[PATCH V2] powerpc/kernel: Add 'ibm,thread-groups' property for CPU allocation

[Michael Bringmann <mwb@linux.vnet.ibm.com>]

Add code to parse the new property 'ibm,thread-groups" when it is
present.  The content of this property explicitly defines the number
of threads per core as well as the PowerPC 'threads_core_mask'.
The design provides a common device-tree for both P9 normal core and
P9 fused core systems.  The new property has been observed to be
available on P9 pHyp systems, but it is not always present on
OpenPower BMC systems.

The property updates the kernel to know which CPUs/threads of each
core are actually present, and then use the map when adding cores
to the system at boot, or during hotplug operations.

* Previously, the information about the number of threads per core
  was inferred solely from the "ibm,ppc-interrupt-server#s" property
  in the system device tree.
* Also previous to this property, The mask of threads per CPU was
  inferred to be a strict linear series from 0..(nthreads-1).
* After reading the "ibm,thread-group" property, we can determine
  the number of threads per core to be the 'bitmask weight' of the
  CPU thread mask.
* Also after reading the property, we can determine which of the
  possible threads we are allowed to online for each CPU.  It is no
  longer a simple linear sequence, but may be discontinuous e.g.
  activate threads 1,2,3,5,6,7 on a core instead of 0-5 sequentially.

Implementation of the "ibm,thread-groups" property is spread across
a few files in the powerpc specific code:

* prom.c: Parse the property and create 'ppc_thread_group_mask'.
          Use the mask in operation of early_init_dt_scan_cpus().
* setup-common.c: Import 'ppc_thread_group_mask' and use the value
          in the operation of cpu_init_thread_core_maps(), and
          smp_setup_cpu_maps.
* hotplug-cpu.c: Use 'ppc_thread_group_mask' in several locations
          where the code previously expected to iterate over a
          linear series of active threads (0..nthreads-1).

Note that the "ibm,thread-groups" property also includes semantics
of 'thread-group' i.e. define one or more subgroups of the available
threads, each group of threads to be used for a specific class of
task.  Translating thread group semantics into Linux kernel features
is TBD.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
Changes in V2:
  -- Add more information and examples to the patch description.
  -- Rename 'pseries_thread_group_mask' to 'ppc_thread_group_mask'
  -- Remove unnecessary debug message complaining about absence of
     property.
  -- Reduce indent complexity of early_init_dt_scan_cpus().
---
 arch/powerpc/include/asm/cputhreads.h        |    2 +
 arch/powerpc/kernel/prom.c                   |   74 ++++++++++++++++++++++++++
 arch/powerpc/kernel/setup-common.c           |   30 +++++++----
 arch/powerpc/platforms/pseries/hotplug-cpu.c |   13 ++++-
 4 files changed, 107 insertions(+), 12 deletions(-)

diff --git a/arch/powerpc/include/asm/cputhreads.h b/arch/powerpc/include/asm/cputhreads.h
index d71a909..8e444d4 100644
--- a/arch/powerpc/include/asm/cputhreads.h
+++ b/arch/powerpc/include/asm/cputhreads.h
@@ -31,6 +31,8 @@
 #define threads_core_mask	(*get_cpu_mask(0))
 #endif
 
+extern cpumask_t ppc_thread_group_mask;
+
 /* cpu_thread_mask_to_cores - Return a cpumask of one per cores
  *                            hit by the argument
  *
diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index b15bae2..0a49231 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c
@@ -68,6 +68,9 @@
 #define DBG(fmt...)
 #endif
 
+cpumask_t ppc_thread_group_mask;
+EXPORT_SYMBOL(ppc_thread_group_mask);
+
 #ifdef CONFIG_PPC64
 int __initdata iommu_is_off;
 int __initdata iommu_force_on;
@@ -303,6 +306,71 @@ static void __init check_cpu_feature_properties(unsigned long node)
 	}
 }
 
