@@ -53,6 +53,9 @@ struct mtk_scp {
void __iomem *cpu_addr;
phys_addr_t phys_addr;
size_t dram_size;
+
+ struct platform_device *pdev;
+ struct rproc_subdev *rpmsg_subdev;
};
/**
@@ -13,6 +13,7 @@
#include <linux/platform_data/mtk_scp.h>
#include <linux/platform_device.h>
#include <linux/remoteproc.h>
+#include <linux/rpmsg/mtk_rpmsg.h>
#include "mtk_common.h"
#include "remoteproc_internal.h"
@@ -299,6 +300,23 @@ static int scp_map_memory_region(struct mtk_scp *scp)
return 0;
}
+static void scp_add_rpmsg_subdev(struct mtk_scp *scp)
+{
+ scp->rpmsg_subdev =
+ mtk_rpmsg_create_rproc_subdev(scp->pdev, scp->rproc);
+ if (scp->rpmsg_subdev)
+ rproc_add_subdev(scp->rproc, scp->rpmsg_subdev);
+}
+
+static void scp_remove_rpmsg_subdev(struct mtk_scp *scp)
+{
+ if (scp->rpmsg_subdev) {
+ rproc_remove_subdev(scp->rproc, scp->rpmsg_subdev);
+ mtk_rpmsg_destroy_rproc_subdev(scp->rpmsg_subdev);
+ scp->rpmsg_subdev = NULL;
+ }
+}
+
static int mtk_scp_probe(struct platform_device *pdev)
{
struct device *dev = &pdev->dev;
@@ -322,6 +340,7 @@ static int mtk_scp_probe(struct platform_device *pdev)
scp = (struct mtk_scp *)rproc->priv;
scp->rproc = rproc;
scp->dev = dev;
+ scp->pdev = pdev;
platform_set_drvdata(pdev, scp);
res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "sram");
@@ -392,13 +411,16 @@ static int mtk_scp_probe(struct platform_device *pdev)
init_waitqueue_head(&scp->run.wq);
init_waitqueue_head(&scp->ack_wq);
+ scp_add_rpmsg_subdev(scp);
+
ret = rproc_add(rproc);
if (ret)
- goto destroy_mutex;
+ goto remove_subdev;
- return ret;
+ return 0;
-destroy_mutex:
+remove_subdev:
+ scp_remove_rpmsg_subdev(scp);
mutex_destroy(&scp->scp_mutex);
free_rproc:
rproc_free(rproc);
@@ -410,6 +432,7 @@ static int mtk_scp_remove(struct platform_device *pdev)
{
struct mtk_scp *scp = platform_get_drvdata(pdev);
+ scp_remove_rpmsg_subdev(scp);
rproc_del(scp->rproc);
rproc_free(scp->rproc);
@@ -67,8 +67,9 @@ int scp_ipi_send(struct platform_device *pdev,
int ret;
if (WARN(id <= SCP_IPI_INIT || id >= SCP_IPI_MAX ||
- len > sizeof(send_obj->share_buf) || !buf,
- "failed to send ipi message\n"))
+ id == SCP_IPI_NS_SERVICE ||
+ len > sizeof(send_obj->share_buf) || !buf,
+ "failed to send ipi message\n"))
return -EINVAL;
ret = clk_prepare_enable(scp->clk);
@@ -15,6 +15,14 @@ config RPMSG_CHAR
in /dev. They make it possible for user-space programs to send and
receive rpmsg packets.
+config RPMSG_MTK_SCP
+ tristate "MediaTek SCP"
+ depends on MTK_SCP
+ select RPMSG
+ help
+ Say y here to enable support providing communication channels to
+ remote processors in MediaTek platforms.
+
config RPMSG_QCOM_GLINK_NATIVE
tristate
select RPMSG
@@ -1,6 +1,7 @@
# SPDX-License-Identifier: GPL-2.0
obj-$(CONFIG_RPMSG) += rpmsg_core.o
obj-$(CONFIG_RPMSG_CHAR) += rpmsg_char.o
+obj-$(CONFIG_RPMSG_MTK_SCP) += mtk_rpmsg.o
obj-$(CONFIG_RPMSG_QCOM_GLINK_RPM) += qcom_glink_rpm.o
obj-$(CONFIG_RPMSG_QCOM_GLINK_NATIVE) += qcom_glink_native.o
obj-$(CONFIG_RPMSG_QCOM_GLINK_SMEM) += qcom_glink_smem.o
new file mode 100644
@@ -0,0 +1,395 @@
+// SPDX-License-Identifier: GPL-2.0
+//
+// Copyright 2018 Google LLC.
