ARM: tegra: hardware arbitration semaphore support

add apis to use the hardware arbitration semaphores in order
to share hardware modules between kernel drivers and AVP
firmware (e.g., the BSEA (audio bitstream) engine and
AES block)

Change-Id: I500ef0797223bc702151ad14e0e2156f50644a2a
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>

1 files changed, 242 insertions, 0 deletions

diff --git a/arch/arm/mach-tegra/arb_sema.c b/arch/arm/mach-tegra/arb_sema.c
new file mode 100644
index 000000000000..55af15e935f2
--- /dev/null
+++ b/arch/arm/mach-tegra/arb_sema.c
@@ -0,0 +1,242 @@
+/*
+ * Copyright (C) 2010, NVIDIA Corporation
+ *
+ * This software is licensed under the terms of the GNU General Public
+ * License version 2, as published by the Free Software Foundation, and
+ * may be copied, distributed, and modified under those terms.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/io.h>
+#include <linux/mutex.h>
+#include <linux/err.h>
+#include <linux/delay.h>
+#include <linux/completion.h>
+
+#include <mach/arb_sema.h>
+#include <mach/irqs.h>
+#include <mach/iomap.h>
+
+#define TEGRA_RPC_MAX_SEM 32
+
+/* arb_gnt ictrl */
+#define ARB_CPU_INT_EN		0x4
+
+/* arb_sema */
+#define ARB_GRANT_STATUS	0x0
+#define ARB_GRANT_REQUEST	0x4
+#define ARB_GRANT_RELEASE	0x8
+
+struct tegra_arb_dev {
+	void __iomem	*sema_base;
+	void __iomem	*gnt_base;
+	spinlock_t lock;
+	struct completion arb_gnt_complete[TEGRA_RPC_MAX_SEM];
+	struct mutex mutexes[TEGRA_RPC_MAX_SEM];
+	int irq;
+	int status;
+	bool suspended;
+};
+
+static struct tegra_arb_dev *arb;
+
+static inline u32 arb_sema_read(u32 offset)
+{
+	return readl(arb->sema_base + offset);
+}
+
+static inline void arb_sema_write(u32 value, u32 offset)
+{
+	writel(value, arb->sema_base + offset);
+}
+
+static inline u32 arb_gnt_read(u32 offset)
+{
+	return readl(arb->gnt_base + offset);
+}
+
+static inline void arb_gnt_write(u32 value, u32 offset)
+{
+	writel(value, arb->gnt_base + offset);
+}
+
+static void request_arb_sem(enum tegra_arb_module lock)
+{
+	unsigned long flags;
+	u32 value;
+
+	spin_lock_irqsave(&arb->lock, flags);
+
+	arb_sema_write(1 << lock, ARB_GRANT_REQUEST);
+	value = arb_gnt_read(ARB_CPU_INT_EN);
+	value |= (1 << lock);
+	arb_gnt_write(value, ARB_CPU_INT_EN);
+
+	spin_unlock_irqrestore(&arb->lock, flags);
+}
+
+static void cancel_arb_sem(enum tegra_arb_module lock)
+{
+	unsigned long flags;
+	u32 value;
+
+	spin_lock_irqsave(&arb->lock, flags);
+
+	arb_sema_write(1 << lock, ARB_GRANT_RELEASE);
+	value = arb_gnt_read(ARB_CPU_INT_EN);
+	value &= ~(1 << lock);
+	arb_gnt_write(value, ARB_CPU_INT_EN);
+
+	spin_unlock_irqrestore(&arb->lock, flags);
+}
+
+int tegra_arb_mutex_lock_timeout(enum tegra_arb_module lock, int msecs)
+{
+	int ret;
+
+	if (!arb)
+		return -ENODEV;
+
+	if (arb->suspended) {
+		pr_err("device in suspend\n");
+		return -ETIMEDOUT;
+	}
+
+	mutex_lock(&arb->mutexes[lock]);
