diff mbox series

[v12,2/4] uacce: add uacce driver

Message ID 1579097568-17542-3-git-send-email-zhangfei.gao@linaro.org (mailing list archive)
State Not Applicable
Delegated to: Herbert Xu
Headers show
Series Add uacce module for Accelerator | expand

Commit Message

Zhangfei Gao Jan. 15, 2020, 2:12 p.m. UTC
From: Kenneth Lee <liguozhu@hisilicon.com>

Uacce (Unified/User-space-access-intended Accelerator Framework) targets to
provide Shared Virtual Addressing (SVA) between accelerators and processes.
So accelerator can access any data structure of the main cpu.
This differs from the data sharing between cpu and io device, which share
only data content rather than address.
Since unified address, hardware and user space of process can share the
same virtual address in the communication.

Uacce create a chrdev for every registration, the queue is allocated to
the process when the chrdev is opened. Then the process can access the
hardware resource by interact with the queue file. By mmap the queue
file space to user space, the process can directly put requests to the
hardware without syscall to the kernel space.

The IOMMU core only tracks mm<->device bonds at the moment, because it
only needs to handle IOTLB invalidation and PASID table entries. However
uacce needs a finer granularity since multiple queues from the same
device can be bound to an mm. When the mm exits, all bound queues must
be stopped so that the IOMMU can safely clear the PASID table entry and
reallocate the PASID.

An intermediate struct uacce_mm links uacce devices and queues.
Note that an mm may be bound to multiple devices but an uacce_mm
structure only ever belongs to a single device, because we don't need
anything more complex (if multiple devices are bound to one mm, then
we'll create one uacce_mm for each bond).

        uacce_device --+-- uacce_mm --+-- uacce_queue
                       |              '-- uacce_queue
                       |
                       '-- uacce_mm --+-- uacce_queue
                                      +-- uacce_queue
                                      '-- uacce_queue

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Kenneth Lee <liguozhu@hisilicon.com>
Signed-off-by: Zaibo Xu <xuzaibo@huawei.com>
Signed-off-by: Zhou Wang <wangzhou1@hisilicon.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
---
 Documentation/ABI/testing/sysfs-driver-uacce |  39 ++
 drivers/misc/Kconfig                         |   1 +
 drivers/misc/Makefile                        |   1 +
 drivers/misc/uacce/Kconfig                   |  13 +
 drivers/misc/uacce/Makefile                  |   2 +
 drivers/misc/uacce/uacce.c                   | 617 +++++++++++++++++++++++++++
 include/linux/uacce.h                        | 161 +++++++
 include/uapi/misc/uacce/uacce.h              |  38 ++
 8 files changed, 872 insertions(+)
 create mode 100644 Documentation/ABI/testing/sysfs-driver-uacce
 create mode 100644 drivers/misc/uacce/Kconfig
 create mode 100644 drivers/misc/uacce/Makefile
 create mode 100644 drivers/misc/uacce/uacce.c
 create mode 100644 include/linux/uacce.h
 create mode 100644 include/uapi/misc/uacce/uacce.h

Comments

Greg Kroah-Hartman Feb. 10, 2020, 11:37 p.m. UTC | #1
On Wed, Jan 15, 2020 at 10:12:46PM +0800, Zhangfei Gao wrote:
> From: Kenneth Lee <liguozhu@hisilicon.com>
> 
> Uacce (Unified/User-space-access-intended Accelerator Framework) targets to
> provide Shared Virtual Addressing (SVA) between accelerators and processes.
> So accelerator can access any data structure of the main cpu.
> This differs from the data sharing between cpu and io device, which share
> only data content rather than address.
> Since unified address, hardware and user space of process can share the
> same virtual address in the communication.
> 
> Uacce create a chrdev for every registration, the queue is allocated to
> the process when the chrdev is opened. Then the process can access the
> hardware resource by interact with the queue file. By mmap the queue
> file space to user space, the process can directly put requests to the
> hardware without syscall to the kernel space.
> 
> The IOMMU core only tracks mm<->device bonds at the moment, because it
> only needs to handle IOTLB invalidation and PASID table entries. However
> uacce needs a finer granularity since multiple queues from the same
> device can be bound to an mm. When the mm exits, all bound queues must
> be stopped so that the IOMMU can safely clear the PASID table entry and
> reallocate the PASID.
> 
> An intermediate struct uacce_mm links uacce devices and queues.
> Note that an mm may be bound to multiple devices but an uacce_mm
> structure only ever belongs to a single device, because we don't need
> anything more complex (if multiple devices are bound to one mm, then
> we'll create one uacce_mm for each bond).
> 
>         uacce_device --+-- uacce_mm --+-- uacce_queue
>                        |              '-- uacce_queue
>                        |
>                        '-- uacce_mm --+-- uacce_queue
>                                       +-- uacce_queue
>                                       '-- uacce_queue
> 
> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> Signed-off-by: Kenneth Lee <liguozhu@hisilicon.com>
> Signed-off-by: Zaibo Xu <xuzaibo@huawei.com>
> Signed-off-by: Zhou Wang <wangzhou1@hisilicon.com>
> Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
> Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>

