[PATCH v3 0/5] misc: fastrpc: fixes and map/unmap support

[PATCH v3 0/5] misc: fastrpc: fixes and map/unmap support

[Srinivas Kandagatla]

Hi Greg,

These patches implement a few fixes identified while working on the
QCS404 ML integration plus we now have support for mmap/unmap of
buffers (so the process can be created with less initial memory
requirements).

Thanks,
srini

Jorge Ramirez-Ortiz (4):
  misc: fastrpc: add mmap/unmap support
  misc: fastrpc: do not interrupt kernel calls
  misc: fastrpc: handle interrupted contexts
  misc: fastrpc: revert max init file size back to 2MB

Srinivas Kandagatla (1):
  misc: fastrpc: fix memory leak from miscdev->name

 drivers/misc/fastrpc.c      | 209 ++++++++++++++++++++++++++++++++++--
 include/uapi/misc/fastrpc.h |  15 +++
 2 files changed, 213 insertions(+), 11 deletions(-)

-- 
2.21.0

[PATCH v3 1/5] misc: fastrpc: add mmap/unmap support

[Srinivas Kandagatla]

From: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>

Support the allocation/deallocation of buffers mapped to the DSP.

When the memory mapped to the DSP at process creation is not enough,
the fastrpc library can extend it at runtime. This avoids having to do
large preallocations by default.

Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>
Reviewed-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
---
 drivers/misc/fastrpc.c      | 181 ++++++++++++++++++++++++++++++++++++
 include/uapi/misc/fastrpc.h |  15 +++
 2 files changed, 196 insertions(+)

diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index 47ae84afac2e..aa1249bb581a 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -34,6 +34,7 @@
 #define FASTRPC_CTXID_MASK (0xFF0)
 #define INIT_FILELEN_MAX (64 * 1024 * 1024)
 #define FASTRPC_DEVICE_NAME	"fastrpc"
+#define ADSP_MMAP_ADD_PAGES 0x1000
 
 /* Retrives number of input buffers from the scalars parameter */
 #define REMOTE_SCALARS_INBUFS(sc)	(((sc) >> 16) & 0x0ff)
@@ -66,6 +67,8 @@
 /* Remote Method id table */
 #define FASTRPC_RMID_INIT_ATTACH	0
 #define FASTRPC_RMID_INIT_RELEASE	1
+#define FASTRPC_RMID_INIT_MMAP		4
+#define FASTRPC_RMID_INIT_MUNMAP	5
 #define FASTRPC_RMID_INIT_CREATE	6
 #define FASTRPC_RMID_INIT_CREATE_ATTR	7
 #define FASTRPC_RMID_INIT_CREATE_STATIC	8
@@ -89,6 +92,23 @@ struct fastrpc_remote_arg {
 	u64 len;
 };
 
+struct fastrpc_mmap_rsp_msg {
+	u64 vaddr;
+};
+
+struct fastrpc_mmap_req_msg {
+	s32 pgid;
+	u32 flags;
+	u64 vaddr;
+	s32 num;
+};
+
+struct fastrpc_munmap_req_msg {
+	s32 pgid;
+	u64 vaddr;
+	u64 size;
+};
+
 struct fastrpc_msg {
 	int pid;		/* process group id */
 	int tid;		/* thread id */
@@ -123,6 +143,9 @@ struct fastrpc_buf {
 	/* Lock for dma buf attachments */
 	struct mutex lock;
 	struct list_head attachments;
+	/* mmap support */
+	struct list_head node; /* list of user requested mmaps */
+	uintptr_t raddr;
 };
 
 struct fastrpc_dma_buf_attachment {
@@ -192,6 +215,7 @@ struct fastrpc_user {
 	struct list_head user;
 	struct list_head maps;
 	struct list_head pending;
+	struct list_head mmaps;
 
 	struct fastrpc_channel_ctx *cctx;
 	struct fastrpc_session_ctx *sctx;
@@ -269,6 +293,7 @@ static int fastrpc_buf_alloc(struct fastrpc_user *fl, struct device *dev,
 		return -ENOMEM;
 
