Patch Set #1, Line 17:
these days, so nobody noticed.
It could be that even when implemented for PCIe root ports, CTL_VGA bits for those bridges have no effect when IGD is enabled.
IIRC, they have an effect: the wrath of resource conflicts.
This is why we have a .disable op that should be implemented
for the IGD.
We should also make sure only one PCIe root port will have CTL_VGA set.
We already do that implicitly. We enable it only on the path to
the winning device in set_vga_bridge_bits().
pci_update_config16(dev, PCI_BRIDGE_CONTROL, ~1, 1<<2);
Unconditionally setting 'ISA enable' bit is also.. interesting.
Yeah, interesting. We try to avoid allocating in those ranges (I
learned today) device.c:400 (should work as long as the resource
is <= 0x100). But same as the VGA avoidance below it, it won't
be noticed by static/preallocated resources. oh wait, nvm, is that
an `else if`? dead branch?
To view, visit change 35516. To unsubscribe, or for help writing mail filters, visit settings.