On 2017年11月16日 14:40, Stefan Weil wrote:
Am 15.11.2017 um 14:09 schrieb Michael Nawrocki:
Hi Stefan,
I've got a VxWorks driver binary that explicitly looks for device ID
0x1030 (which is admittedly not ideal). It seems like the "82559
InBusiness 10/100" hardware uses this, though I've had trouble finding
an official source. The following documents reference that ID:
https://pci-ids.ucw.cz/read/PC/8086/1030
http://ks.pams.ncsu.edu/pub/ncsuscyld/i386/misc/src/trees/hdstg2/modules/pcitable
https://cateee.net/lkddb/web-lkddb/E100.html
And I found a similar post on a different mailing list that might shed
some light:
http://www.beowulf.org/pipermail/eepro100/2000-January/000760.html
It looks like the 8255x series of devices have a number of potential
IDs; maybe a property to set a specific PCI device ID would work?
Thanks,
Mike
Yes, that might be a very general solution which could be applied to all
PCI devices.
It could even be extended to include the vendor ID or any PCI
configuration value
as well.
Nevertheless the technically correct solution would be a full emulation
of the
EEPROM. Then we could provide an EEPROM for the 82559 InBusiness 10/100",
and the data from that EEPROM would set the right PCI device ID.
Jason, until we get a better solution, the last commit should be
reverted before
the new QEMU version is made. That commit also added a wrong help text.
Regards
Stefan
Let me post a patch soon.
Thanks
