Rework SLB trap handling so that double-faults into an SLB trap handler are

possible, and double faults within an SLB trap handler are not. The result
is that it possible to take an SLB fault at any time, on any address, for
any reason, at any point in the kernel.

This lets us do two important things. First, it removes the (soft) 16 GB RAM
ceiling on PPC64 as well as any architectural limitations on KVA space.
Second, it lets the kernel tolerate poorly designed hypervisors that
have a tendency to fail to restore the SLB properly after a hypervisor
context switch.

MFC after:	6 weeks

1 files changed, 4 insertions, 14 deletions

diff --git a/sys/powerpc/aim/slb.c b/sys/powerpc/aim/slb.c
index 1a5ce65..7f4b2ef 100644
--- a/sys/powerpc/aim/slb.c
+++ b/sys/powerpc/aim/slb.c
@@ -409,15 +409,11 @@ slb_alloc_tree(void)
 
 /* Lock entries mapping kernel text and stacks */
 
-#define SLB_SPILLABLE(slbe) \
-	(((slbe & SLBE_ESID_MASK) < VM_MIN_KERNEL_ADDRESS && \
-	    (slbe & SLBE_ESID_MASK) > 16*SEGMENT_LENGTH) || \
-	    (slbe & SLBE_ESID_MASK) > VM_MAX_KERNEL_ADDRESS)
 void
 slb_insert_kernel(uint64_t slbe, uint64_t slbv)
 {
 	struct slb *slbcache;
-	int i, j;
+	int i;
 
 	/* We don't want to be preempted while modifying the kernel map */
 	critical_enter();
@@ -437,15 +433,9 @@ slb_insert_kernel(uint64_t slbe, uint64_t slbv)
 			slbcache[USER_SLB_SLOT].slbe = 1;
 	}
 
-	for (i = mftb() % n_slbs, j = 0; j < n_slbs; j++, i = (i+1) % n_slbs) {
-		if (i == USER_SLB_SLOT)
-			continue;
-
-		if (SLB_SPILLABLE(slbcache[i].slbe))
-			break;
-	}
-
-	KASSERT(j < n_slbs, ("All kernel SLB slots locked!"));
+	i = mftb() % n_slbs;
+	if (i == USER_SLB_SLOT)
+			i = (i+1) % n_slbs;
 
 fillkernslb:
 	KASSERT(i != USER_SLB_SLOT,