 	INIT_LIST_HEAD(&buf->attachments);
+	INIT_LIST_HEAD(&buf->node);
 	mutex_init(&buf->lock);
 
 	buf->fl = fl;
@@ -276,6 +301,7 @@ static int fastrpc_buf_alloc(struct fastrpc_user *fl, struct device *dev,
 	buf->phys = 0;
 	buf->size = size;
 	buf->dev = dev;
+	buf->raddr = 0;
 
 	buf->virt = dma_alloc_coherent(dev, buf->size, (dma_addr_t *)&buf->phys,
 				       GFP_KERNEL);
@@ -1130,6 +1156,7 @@ static int fastrpc_device_release(struct inode *inode, struct file *file)
 	struct fastrpc_channel_ctx *cctx = fl->cctx;
 	struct fastrpc_invoke_ctx *ctx, *n;
 	struct fastrpc_map *map, *m;
+	struct fastrpc_buf *buf, *b;
 	unsigned long flags;
 
 	fastrpc_release_current_dsp_process(fl);
@@ -1151,6 +1178,11 @@ static int fastrpc_device_release(struct inode *inode, struct file *file)
 		fastrpc_map_put(map);
 	}
 
+	list_for_each_entry_safe(buf, b, &fl->mmaps, node) {
+		list_del(&buf->node);
+		fastrpc_buf_free(buf);
+	}
+
 	fastrpc_session_free(cctx, fl->sctx);
 	fastrpc_channel_ctx_put(cctx);
 
@@ -1179,6 +1211,7 @@ static int fastrpc_device_open(struct inode *inode, struct file *filp)
 	mutex_init(&fl->mutex);
 	INIT_LIST_HEAD(&fl->pending);
 	INIT_LIST_HEAD(&fl->maps);
+	INIT_LIST_HEAD(&fl->mmaps);
 	INIT_LIST_HEAD(&fl->user);
 	fl->tgid = current->tgid;
 	fl->cctx = cctx;
@@ -1284,6 +1317,148 @@ static int fastrpc_invoke(struct fastrpc_user *fl, char __user *argp)
 	return err;
 }
 