Looks much saner now, thanks for all of the work on this:

Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Or am I supposed to take this in my tree?  If so, I can, but I need an
ack for the crypto parts.

thanks,

greg k-h
Zhangfei Gao Feb. 11, 2020, 7:59 a.m. UTC | #2
On 2020/2/11 上午7:37, Greg Kroah-Hartman wrote:
> On Wed, Jan 15, 2020 at 10:12:46PM +0800, Zhangfei Gao wrote:
>> From: Kenneth Lee <liguozhu@hisilicon.com>
>>
>> Uacce (Unified/User-space-access-intended Accelerator Framework) targets to
>> provide Shared Virtual Addressing (SVA) between accelerators and processes.
>> So accelerator can access any data structure of the main cpu.
>> This differs from the data sharing between cpu and io device, which share
>> only data content rather than address.
>> Since unified address, hardware and user space of process can share the
>> same virtual address in the communication.
>>
>> Uacce create a chrdev for every registration, the queue is allocated to
>> the process when the chrdev is opened. Then the process can access the
>> hardware resource by interact with the queue file. By mmap the queue
>> file space to user space, the process can directly put requests to the
>> hardware without syscall to the kernel space.
>>
>> The IOMMU core only tracks mm<->device bonds at the moment, because it
>> only needs to handle IOTLB invalidation and PASID table entries. However
>> uacce needs a finer granularity since multiple queues from the same
>> device can be bound to an mm. When the mm exits, all bound queues must
>> be stopped so that the IOMMU can safely clear the PASID table entry and
>> reallocate the PASID.
>>
>> An intermediate struct uacce_mm links uacce devices and queues.
>> Note that an mm may be bound to multiple devices but an uacce_mm
>> structure only ever belongs to a single device, because we don't need
>> anything more complex (if multiple devices are bound to one mm, then
>> we'll create one uacce_mm for each bond).
>>
>>          uacce_device --+-- uacce_mm --+-- uacce_queue
>>                         |              '-- uacce_queue
>>                         |
>>                         '-- uacce_mm --+-- uacce_queue
>>                                        +-- uacce_queue
>>                                        '-- uacce_queue
>>
>> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
>> Signed-off-by: Kenneth Lee <liguozhu@hisilicon.com>
>> Signed-off-by: Zaibo Xu <xuzaibo@huawei.com>
>> Signed-off-by: Zhou Wang <wangzhou1@hisilicon.com>
>> Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
>> Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
> Looks much saner now, thanks for all of the work on this:
>
> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>
> Or am I supposed to take this in my tree?  If so, I can, but I need an
> ack for the crypto parts.
>
>
That's Great, thanks Greg.

For the convenience, I rebase the patchset on 5.6-rc1.
Not sure is there any conflict to crypto tree.
How about just pick the uacce part, patch 1 , 2.
We can resend the crypto part to crypto tree.

Thanks
Herbert Xu Feb. 13, 2020, 9:15 a.m. UTC | #3
On Mon, Feb 10, 2020 at 03:37:11PM -0800, Greg Kroah-Hartman wrote:
>
> Looks much saner now, thanks for all of the work on this:
> 
> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> 
> Or am I supposed to take this in my tree?  If so, I can, but I need an
> ack for the crypto parts.

I can take this series through the crypto tree if that's fine with
you.

Thank,
Zhangfei Gao Feb. 13, 2020, 9:42 a.m. UTC | #4
On 2020/2/13 下午5:15, Herbert Xu wrote:
> On Mon, Feb 10, 2020 at 03:37:11PM -0800, Greg Kroah-Hartman wrote:
>> Looks much saner now, thanks for all of the work on this:
>>
>> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>>
>> Or am I supposed to take this in my tree?  If so, I can, but I need an
>> ack for the crypto parts.
> I can take this series through the crypto tree if that's fine with
> you.

Thanks Herbert
That's a good idea, otherwise there may be build issue if taken separately.