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_data/mtk_scp.h>
+#include <linux/platform_device.h>
+#include <linux/remoteproc.h>
+#include <linux/rpmsg/mtk_rpmsg.h>
+#include <linux/workqueue.h>
+
+#include "rpmsg_internal.h"
+
+/*
+ * TODO: This is built on top of scp_ipi_register / scp_ipi_send in mtk_scp.h.
+ * It's probably better to move the implementation of register / send to here
+ * instead of in remoteproc/mtk_scp_ipi.c
+ */
+/*
+ * TODO: Do we need some sort of vring for performance? We may be able to get
+ * rid of this file if so, but that would require SCP firmware support too.
+ */
+
+struct mtk_rpmsg_device {
+ struct rpmsg_device rpdev;
+ struct platform_device *scp_pdev;
+};
+
+struct mtk_rpmsg_endpoint {
+ struct rpmsg_endpoint ept;
+ struct platform_device *scp_pdev;
+};
+
+/* TODO: Naming is hard... */
+struct mtk_rpmsg_rproc_subdev {
+ struct rproc *scp_rproc;
+ struct platform_device *scp_pdev;
+ struct rpmsg_endpoint *ns_ept;
+ struct rproc_subdev subdev;
+
+ struct work_struct register_work;
+ struct list_head channels;
+ spinlock_t channels_lock;
+};
+
+#define to_mtk_subdev(d) container_of(d, struct mtk_rpmsg_rproc_subdev, subdev)
+
+struct mtk_rpmsg_channel_info {
+ struct rpmsg_channel_info info;
+ bool registered;
+ struct list_head list;
+};
+
+/**
+ * TODO: This is copied from virtio_rpmsg_bus.
+ * struct rpmsg_ns_msg - dynamic name service announcement message
+ * @name: name of remote service that is published
+ * @addr: address of remote service that is published
+ *
+ * This message is sent across to publish a new service, or announce
+ * about its removal. When we receive these messages, an appropriate
+ * rpmsg channel (i.e device) is created/destroyed. In turn, the ->probe()
+ * or ->remove() handler of the appropriate rpmsg driver will be invoked
+ * (if/as-soon-as one is registered).
+ */
+struct rpmsg_ns_msg {
+ char name[RPMSG_NAME_SIZE];
+ u32 addr;
+} __packed;
+
+#define to_scp_device(r) container_of(r, struct mtk_rpmsg_device, rpdev)
+#define to_scp_endpoint(r) container_of(r, struct mtk_rpmsg_endpoint, ept)
+
+static const struct rpmsg_endpoint_ops mtk_rpmsg_endpoint_ops;
+
+static void __ept_release(struct kref *kref)
+{
+ struct rpmsg_endpoint *ept = container_of(kref, struct rpmsg_endpoint,
+ refcount);
+ kfree(to_scp_endpoint(ept));
+}
+
+static void mtk_rpmsg_ipi_handler(void *data, unsigned int len, void *priv)
+{
+ struct mtk_rpmsg_endpoint *mept = priv;
+ struct rpmsg_endpoint *ept = &mept->ept;
+
+ /* TODO: What if the cb() returns error? */
+ (*ept->cb)(ept->rpdev, data, len, ept->priv, ept->addr);
+}
+
+static struct rpmsg_endpoint *
+__rpmsg_create_ept(struct platform_device *scp_pdev, struct rpmsg_device *rpdev,
+ rpmsg_rx_cb_t cb, void *priv,
+ u32 id)
+{
+ struct mtk_rpmsg_endpoint *mept;
+ struct rpmsg_endpoint *ept;
+
+ int ret;
+
+ mept = kzalloc(sizeof(*mept), GFP_KERNEL);
+ if (!mept)
+ return NULL;
+ mept->scp_pdev = scp_pdev;
+
+ ept = &mept->ept;
+ kref_init(&ept->refcount);
+
+ ept->rpdev = rpdev;
+ ept->cb = cb;
+ ept->priv = priv;
+ ept->ops = &mtk_rpmsg_endpoint_ops;