+static int fastrpc_req_munmap_impl(struct fastrpc_user *fl,
+				   struct fastrpc_req_munmap *req)
+{
+	struct fastrpc_invoke_args args[1] = { [0] = { 0 } };
+	struct fastrpc_buf *buf, *b;
+	struct fastrpc_munmap_req_msg req_msg;
+	struct device *dev = fl->sctx->dev;
+	int err;
+	u32 sc;
+
+	spin_lock(&fl->lock);
+	list_for_each_entry_safe(buf, b, &fl->mmaps, node) {
+		if ((buf->raddr == req->vaddrout) && (buf->size == req->size))
+			break;
+		buf = NULL;
+	}
+	spin_unlock(&fl->lock);
+
+	if (!buf) {
+		dev_err(dev, "mmap not in list\n");
+		return -EINVAL;
+	}
+
+	req_msg.pgid = fl->tgid;
+	req_msg.size = buf->size;
+	req_msg.vaddr = buf->raddr;
+
+	args[0].ptr = (u64) (uintptr_t) &req_msg;
+	args[0].length = sizeof(req_msg);
+
+	sc = FASTRPC_SCALARS(FASTRPC_RMID_INIT_MUNMAP, 1, 0);
+	err = fastrpc_internal_invoke(fl, true, FASTRPC_INIT_HANDLE, sc,
+				      &args[0]);
+	if (!err) {
+		dev_dbg(dev, "unmmap\tpt 0x%09lx OK\n", buf->raddr);
+		spin_lock(&fl->lock);
+		list_del(&buf->node);
+		spin_unlock(&fl->lock);
+		fastrpc_buf_free(buf);
+	} else {
+		dev_err(dev, "unmmap\tpt 0x%09lx ERROR\n", buf->raddr);
+	}
+
+	return err;
+}
+
+static int fastrpc_req_munmap(struct fastrpc_user *fl, char __user *argp)
+{
+	struct fastrpc_req_munmap req;
+
+	if (copy_from_user(&req, argp, sizeof(req)))
+		return -EFAULT;
+
+	return fastrpc_req_munmap_impl(fl, &req);
+}
+
+static int fastrpc_req_mmap(struct fastrpc_user *fl, char __user *argp)
+{
+	struct fastrpc_invoke_args args[3] = { [0 ... 2] = { 0 } };
+	struct fastrpc_buf *buf = NULL;
+	struct fastrpc_mmap_req_msg req_msg;
+	struct fastrpc_mmap_rsp_msg rsp_msg;
+	struct fastrpc_req_munmap req_unmap;
+	struct fastrpc_phy_page pages;
+	struct fastrpc_req_mmap req;
+	struct device *dev = fl->sctx->dev;
+	int err;
+	u32 sc;
+
+	if (copy_from_user(&req, argp, sizeof(req)))
+		return -EFAULT;
+
+	if (req.flags != ADSP_MMAP_ADD_PAGES) {
+		dev_err(dev, "flag not supported 0x%x\n", req.flags);
+		return -EINVAL;
+	}
+
+	if (req.vaddrin) {
+		dev_err(dev, "adding user allocated pages is not supported\n");
+		return -EINVAL;
+	}
+
+	err = fastrpc_buf_alloc(fl, fl->sctx->dev, req.size, &buf);
+	if (err) {
+		dev_err(dev, "failed to allocate buffer\n");
+		return err;
+	}
+
+	req_msg.pgid = fl->tgid;
+	req_msg.flags = req.flags;
+	req_msg.vaddr = req.vaddrin;
+	req_msg.num = sizeof(pages);
+
+	args[0].ptr = (u64) (uintptr_t) &req_msg;
+	args[0].length = sizeof(req_msg);
+
+	pages.addr = buf->phys;
+	pages.size = buf->size;
+
+	args[1].ptr = (u64) (uintptr_t) &pages;
+	args[1].length = sizeof(pages);
+
+	args[2].ptr = (u64) (uintptr_t) &rsp_msg;
+	args[2].length = sizeof(rsp_msg);
+
+	sc = FASTRPC_SCALARS(FASTRPC_RMID_INIT_MMAP, 2, 1);
+	err = fastrpc_internal_invoke(fl, true, FASTRPC_INIT_HANDLE, sc,
+				      &args[0]);
+	if (err) {
+		dev_err(dev, "mmap error (len 0x%08llx)\n", buf->size);
+		goto err_invoke;
+	}
+
+	/* update the buffer to be able to deallocate the memory on the DSP */
+	buf->raddr = (uintptr_t) rsp_msg.vaddr;
+
+	/* let the client know the address to use */
+	req.vaddrout = rsp_msg.vaddr;
+
+	spin_lock(&fl->lock);
+	list_add_tail(&buf->node, &fl->mmaps);
+	spin_unlock(&fl->lock);
+
+	if (copy_to_user((void __user *)argp, &req, sizeof(req))) {
+		/* unmap the memory and release the buffer */
+		req_unmap.vaddrout = buf->raddr;
+		req_unmap.size = buf->size;
+		fastrpc_req_munmap_impl(fl, &req_unmap);
+		return -EFAULT;
+	}
+
+	dev_dbg(dev, "mmap\t\tpt 0x%09lx OK [len 0x%08llx]\n",
+		buf->raddr, buf->size);
+
+	return 0;
+
+err_invoke:
+	fastrpc_buf_free(buf);
+
+	return err;
+}
+
 static long fastrpc_device_ioctl(struct file *file, unsigned int cmd,
 				 unsigned long arg)
 {
@@ -1304,6 +1479,12 @@ static long fastrpc_device_ioctl(struct file *file, unsigned int cmd,
 	case FASTRPC_IOCTL_ALLOC_DMA_BUFF:
 		err = fastrpc_dmabuf_alloc(fl, argp);
 		break;
+	case FASTRPC_IOCTL_MMAP:
+		err = fastrpc_req_mmap(fl, argp);
+		break;
+	case FASTRPC_IOCTL_MUNMAP:
+		err = fastrpc_req_munmap(fl, argp);
+		break;
 	default:
 		err = -ENOTTY;
 		break;
diff --git a/include/uapi/misc/fastrpc.h b/include/uapi/misc/fastrpc.h
index fb792e882cef..07de2b7aac85 100644
--- a/include/uapi/misc/fastrpc.h
+++ b/include/uapi/misc/fastrpc.h
@@ -10,6 +10,8 @@
 #define FASTRPC_IOCTL_INVOKE		_IOWR('R', 3, struct fastrpc_invoke)
 #define FASTRPC_IOCTL_INIT_ATTACH	_IO('R', 4)
 #define FASTRPC_IOCTL_INIT_CREATE	_IOWR('R', 5, struct fastrpc_init_create)
+#define FASTRPC_IOCTL_MMAP              _IOWR('R', 6, struct fastrpc_req_mmap)
+#define FASTRPC_IOCTL_MUNMAP            _IOWR('R', 7, struct fastrpc_req_munmap)
 
