On 26/11/12 16:25, Christoffer Dall wrote: > On Mon, Nov 26, 2012 at 9:43 AM, Marc Zyngier <maz@xxxxxxxxxxxxxxx> wrote: >> On Sat, 24 Nov 2012 13:37:18 -0500, Christoffer Dall >> <c.dall@xxxxxxxxxxxxxxxxxxxxxx> wrote: >>> kvm_free_stage2_pgd and its child functions perform the exact same job >>> as unmap_stage2_range, just assuming the entire range needs to be >>> unmapped. Therefore, turn kvm_free_stage2_pgd into a call to >>> unmap_stage2_range that covers the entire physical memory space. >> >> Looks like we've lost all the page_count() checks in the process. >> They were very useful in tracking some ugly bugs, and given how easy it >> is to break the whole VM thing, I'd be inclined to try to keep them >> around. >> >> Maybe a "check" flag on the unmap path? >> > > I figured we would have caught any issues by now :) Please think of us who have to deal with these silly new architectures with additional bits on the side... Plenty of new fancy bugs to catch ;-) Though not that many have been caught in this part of the code yet... > But, I can add some VM_BUG_ON statements back. Thanks. M. -- Jazz is not dead. It just smells funny... _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