By the way, the latest v13 is on v5.6-rc1
https://lkml.org/lkml/2020/2/11/54

Thanks
Greg Kroah-Hartman Feb. 13, 2020, 1 p.m. UTC | #5
On Thu, Feb 13, 2020 at 05:15:10PM +0800, Herbert Xu wrote:
> On Mon, Feb 10, 2020 at 03:37:11PM -0800, Greg Kroah-Hartman wrote:
> >
> > Looks much saner now, thanks for all of the work on this:
> > 
> > Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > 
> > Or am I supposed to take this in my tree?  If so, I can, but I need an
> > ack for the crypto parts.
> 
> I can take this series through the crypto tree if that's fine with
> you.

Please do, thanks!

greg k-h
Raj, Ashok Feb. 24, 2020, 6:22 p.m. UTC | #6
Hi Kenneth,

sorry for waking up late on this patchset.


On Wed, Jan 15, 2020 at 10:12:46PM +0800, Zhangfei Gao wrote:
[... trimmed]

> +
> +static int uacce_fops_open(struct inode *inode, struct file *filep)
> +{
> +	struct uacce_mm *uacce_mm = NULL;
> +	struct uacce_device *uacce;
> +	struct uacce_queue *q;
> +	int ret = 0;
> +
> +	uacce = xa_load(&uacce_xa, iminor(inode));
> +	if (!uacce)
> +		return -ENODEV;
> +
> +	q = kzalloc(sizeof(struct uacce_queue), GFP_KERNEL);
> +	if (!q)
> +		return -ENOMEM;
> +
> +	mutex_lock(&uacce->mm_lock);
> +	uacce_mm = uacce_mm_get(uacce, q, current->mm);

I think having this at open time is a bit unnatural. Since when a process
does fork, we do not inherit the PASID. Although it inherits the fd
but cannot use the mmaped address in the child.

If you move this to the mmap time, its more natural. The child could
do a mmap() get a new PASID + mmio space to work with the hardware.


> +	mutex_unlock(&uacce->mm_lock);
> +	if (!uacce_mm) {
> +		ret = -ENOMEM;
> +		goto out_with_mem;
> +	}
> +
Zhangfei Gao Feb. 25, 2020, 8:36 a.m. UTC | #7
Hi, Raj

On 2020/2/25 上午2:22, Raj, Ashok wrote:
> Hi Kenneth,
>
> sorry for waking up late on this patchset.
>
> +
> +static int uacce_fops_open(struct inode *inode, struct file *filep)
> +{
> +	struct uacce_mm *uacce_mm = NULL;
> +	struct uacce_device *uacce;
> +	struct uacce_queue *q;
> +	int ret = 0;
> +
> +	uacce = xa_load(&uacce_xa, iminor(inode));
> +	if (!uacce)
> +		return -ENODEV;
> +
> +	q = kzalloc(sizeof(struct uacce_queue), GFP_KERNEL);
> +	if (!q)
> +		return -ENOMEM;
> +
> +	mutex_lock(&uacce->mm_lock);
> +	uacce_mm = uacce_mm_get(uacce, q, current->mm);
> I think having this at open time is a bit unnatural. Since when a process
> does fork, we do not inherit the PASID. Although it inherits the fd
> but cannot use the mmaped address in the child.
>
> If you move this to the mmap time, its more natural. The child could
> do a mmap() get a new PASID + mmio space to work with the hardware.
>
Thanks for the suggestion.
We will consider fork in the next step, may need some time.

Thanks
Jean-Philippe Brucker March 3, 2020, 11:45 a.m. UTC | #8
On Mon, Feb 24, 2020 at 10:22:02AM -0800, Raj, Ashok wrote:
> Hi Kenneth,
> 
> sorry for waking up late on this patchset.
> 
> 
> On Wed, Jan 15, 2020 at 10:12:46PM +0800, Zhangfei Gao wrote:
> [... trimmed]
> 
> > +
> > +static int uacce_fops_open(struct inode *inode, struct file *filep)
> > +{
> > +	struct uacce_mm *uacce_mm = NULL;
> > +	struct uacce_device *uacce;
> > +	struct uacce_queue *q;
> > +	int ret = 0;
> > +
> > +	uacce = xa_load(&uacce_xa, iminor(inode));
> > +	if (!uacce)
> > +		return -ENODEV;
> > +
> > +	q = kzalloc(sizeof(struct uacce_queue), GFP_KERNEL);
> > +	if (!q)
> > +		return -ENOMEM;
> > +
> > +	mutex_lock(&uacce->mm_lock);
> > +	uacce_mm = uacce_mm_get(uacce, q, current->mm);
> 
> I think having this at open time is a bit unnatural. Since when a process
> does fork, we do not inherit the PASID. Although it inherits the fd
> but cannot use the mmaped address in the child.

Both the queue and the PASID are tied to a single address space. When it
disappears, the queue is stopped (zombie state) and the PASID is freed.
The fd is not usable nor recoverable at this point, it's just waiting to
be released.

