I am attempting to interface to a SBS-Bit3 Model 616/617 PCI to VME bridge.
The pci-bios correctly sees the card and it shows up under “pci -v”
However, pci-bios does not allocate PCI memory windows for the device. (It
needs 1 I/O window and 3 memory windows.) This is evident by the output of
“pci -v”.
This is also confirmed when I do a pci_attach_device and get the info back.
All of the address and size fields have 0x0 in them. (Note: When I do a
pci_read_config the device properly reports back what addresses it would
like to have.) The system is not creating windows to the PCI device.
In /etc/system/enum/devices/bridge there is a catch all entry for unknown
things that echoes a message to /etc/system/trap/unknown.
I can put an entry in /bridge that catches my device. I know this because I
changed the message echoed to /unknown.
When I removed the echo and try to setup an entry similar to the other ones
(PCMIA and others) I still can not get pci-bios to give me PCI windows.
What do I have to do to get PCI windows?
Why doesn’t pci-bios default to a mode where it creates the windows even
if it doesn’t know what the device is?
Note: I have used the same hardware and the exact same bios settings under
Linux and it works just fine.
FYI: I am a current QNX 4.25 user and think QNX RTP is a definite
improvement (better video support and GNU is better than the Watcom tools).
Side Note:
I am running QNX RTP Jan 2001, I can not access the WWW from my machine in
order to apply Patch B. I guess that raises another question. Why is Patch
B only available via Package Mgr access on the internet? Why isn’t it
available for download and application?
In answer to your last question (I know, this is not your chief problem, but it
is the easier to answer…), I have the same problem and it was suggested that
I download updates using package manager on a connected machine, and check the
box that says save to archive, than move the archive over to the other machine
and point package mananger to it. See a thread under my name on the
installation group for more.
As for PCI problems, I ran inot some also, but not quite like yours. I did find
there was a ‘feature’ in QRTP to be changed soon, that made it hard to access
PCI boards on other than the first PCI bus, and another that mishandled PCI
boards where the type was 06 (bridge). See the thread pci_attach() problems
with Acromag card in this group.
Tony Ganote wrote:
Greetings,
I am attempting to interface to a SBS-Bit3 Model 616/617 PCI to VME bridge.
The pci-bios correctly sees the card and it shows up under “pci -v”
However, pci-bios does not allocate PCI memory windows for the device. (It
needs 1 I/O window and 3 memory windows.) This is evident by the output of
“pci -v”.
This is also confirmed when I do a pci_attach_device and get the info back.
All of the address and size fields have 0x0 in them. (Note: When I do a
pci_read_config the device properly reports back what addresses it would
like to have.) The system is not creating windows to the PCI device.
In /etc/system/enum/devices/bridge there is a catch all entry for unknown
things that echoes a message to /etc/system/trap/unknown.
I can put an entry in /bridge that catches my device. I know this because I
changed the message echoed to /unknown.
When I removed the echo and try to setup an entry similar to the other ones
(PCMIA and others) I still can not get pci-bios to give me PCI windows.
What do I have to do to get PCI windows?
Why doesn’t pci-bios default to a mode where it creates the windows even
if it doesn’t know what the device is?
Note: I have used the same hardware and the exact same bios settings under
Linux and it works just fine.
FYI: I am a current QNX 4.25 user and think QNX RTP is a definite
improvement (better video support and GNU is better than the Watcom tools).
Side Note:
I am running QNX RTP Jan 2001, I can not access the WWW from my machine in
order to apply Patch B. I guess that raises another question. Why is Patch
B only available via Package Mgr access on the internet? Why isn’t it
available for download and application?
I have a very similar problem. I have a PC-104 system consisting of a DSP Geode GXm based CPU card and an Imagenation PXC200 frame grabber. The “pci -v” command shows the card but memory address is 0h so the Imagenation-written test app that uses the library that QNX ported for them, not surprisingly, doesn’t work. Hopefully someone at QNX will soon give us some assistance.
Previously, Tony Ganote wrote in qdn.public.qnxrtp.os:
Greetings,
I am attempting to interface to a SBS-Bit3 Model 616/617 PCI to VME bridge.
The pci-bios correctly sees the card and it shows up under “pci -v”
However, pci-bios does not allocate PCI memory windows for the device. (It
needs 1 I/O window and 3 memory windows.) This is evident by the output of
“pci -v”.
This is also confirmed when I do a pci_attach_device and get the info back.
All of the address and size fields have 0x0 in them. (Note: When I do a
pci_read_config the device properly reports back what addresses it would
like to have.) The system is not creating windows to the PCI device.
In /etc/system/enum/devices/bridge there is a catch all entry for unknown
things that echoes a message to /etc/system/trap/unknown.
I can put an entry in /bridge that catches my device. I know this because I
changed the message echoed to /unknown.
When I removed the echo and try to setup an entry similar to the other ones
(PCMIA and others) I still can not get pci-bios to give me PCI windows.
What do I have to do to get PCI windows?
Why doesn’t pci-bios default to a mode where it creates the windows even
if it doesn’t know what the device is?
Note: I have used the same hardware and the exact same bios settings under
Linux and it works just fine.
FYI: I am a current QNX 4.25 user and think QNX RTP is a definite
improvement (better video support and GNU is better than the Watcom tools).
Side Note:
I am running QNX RTP Jan 2001, I can not access the WWW from my machine in
order to apply Patch B. I guess that raises another question. Why is Patch
B only available via Package Mgr access on the internet? Why isn’t it
available for download and application?
Use pci_attach_device to find the device. Ignore the stuff in “info”.
Then use the address values reported by pci_read_config.
Thx!
Tony Ganote wrote:
Greetings,
I am attempting to interface to a SBS-Bit3 Model 616/617 PCI to VME bridge.
The pci-bios correctly sees the card and it shows up under “pci -v”
However, pci-bios does not allocate PCI memory windows for the device. (It
needs 1 I/O window and 3 memory windows.) This is evident by the output of
“pci -v”.
This is also confirmed when I do a pci_attach_device and get the info back.
All of the address and size fields have 0x0 in them. (Note: When I do a
pci_read_config the device properly reports back what addresses it would
like to have.) The system is not creating windows to the PCI device.
In /etc/system/enum/devices/bridge there is a catch all entry for unknown
things that echoes a message to /etc/system/trap/unknown.
I can put an entry in /bridge that catches my device. I know this because I
changed the message echoed to /unknown.
When I removed the echo and try to setup an entry similar to the other ones
(PCMIA and others) I still can not get pci-bios to give me PCI windows.
What do I have to do to get PCI windows?
Why doesn’t pci-bios default to a mode where it creates the windows even
if it doesn’t know what the device is?
Note: I have used the same hardware and the exact same bios settings under
Linux and it works just fine.
FYI: I am a current QNX 4.25 user and think QNX RTP is a definite
improvement (better video support and GNU is better than the Watcom tools).
Side Note:
I am running QNX RTP Jan 2001, I can not access the WWW from my machine in
order to apply Patch B. I guess that raises another question. Why is Patch
B only available via Package Mgr access on the internet? Why isn’t it
available for download and application?