+ ept->addr = id;
+
+ ret = scp_ipi_register(scp_pdev, id, mtk_rpmsg_ipi_handler, mept);
+ if (ret) {
+ dev_err(&scp_pdev->dev, "ipi register failed, id = %d", id);
+ kref_put(&ept->refcount, __ept_release);
+ return NULL;
+ }
+
+ return ept;
+}
+
+static struct rpmsg_endpoint *
+mtk_rpmsg_create_ept(struct rpmsg_device *rpdev, rpmsg_rx_cb_t cb, void *priv,
+ struct rpmsg_channel_info chinfo)
+{
+ struct platform_device *scp_pdev = to_scp_device(rpdev)->scp_pdev;
+ /* TODO: Is using src as IPI id "correct"? */
+ return __rpmsg_create_ept(scp_pdev, rpdev, cb, priv, chinfo.src);
+}
+
+static const struct rpmsg_device_ops mtk_rpmsg_device_ops = {
+ .create_ept = mtk_rpmsg_create_ept,
+};
+
+static void mtk_rpmsg_destroy_ept(struct rpmsg_endpoint *ept)
+{
+ struct mtk_rpmsg_endpoint *mept = to_scp_endpoint(ept);
+
+ scp_ipi_unregister(mept->scp_pdev, ept->addr);
+ kref_put(&ept->refcount, __ept_release);
+}
+
+static int __mtk_rpmsg_send(struct mtk_rpmsg_endpoint *mept, void *data,
+ int len, bool wait)
+{
+ /* TODO: This currently always wait for SCP to receive last command. */
+ return scp_ipi_send(mept->scp_pdev, mept->ept.addr, data, len, 0);
+}
+
+static int mtk_rpmsg_send(struct rpmsg_endpoint *ept, void *data, int len)
+{
+ struct mtk_rpmsg_endpoint *mept = to_scp_endpoint(ept);
+
+ return __mtk_rpmsg_send(mept, data, len, true);
+}
+
+static int mtk_rpmsg_trysend(struct rpmsg_endpoint *ept, void *data, int len)
+{
+ struct mtk_rpmsg_endpoint *mept = to_scp_endpoint(ept);
+
+ return __mtk_rpmsg_send(mept, data, len, false);
+}
+
+static const struct rpmsg_endpoint_ops mtk_rpmsg_endpoint_ops = {
+ .destroy_ept = mtk_rpmsg_destroy_ept,
+ .send = mtk_rpmsg_send,
+ .trysend = mtk_rpmsg_trysend,
+};
+
+static void mtk_rpmsg_release_device(struct device *dev)
+{
+ struct rpmsg_device *rpdev = to_rpmsg_device(dev);
+ struct mtk_rpmsg_device *mdev = to_scp_device(rpdev);
+
+ kfree(mdev);
+}
+
+static int mtk_rpmsg_register_device(struct platform_device *scp_pdev,
+ struct rpmsg_channel_info *info)
+{
+ struct rpmsg_device *rpdev;
+ struct mtk_rpmsg_device *mdev;
+ int ret;
+
+ mdev = kzalloc(sizeof(*mdev), GFP_KERNEL);
+ if (!mdev)
+ return -ENOMEM;
+
+ mdev->scp_pdev = scp_pdev;
+
+ rpdev = &mdev->rpdev;
+ rpdev->ops = &mtk_rpmsg_device_ops;
+ rpdev->src = info->src;
+ rpdev->dst = info->dst;
+ strncpy(rpdev->id.name, info->name, RPMSG_NAME_SIZE);
+
+ rpdev->dev.parent = &scp_pdev->dev;
+ rpdev->dev.release = mtk_rpmsg_release_device;
+
+ ret = rpmsg_register_device(rpdev);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static void mtk_register_device_work_function(struct work_struct *register_work)
+{
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev = container_of(
+ register_work, struct mtk_rpmsg_rproc_subdev, register_work);
+ struct platform_device *scp_pdev = mtk_subdev->scp_pdev;
+ struct mtk_rpmsg_channel_info *info;
+
+ int ret;
+ unsigned long flags;
+
+ spin_lock_irqsave(&mtk_subdev->channels_lock, flags);
+ list_for_each_entry(info, &mtk_subdev->channels, list) {
+ if (info->registered)
+ continue;