+static void __init early_init_setup_thread_group_mask(unsigned long node,
+						cpumask_t *thread_group_mask)
+{
+	const __be32 *thrgrp;
+	int len, rc = 0;
+	u32 cc_type = 0, no_split = 0, thr_per_split = 0;
+	int j, k;
+
+	cpumask_clear(thread_group_mask);
+
+	thrgrp = of_get_flat_dt_prop(node, "ibm,thread-groups", &len);
+	if (!thrgrp)
+		return;
+
+	/* Process the thread groups for the Core thread mask */
+	/* Characteristic type per table */
+	cc_type = of_read_number(thrgrp++, 1);
+
+	/*
+	 * 1 : Group shares common L1, translation cache, and
+	 *     instruction data flow
+	 * >1 : Reserved
+	 */
+	if (cc_type != 1) {
+		rc = -EINVAL;
+		goto endit;
+	}
+
+	/* No. splits */
+	no_split = of_read_number(thrgrp++, 1);
+	if (no_split == 0) {
+		rc = -EINVAL;
+		goto endit;
+	}
+
+	/* Threads per split */
+	thr_per_split = of_read_number(thrgrp++, 1);
+	if (thr_per_split == 0) {
+		rc = -EINVAL;
+		goto endit;
+	}
+
+	DBG("INFO: Node %d; ibm,thread-group "
+		"(cc_t=%d, no_spl=%d, thr_p_spl=%d)\n",
+		(int)node, (int)cc_type, (int)no_split,
+		(int)thr_per_split);
+
+	for (j = 0; j < no_split; j++) {
+		for (k = 0; k < thr_per_split; k++) {
+			u32 t = of_read_number(thrgrp++, 1);
+
+			cpumask_set_cpu(t, thread_group_mask);
+			DBG("INFO: Node %d; enable thread %d\n",
+				(int)node, (int)t);
+		}
+	}
+
+endit:
+	if (rc) {
+		DBG("WARNING: Node %d; error processing "
+		    "ibm,thread-group property\n", (int)node);
+		cpumask_setall(thread_group_mask);
+	}
+}
+
 static int __init early_init_dt_scan_cpus(unsigned long node,
 					  const char *uname, int depth,
 					  void *data)
@@ -326,11 +394,17 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
 
 	nthreads = len / sizeof(int);
 
+	/* Figure out the thread subset */
+	early_init_setup_thread_group_mask(node, &ppc_thread_group_mask);
+
 	/*
 	 * Now see if any of these threads match our boot cpu.
 	 * NOTE: This must match the parsing done in smp_setup_cpu_maps.
 	 */
 	for (i = 0; i < nthreads; i++) {
+		if (!cpumask_test_cpu(i % nthreads, &ppc_thread_group_mask))
+			continue;
+
 		/*
 		 * version 2 of the kexec param format adds the phys cpuid of
 		 * booted proc.
diff --git a/arch/powerpc/kernel/setup-common.c b/arch/powerpc/kernel/setup-common.c
index 2075322..53cadcd 100644
--- a/arch/powerpc/kernel/setup-common.c
+++ b/arch/powerpc/kernel/setup-common.c
@@ -427,13 +427,16 @@ void __init check_for_initrd(void)
 EXPORT_SYMBOL_GPL(threads_shift);
 EXPORT_SYMBOL_GPL(threads_core_mask);
 
-static void __init cpu_init_thread_core_maps(int tpc)
+static void __init cpu_init_thread_core_maps(int tpc,
+				cpumask_t *thread_group_mask)
 {
+	cpumask_t work_mask;
 	int i;
 
 	threads_per_core = tpc;
 	threads_per_subcore = tpc;
 	cpumask_clear(&threads_core_mask);
+	cpumask_clear(&work_mask);
 
 	/* This implementation only supports power of 2 number of threads
 	 * for simplicity and performance
@@ -442,14 +445,14 @@ static void __init cpu_init_thread_core_maps(int tpc)
 	BUG_ON(tpc != (1 << threads_shift));
 
 	for (i = 0; i < tpc; i++)
-		cpumask_set_cpu(i, &threads_core_mask);
+		cpumask_set_cpu(i, &work_mask);
+	cpumask_and(&threads_core_mask, &work_mask, thread_group_mask);
 
 	printk(KERN_INFO "CPU maps initialized for %d thread%s per core\n",
 	       tpc, tpc > 1 ? "s" : "");
 	printk(KERN_DEBUG " (thread shift is %d)\n", threads_shift);
 }
 
