On Wed, Jul 05, 2023 at 06:22:59PM +0900, David Stevens wrote:

On Wed, Jul 5, 2023 at 12:10 PM Yu Zhang [off-list ref] wrote:

quoted

quoted

@@ -2514,35 +2512,26 @@ static bool hva_to_pfn_fast(unsigned long addr, bool write_fault,
  * The slow path to get the pfn of the specified host virtual address,
  * 1 indicates success, -errno is returned if error is detected.
  */
-static int hva_to_pfn_slow(unsigned long addr, bool *async, bool write_fault,
-                        bool interruptible, bool *writable, kvm_pfn_t *pfn)
+static int hva_to_pfn_slow(struct kvm_follow_pfn *foll, kvm_pfn_t *pfn)
 {
-     unsigned int flags = FOLL_HWPOISON;
+     unsigned int flags = FOLL_HWPOISON | FOLL_GET | foll->flags;
      struct page *page;
      int npages;

      might_sleep();

-     if (writable)
-             *writable = write_fault;
-
-     if (write_fault)
-             flags |= FOLL_WRITE;
-     if (async)
-             flags |= FOLL_NOWAIT;
-     if (interruptible)
-             flags |= FOLL_INTERRUPTIBLE;
-
-     npages = get_user_pages_unlocked(addr, 1, &page, flags);
+     npages = get_user_pages_unlocked(foll->hva, 1, &page, flags);
      if (npages != 1)
              return npages;

+     foll->writable = (foll->flags & FOLL_WRITE) && foll->allow_write_mapping;
+
      /* map read fault as writable if possible */
-     if (unlikely(!write_fault) && writable) {
+     if (unlikely(!foll->writable) && foll->allow_write_mapping) {

I guess !foll->writable should be !(foll->flags & FOLL_WRITE) here.

The two statements are logically equivalent, although I guess using
!(foll->flags & FOLL_WRITE) may be a little clearer, if a little more
verbose.

Well, as the comment says, we wanna try to map the read fault as writable
whenever possible. And __gfn_to_pfn_memslot() will only set the FOLL_WRITE
for write faults. So I guess using !foll->writable will not allow this.
Did I miss anything?

quoted

quoted

+kvm_pfn_t __gfn_to_pfn_memslot(const struct kvm_memory_slot *slot, gfn_t gfn,
+                            bool atomic, bool interruptible, bool *async,
+                            bool write_fault, bool *writable, hva_t *hva)
+{
+     kvm_pfn_t pfn;
+     struct kvm_follow_pfn foll = {
+             .slot = slot,
+             .gfn = gfn,
+             .flags = 0,
+             .atomic = atomic,
+             .allow_write_mapping = !!writable,
+     };
+
+     if (write_fault)
+             foll.flags |= FOLL_WRITE;
+     if (async)
+             foll.flags |= FOLL_NOWAIT;
+     if (interruptible)
+             foll.flags |= FOLL_INTERRUPTIBLE;
+
+     pfn = __kvm_follow_pfn(&foll);
+     if (pfn == KVM_PFN_ERR_NEEDS_IO) {

Could we just use KVM_PFN_ERR_FAULT and foll.flags here? I.e.,
        if (pfn == KVM_PFN_ERR_FAULT && (foll.flags & FOLL_NOWAIT))?
Setting pfn to KVM_PFN_ERR_NEEDS_IO just to indicate an async fault
seems unnecessary.

There are the cases where the fault does not fall within a vma or when
the target vma's flags don't support the fault's access permissions.
In those cases, continuing to try to resolve the fault won't cause
problems per-se, but it's wasteful and a bit confusing. Having
hva_to_pfn detect whether or not it may be possible to resolve the
fault asynchronously and return KVM_PFN_ERR_NEEDS_IO if so seems like
a good idea. It also matches what the existing code does.

Got it. Sounds reasonable. And thanks! :)

B.R.
Yu