powerpc/pseries: PAPR persistent memory support

This patch implements support for discovering storage class memory
devices at boot and for handling hotplug of new regions via RTAS
hotplug events.

Signed-off-by: Oliver O'Halloran <[e-mail]>
[mpe: Fix CONFIG_MEMORY_HOTPLUG=n build]
Signed-off-by: Michael Ellerman <[e-mail]>

arch/powerpc/include/asm/firmware.h

···
 #define FW_FEATURE_DRMEM_V2	ASM_CONST(0x0000000400000000)
 #define FW_FEATURE_DRC_INFO	ASM_CONST(0x0000000800000000)
 #define FW_FEATURE_BLOCK_REMOVE ASM_CONST(0x0000001000000000)
+#define FW_FEATURE_PAPR_SCM 	ASM_CONST(0x0000002000000000)
 
 #ifndef __ASSEMBLY__
 
···
 		FW_FEATURE_SET_MODE | FW_FEATURE_BEST_ENERGY |
 		FW_FEATURE_TYPE1_AFFINITY | FW_FEATURE_PRRN |
 		FW_FEATURE_HPT_RESIZE | FW_FEATURE_DRMEM_V2 |
-		FW_FEATURE_DRC_INFO | FW_FEATURE_BLOCK_REMOVE,
+		FW_FEATURE_DRC_INFO | FW_FEATURE_BLOCK_REMOVE |
+		FW_FEATURE_PAPR_SCM,
 	FW_FEATURE_PSERIES_ALWAYS = 0,
 	FW_FEATURE_POWERNV_POSSIBLE = FW_FEATURE_OPAL,
 	FW_FEATURE_POWERNV_ALWAYS = 0,

arch/powerpc/include/asm/hvcall.h

···
 #define H_INT_ESB               0x3C8
 #define H_INT_SYNC              0x3CC
 #define H_INT_RESET             0x3D0
-#define MAX_HCALL_OPCODE	H_INT_RESET
+#define H_SCM_READ_METADATA     0x3E4
+#define H_SCM_WRITE_METADATA    0x3E8
+#define H_SCM_BIND_MEM          0x3EC
+#define H_SCM_UNBIND_MEM        0x3F0
+#define H_SCM_QUERY_BLOCK_MEM_BINDING 0x3F4
+#define H_SCM_QUERY_LOGICAL_MEM_BINDING 0x3F8
+#define H_SCM_MEM_QUERY	        0x3FC
+#define H_SCM_BLOCK_CLEAR       0x400
+#define MAX_HCALL_OPCODE	H_SCM_BLOCK_CLEAR
 
 /* H_VIOCTL functions */
 #define H_GET_VIOA_DUMP_SIZE	0x01

arch/powerpc/include/asm/rtas.h

···
 #define RTAS_TYPE_INFO			0xE2
 #define RTAS_TYPE_DEALLOC		0xE3
 #define RTAS_TYPE_DUMP			0xE4
+#define RTAS_TYPE_HOTPLUG		0xE5
 /* I don't add PowerMGM events right now, this is a different topic */ 
 #define RTAS_TYPE_PMGM_POWER_SW_ON	0x60
 #define RTAS_TYPE_PMGM_POWER_SW_OFF	0x61
···
 #define PSERIES_HP_ELOG_RESOURCE_MEM	2
 #define PSERIES_HP_ELOG_RESOURCE_SLOT	3
 #define PSERIES_HP_ELOG_RESOURCE_PHB	4
+#define PSERIES_HP_ELOG_RESOURCE_PMEM   6
 