> If you move this to the mmap time, its more natural. The child could
> do a mmap() get a new PASID + mmio space to work with the hardware.

I like the idea, as it ties the lifetime of the bond to that of the queue
mapping, but I have two small concerns:

* It adds a lot of side-effect to mmap(). In addition to mapping the MMIO
  region it would now create both the bond and the queue. For userspace,
  figuring out why the mmap() fails would be more difficult.

* It forces uacce drivers to implement an mmap() interface, and have MMIO
  regions to share. I suspect it's going to be the norm but at the moment
  it's not mandatory, drivers could just implement ioctls ops.

I guess the main benefit would be reusing an fd after the original address
space dies, but is it a use-case?

I'd rather go one step further in the other direction, declare that an fd
is a queue and is exclusive to an address space, by preventing any
operation (ioctl and mmap) from an mm other than the one that opened the
fd. It's not natural but it'd keep the kernel driver simple as we wouldn't
have to reconfigure the queue during the lifetime of the fd.

Thanks,
Jean
diff mbox series

Patch

diff --git a/Documentation/ABI/testing/sysfs-driver-uacce b/Documentation/ABI/testing/sysfs-driver-uacce
new file mode 100644
index 0000000..ef4003a
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-driver-uacce
@@ -0,0 +1,39 @@ 
+What:           /sys/class/uacce/<dev_name>/api
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Api of the device
+                Can be any string and up to userspace to parse.
+                Application use the api to match the correct driver
+
+What:           /sys/class/uacce/<dev_name>/flags
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Attributes of the device, see UACCE_DEV_xxx flag defined in uacce.h
+
+What:           /sys/class/uacce/<dev_name>/available_instances
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Available instances left of the device
+                Return -ENODEV if uacce_ops get_available_instances is not provided
+
+What:           /sys/class/uacce/<dev_name>/algorithms
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Algorithms supported by this accelerator, separated by new line.
+                Can be any string and up to userspace to parse.
+
+What:           /sys/class/uacce/<dev_name>/region_mmio_size
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Size (bytes) of mmio region queue file
+
+What:           /sys/class/uacce/<dev_name>/region_dus_size
+Date:           Jan 2020
+KernelVersion:  5.6
+Contact:        linux-accelerators@lists.ozlabs.org
+Description:    Size (bytes) of dus region queue file
diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
index 7f0d48f..99e1514 100644
--- a/drivers/misc/Kconfig
+++ b/drivers/misc/Kconfig
@@ -480,4 +480,5 @@  source "drivers/misc/cxl/Kconfig"
 source "drivers/misc/ocxl/Kconfig"
 source "drivers/misc/cardreader/Kconfig"
 source "drivers/misc/habanalabs/Kconfig"
+source "drivers/misc/uacce/Kconfig"
 endmenu
diff --git a/drivers/misc/Makefile b/drivers/misc/Makefile
index c1860d3..9abf292 100644
--- a/drivers/misc/Makefile
+++ b/drivers/misc/Makefile
@@ -56,4 +56,5 @@  obj-$(CONFIG_OCXL)		+= ocxl/
 obj-y				+= cardreader/
 obj-$(CONFIG_PVPANIC)   	+= pvpanic.o
 obj-$(CONFIG_HABANA_AI)		+= habanalabs/
+obj-$(CONFIG_UACCE)		+= uacce/
 obj-$(CONFIG_XILINX_SDFEC)	+= xilinx_sdfec.o
diff --git a/drivers/misc/uacce/Kconfig b/drivers/misc/uacce/Kconfig
new file mode 100644
index 0000000..5e39b60