 struct fastrpc_invoke_args {
 	__u64 ptr;
@@ -38,4 +40,17 @@ struct fastrpc_alloc_dma_buf {
 	__u64 size;	/* size */
 };
 
+struct fastrpc_req_mmap {
+	__s32 fd;
+	__u32 flags;	/* flags for dsp to map with */
+	__u64 vaddrin;	/* optional virtual address */
+	__u64 size;	/* size */
+	__u64 vaddrout;	/* dsp virtual address */
+};
+
+struct fastrpc_req_munmap {
+	__u64 vaddrout;	/* address to unmap */
+	__u64 size;	/* size */
+};
+
 #endif /* __QCOM_FASTRPC_H__ */
-- 
2.21.0

[PATCH v3 2/5] misc: fastrpc: fix memory leak from miscdev->name

[Srinivas Kandagatla]

Fix a memory leak in miscdev->name by using devm_variant

Orignally reported by kmemleak:
    [<ffffff80088b74d8>] kmemleak_alloc+0x50/0x84
    [<ffffff80081e015c>] __kmalloc_track_caller+0xe8/0x168
    [<ffffff8008371ab0>] kvasprintf+0x78/0x100
    [<ffffff8008371c6c>] kasprintf+0x50/0x74
    [<ffffff8008507f2c>] fastrpc_rpmsg_probe+0xd8/0x20c
    [<ffffff80086b63b4>] rpmsg_dev_probe+0xa8/0x148
    [<ffffff80084de50c>] really_probe+0x208/0x248
    [<ffffff80084de2dc>] driver_probe_device+0x98/0xc0
    [<ffffff80084dec6c>] __device_attach_driver+0x9c/0xac
    [<ffffff80084dca8c>] bus_for_each_drv+0x60/0x8c
    [<ffffff80084de64c>] __device_attach+0x8c/0x100
    [<ffffff80084de6e0>] device_initial_probe+0x20/0x28
    [<ffffff80084dcbd0>] bus_probe_device+0x34/0x7c
    [<ffffff80084da32c>] device_add+0x420/0x498
    [<ffffff80084da680>] device_register+0x24/0x2c