 #define PSERIES_HP_ELOG_ACTION_ADD	1
 #define PSERIES_HP_ELOG_ACTION_REMOVE	2

arch/powerpc/kernel/rtasd.c

···
 			return "Dump Notification Event";
 		case RTAS_TYPE_PRRN:
 			return "Platform Resource Reassignment Event";
+		case RTAS_TYPE_HOTPLUG:
+			return "Hotplug Event";
 	}
 
 	return rtas_type[0];

arch/powerpc/platforms/pseries/Makefile

···
 obj-$(CONFIG_PSERIES_ENERGY)	+= pseries_energy.o
 
 obj-$(CONFIG_HOTPLUG_CPU)	+= hotplug-cpu.o
-obj-$(CONFIG_MEMORY_HOTPLUG)	+= hotplug-memory.o
+obj-$(CONFIG_MEMORY_HOTPLUG)	+= hotplug-memory.o pmem.o
 
 obj-$(CONFIG_HVC_CONSOLE)	+= hvconsole.o
 obj-$(CONFIG_HVCS)		+= hvcserver.o

arch/powerpc/platforms/pseries/dlpar.c

···
 	case PSERIES_HP_ELOG_RESOURCE_CPU:
 		rc = dlpar_cpu(hp_elog);
 		break;
+	case PSERIES_HP_ELOG_RESOURCE_PMEM:
+		rc = dlpar_hp_pmem(hp_elog);
+		break;
+
 	default:
 		pr_warn_ratelimited("Invalid resource (%d) specified\n",
 				    hp_elog->resource);

arch/powerpc/platforms/pseries/firmware.c

···
 	{FW_FEATURE_BEST_ENERGY,	"hcall-best-energy-1*"},
 	{FW_FEATURE_HPT_RESIZE,		"hcall-hpt-resize"},
 	{FW_FEATURE_BLOCK_REMOVE,	"hcall-block-remove"},
+	{FW_FEATURE_PAPR_SCM,		"hcall-scm"},
 };
 