--- /dev/null
+++ b/drivers/misc/uacce/Kconfig
@@ -0,0 +1,13 @@ 
+config UACCE
+	tristate "Accelerator Framework for User Land"
+	depends on IOMMU_API
+	help
+	  UACCE provides interface for the user process to access the hardware
+	  without interaction with the kernel space in data path.
+
+	  The user-space interface is described in
+	  include/uapi/misc/uacce/uacce.h
+
+	  See Documentation/misc-devices/uacce.rst for more details.
+
+	  If you don't know what to do here, say N.
diff --git a/drivers/misc/uacce/Makefile b/drivers/misc/uacce/Makefile
new file mode 100644
index 0000000..5b4374e
--- /dev/null
+++ b/drivers/misc/uacce/Makefile
@@ -0,0 +1,2 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+obj-$(CONFIG_UACCE) += uacce.o
diff --git a/drivers/misc/uacce/uacce.c b/drivers/misc/uacce/uacce.c
new file mode 100644
index 0000000..ffced4d
--- /dev/null
+++ b/drivers/misc/uacce/uacce.c
@@ -0,0 +1,617 @@ 
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/compat.h>
+#include <linux/dma-mapping.h>
+#include <linux/iommu.h>
+#include <linux/module.h>
+#include <linux/poll.h>
+#include <linux/uacce.h>
+
+static struct class *uacce_class;
+static dev_t uacce_devt;
+static DEFINE_MUTEX(uacce_mutex);
+static DEFINE_XARRAY_ALLOC(uacce_xa);
+
+static int uacce_start_queue(struct uacce_queue *q)
+{
+	int ret = 0;
+
+	mutex_lock(&uacce_mutex);
+
+	if (q->state != UACCE_Q_INIT) {
+		ret = -EINVAL;
+		goto out_with_lock;
+	}
+
+	if (q->uacce->ops->start_queue) {
+		ret = q->uacce->ops->start_queue(q);
+		if (ret < 0)
+			goto out_with_lock;
+	}
+
+	q->state = UACCE_Q_STARTED;
+
+out_with_lock:
+	mutex_unlock(&uacce_mutex);
+
+	return ret;
+}
+
+static int uacce_put_queue(struct uacce_queue *q)
+{
+	struct uacce_device *uacce = q->uacce;
+
+	mutex_lock(&uacce_mutex);
+
+	if (q->state == UACCE_Q_ZOMBIE)
+		goto out;
+
+	if ((q->state == UACCE_Q_STARTED) && uacce->ops->stop_queue)
+		uacce->ops->stop_queue(q);
+
+	if ((q->state == UACCE_Q_INIT || q->state == UACCE_Q_STARTED) &&
+	     uacce->ops->put_queue)
+		uacce->ops->put_queue(q);
+
+	q->state = UACCE_Q_ZOMBIE;
+out:
+	mutex_unlock(&uacce_mutex);
+
+	return 0;
+}
+
+static long uacce_fops_unl_ioctl(struct file *filep,
+				 unsigned int cmd, unsigned long arg)
+{
+	struct uacce_queue *q = filep->private_data;
+	struct uacce_device *uacce = q->uacce;
+
+	switch (cmd) {
+	case UACCE_CMD_START_Q:
+		return uacce_start_queue(q);
+
+	case UACCE_CMD_PUT_Q:
+		return uacce_put_queue(q);
+
+	default:
+		if (!uacce->ops->ioctl)
+			return -EINVAL;
+
+		return uacce->ops->ioctl(q, cmd, arg);
+	}
+}
+
+#ifdef CONFIG_COMPAT
+static long uacce_fops_compat_ioctl(struct file *filep,
+				   unsigned int cmd, unsigned long arg)
+{
+	arg = (unsigned long)compat_ptr(arg);
+
+	return uacce_fops_unl_ioctl(filep, cmd, arg);
+}
+#endif
+
+static int uacce_sva_exit(struct device *dev, struct iommu_sva *handle,
+			  void *data)
+{
+	struct uacce_mm *uacce_mm = data;
+	struct uacce_queue *q;
+
+	/*
+	 * No new queue can be added concurrently because no caller can have a
+	 * reference to this mm. But there may be concurrent calls to
+	 * uacce_mm_put(), so we need the lock.
+	 */
+	mutex_lock(&uacce_mm->lock);
+	list_for_each_entry(q, &uacce_mm->queues, list)
+		uacce_put_queue(q);
+	uacce_mm->mm = NULL;
+	mutex_unlock(&uacce_mm->lock);
+
+	return 0;
+}
+
+static struct iommu_sva_ops uacce_sva_ops = {
+	.mm_exit = uacce_sva_exit,
+};
+
+static struct uacce_mm *uacce_mm_get(struct uacce_device *uacce,
+				     struct uacce_queue *q,
