coherency.c 6.8 KB
Newer Older
1
/*
2 3
 * Coherency fabric (Aurora) support for Armada 370, 375, 38x and XP
 * platforms.
4 5 6 7 8 9 10 11 12 13 14
 *
 * Copyright (C) 2012 Marvell
 *
 * Yehuda Yitschak <yehuday@marvell.com>
 * Gregory Clement <gregory.clement@free-electrons.com>
 * Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
 *
 * This file is licensed under the terms of the GNU General Public
 * License version 2.  This program is licensed "as is" without any
 * warranty of any kind, whether express or implied.
 *
15
 * The Armada 370, 375, 38x and XP SOCs have a coherency fabric which is
16 17 18 19 20
 * responsible for ensuring hardware coherency between all CPUs and between
 * CPUs and I/O masters. This file initializes the coherency fabric and
 * supplies basic routines for configuring and controlling hardware coherency
 */

21 22
#define pr_fmt(fmt) "mvebu-coherency: " fmt

23 24 25 26 27
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/of_address.h>
#include <linux/io.h>
#include <linux/smp.h>
28 29
#include <linux/dma-mapping.h>
#include <linux/platform_device.h>
30 31
#include <linux/slab.h>
#include <linux/mbus.h>
32
#include <linux/pci.h>
33
#include <asm/smp_plat.h>
34
#include <asm/cacheflush.h>
35
#include <asm/mach/map.h>
36
#include <asm/dma-mapping.h>
37
#include "coherency.h"
38
#include "mvebu-soc-id.h"
39

40
unsigned long coherency_phys_base;
41
void __iomem *coherency_base;
42
static void __iomem *coherency_cpu_base;
43 44

/* Coherency fabric registers */
45 46
#define IO_SYNC_BARRIER_CTL_OFFSET		   0x0

47
enum {
48
	COHERENCY_FABRIC_TYPE_NONE,
49
	COHERENCY_FABRIC_TYPE_ARMADA_370_XP,
50
	COHERENCY_FABRIC_TYPE_ARMADA_375,
51
	COHERENCY_FABRIC_TYPE_ARMADA_380,
52 53
};

54
static struct of_device_id of_coherency_table[] = {
55 56
	{.compatible = "marvell,coherency-fabric",
	 .data = (void *) COHERENCY_FABRIC_TYPE_ARMADA_370_XP },
57 58
	{.compatible = "marvell,armada-375-coherency-fabric",
	 .data = (void *) COHERENCY_FABRIC_TYPE_ARMADA_375 },
59 60
	{.compatible = "marvell,armada-380-coherency-fabric",
	 .data = (void *) COHERENCY_FABRIC_TYPE_ARMADA_380 },
61 62 63
	{ /* end of list */ },
};

64 65 66
/* Functions defined in coherency_ll.S */
int ll_enable_coherency(void);
void ll_add_cpu_to_smp_group(void);
67

68
int set_cpu_coherent(void)
69 70
{
	if (!coherency_base) {
71
		pr_warn("Can't make current CPU cache coherent.\n");
72 73 74 75
		pr_warn("Coherency fabric is not initialized\n");
		return 1;
	}

76 77
	ll_add_cpu_to_smp_group();
	return ll_enable_coherency();
78 79
}

80 81
static int mvebu_hwcc_notifier(struct notifier_block *nb,
			       unsigned long event, void *__dev)
82 83 84 85 86
{
	struct device *dev = __dev;

	if (event != BUS_NOTIFY_ADD_DEVICE)
		return NOTIFY_DONE;
87
	set_dma_ops(dev, &arm_coherent_dma_ops);
88 89 90 91

	return NOTIFY_OK;
}

92 93
static struct notifier_block mvebu_hwcc_nb = {
	.notifier_call = mvebu_hwcc_notifier,
94 95
};

96 97 98 99
static struct notifier_block mvebu_hwcc_pci_nb = {
	.notifier_call = mvebu_hwcc_notifier,
};

100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
static void __init armada_370_coherency_init(struct device_node *np)
{
	struct resource res;

	of_address_to_resource(np, 0, &res);
	coherency_phys_base = res.start;
	/*
	 * Ensure secondary CPUs will see the updated value,
	 * which they read before they join the coherency
	 * fabric, and therefore before they are coherent with
	 * the boot CPU cache.
	 */
	sync_cache_w(&coherency_phys_base);
	coherency_base = of_iomap(np, 0);
	coherency_cpu_base = of_iomap(np, 1);
115
	set_cpu_coherent();
116 117
}

