David Brownell wrote: > On Tuesday 15 May 2007, Tzachi Perelstein wrote: >> David Brownell wrote: >>> On Sunday 13 May 2007, Tzachi Perelstein wrote: >>>> Hello, >>>> >>>> I'm working on EHCI HC glue port for Marvell Orion SoC. The Orion has >>>> a PCI/PCIE interfaces as well as an integrated USB controller. >>>> >>>> While looking at ehci-hcd.c it looks like there's some kind of dependency >>>> between PCI_DRIVER and PLATFORM_DRIVER. I saw some previous threads >>>> about the ehci-fsl.c regarding this issue, but I'm not sure I understand, >>>> for example, why in module_init if pci_register_driver fails the platform >>>> driver is unregistered? Shouldn't these two types of drivers be >>>> independent? >>> If the platform has PCI but you can't register a PCI driver, it's >>> quite broken. How broken must a platform be before you fix those >>> bugs, rather than expect drivers to work around them? >> I have no problem with my PCI, it works just fine! Who said anything about >> bugs workaround??? > > You -- by asking why a bug (pci_register_driver fails) should > be handled differently (working around the failure).
Asking about some existing implementation does not necessarily mean one need a workaround. I just wanted to better understand this dependency and perhaps improving it. That's all. >> I was just wondering if there is a _real_ reason why *if* pci_register_driver >> fails then the platform driver will be unregistered. Theoretically it could >> be that >> PCI will fail but platform driver will work ok, so why unregistering it?? >> Why not >> just dumping an error message and continuing?? > > No; "theoretically" pci_register_driver() has no reason to fail > on a correctly working system. I agree with that. However, the existing implementation _does_ handle this theoretical pci_register_driver failure by unregistering the *platform* driver (?). Besides arguing with me if I have a buggy PCI or not, you couldn't explain _why_ this is the appropriate handling. It's not that important, so you can just forget it. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