+				     struct mm_struct *mm)
+{
+	struct uacce_mm *uacce_mm = NULL;
+	struct iommu_sva *handle = NULL;
+	int ret;
+
+	lockdep_assert_held(&uacce->mm_lock);
+
+	list_for_each_entry(uacce_mm, &uacce->mm_list, list) {
+		if (uacce_mm->mm == mm) {
+			mutex_lock(&uacce_mm->lock);
+			list_add(&q->list, &uacce_mm->queues);
+			mutex_unlock(&uacce_mm->lock);
+			return uacce_mm;
+		}
+	}
+
+	uacce_mm = kzalloc(sizeof(*uacce_mm), GFP_KERNEL);
+	if (!uacce_mm)
+		return NULL;
+
+	if (uacce->flags & UACCE_DEV_SVA) {
+		/*
+		 * Safe to pass an incomplete uacce_mm, since mm_exit cannot
+		 * fire while we hold a reference to the mm.
+		 */
+		handle = iommu_sva_bind_device(uacce->parent, mm, uacce_mm);
+		if (IS_ERR(handle))
+			goto err_free;
+
+		ret = iommu_sva_set_ops(handle, &uacce_sva_ops);
+		if (ret)
+			goto err_unbind;
+
+		uacce_mm->pasid = iommu_sva_get_pasid(handle);
+		if (uacce_mm->pasid == IOMMU_PASID_INVALID)
+			goto err_unbind;
+	}
+
+	uacce_mm->mm = mm;
+	uacce_mm->handle = handle;
+	INIT_LIST_HEAD(&uacce_mm->queues);
+	mutex_init(&uacce_mm->lock);
+	list_add(&q->list, &uacce_mm->queues);
+	list_add(&uacce_mm->list, &uacce->mm_list);
+
+	return uacce_mm;
+
+err_unbind:
+	if (handle)
+		iommu_sva_unbind_device(handle);
+err_free:
+	kfree(uacce_mm);
+	return NULL;
+}
+
+static void uacce_mm_put(struct uacce_queue *q)
+{
+	struct uacce_mm *uacce_mm = q->uacce_mm;
+
+	lockdep_assert_held(&q->uacce->mm_lock);
+
+	mutex_lock(&uacce_mm->lock);
+	list_del(&q->list);
+	mutex_unlock(&uacce_mm->lock);
+
+	if (list_empty(&uacce_mm->queues)) {
+		if (uacce_mm->handle)
+			iommu_sva_unbind_device(uacce_mm->handle);
+		list_del(&uacce_mm->list);
+		kfree(uacce_mm);
+	}
+}
+
+static int uacce_fops_open(struct inode *inode, struct file *filep)
+{
+	struct uacce_mm *uacce_mm = NULL;
+	struct uacce_device *uacce;
+	struct uacce_queue *q;
+	int ret = 0;
+
+	uacce = xa_load(&uacce_xa, iminor(inode));
+	if (!uacce)
+		return -ENODEV;
+
+	q = kzalloc(sizeof(struct uacce_queue), GFP_KERNEL);
+	if (!q)
+		return -ENOMEM;
+
+	mutex_lock(&uacce->mm_lock);
+	uacce_mm = uacce_mm_get(uacce, q, current->mm);
+	mutex_unlock(&uacce->mm_lock);
+	if (!uacce_mm) {
+		ret = -ENOMEM;
+		goto out_with_mem;
+	}
+
+	q->uacce = uacce;
+	q->uacce_mm = uacce_mm;
+
+	if (uacce->ops->get_queue) {
+		ret = uacce->ops->get_queue(uacce, uacce_mm->pasid, q);
+		if (ret < 0)
+			goto out_with_mm;
+	}
+
+	init_waitqueue_head(&q->wait);
+	filep->private_data = q;
+	q->state = UACCE_Q_INIT;
+
+	return 0;
+
+out_with_mm:
+	mutex_lock(&uacce->mm_lock);
+	uacce_mm_put(q);
+	mutex_unlock(&uacce->mm_lock);
+out_with_mem:
+	kfree(q);
+	return ret;
+}
+
+static int uacce_fops_release(struct inode *inode, struct file *filep)
+{
+	struct uacce_queue *q = filep->private_data;
+	struct uacce_device *uacce = q->uacce;
+
+	uacce_put_queue(q);
+
+	mutex_lock(&uacce->mm_lock);
+	uacce_mm_put(q);
+	mutex_unlock(&uacce->mm_lock);
+
+	kfree(q);
+
+	return 0;
+}
+
+static void uacce_vma_close(struct vm_area_struct *vma)
+{
+	struct uacce_queue *q = vma->vm_private_data;
+	struct uacce_qfile_region *qfr = NULL;
+
+	if (vma->vm_pgoff < UACCE_MAX_REGION)
+		qfr = q->qfrs[vma->vm_pgoff];
+
+	kfree(qfr);
+}
+
+static const struct vm_operations_struct uacce_vm_ops = {
+	.close = uacce_vma_close,
+};
+
+static int uacce_fops_mmap(struct file *filep, struct vm_area_struct *vma)
+{
+	struct uacce_queue *q = filep->private_data;
+	struct uacce_device *uacce = q->uacce;
+	struct uacce_qfile_region *qfr;
+	enum uacce_qfrt type = UACCE_MAX_REGION;
+	int ret = 0;