 /* Build up the firmware features bitmask using the contents of

arch/powerpc/platforms/pseries/pmem.c

···
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Handles hot and cold plug of persistent memory regions on pseries.
+ */
+
+#define pr_fmt(fmt)     "pseries-pmem: " fmt
+
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/sched.h>	/* for idle_task_exit */
+#include <linux/sched/hotplug.h>
+#include <linux/cpu.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/slab.h>
+#include <asm/prom.h>
+#include <asm/rtas.h>
+#include <asm/firmware.h>
+#include <asm/machdep.h>
+#include <asm/vdso_datapage.h>
+#include <asm/plpar_wrappers.h>
+#include <asm/topology.h>
+
+#include "pseries.h"
+#include "offline_states.h"
+
+static struct device_node *pmem_node;
+
+static ssize_t pmem_drc_add_node(u32 drc_index)
+{
+	struct device_node *dn;
+	int rc;
+
+	pr_debug("Attempting to add pmem node, drc index: %x\n", drc_index);
+
+	rc = dlpar_acquire_drc(drc_index);
+	if (rc) {
+		pr_err("Failed to acquire DRC, rc: %d, drc index: %x\n",
+			rc, drc_index);
+		return -EINVAL;
+	}
+
+	dn = dlpar_configure_connector(cpu_to_be32(drc_index), pmem_node);
+	if (!dn) {
+		pr_err("configure-connector failed for drc %x\n", drc_index);
+		dlpar_release_drc(drc_index);
+		return -EINVAL;
+	}
+
+	/* NB: The of reconfig notifier creates platform device from the node */
+	rc = dlpar_attach_node(dn, pmem_node);
+	if (rc) {
+		pr_err("Failed to attach node %s, rc: %d, drc index: %x\n",
+			dn->name, rc, drc_index);
+
+		if (dlpar_release_drc(drc_index))
+			dlpar_free_cc_nodes(dn);
+
+		return rc;
+	}
+
+	pr_info("Successfully added %pOF, drc index: %x\n", dn, drc_index);
+
+	return 0;
+}
+
+static ssize_t pmem_drc_remove_node(u32 drc_index)
+{
+	struct device_node *dn;
+	uint32_t index;
+	int rc;
+
+	for_each_child_of_node(pmem_node, dn) {
+		if (of_property_read_u32(dn, "ibm,my-drc-index", &index))
+			continue;
+		if (index == drc_index)
+			break;
+	}
+
+	if (!dn) {
+		pr_err("Attempting to remove unused DRC index %x\n", drc_index);
+		return -ENODEV;
+	}
+
+	pr_debug("Attempting to remove %pOF, drc index: %x\n", dn, drc_index);
+
+	/* * NB: tears down the ibm,pmemory device as a side-effect */
+	rc = dlpar_detach_node(dn);
+	if (rc)
+		return rc;
+
+	rc = dlpar_release_drc(drc_index);
+	if (rc) {
+		pr_err("Failed to release drc (%x) for CPU %s, rc: %d\n",
+			drc_index, dn->name, rc);
+		dlpar_attach_node(dn, pmem_node);
+		return rc;
+	}
+
+	pr_info("Successfully removed PMEM with drc index: %x\n", drc_index);
+
+	return 0;
+}
+
+int dlpar_hp_pmem(struct pseries_hp_errorlog *hp_elog)
+{
+	u32 count, drc_index;
+	int rc;
+
+	/* slim chance, but we might get a hotplug event while booting */
+	if (!pmem_node)
+		pmem_node = of_find_node_by_type(NULL, "ibm,persistent-memory");
+	if (!pmem_node) {
+		pr_err("Hotplug event for a pmem device, but none exists\n");
+		return -ENODEV;
+	}
+
+	if (hp_elog->id_type != PSERIES_HP_ELOG_ID_DRC_INDEX) {
+		pr_err("Unsupported hotplug event type %d\n",
+				hp_elog->id_type);
+		return -EINVAL;
+	}
+
+	count = hp_elog->_drc_u.drc_count;
+	drc_index = hp_elog->_drc_u.drc_index;
+
+	lock_device_hotplug();
+
+	if (hp_elog->action == PSERIES_HP_ELOG_ACTION_ADD) {
+		rc = pmem_drc_add_node(drc_index);
+	} else if (hp_elog->action == PSERIES_HP_ELOG_ACTION_REMOVE) {
+		rc = pmem_drc_remove_node(drc_index);
+	} else {
+		pr_err("Unsupported hotplug action (%d)\n", hp_elog->action);
+		rc = -EINVAL;
+	}
+
+	unlock_device_hotplug();
+	return rc;
+}
+
+const struct of_device_id drc_pmem_match[] = {
+	{ .type = "ibm,persistent-memory", },
+	{}
+};
+
+static int pseries_pmem_init(void)
+{
+	pmem_node = of_find_node_by_type(NULL, "ibm,persistent-memory");
+	if (!pmem_node)
+		return 0;
+
+	/*
+	 * The generic OF bus probe/populate handles creating platform devices
+	 * from the child (ibm,pmemory) nodes. The generic code registers an of
+	 * reconfig notifier to handle the hot-add/remove cases too.
+	 */
+	of_platform_bus_probe(pmem_node, drc_pmem_match, NULL);
+
+	return 0;
+}
+machine_arch_initcall(pseries, pseries_pmem_init);

arch/powerpc/platforms/pseries/pseries.h

···
 
 #ifdef CONFIG_MEMORY_HOTPLUG
 int dlpar_memory(struct pseries_hp_errorlog *hp_elog);
+int dlpar_hp_pmem(struct pseries_hp_errorlog *hp_elog);
 #else
 static inline int dlpar_memory(struct pseries_hp_errorlog *hp_elog)
+{
+	return -EOPNOTSUPP;
+}
+static inline int dlpar_hp_pmem(struct pseries_hp_errorlog *hp_elog)
 {
 	return -EOPNOTSUPP;
 }

arch/powerpc/platforms/pseries/ras.c

···
 	 * hotplug events on the ras_log_buf to be handled by rtas_errd.
 	 */
 	if (hp_elog->resource == PSERIES_HP_ELOG_RESOURCE_MEM ||
-	    hp_elog->resource == PSERIES_HP_ELOG_RESOURCE_CPU)
+	    hp_elog->resource == PSERIES_HP_ELOG_RESOURCE_CPU ||
+	    hp_elog->resource == PSERIES_HP_ELOG_RESOURCE_PMEM)
 		queue_hotplug_event(hp_elog);
 	else
 		log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, 0);