1 files changed, 264 insertions, 29 deletions

diff --git a/drivers/hwtracing/coresight/coresight-etb10.c b/drivers/hwtracing/coresight/coresight-etb10.c
index 77d0f9c1118d..acbce79934d6 100644
--- a/drivers/hwtracing/coresight/coresight-etb10.c
+++ b/drivers/hwtracing/coresight/coresight-etb10.c
@@ -1,5 +1,7 @@
 /* Copyright (c) 2011-2012, The Linux Foundation. All rights reserved.
  *
+ * Description: CoreSight Embedded Trace Buffer driver
+ *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 and
  * only version 2 as published by the Free Software Foundation.
@@ -10,8 +12,8 @@
  * GNU General Public License for more details.
  */
 
+#include <asm/local.h>
 #include <linux/kernel.h>
-#include <linux/module.h>
 #include <linux/init.h>
 #include <linux/types.h>
 #include <linux/device.h>
@@ -27,6 +29,11 @@
 #include <linux/coresight.h>
 #include <linux/amba/bus.h>
 #include <linux/clk.h>
+#include <linux/circ_buf.h>
+#include <linux/mm.h>
+#include <linux/perf_event.h>
+
+#include <asm/local.h>
 
 #include "coresight-priv.h"
 
@@ -64,6 +71,26 @@
 #define ETB_FRAME_SIZE_WORDS	4
 
 /**
+ * struct cs_buffer - keep track of a recording session' specifics
+ * @cur:	index of the current buffer
+ * @nr_pages:	max number of pages granted to us
+ * @offset:	offset within the current buffer
+ * @data_size:	how much we collected in this run
+ * @lost:	other than zero if we had a HW buffer wrap around
+ * @snapshot:	is this run in snapshot mode
+ * @data_pages:	a handle the ring buffer
+ */
+struct cs_buffers {
+	unsigned int		cur;
+	unsigned int		nr_pages;
+	unsigned long		offset;
+	local_t			data_size;
+	local_t			lost;
+	bool			snapshot;
+	void			**data_pages;
+};
+
+/**
  * struct etb_drvdata - specifics associated to an ETB component
  * @base:	memory mapped base address for this component.
  * @dev:	the device entity associated to this component.
@@ -71,10 +98,10 @@
  * @csdev:	component vitals needed by the framework.
  * @miscdev:	specifics to handle "/dev/xyz.etb" entry.
  * @spinlock:	only one at a time pls.
- * @in_use:	synchronise user space access to etb buffer.
+ * @reading:	synchronise user space access to etb buffer.
+ * @mode:	this ETB is being used.
  * @buf:	area of memory where ETB buffer content gets sent.
  * @buffer_depth: size of @buf.
- * @enable:	this ETB is being used.
  * @trigger_cntr: amount of words to store after a trigger.
  */
 struct etb_drvdata {
@@ -84,10 +111,10 @@ struct etb_drvdata {
 	struct coresight_device	*csdev;
 	struct miscdevice	miscdev;
 	spinlock_t		spinlock;
-	atomic_t		in_use;
+	local_t			reading;
+	local_t			mode;
 	u8			*buf;
 	u32			buffer_depth;
-	bool			enable;
 	u32			trigger_cntr;
 };
 
@@ -132,18 +159,31 @@ static void etb_enable_hw(struct etb_drvdata *drvdata)
 	CS_LOCK(drvdata->base);
 }
 
-static int etb_enable(struct coresight_device *csdev)
+static int etb_enable(struct coresight_device *csdev, u32 mode)
 {
-	struct etb_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
+	u32 val;
 	unsigned long flags;
+	struct etb_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
 
-	pm_runtime_get_sync(drvdata->dev);
+	val = local_cmpxchg(&drvdata->mode,
+			    CS_MODE_DISABLED, mode);
+	/*
+	 * When accessing from Perf, a HW buffer can be handled
+	 * by a single trace entity.  In sysFS mode many tracers
+	 * can be logging to the same HW buffer.
+	 */
+	if (val == CS_MODE_PERF)
+		return -EBUSY;
+
+	/* Nothing to do, the tracer is already enabled. */
+	if (val == CS_MODE_SYSFS)
+		goto out;
 
 	spin_lock_irqsave(&drvdata->spinlock, flags);
 	etb_enable_hw(drvdata);
-	drvdata->enable = true;
 	spin_unlock_irqrestore(&drvdata->spinlock, flags);
 
+out:
 	dev_info(drvdata->dev, "ETB enabled\n");
 	return 0;
 }
@@ -244,17 +284,225 @@ static void etb_disable(struct coresight_device *csdev)
 	spin_lock_irqsave(&drvdata->spinlock, flags);
 	etb_disable_hw(drvdata);
 	etb_dump_hw(drvdata);
-	drvdata->enable = false;
 	spin_unlock_irqrestore(&drvdata->spinlock, flags);
 
