- Remove the old smp cpu topology specification with a new, more flexible

   tree structure that encodes the level of cache sharing and other
   properties.
 - Provide several convenience functions for creating one and two level
   cpu trees as well as a default flat topology.  The system now always
   has some topology.
 - On i386 and amd64 create a seperate level in the hierarchy for HTT
   and multi-core cpus.  This will allow the scheduler to intelligently
   load balance non-uniform cores.  Presently we don't detect what level
   of the cache hierarchy is shared at each level in the topology.
 - Add a mechanism for testing common topologies that have more information
   than the MD code is able to provide via the kern.smp.topology tunable.
   This should be considered a debugging tool only and not a stable api.

Sponsored by:	Nokia

3 files changed, 41 insertions, 43 deletions

diff --git a/sys/amd64/amd64/identcpu.c b/sys/amd64/amd64/identcpu.c
index a51f02e..f0afb9f 100644
--- a/sys/amd64/amd64/identcpu.c
+++ b/sys/amd64/amd64/identcpu.c
@@ -97,6 +97,10 @@ static struct {
 	{ "Sledgehammer",	CPUCLASS_K8 },		/* CPU_SLEDGEHAMMER */
 };
 
+int cpu_cores;
+int cpu_logical;
+
+
 extern int pq_l2size;
 extern int pq_l2nways;
 
@@ -360,11 +364,13 @@ printcpuinfo(void)
 				if ((regs[0] & 0x1f) != 0)
 					cmp = ((regs[0] >> 26) & 0x3f) + 1;
 			}
+			cpu_cores = cmp;
+			cpu_logical = htt / cmp;
 			if (cmp > 1)
 				printf("\n  Cores per package: %d", cmp);
 			if ((htt / cmp) > 1)
 				printf("\n  Logical CPUs per core: %d",
-				    htt / cmp);
+				    cpu_logical);
 		}
 	}
 	/* Avoid ugly blank lines: only print newline when we have to. */
diff --git a/sys/amd64/amd64/mp_machdep.c b/sys/amd64/amd64/mp_machdep.c
index a827137..83eab2e 100644
--- a/sys/amd64/amd64/mp_machdep.c
+++ b/sys/amd64/amd64/mp_machdep.c
@@ -83,12 +83,6 @@ extern	int nkpt;
 
 extern  struct pcpu __pcpu[];
 
-/*
- * CPU topology map datastructures for HTT.
- */
-static struct cpu_group mp_groups[MAXCPU];
-static struct cpu_top mp_top;
-
 /* AP uses this during bootstrap.  Do not staticize.  */
 char *bootSTK;
 static int bootAP;
@@ -182,40 +176,38 @@ mem_range_AP_init(void)
 		mem_range_softc.mr_op->initAP(&mem_range_softc);
 }
 
-void
-mp_topology(void)
+struct cpu_group *
+cpu_topo(void)
 {
-	struct cpu_group *group;
-	int apic_id;
-	int groups;
-	int cpu;
-
-	/* Build the smp_topology map. */
-	/* Nothing to do if there is no HTT support. */
-	if (hyperthreading_cpus <= 1)
-		return;
-	group = &mp_groups[0];
-	groups = 1;
-	for (cpu = 0, apic_id = 0; apic_id <= MAX_APIC_ID; apic_id++) {
-		if (!cpu_info[apic_id].cpu_present)
-			continue;
-		/*
-		 * If the current group has members and we're not a logical
-		 * cpu, create a new group.
-		 */
-		if (group->cg_count != 0 &&
-		    (apic_id % hyperthreading_cpus) == 0) {
-			group++;
-			groups++;
-		}
-		group->cg_count++;
-		group->cg_mask |= 1 << cpu;
-		cpu++;
+	if (cpu_cores == 0)
+		cpu_cores = 1;
+	if (cpu_logical == 0)
+		cpu_logical = 1;
+	if (mp_ncpus % (cpu_cores * cpu_logical) != 0) {
+		printf("WARNING: Non-uniform processors.\n");
+		printf("WARNING: Using suboptimal topology.\n");
+		return (smp_topo_none());
 	}
-
-	mp_top.ct_count = groups;
-	mp_top.ct_group = mp_groups;
-	smp_topology = &mp_top;
+	/*
+	 * No multi-core or hyper-threaded.
+	 */
+	if (cpu_logical * cpu_cores == 1)
+		return (smp_topo_none());
+	/*
+	 * Only HTT no multi-core.
+	 */
+	if (cpu_logical > 1 && cpu_cores == 1)
+		return (smp_topo_1level(CG_SHARE_L1, cpu_logical, CG_FLAG_HTT));
+	/*
+	 * Only multi-core no HTT.
+	 */
+	if (cpu_cores > 1 && cpu_logical == 1)
+		return (smp_topo_1level(CG_SHARE_NONE, cpu_cores, 0));
+	/*
+	 * Both HTT and multi-core.
+	 */
+	return (smp_topo_2level(CG_SHARE_NONE, cpu_cores,
+	    CG_SHARE_L1, cpu_logical, CG_FLAG_HTT));
 }
 
 /*
@@ -409,9 +401,6 @@ cpu_mp_start(void)
 	}
 
 	set_interrupt_apic_ids();
-
-	/* Last, setup the cpu topology now that we have probed CPUs */
-	mp_topology();
 }
 
 
diff --git a/sys/amd64/include/smp.h b/sys/amd64/include/smp.h
index d85c55e..f961d6b 100644
--- a/sys/amd64/include/smp.h
+++ b/sys/amd64/include/smp.h
@@ -36,6 +36,10 @@ extern int			boot_cpu_id;
 extern struct pcb		stoppcbs[];
 extern int			cpu_apic_ids[];
 
+/* global data in identcpu.c */
+extern int			cpu_cores;
+extern int			cpu_logical;
+
 /* IPI handlers */
 inthand_t
 	IDTVEC(invltlb),	/* TLB shootdowns - global */
@@ -57,7 +61,6 @@ void	ipi_self(u_int ipi);
 void 	ipi_bitmap_handler(struct trapframe frame);
 u_int	mp_bootaddress(u_int);
 int	mp_grab_cpu_hlt(void);
-void	mp_topology(void);
 void	smp_cache_flush(void);
 void	smp_invlpg(vm_offset_t addr);
 void	smp_masked_invlpg(u_int mask, vm_offset_t addr);