+
+ spin_unlock_irqrestore(&mtk_subdev->channels_lock, flags);
+ ret = mtk_rpmsg_register_device(scp_pdev, &info->info);
+ spin_lock_irqsave(&mtk_subdev->channels_lock, flags);
+
+ if (ret) {
+ dev_err(&scp_pdev->dev, "Can't create rpmsg_device\n");
+ continue;
+ }
+
+ info->registered = true;
+ }
+ spin_unlock_irqrestore(&mtk_subdev->channels_lock, flags);
+}
+
+static int mtk_rpmsg_create_device(struct mtk_rpmsg_rproc_subdev *mtk_subdev,
+ char *name, u32 addr)
+{
+ struct mtk_rpmsg_channel_info *info;
+ unsigned long flags;
+
+ /* This is called in interrupt context from name service callback. */
+ info = kzalloc(sizeof(*info), GFP_ATOMIC);
+ if (!info)
+ return -ENOMEM;
+
+ strncpy(info->info.name, name, RPMSG_NAME_SIZE);
+ info->info.src = addr;
+ info->info.dst = RPMSG_ADDR_ANY;
+ spin_lock_irqsave(&mtk_subdev->channels_lock, flags);
+ list_add(&info->list, &mtk_subdev->channels);
+ spin_unlock_irqrestore(&mtk_subdev->channels_lock, flags);
+
+ schedule_work(&mtk_subdev->register_work);
+ return 0;
+}
+
+static int mtk_rpmsg_ns_cb(struct rpmsg_device *rpdev, void *data, int len,
+ void *priv, u32 src)
+{
+ struct rpmsg_ns_msg *msg = data;
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev = priv;
+ struct device *dev = &mtk_subdev->scp_pdev->dev;
+
+ int ret;
+
+ if (len != sizeof(*msg)) {
+ dev_err(dev, "malformed ns msg (%d)\n", len);
+ return -EINVAL;
+ }
+
+ /*
+ * the name service ept does _not_ belong to a real rpmsg channel,
+ * and is handled by the rpmsg bus itself.
+ * for sanity reasons, make sure a valid rpdev has _not_ sneaked
+ * in somehow.
+ */
+ if (rpdev) {
+ dev_err(dev, "anomaly: ns ept has an rpdev handle\n");
+ return -EINVAL;
+ }
+
+ /* don't trust the remote processor for null terminating the name */
+ msg->name[RPMSG_NAME_SIZE - 1] = '\0';
+
+ dev_info(dev, "creating channel %s addr 0x%x\n", msg->name, msg->addr);
+
+ ret = mtk_rpmsg_create_device(mtk_subdev, msg->name, msg->addr);
+ if (ret) {
+ dev_err(dev, "create rpmsg device failed\n");
+ return ret;
+ }
+
+ return 0;
+}
+
+int mtk_rpmsg_prepare(struct rproc_subdev *subdev)
+{
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev;
+ struct platform_device *scp_pdev;
+
+ mtk_subdev = to_mtk_subdev(subdev);
+ scp_pdev = mtk_subdev->scp_pdev;
+
+ /* a dedicated endpoint handles the name service msgs */
+ mtk_subdev->ns_ept =
+ __rpmsg_create_ept(scp_pdev, NULL, mtk_rpmsg_ns_cb, mtk_subdev,
+ SCP_IPI_NS_SERVICE);
+ if (!mtk_subdev->ns_ept) {
+ dev_err(&scp_pdev->dev,
+ "failed to create name service endpoint\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+void mtk_rpmsg_unprepare(struct rproc_subdev *subdev)
+{
+ struct mtk_rpmsg_channel_info *info, *next;
+ struct device *dev;
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev = to_mtk_subdev(subdev);
+ unsigned long flags;
+
+ dev = &mtk_subdev->scp_pdev->dev;
+
+ cancel_work_sync(&mtk_subdev->register_work);
+ /*
+ * TODO: It feels like there's still possible race condition for some
+ * callback to schedule new work after the cancel_work_sync. How's this
+ * typically handled?
