Quoting ron minnich rminnich@gmail.com:
On 9/25/07, Marc Jones marc.jones@amd.com wrote:
All this seems like a good reason to let the driver and/or system level software enable PERR# and SERR# and for LinuxBIOS to leave them alone.
you're right. I don't even remember when those started getting set, and had not noticed it, but it's a mistake.
I'm almost inclined to say "leave that line in there commented out, with a warning: NEVER DO THIS!". That's up to you.
Acked-by: Ronald G. Minnich rminnich@gmail.com
Marc, your the man! This is the exact problem I am having with the Intel 82801DB. When the PCI Bridge goes to "Enabling resources..." it just freezes. I have traced it back to the "command |= (PCI_COMMAND_PARITY + PCI_COMMAND_SERR); /* error check */" line. Yeh I have done alot of reading about parity errors and it seems to be something software drivers on the OS level use, not at the bios level. Parity error checking also seems to be very old method. So, I was starting to question why this line was even there. These bits are not set with the factory bios. Also, what about the pci_bus_enable_resources() function in pci_device.c? Do we want to comment out the "ctrl |= (PCI_BRIDGE_CTL_PARITY + PCI_BRIDGE_CTL_SERR); /* error check */" line also?? Anyways nice work Marc:-)
Acked-by: Joseph Smith joe@smittys.pointclark.net
Thanks - Joe