Hi Jason, On 30 April 2014 16:44, Jason wrote: > On Wed, Apr 30, 2014 at 10:33:08AM +0000, Phil Edworthy wrote: > > > > You set the value of the integrated bus based on the bus number in the > > > TYPE0 accesses the driver performs and it shows up in the PCI config > > > space of the root port bridge: > > > > > > root@koelsch:~# lspci -vv > > > 00:00.0 PCI bridge: Renesas Technology Corp. Device 001f (prog-if 00 > > > [Normal decode]) > > > Bus: primary=00, secondary=01, subordinate=01, sec-latency=0 > > > > > > The value here will be also be exported on-the-wire in certain PCI-E > > > TLPs. > > > > > > So it is important, and must be set properly. > > I am not so sure about this... > > > > There are three things we need to consider: > > 1) talking to the host bridge itself > > 2) talking to immediate device (connected directly to host bridge) > > 3) talking to other devices > > > > The PCI standard tells us we use type 1's for 3) and type 0's for > > 2). The driver uses root_bus_nr to make this decision. For 1), > > there is a note in the driver that the hardware cannot target > > itself. The driver has code in rcar_pcie_config_access() to check if > > it's an access to the host bridge, and if so, use I/O (readl/writel) > > to perform the config access. > > The logic should be: > if (bus == primary) > do io access to host bridge > else if (bus == secondary) > issue type 0 TLP on the wire > else if (bus > secondary && bus <= subordinate) > issue type 1 TLP on the wire > else > fail, invalid bus number > Where the three values come from the register in the PCI host bridge's > configuration space, and are kept in sync with the programming from > the Linux PCI core. > > It is just a happy hapenstance that root_bus_nr equals the value the > PCI core programmed into secondary - that is not guarenteed, you must > use the primary value directly. For type0 TLPs, we are not checking that root_bus_nr equals the value the PCI core programmed into secondary, we are checking that the (root_bus_nr == bus->parent->number). The only way this wouldn't work is if root_bus_nr was not the root bus number. Since the Synopsys DW driver also saves off sys->busnr and later uses this to determine if accesses are for the host bridge, I guess that means it won't always work either, right? Or is it ok because the DW driver saves sys->busnr in its .scan function? When would the PCI core change the root bus number to something other than set in sys->busnr? > I recommend capturing it on write to the host bridge config > space. Thanks Phil -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
