| Message ID | 20200416155903.267414-1-peterx@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | KVM: Remove async parameter for hva_to_pfn_remapped() | expand |
On 16/04/20 17:59, Peter Xu wrote: > We always do synchronous fault in for those pages. > > Signed-off-by: Peter Xu <peterx@redhat.com> > --- > > Or, does it make sense to allow async pf for PFNMAP|IO too? I just > didn't figure out why not... > --- I think async pf would use FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT. On failure you would set *async = true. In practice I don't think fixup_user_fault is likely to do anything asynchronously. Paolo
On Thu, Apr 16, 2020 at 06:13:57PM +0200, Paolo Bonzini wrote: > On 16/04/20 17:59, Peter Xu wrote: > > We always do synchronous fault in for those pages. > > > > Signed-off-by: Peter Xu <peterx@redhat.com> > > --- > > > > Or, does it make sense to allow async pf for PFNMAP|IO too? I just > > didn't figure out why not... > > --- > > I think async pf would use FAULT_FLAG_ALLOW_RETRY | > FAULT_FLAG_RETRY_NOWAIT. On failure you would set *async = true. > > In practice I don't think fixup_user_fault is likely to do anything > asynchronously. Yeah, I'm thinking whether the fixup_user_fault() could need time to finish sometimes (IIUC that's major for the GPU drivers? I have no idea, e.g., how long time it'll take normally for a major fault), then whether we can also do that in another thread like what we do with normal anonymous pages when it was swapped out. Thanks,
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 74bdb7bf3295..2f1f2f56e93d 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -1807,7 +1807,7 @@ static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault) } static int hva_to_pfn_remapped(struct vm_area_struct *vma, - unsigned long addr, bool *async, + unsigned long addr, bool write_fault, bool *writable, kvm_pfn_t *p_pfn) { @@ -1902,7 +1902,7 @@ static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async, if (vma == NULL) pfn = KVM_PFN_ERR_FAULT; else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) { - r = hva_to_pfn_remapped(vma, addr, async, write_fault, writable, &pfn); + r = hva_to_pfn_remapped(vma, addr, write_fault, writable, &pfn); if (r == -EAGAIN) goto retry; if (r < 0)
We always do synchronous fault in for those pages. Signed-off-by: Peter Xu <peterx@redhat.com> --- Or, does it make sense to allow async pf for PFNMAP|IO too? I just didn't figure out why not... --- virt/kvm/kvm_main.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)