Messages in this thread | | | Date | Wed, 20 Jun 2012 21:21:07 +0800 | From | Xiao Guangrong <> | Subject | Re: [PATCH v7 02/10] KVM: MMU: abstract spte write-protect |
| |
On 06/20/2012 08:56 PM, Takuya Yoshikawa wrote:
> On Wed, 20 Jun 2012 17:11:06 +0800 > Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com> wrote: > >> Strange! Why do you think it is wrong? It is just debug code. > > kvm_mmu_slot_remove_write_access() does not use rmap but the debug code says: > rmap_printk("rmap_write_protect: spte %p %llx\n", sptep, *sptep);
It is not a problem since all sptes which are pointing to gfn is existed in rmap.
> >>> If you think it is not a problem, please explain why you think so in >>> the changelog. >> >> >> It is a from the first place and it is used to debug and not compiled at all. > > It was not in kvm_mmu_slot_remove_write_access() before, no? > > This patch says that the write protection code becomes commonly usable > function, but it still has rmap_write_protect specific debug code in it; > using it in kvm_mmu_slot_remove_write_access(), which is not at all related > to rmap_write_protect, is strange. > > As you say, this is just debug code and does not have any practical problem. > But randomly putting debug code is not a good thing. >
Again, "rmap" does not break the logic, the spte we handle in this function must be in rmap.
| |