+
+	if (vma->vm_pgoff < UACCE_MAX_REGION)
+		type = vma->vm_pgoff;
+	else
+		return -EINVAL;
+
+	qfr = kzalloc(sizeof(*qfr), GFP_KERNEL);
+	if (!qfr)
+		return -ENOMEM;
+
+	vma->vm_flags |= VM_DONTCOPY | VM_DONTEXPAND | VM_WIPEONFORK;
+	vma->vm_ops = &uacce_vm_ops;
+	vma->vm_private_data = q;
+	qfr->type = type;
+
+	mutex_lock(&uacce_mutex);
+
+	if (q->state != UACCE_Q_INIT && q->state != UACCE_Q_STARTED) {
+		ret = -EINVAL;
+		goto out_with_lock;
+	}
+
+	if (q->qfrs[type]) {
+		ret = -EEXIST;
+		goto out_with_lock;
+	}
+
+	switch (type) {
+	case UACCE_QFRT_MMIO:
+		if (!uacce->ops->mmap) {
+			ret = -EINVAL;
+			goto out_with_lock;
+		}
+
+		ret = uacce->ops->mmap(q, vma, qfr);
+		if (ret)
+			goto out_with_lock;
+
+		break;
+
+	case UACCE_QFRT_DUS:
+		if (!uacce->ops->mmap) {
+			ret = -EINVAL;
+			goto out_with_lock;
+		}
+
+		ret = uacce->ops->mmap(q, vma, qfr);
+		if (ret)
+			goto out_with_lock;
+		break;
+
+	default:
+		ret = -EINVAL;
+		goto out_with_lock;
+	}
+
+	q->qfrs[type] = qfr;
+	mutex_unlock(&uacce_mutex);
+
+	return ret;
+
+out_with_lock:
+	mutex_unlock(&uacce_mutex);
+	kfree(qfr);
+	return ret;
+}
+
+static __poll_t uacce_fops_poll(struct file *file, poll_table *wait)
+{
+	struct uacce_queue *q = file->private_data;
+	struct uacce_device *uacce = q->uacce;
+
+	poll_wait(file, &q->wait, wait);
+	if (uacce->ops->is_q_updated && uacce->ops->is_q_updated(q))
+		return EPOLLIN | EPOLLRDNORM;
+
+	return 0;
+}
+
+static const struct file_operations uacce_fops = {
+	.owner		= THIS_MODULE,
+	.open		= uacce_fops_open,
+	.release	= uacce_fops_release,
+	.unlocked_ioctl	= uacce_fops_unl_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= uacce_fops_compat_ioctl,
+#endif
+	.mmap		= uacce_fops_mmap,
+	.poll		= uacce_fops_poll,
+};
+
+#define to_uacce_device(dev) container_of(dev, struct uacce_device, dev)
+
+static ssize_t api_show(struct device *dev,
+			struct device_attribute *attr, char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	return sprintf(buf, "%s\n", uacce->api_ver);
+}
+
+static ssize_t flags_show(struct device *dev,
+			  struct device_attribute *attr, char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	return sprintf(buf, "%u\n", uacce->flags);
+}
+
+static ssize_t available_instances_show(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	if (!uacce->ops->get_available_instances)
+		return -ENODEV;
+
+	return sprintf(buf, "%d\n",
+		       uacce->ops->get_available_instances(uacce));
+}
+
+static ssize_t algorithms_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	return sprintf(buf, "%s\n", uacce->algs);
+}
+
+static ssize_t region_mmio_size_show(struct device *dev,
+				     struct device_attribute *attr, char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	return sprintf(buf, "%lu\n",
+		       uacce->qf_pg_num[UACCE_QFRT_MMIO] << PAGE_SHIFT);
+}
+
+static ssize_t region_dus_size_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	return sprintf(buf, "%lu\n",
+		       uacce->qf_pg_num[UACCE_QFRT_DUS] << PAGE_SHIFT);
+}
+
+static DEVICE_ATTR_RO(api);
+static DEVICE_ATTR_RO(flags);
+static DEVICE_ATTR_RO(available_instances);
+static DEVICE_ATTR_RO(algorithms);
+static DEVICE_ATTR_RO(region_mmio_size);
+static DEVICE_ATTR_RO(region_dus_size);
+
+static struct attribute *uacce_dev_attrs[] = {
+	&dev_attr_api.attr,
+	&dev_attr_flags.attr,
+	&dev_attr_available_instances.attr,
+	&dev_attr_algorithms.attr,
+	&dev_attr_region_mmio_size.attr,
+	&dev_attr_region_dus_size.attr,
+	NULL,
+};
+
+static umode_t uacce_dev_is_visible(struct kobject *kobj,