118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
/*
 * This ioremap hook is used on Armada 375/38x to ensure that PCIe
 * memory areas are mapped as MT_UNCACHED instead of MT_DEVICE. This
 * is needed as a workaround for a deadlock issue between the PCIe
 * interface and the cache controller.
 */
static void __iomem *
armada_pcie_wa_ioremap_caller(phys_addr_t phys_addr, size_t size,
			      unsigned int mtype, void *caller)
{
	struct resource pcie_mem;

	mvebu_mbus_get_pcie_mem_aperture(&pcie_mem);

	if (pcie_mem.start <= phys_addr && (phys_addr + size) <= pcie_mem.end)
		mtype = MT_UNCACHED;

	return __arm_ioremap_caller(phys_addr, size, mtype, caller);
}

138
static void __init armada_375_380_coherency_init(struct device_node *np)
139
{
140 141
	struct device_node *cache_dn;

142
	coherency_cpu_base = of_iomap(np, 0);
143 144
	arch_ioremap_caller = armada_pcie_wa_ioremap_caller;

145 146 147 148 149 150 151
	/*
	 * We should switch the PL310 to I/O coherency mode only if
	 * I/O coherency is actually enabled.
	 */
	if (!coherency_available())
		return;

152 153 154 155 156 157 158 159 160 161 162 163 164 165
	/*
	 * Add the PL310 property "arm,io-coherent". This makes sure the
	 * outer sync operation is not used, which allows to
	 * workaround the system erratum that causes deadlocks when
	 * doing PCIe in an SMP situation on Armada 375 and Armada
	 * 38x.
	 */
	for_each_compatible_node(cache_dn, NULL, "arm,pl310-cache") {
		struct property *p;

		p = kzalloc(sizeof(*p), GFP_KERNEL);
		p->name = kstrdup("arm,io-coherent", GFP_KERNEL);
		of_add_property(cache_dn, p);
	}
166 167
}

168
static int coherency_type(void)
169 170
{
	struct device_node *np;
171
	const struct of_device_id *match;
172
	int type;
173

174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
	/*
	 * The coherency fabric is needed:
	 * - For coherency between processors on Armada XP, so only
	 *   when SMP is enabled.
	 * - For coherency between the processor and I/O devices, but
	 *   this coherency requires many pre-requisites (write
	 *   allocate cache policy, shareable pages, SMP bit set) that
	 *   are only meant in SMP situations.
	 *
	 * Note that this means that on Armada 370, there is currently
	 * no way to use hardware I/O coherency, because even when
	 * CONFIG_SMP is enabled, is_smp() returns false due to the
	 * Armada 370 being a single-core processor. To lift this
	 * limitation, we would have to find a way to make the cache
	 * policy set to write-allocate (on all Armada SoCs), and to
	 * set the shareable attribute in page tables (on all Armada
	 * SoCs except the Armada 370). Unfortunately, such decisions
	 * are taken very early in the kernel boot process, at a point
	 * where we don't know yet on which SoC we are running.

	 */
	if (!is_smp())
		return COHERENCY_FABRIC_TYPE_NONE;
197

198 199 200
	np = of_find_matching_node_and_match(NULL, of_coherency_table, &match);
	if (!np)
		return COHERENCY_FABRIC_TYPE_NONE;
201

202
	type = (int) match->data;
203

204
	of_node_put(np);
205

206
	return type;
207
}
208

209
int coherency_available(void)
210
{
211
	return coherency_type() != COHERENCY_FABRIC_TYPE_NONE;
212 213 214 215 216
}

int __init coherency_init(void)
{
	int type = coherency_type();
217 218 219
	struct device_node *np;

	np = of_find_matching_node(NULL, of_coherency_table);
220 221 222

	if (type == COHERENCY_FABRIC_TYPE_ARMADA_370_XP)
		armada_370_coherency_init(np);
223 224 225
	else if (type == COHERENCY_FABRIC_TYPE_ARMADA_375 ||
		 type == COHERENCY_FABRIC_TYPE_ARMADA_380)
		armada_375_380_coherency_init(np);
226

227 228
	of_node_put(np);

229 230 231 232 233
	return 0;
}

static int __init coherency_late_init(void)
{
234 235 236
	if (coherency_available())
		bus_register_notifier(&platform_bus_type,
				      &mvebu_hwcc_nb);
237 238 239 240
	return 0;
}

postcore_initcall(coherency_late_init);
241

242
#if IS_ENABLED(CONFIG_PCI)
243 244 245 246
static int __init coherency_pci_init(void)
{
	if (coherency_available())
		bus_register_notifier(&pci_bus_type,
247
				       &mvebu_hwcc_pci_nb);
248 249 250 251
	return 0;
}

arch_initcall(coherency_pci_init);
252
#endif