vmm svm: Use malloc_aligned() instead of contigmalloc().

[jhb]

Summary

The MSR and I/O bitmaps just need to be page-aligned. They do
not have other constraints, so malloc_aligned() is sufficient.

[markj]

sys/amd64/vmm/amd/svm.c

574

I don't quite follow. svm_init() stores the PAs of these bitmaps in the VMCB, and they're both larger than PAGE_SIZE, so don't they need to be physically contiguous?

[jhb]

sys/amd64/vmm/amd/svm.c

574

Oof, yes, that's a fail on my part.

sys/amd64/vmm/amd/svm.c

 void
 mrt_stats()
 {
         struct mrtstat mrtstat;
         u_long mstaddr;
 
         mstaddr = nl[N_MRTSTAT].n_value;
 
-        if (mstaddr == 0) {
+        if (fetch_stats("net.inet.ip.mrtstat", mstaddr, &mrtstat,
+            sizeof(mrtstat), kread_counters) != 0) {
+                if ((live && errno == ENOENT) || (!live && mstaddr == 0))
                         fprintf(stderr, "No IPv4 MROUTING kernel support.\n");
                 return;
         }
-
-        if (fetch_stats("net.inet.ip.mrtstat", mstaddr, &mrtstat,
-            sizeof(mrtstat), kread_counters) != 0)
-                return;
 
         xo_emit("{T:IPv4 multicast forwarding}:\n");
 
 #define p(f, m) if (mrtstat.f || sflag <= 1) \
         xo_emit(m, (uintmax_t)mrtstat.f, plural(mrtstat.f))
 #define p2(f, m) if (mrtstat.f || sflag <= 1) \
         xo_emit(m, (uintmax_t)mrtstat.f, plurales(mrtstat.f))