <linear 0x800000 0x1000000>: not enough free address space
rio: can't open display: initdisplay: /dev/draw/new: no frame buffer

The buildin vga is an ATI Mach64. I've attached the vgginfo.txt and vgadb
(I added
the entry for 'Aptiva'). Has anyone else seen this?

P.S. I've looked through the old mail and the only explanation to a similar
problem
is the following:



===> 2/ (text/plain) [file]
        cp /mail/fs/mbox/1048/2/body /usr/forsyth/VGADB

# VGA database.
#
# ctlr:
# These entries are used to identify a particular controller.
# 'numeric offset'="string" identifies the controller by searching
# the BIOS for "string" at 'offset'. There may be multiple identifier
# attributes per ctlr.
# Strings can be found in the BIOS ROM by using the DOS 'debug'
# command and typing 'd c000:0' to obtain a dump of the first part of
# the ROM. Note that some machines (e.g. AT&T/NCR laptops) have the
# ROM mapped at 0xE0000, not 0xC0000.
# The "clock", "ctlr", "link" and "ramdac" entries give a chain in order
# of functions to be called for this controller for initialisation, etc.
# The "clock", "ctlr", "link" and "ramdac" values can all take an extension
# following a '-'. This can be used as a speed-grade or subtype; matching
# is done without the extension.
#
ctlr
        0xC006F="8C801"                                       # STB PowerGraph X-24
        link=vga
        ramdac=att20c491-100
        hwgc=s3hwgc
        link=s3clock clock=icd2061a
        ctlr=s3801 link=ibm8514
ctlr
        0xC004F="86C805"                              # Orchid Fahrenheit 1280
        0XC004F="86C801"                              # Quadtel S3 86C801/805
        link=vga
        hwgc=s3hwgc
        ramdac=sc15025-80
        link=s3clock clock=ch9294-g
        ctlr=s3805 link=ibm8514
ctlr
        0xC0067="#9GXE"                                       # #9GXE Level-1[12]
        0xC0045="Stealth 64 Vers. 1.05"
        link=vga
        ramdac=bt485-135
        hwgc=bt485hwgc
        ctlr=s3928 linear=1
        link=ibm8514
        clock=icd2061a link=s3clock
ctlr
        0xC0064="#9GXE"                                       # #9GXE Level-16
        link=vga
        hwgc=tvp3020hwgc
        ctlr=s3928 linear=1
        link=ibm8514
        ramdac=tvp3020-200
        clock=icd2061a link=s3clock
ctlr
        0xC0094="#9-864"                              # #9GXE64
        0xC012C="V864 PCI"                            # GIS Globalyst 550
        link=vga
        hwgc=s3hwgc
        ctlr=vision864 linear=1
        link=ibm8514
        ramdac=att21c498-135
        clock=icd2061a link=s3clock
ctlr
        0xC0094="#9-964 BIOS Version 1.03.03"         # #9GXE64pro

that fix is in vgas3.c now

ATI does make other bits, and vgamach64xx.c checks a list of
known device IDs, but unfortunately checks only the first ATI device it finds.
that's why i asked about any other messages, because it
distinguishes two cases in mach64xxlinear.
i also wonder what happens with multifunction devices.
if the Aptiva boots showing the PCI device list, seeing that might help
to work out what happened.

Does the machine in question have other (non-VGA) PCI hardware made by
ATI?  If so, you may indeed be seeing a problem similar to what I saw.
The same technique may be used to fix the problem.

Martin

was nothing else printed before that?
        vgamach64xx: aperture not found
for instance?