+ */
+ spin_lock_irqsave(&mtk_subdev->channels_lock, flags);
+ list_for_each_entry(info, &mtk_subdev->channels, list) {
+ if (!info->registered)
+ continue;
+ spin_unlock_irqrestore(&mtk_subdev->channels_lock, flags);
+ if (rpmsg_unregister_device(dev, &info->info)) {
+ dev_warn(
+ dev,
+ "rpmsg_unregister_device failed for %s.%d.%d\n",
+ info->info.name, info->info.src,
+ info->info.dst);
+ }
+ spin_lock_irqsave(&mtk_subdev->channels_lock, flags);
+ }
+
+ list_for_each_entry_safe(info, next,
+ &mtk_subdev->channels, list) {
+ list_del(&info->list);
+ kfree(info);
+ }
+ spin_unlock_irqrestore(&mtk_subdev->channels_lock, flags);
+
+ kref_put(&mtk_subdev->ns_ept->refcount, __ept_release);
+}
+
+struct rproc_subdev *
+mtk_rpmsg_create_rproc_subdev(struct platform_device *scp_pdev,
+ struct rproc *scp_rproc)
+{
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev;
+
+ mtk_subdev = kzalloc(sizeof(*mtk_subdev), GFP_KERNEL);
+ if (!mtk_subdev)
+ return NULL;
+
+ mtk_subdev->scp_pdev = scp_pdev;
+ mtk_subdev->scp_rproc = scp_rproc;
+ mtk_subdev->subdev.prepare = mtk_rpmsg_prepare;
+ mtk_subdev->subdev.unprepare = mtk_rpmsg_unprepare;
+ INIT_LIST_HEAD(&mtk_subdev->channels);
+ INIT_WORK(&mtk_subdev->register_work,
+ mtk_register_device_work_function);
+
+ return &mtk_subdev->subdev;
+}
+EXPORT_SYMBOL_GPL(mtk_rpmsg_create_rproc_subdev);
+
+void mtk_rpmsg_destroy_rproc_subdev(struct rproc_subdev *subdev)
+{
+ struct mtk_rpmsg_rproc_subdev *mtk_subdev = to_mtk_subdev(subdev);
+
+ kfree(mtk_subdev);
+}
+EXPORT_SYMBOL_GPL(mtk_rpmsg_destroy_rproc_subdev);
+
+MODULE_LICENSE("GPL v2");
+MODULE_DESCRIPTION("MediaTek scp rpmsg driver");
@@ -22,7 +22,7 @@ typedef void (*scp_ipi_handler_t) (void *data,
* command to SCP.
* For other IPI below, AP should send the request
* to SCP to trigger the interrupt.
- * @IPI_MAX: The maximum IPI number
+ * @SCP_IPI_MAX: The maximum IPI number
*/
enum scp_ipi_id {
@@ -34,9 +34,11 @@ enum scp_ipi_id {
SCP_IPI_VENC_VP8,
SCP_IPI_MDP,
SCP_IPI_CROS_HOST_CMD,
- SCP_IPI_MAX,
};
+#define SCP_IPI_NS_SERVICE 0xFF
+#define SCP_IPI_MAX 0x100
+
/**
* scp_ipi_register - register an ipi function
*
new file mode 100644
@@ -0,0 +1,35 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright 2018 Google LLC.
+ */
+
+#ifndef __LINUX_RPMSG_MTK_RPMSG_H
+#define __LINUX_RPMSG_MTK_RPMSG_H
+
+#include <linux/device.h>
+#include <linux/remoteproc.h>
+
+#if IS_ENABLED(CONFIG_RPMSG_MTK_SCP)
+
+struct rproc_subdev *
+mtk_rpmsg_create_rproc_subdev(struct platform_device *scp_pdev,
+ struct rproc *scp_rproc);
+
+void mtk_rpmsg_destroy_rproc_subdev(struct rproc_subdev *subdev);
+
+#else
+
+static inline struct rproc_subdev *
+mtk_rpmsg_create_rproc_subdev(struct platform_device *scp_pdev,
+ struct rproc *scp_rproc)
+{
+ return NULL;
+}
+
+static inline void mtk_rpmsg_destroy_rproc_subdev(struct rproc_subdev *subdev)
+{
+}
+
+#endif
+
+#endif
Add a simple rpmsg support for mt8183 SCP, that use IPI / IPC directly. Lots of TODO, and I'm not sure on all file / type / variable namings. Signed-off-by: Pi-Hsun Shih <pihsun@chromium.org> --- Changes from v2: - Unregiser IPI handler on unprepare. - Lock the channel list on operations. - Move SCP_IPI_NS_SERVICE to 0xFF. Changes from v1: - Do cleanup properly in mtk_rpmsg.c, which also removes the problem of short-lived work items. - Fix several issues checkpatch found. --- drivers/remoteproc/mtk_common.h | 3 + drivers/remoteproc/mtk_scp.c | 29 +- drivers/remoteproc/mtk_scp_ipi.c | 5 +- drivers/rpmsg/Kconfig | 8 + drivers/rpmsg/Makefile | 1 + drivers/rpmsg/mtk_rpmsg.c | 395 ++++++++++++++++++++++++++ include/linux/platform_data/mtk_scp.h | 6 +- include/linux/rpmsg/mtk_rpmsg.h | 35 +++ 8 files changed, 475 insertions(+), 7 deletions(-) create mode 100644 drivers/rpmsg/mtk_rpmsg.c create mode 100644 include/linux/rpmsg/mtk_rpmsg.h