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
---
 drivers/misc/fastrpc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index aa1249bb581a..c0fe3419c685 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -1610,8 +1610,8 @@ static int fastrpc_rpmsg_probe(struct rpmsg_device *rpdev)
 		return -ENOMEM;
 
 	data->miscdev.minor = MISC_DYNAMIC_MINOR;
-	data->miscdev.name = kasprintf(GFP_KERNEL, "fastrpc-%s",
-				domains[domain_id]);
+	data->miscdev.name = devm_kasprintf(rdev, GFP_KERNEL, "fastrpc-%s",
+					    domains[domain_id]);
 	data->miscdev.fops = &fastrpc_fops;
 	err = misc_register(&data->miscdev);
 	if (err)
-- 
2.21.0

[PATCH v3 3/5] misc: fastrpc: do not interrupt kernel calls

[Srinivas Kandagatla]

From: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>

the DSP firmware requires some calls to be held until processing has
completed: this is to guarantee that memory continues to be
accessible.

Nevertheless, the fastrpc driver chooses not support the case were
requests need to be held for unbounded amounts of time. If such a
use-case becomes necessary, this timeout will need to be revisited.

Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>
Reviewed-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
---
 drivers/misc/fastrpc.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index c0fe3419c685..666c431380ce 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -959,8 +959,13 @@ static int fastrpc_internal_invoke(struct fastrpc_user *fl,  u32 kernel,
 	if (err)
 		goto bail;
 
-	/* Wait for remote dsp to respond or time out */
-	err = wait_for_completion_interruptible(&ctx->work);
+	if (kernel) {
+		if (!wait_for_completion_timeout(&ctx->work, 10 * HZ))
+			err = -ETIMEDOUT;
+	} else {
+		err = wait_for_completion_interruptible(&ctx->work);
+	}
+
 	if (err)
 		goto bail;
 
-- 
2.21.0

[PATCH v3 4/5] misc: fastrpc: handle interrupted contexts

[Srinivas Kandagatla]

From: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>

Buffers owned by a context that has been interrupted either by a
signal or a timeout might still be being accessed by the DSP.

delegate returning the associated memory to a later time when the
device is released.

Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>
Reviewed-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
---
 drivers/misc/fastrpc.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index 666c431380ce..eef2cdc00672 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -984,12 +984,13 @@ static int fastrpc_internal_invoke(struct fastrpc_user *fl,  u32 kernel,
 	}
 
 bail:
-	/* We are done with this compute context, remove it from pending list */
-	spin_lock(&fl->lock);
-	list_del(&ctx->node);
-	spin_unlock(&fl->lock);
-	fastrpc_context_put(ctx);
-
+	if (err != -ERESTARTSYS && err != -ETIMEDOUT) {
+		/* We are done with this compute context */
+		spin_lock(&fl->lock);
+		list_del(&ctx->node);
+		spin_unlock(&fl->lock);
+		fastrpc_context_put(ctx);
+	}
 	if (err)
 		dev_dbg(fl->sctx->dev, "Error: Invoke Failed %d\n", err);
 
-- 
2.21.0

[PATCH v3 5/5] misc: fastrpc: revert max init file size back to 2MB

[Srinivas Kandagatla]

From: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>

With the integration of the mmap/unmap functionality, it is no longer
necessary to allow large memory allocations upfront since they can be
handled during runtime.

Tested on QCS404 with CDSP Neural Processing test suite.

Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>
Reviewed-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
---
 drivers/misc/fastrpc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index eef2cdc00672..b6420aae45b9 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -32,7 +32,7 @@
 #define FASTRPC_CTX_MAX (256)
 #define FASTRPC_INIT_HANDLE	1
 #define FASTRPC_CTXID_MASK (0xFF0)
-#define INIT_FILELEN_MAX (64 * 1024 * 1024)
+#define INIT_FILELEN_MAX (2 * 1024 * 1024)
 #define FASTRPC_DEVICE_NAME	"fastrpc"
 #define ADSP_MMAP_ADD_PAGES 0x1000
 
-- 
2.21.0