On 05/13/2010 11:56 AM, Bjorn Helgaas wrote:
>
> I'm a little bit nervous about Linux's current strategy of assigning
> resources to things before we even know whether we're going to use
> them. We don't support dynamic PCI resource reassignment, so maybe
> we don't have any choice in this case, but generally I prefer the
> lazy approach.
>
Lazy has its own pitfalls. In particular, the issue here is about
allocation of bridge devices, which may cause all kinds of issues with
bridges further up. Consider:
A
/ \
B C
| |
D E
... where A, B, and C are bridges, and D and E are devices. Now the
driver initializes D, and requests address space. As a result, the OS
configures bridge B to have a 4K window, and accordingly the same for
bridge A.
Now you want to initialize device E. This means the window for bridge A
has to be widened, because bridge C is going to need its own 4K window.
If the linearly consecutive address space is not available, it now
means reconfiguring bridge B and device D.
4K granularity really hurts. It made sense when bridges were relatively
rare, but in PCI-express world that is no longer the case...
-hpa
--
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