-
 /**
  * setup_cpu_maps - initialize the following cpu maps:
  *                  cpu_possible_mask
@@ -503,17 +506,24 @@ void __init smp_setup_cpu_maps(void)
 		for (j = 0; j < nthreads && cpu < nr_cpu_ids; j++) {
 			bool avail;
 
-			DBG("    thread %d -> cpu %d (hard id %d)\n",
-			    j, cpu, be32_to_cpu(intserv[j]));
-
 			avail = of_device_is_available(dn);
 			if (!avail)
 				avail = !of_property_match_string(dn,
 						"enable-method", "spin-table");
 
-			set_cpu_present(cpu, avail);
-			set_hard_smp_processor_id(cpu, be32_to_cpu(intserv[j]));
-			set_cpu_possible(cpu, true);
+			DBG("    thread %d -> cpu %d (hard id %d)\n",
+			    j, cpu, be32_to_cpu(intserv[j]));
+
+			if (cpumask_test_cpu(cpu % nthreads,
+						&ppc_thread_group_mask)) {
+				set_cpu_present(cpu, avail);
+				set_hard_smp_processor_id(cpu,
+						be32_to_cpu(intserv[j]));
+				set_cpu_possible(cpu, true);
+			} else {
+				set_cpu_present(cpu, false);
+				set_cpu_possible(cpu, false);
+			}
 			cpu++;
 		}
 	}
@@ -572,7 +582,7 @@ void __init smp_setup_cpu_maps(void)
 	 * every CPU in the system. If that is not the case, then some code
 	 * here will have to be reworked
 	 */
-	cpu_init_thread_core_maps(nthreads);
+	cpu_init_thread_core_maps(nthreads, &ppc_thread_group_mask);
 
 	/* Now that possible cpus are set, set nr_cpu_ids for later use */
 	setup_nr_cpu_ids();
diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
index a7d14aa7..4125eaa 100644
--- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
+++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
@@ -36,6 +36,7 @@
 #include <asm/xics.h>
 #include <asm/xive.h>
 #include <asm/plpar_wrappers.h>
+#include <asm/cputhreads.h>
 
 #include "pseries.h"
 #include "offline_states.h"
@@ -258,8 +259,10 @@ static int pseries_add_processor(struct device_node *np)
 	zalloc_cpumask_var(&tmp, GFP_KERNEL);
 
 	nthreads = len / sizeof(u32);
-	for (i = 0; i < nthreads; i++)
-		cpumask_set_cpu(i, tmp);
+	for (i = 0; i < nthreads; i++) {
+		if (cpumask_test_cpu(i % nthreads, &ppc_thread_group_mask))
+			cpumask_set_cpu(i, tmp);
+	}
 
 	cpu_maps_update_begin();
 
@@ -324,6 +327,8 @@ static void pseries_remove_processor(struct device_node *np)
 
 	cpu_maps_update_begin();
 	for (i = 0; i < nthreads; i++) {
+		if (!cpumask_test_cpu(i % nthreads, &ppc_thread_group_mask))
+			continue;
 		thread = be32_to_cpu(intserv[i]);
 		for_each_present_cpu(cpu) {
 			if (get_hard_smp_processor_id(cpu) != thread)
@@ -356,6 +361,8 @@ static int dlpar_online_cpu(struct device_node *dn)
 
 	cpu_maps_update_begin();
 	for (i = 0; i < nthreads; i++) {
+		if (!cpumask_test_cpu(i % nthreads, &ppc_thread_group_mask))
+			continue;
 		thread = be32_to_cpu(intserv[i]);
 		for_each_present_cpu(cpu) {
 			if (get_hard_smp_processor_id(cpu) != thread)
@@ -522,6 +529,8 @@ static int dlpar_offline_cpu(struct device_node *dn)
 
 	cpu_maps_update_begin();
 	for (i = 0; i < nthreads; i++) {
+		if (!cpumask_test_cpu(i % nthreads, &ppc_thread_group_mask))
+			continue;
 		thread = be32_to_cpu(intserv[i]);
 		for_each_present_cpu(cpu) {
 			if (get_hard_smp_processor_id(cpu) != thread)