-added the NV40 arch screen connections limitations to the doc;

-added a fixme comment in the kerneldriver: it seems we must try to map cardBIOSes to memory ourselves if possible after all (should be PCI config manager work I'd say): encountered two systems now that don't have that done (by BIOS I guess). A old PCI-only system (Pentium200), and a brand new PCIe system: Both PCI and PCIe cardBIOSes don't get mapped here which prevents us from using coldstarts on secondary cards :-/ git-svn-id: file:///srv/svn/repos/haiku/trunk/current@11400 a95241bf-73f2-0310-859d-f6bbb57e9c96
2005-02-17 11:53:25 +00:00
parent 36ad4411a8
commit 4157b3aeec
2 changed files with 8 additions and 2 deletions
@@ -606,6 +606,10 @@ static status_t map_device(device_info *di)

 	/* get ROM memory mapped base adress - this is defined in the PCI standard */
 	tmpUlong = get_pci(PCI_rom_base, 4);
+	//fixme?: if (!tmpUlong) try to map the ROM ourselves. Confirmed a PCIe system not
+	//having the ROM mapped on PCI and PCIe cards. Falling back to fetching from ISA
+	//legacy space will get us into trouble if we aren't the primary graphics card!!
+	//(as legacy space always has the primary card's ROM 'mapped'!)
 	if (tmpUlong)
 	{
 		/* ROM was assigned an adress, so enable ROM decoding - see PCI standard */