A new Customer is evaluating QNX for a new project.
They are using hardware based on the PLX9050
PCI interface chip.
The chip’s class is 68000, which looks like
a bridge to QNX, so all the memory, IO & IRQ resources
aren’t populated.
Apparently there is a bug in “pci(?)” which causes
this.
http://www.openqnx.com/index.php?name=PNphpBB2&file=viewtopic&p=5412
http://groups-beta.google.com/group/comp.os.qnx/browse_thread/thread/998f6ae9acc47d88/28e87a8f3b8caa10?q=plx+qnx+6.2&rnum=2&hl=en#28e87a8f3b8caa10
http://groups-beta.google.com/group/comp.os.ms-windows.programmer.nt.kernel-mode/browse_thread/thread/210acda307bed9a8/3c00cd081a8d7c26?q=plx+qnx&rnum=4&hl=en#3c00cd081a8d7c26
According to PLX (last thread):
“Under Windows the PCI 8080 default Class Code (Other PCI bridge
device) can be used. If your device might be used under the QNX OS
then the Class Code should be changed (in the serial EEPROM), as QNX
does not report address space mappings to the driver if the PCI Class
Code is the default Other Bridge (068000h) class.”
Is there a “fixed” version of pci available the customer can try?
Is there a workaround?
I had a look in my PE installation, but can see no
pci.c. If the source to pci is available, where can it
be accessed? Will fixing pci help or is there another
server involved (pci-bios maybe)?
PS The hardware resources are identified
correctly under DOS, Windows & Linux so it
is definitely a QNX issue.
The target is QNX6 x86, with SP1 loaded.
\
Using Opera’s revolutionary e-mail client: http://www.opera.com/m2/
Alex/Systems 104 wrote:
A new Customer is evaluating QNX for a new project.
They are using hardware based on the PLX9050
PCI interface chip.
The chip’s class is 68000, which looks like
a bridge to QNX, so all the memory, IO & IRQ resources
aren’t populated.
Apparently there is a bug in “pci(?)” which causes
this.
http://www.openqnx.com/index.php?name=PNphpBB2&file=viewtopic&p=5412
http://groups-beta.google.com/group/comp.os.qnx/browse_thread/thread/998f6ae9acc47d88/28e87a8f3b8caa10?q=plx+qnx+6.2&rnum=2&hl=en#28e87a8f3b8caa10
http://groups-beta.google.com/group/comp.os.ms-windows.programmer.nt.kernel-mode/browse_thread/thread/210acda307bed9a8/3c00cd081a8d7c26?q=plx+qnx&rnum=4&hl=en#3c00cd081a8d7c26
According to PLX (last thread):
“Under Windows the PCI 8080 default Class Code (Other PCI bridge
device) can be used. If your device might be used under the QNX OS
then the Class Code should be changed (in the serial EEPROM), as QNX
does not report address space mappings to the driver if the PCI Class
Code is the default Other Bridge (068000h) class.”
Is there a “fixed” version of pci available the customer can try?
Is there a workaround?
I had a look in my PE installation, but can see no
pci.c. If the source to pci is available, where can it
be accessed? Will fixing pci help or is there another
server involved (pci-bios maybe)?
PS The hardware resources are identified
correctly under DOS, Windows & Linux so it
is definitely a QNX issue.
The target is QNX6 x86, with SP1 loaded.
I use the PCI9054 and I had to change the class code in hardware for QNX
to populate it properly. Otherwise you can fill in the pci_dev_info
structure yourself after pci_attach_device.
Francois
On Fri, 22 Apr 2005 08:57:19 +0200, Francois Joubert
<sommerfj@webmail.co.za> wrote:
I use the PCI9054 and I had to change the class code in hardware for QNX
to populate it properly. Otherwise you can fill in the pci_dev_info
structure yourself after pci_attach_device.
Thank you - will have advised the customer on both
issues.
FYI, a patch for pci-bios server for x86 platform was posted in newsgroups.
I think it worked fine for everyone who tried it.
– igor
“Alex/Systems 104” <acellarius@yah0o.lsd.com> wrote in message
news:opspsxv1kvbinb6v@alex-pentium-m…
On Fri, 22 Apr 2005 08:57:19 +0200, Francois Joubert
sommerfj@webmail.co.za> > wrote:
I use the PCI9054 and I had to change the class code in hardware for QNX
to populate it properly. Otherwise you can fill in the pci_dev_info
structure yourself after pci_attach_device.
Thank you - will have advised the customer on both
issues.
On Mon, 25 Apr 2005 20:32:19 -0500, Igor Kovalenko <kovalenko@comcast.net>
wrote:
FYI, a patch for pci-bios server for x86 platform was posted in
newsgroups.
I think it worked fine for everyone who tried it.
Thank Igor
I saw mention of the patch but never saw it.
Also the source to pci-bios doesn’t appear to
be part of the default PE install
(it’s not on my machine anyway)