On Fri, Mar 16, 2012 at 12:36 PM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote: > I've objected to this before on the grounds that it's not safe to > update res->end because res is already linked into the resource tree, > but I just noticed another problem. > > Tree updates are protected by resource_lock, but we don't hold that > here. I'm sure you "know" that this update can't race with any other > conflicting resource operations, but its safety certainly is not > obvious from the patch. > > I think a better implementation would be to add something like > "resource_extend()" that could live in kernel/resource.c so it could > do the appropriate locking and checking. Such an interface could also > be independent of the resource type, which may be useful when doing > mem/io resource reassignment. Thanks for point it out. Let me thinking about it. > > This patch is already in Jesse's linux-next branch (commit > c901d4c0407), so I'm leaving it up to you and Jesse to figure out > what, if anything, to do about this. No, this patch is one in busn_alloc patchset. Jesse did NOT pick that patchset yet. Yinghai -- 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