-	pm_runtime_put(drvdata->dev);
+	local_set(&drvdata->mode, CS_MODE_DISABLED);
 
 	dev_info(drvdata->dev, "ETB disabled\n");
 }
 
+static void *etb_alloc_buffer(struct coresight_device *csdev, int cpu,
+			      void **pages, int nr_pages, bool overwrite)
+{
+	int node;
+	struct cs_buffers *buf;
+
+	if (cpu == -1)
+		cpu = smp_processor_id();
+	node = cpu_to_node(cpu);
+
+	buf = kzalloc_node(sizeof(struct cs_buffers), GFP_KERNEL, node);
+	if (!buf)
+		return NULL;
+
+	buf->snapshot = overwrite;
+	buf->nr_pages = nr_pages;
+	buf->data_pages = pages;
+
+	return buf;
+}
+
+static void etb_free_buffer(void *config)
+{
+	struct cs_buffers *buf = config;
+
+	kfree(buf);
+}
+
+static int etb_set_buffer(struct coresight_device *csdev,
+			  struct perf_output_handle *handle,
+			  void *sink_config)
+{
+	int ret = 0;
+	unsigned long head;
+	struct cs_buffers *buf = sink_config;
+
+	/* wrap head around to the amount of space we have */
+	head = handle->head & ((buf->nr_pages << PAGE_SHIFT) - 1);
+
+	/* find the page to write to */
+	buf->cur = head / PAGE_SIZE;
+
+	/* and offset within that page */
+	buf->offset = head % PAGE_SIZE;
+
+	local_set(&buf->data_size, 0);
+
+	return ret;
+}
+
+static unsigned long etb_reset_buffer(struct coresight_device *csdev,
+				      struct perf_output_handle *handle,
+				      void *sink_config, bool *lost)
+{
+	unsigned long size = 0;
+	struct cs_buffers *buf = sink_config;
+
+	if (buf) {
+		/*
+		 * In snapshot mode ->data_size holds the new address of the
+		 * ring buffer's head.  The size itself is the whole address
+		 * range since we want the latest information.
+		 */
+		if (buf->snapshot)
+			handle->head = local_xchg(&buf->data_size,
+						  buf->nr_pages << PAGE_SHIFT);
+
+		/*
+		 * Tell the tracer PMU how much we got in this run and if
+		 * something went wrong along the way.  Nobody else can use
+		 * this cs_buffers instance until we are done.  As such
+		 * resetting parameters here and squaring off with the ring
+		 * buffer API in the tracer PMU is fine.
+		 */
+		*lost = !!local_xchg(&buf->lost, 0);
+		size = local_xchg(&buf->data_size, 0);
+	}
+
+	return size;
+}
+
+static void etb_update_buffer(struct coresight_device *csdev,
+			      struct perf_output_handle *handle,
+			      void *sink_config)
+{
+	int i, cur;
+	u8 *buf_ptr;
+	u32 read_ptr, write_ptr, capacity;
+	u32 status, read_data, to_read;
+	unsigned long offset;
+	struct cs_buffers *buf = sink_config;
+	struct etb_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
+
+	if (!buf)
+		return;
+
+	capacity = drvdata->buffer_depth * ETB_FRAME_SIZE_WORDS;
+
+	CS_UNLOCK(drvdata->base);
+	etb_disable_hw(drvdata);
+
+	/* unit is in words, not bytes */
+	read_ptr = readl_relaxed(drvdata->base + ETB_RAM_READ_POINTER);
+	write_ptr = readl_relaxed(drvdata->base + ETB_RAM_WRITE_POINTER);
+
+	/*
+	 * Entries should be aligned to the frame size.  If they are not
+	 * go back to the last alignement point to give decoding tools a
+	 * chance to fix things.
+	 */
+	if (write_ptr % ETB_FRAME_SIZE_WORDS) {
+		dev_err(drvdata->dev,
+			"write_ptr: %lu not aligned to formatter frame size\n",
+			(unsigned long)write_ptr);
+
+		write_ptr &= ~(ETB_FRAME_SIZE_WORDS - 1);
+		local_inc(&buf->lost);
+	}
+
+	/*
+	 * Get a hold of the status register and see if a wrap around
+	 * has occurred.  If so adjust things accordingly.  Otherwise
+	 * start at the beginning and go until the write pointer has
+	 * been reached.
+	 */
+	status = readl_relaxed(drvdata->base + ETB_STATUS_REG);
+	if (status & ETB_STATUS_RAM_FULL) {
+		local_inc(&buf->lost);
+		to_read = capacity;
+		read_ptr = write_ptr;
+	} else {
+		to_read = CIRC_CNT(write_ptr, read_ptr, drvdata->buffer_depth);
+		to_read *= ETB_FRAME_SIZE_WORDS;
+	}
+
+	/*
+	 * Make sure we don't overwrite data that hasn't been consumed yet.
+	 * It is entirely possible that the HW buffer has more data than the
+	 * ring buffer can currently handle.  If so adjust the start address
+	 * to take only the last traces.
+	 *
+	 * In snapshot mode we are looking to get the latest traces only and as
+	 * such, we don't care about not overwriting data that hasn't been
+	 * processed by user space.
+	 */
+	if (!buf->snapshot && to_read > handle->size) {
+		u32 mask = ~(ETB_FRAME_SIZE_WORDS - 1);
+
+		/* The new read pointer must be frame size aligned */
+		to_read -= handle->size & mask;
+		/*
+		 * Move the RAM read pointer up, keeping in mind that
+		 * everything is in frame size units.
+		 */
+		read_ptr = (write_ptr + drvdata->buffer_depth) -
+					to_read / ETB_FRAME_SIZE_WORDS;
+		/* Wrap around if need be*/
+		read_ptr &= ~(drvdata->buffer_depth - 1);
+		/* let the decoder know we've skipped ahead */
+		local_inc(&buf->lost);
+	}
+
+	/* finally tell HW where we want to start reading from */
+	writel_relaxed(read_ptr, drvdata->base + ETB_RAM_READ_POINTER);
+
+	cur = buf->cur;
+	offset = buf->offset;
+	for (i = 0; i < to_read; i += 4) {
+		buf_ptr = buf->data_pages[cur] + offset;
+		read_data = readl_relaxed(drvdata->base +
+					  ETB_RAM_READ_DATA_REG);
+		*buf_ptr++ = read_data >> 0;
+		*buf_ptr++ = read_data >> 8;
+		*buf_ptr++ = read_data >> 16;
+		*buf_ptr++ = read_data >> 24;
+
+		offset += 4;
+		if (offset >= PAGE_SIZE) {
+			offset = 0;
+			cur++;
+			/* wrap around at the end of the buffer */
+			cur &= buf->nr_pages - 1;
+		}
+	}
+
+	/* reset ETB buffer for next run */
+	writel_relaxed(0x0, drvdata->base + ETB_RAM_READ_POINTER);
+	writel_relaxed(0x0, drvdata->base + ETB_RAM_WRITE_POINTER);
+
+	/*
+	 * In snapshot mode all we have to do is communicate to
+	 * perf_aux_output_end() the address of the current head.  In full
+	 * trace mode the same function expects a size to move rb->aux_head
+	 * forward.
+	 */
+	if (buf->snapshot)
+		local_set(&buf->data_size, (cur * PAGE_SIZE) + offset);
+	else
+		local_add(to_read, &buf->data_size);
+
+	etb_enable_hw(drvdata);
+	CS_LOCK(drvdata->base);
+}
+
 static const struct coresight_ops_sink etb_sink_ops = {
 	.enable		= etb_enable,
 	.disable	= etb_disable,
+	.alloc_buffer	= etb_alloc_buffer,
+	.free_buffer	= etb_free_buffer,
+	.set_buffer	= etb_set_buffer,
+	.reset_buffer	= etb_reset_buffer,
+	.update_buffer	= etb_update_buffer,
 };
 