+	INIT_COMPLETION(arb->arb_gnt_complete[lock]);
+	request_arb_sem(lock);
+	ret = wait_for_completion_timeout(&arb->arb_gnt_complete[lock], msecs_to_jiffies(msecs));
+	if (ret == 0) {
+		pr_err("timed out.\n");
+		cancel_arb_sem(lock);
+		mutex_unlock(&arb->mutexes[lock]);
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(tegra_arb_mutex_lock_timeout);
+
+int tegra_arb_mutex_unlock(enum tegra_arb_module lock)
+{
+	if (!arb)
+		return -ENODEV;
+
+	if (arb->suspended) {
+		pr_err("device in suspend\n");
+		return -ETIMEDOUT;
+	}
+
+	cancel_arb_sem(lock);
+	mutex_unlock(&arb->mutexes[lock]);
+	return 0;
+}
+EXPORT_SYMBOL(tegra_arb_mutex_unlock);
+
+static irqreturn_t arb_gnt_isr(int irq, void *dev_id)
+{
+	struct tegra_arb_dev *dev = dev_id;
+	unsigned long status;
+	u32 cpu_int_en;
+	unsigned int bit;
+	unsigned long flags;
+
+	spin_lock_irqsave(&arb->lock, flags);
+
+	status = arb_sema_read(ARB_GRANT_STATUS);
+	pr_debug("%s: 0x%lx\n", __func__, status);
+
+	/* disable the arb semaphores which were signalled */
+	cpu_int_en = arb_gnt_read(ARB_CPU_INT_EN);
+	arb_gnt_write((cpu_int_en & ~(status & cpu_int_en)),
+		ARB_CPU_INT_EN);
+
+	status &= cpu_int_en;
+	for_each_set_bit(bit, &status, BITS_PER_LONG)
+		complete(&dev->arb_gnt_complete[bit]);
+
+	spin_unlock_irqrestore(&arb->lock, flags);
+	return IRQ_HANDLED;
+}
+
+int tegra_arb_suspend(void)
+{
+	unsigned long status = arb_sema_read(ARB_GRANT_STATUS);
+
+	if (WARN_ON(status != 0)) {
+		pr_err("%s: suspending while holding arbitration "
+		       "semaphore: %08lx\n", __func__, status);
+	}
+	arb->suspended = true;
+
+	return status ? -EBUSY : 0;
+}
+
+int tegra_arb_resume(void)
+{
+	arb->suspended = false;
+	return 0;
+}
+
+static int __init tegra_arb_init(void)
+{
+	struct tegra_arb_dev *dev = NULL;
+	int err, i;
+
+	dev = kzalloc(sizeof(struct tegra_arb_dev), GFP_KERNEL);
+	if (dev == NULL) {
+		pr_err("%s: unable to alloc data struct.\n", __func__);
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < TEGRA_RPC_MAX_SEM; i++) {
+		mutex_init(&dev->mutexes[i]);
+		init_completion(&dev->arb_gnt_complete[i]);
+	}
+
+	dev->sema_base = IO_ADDRESS(TEGRA_ARB_SEMA_BASE);
+	if (!dev->sema_base) {
+		pr_err("%s: can't get arb sema_base\n", __func__);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	dev->gnt_base = IO_ADDRESS(TEGRA_ARBGNT_ICTLR_BASE);
+	if (!dev->gnt_base) {
+		pr_err("%s: can't ioremap gnt_base\n", __func__);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	dev->irq = INT_GNT_1;
+	err = request_irq(dev->irq, arb_gnt_isr, 0, "rpc-arbsema", dev);
+	if (err) {
+		pr_err("%s: request_irq(%d) failed(%d)\n", __func__,
+			dev->irq, err);
+		goto out;
+	}
+
+	spin_lock_init(&dev->lock);
+	arb = dev;
+
+	pr_info("%s: initialized\n", __func__);
+	return 0;
+
+out:
+	kfree(dev);
+	pr_err("%s: initialization failed.\n", __func__);
+	return err;
+}
+subsys_initcall(tegra_arb_init);
+
+MODULE_LICENSE("GPLv2");