Messages in this thread | | | Subject | Re: Possible issue with dangling PCI BARs | From | Benjamin Herrenschmidt <> | Date | Thu, 13 Dec 2007 20:29:23 +1100 |
| |
On Thu, 2007-12-13 at 12:14 +0300, Ivan Kokshaysky wrote: > On Thu, Dec 13, 2007 at 04:26:42PM +1100, Benjamin Herrenschmidt wrote: > > I can try to whip up some code tomorrow I suppose, though I'm always > > afraid some dodgy x86 setup will blow up... > > That scares me too, but something like pci_dangling_bar(dev, idx) with > a default (for now) no-op implementation in asm-generic/pci.h should > be safe...
Well, the code that detects collisions and "unsets" a resource is arch in the first place, and so is pcibios_enable_device().
So I'm going to do a powerpc implementation & test it a bit with what I have around, and then do an x86 since with my latest (pending for .25) patches. powerpc PCI code is -very- similar to x86.
Cheers, Ben.
| |