 static const struct coresight_ops etb_cs_ops = {
@@ -266,7 +514,7 @@ static void etb_dump(struct etb_drvdata *drvdata)
 	unsigned long flags;
 
 	spin_lock_irqsave(&drvdata->spinlock, flags);
-	if (drvdata->enable) {
+	if (local_read(&drvdata->mode) == CS_MODE_SYSFS) {
 		etb_disable_hw(drvdata);
 		etb_dump_hw(drvdata);
 		etb_enable_hw(drvdata);
@@ -281,7 +529,7 @@ static int etb_open(struct inode *inode, struct file *file)
 	struct etb_drvdata *drvdata = container_of(file->private_data,
 						   struct etb_drvdata, miscdev);
 
-	if (atomic_cmpxchg(&drvdata->in_use, 0, 1))
+	if (local_cmpxchg(&drvdata->reading, 0, 1))
 		return -EBUSY;
 
 	dev_dbg(drvdata->dev, "%s: successfully opened\n", __func__);
@@ -317,7 +565,7 @@ static int etb_release(struct inode *inode, struct file *file)
 {
 	struct etb_drvdata *drvdata = container_of(file->private_data,
 						   struct etb_drvdata, miscdev);
-	atomic_set(&drvdata->in_use, 0);
+	local_set(&drvdata->reading, 0);
 
 	dev_dbg(drvdata->dev, "%s: released\n", __func__);
 	return 0;
@@ -489,15 +737,6 @@ err_misc_register:
 	return ret;
 }
 
-static int etb_remove(struct amba_device *adev)
-{
-	struct etb_drvdata *drvdata = amba_get_drvdata(adev);
-
-	misc_deregister(&drvdata->miscdev);
-	coresight_unregister(drvdata->csdev);
-	return 0;
-}
-
 #ifdef CONFIG_PM
 static int etb_runtime_suspend(struct device *dev)
 {
@@ -537,14 +776,10 @@ static struct amba_driver etb_driver = {
 		.name	= "coresight-etb10",
 		.owner	= THIS_MODULE,
 		.pm	= &etb_dev_pm_ops,
+		.suppress_bind_attrs = true,
 
 	},
 	.probe		= etb_probe,
-	.remove		= etb_remove,
 	.id_table	= etb_ids,
 };
-
-module_amba_driver(etb_driver);
-
-MODULE_LICENSE("GPL v2");
-MODULE_DESCRIPTION("CoreSight Embedded Trace Buffer driver");
+builtin_amba_driver(etb_driver);