+				    struct attribute *attr, int n)
+{
+	struct device *dev = container_of(kobj, struct device, kobj);
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	if (((attr == &dev_attr_region_mmio_size.attr) &&
+	    (!uacce->qf_pg_num[UACCE_QFRT_MMIO])) ||
+	    ((attr == &dev_attr_region_dus_size.attr) &&
+	    (!uacce->qf_pg_num[UACCE_QFRT_DUS])))
+		return 0;
+
+	return attr->mode;
+}
+
+static struct attribute_group uacce_dev_group = {
+	.is_visible	= uacce_dev_is_visible,
+	.attrs		= uacce_dev_attrs,
+};
+
+__ATTRIBUTE_GROUPS(uacce_dev);
+
+static void uacce_release(struct device *dev)
+{
+	struct uacce_device *uacce = to_uacce_device(dev);
+
+	kfree(uacce);
+}
+
+/**
+ * uacce_alloc() - alloc an accelerator
+ * @parent: pointer of uacce parent device
+ * @interface: pointer of uacce_interface for register
+ *
+ * Returns uacce pointer if success and ERR_PTR if not
+ * Need check returned negotiated uacce->flags
+ */
+struct uacce_device *uacce_alloc(struct device *parent,
+				 struct uacce_interface *interface)
+{
+	unsigned int flags = interface->flags;
+	struct uacce_device *uacce;
+	int ret;
+
+	uacce = kzalloc(sizeof(struct uacce_device), GFP_KERNEL);
+	if (!uacce)
+		return ERR_PTR(-ENOMEM);
+
+	if (flags & UACCE_DEV_SVA) {
+		ret = iommu_dev_enable_feature(parent, IOMMU_DEV_FEAT_SVA);
+		if (ret)
+			flags &= ~UACCE_DEV_SVA;
+	}
+
+	uacce->parent = parent;
+	uacce->flags = flags;
+	uacce->ops = interface->ops;
+
+	ret = xa_alloc(&uacce_xa, &uacce->dev_id, uacce, xa_limit_32b,
+		       GFP_KERNEL);
+	if (ret < 0)
+		goto err_with_uacce;
+
+	INIT_LIST_HEAD(&uacce->mm_list);
+	mutex_init(&uacce->mm_lock);
+	device_initialize(&uacce->dev);
+	uacce->dev.devt = MKDEV(MAJOR(uacce_devt), uacce->dev_id);
+	uacce->dev.class = uacce_class;
+	uacce->dev.groups = uacce_dev_groups;
+	uacce->dev.parent = uacce->parent;
+	uacce->dev.release = uacce_release;
+	dev_set_name(&uacce->dev, "%s-%d", interface->name, uacce->dev_id);
+
+	return uacce;
+
+err_with_uacce:
+	if (flags & UACCE_DEV_SVA)
+		iommu_dev_disable_feature(uacce->parent, IOMMU_DEV_FEAT_SVA);
+	kfree(uacce);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL_GPL(uacce_alloc);
+
+/**
+ * uacce_register() - add the accelerator to cdev and export to user space
+ * @uacce: The initialized uacce device
+ *
+ * Return 0 if register succeeded, or an error.
+ */
+int uacce_register(struct uacce_device *uacce)
+{
+	if (!uacce)
+		return -ENODEV;
+
+	uacce->cdev = cdev_alloc();
+	if (!uacce->cdev)
+		return -ENOMEM;
+
+	uacce->cdev->ops = &uacce_fops;
+	uacce->cdev->owner = THIS_MODULE;
+
+	return cdev_device_add(uacce->cdev, &uacce->dev);
+}
+EXPORT_SYMBOL_GPL(uacce_register);
+
+/**
+ * uacce_remove() - remove the accelerator
+ * @uacce: the accelerator to remove
+ */
+void uacce_remove(struct uacce_device *uacce)
+{
+	struct uacce_mm *uacce_mm;
+	struct uacce_queue *q;
+
+	if (!uacce)
+		return;
+
+	/* ensure no open queue remains */
+	mutex_lock(&uacce->mm_lock);
+	list_for_each_entry(uacce_mm, &uacce->mm_list, list) {
+		/*
+		 * We don't take the uacce_mm->lock here. Since we hold the
+		 * device's mm_lock, no queue can be added to or removed from
+		 * this uacce_mm. We may run concurrently with mm_exit, but
+		 * uacce_put_queue() is serialized and iommu_sva_unbind_device()
+		 * waits for the lock that mm_exit is holding.
+		 */
+		list_for_each_entry(q, &uacce_mm->queues, list)
+			uacce_put_queue(q);
+
+		if (uacce->flags & UACCE_DEV_SVA) {
+			iommu_sva_unbind_device(uacce_mm->handle);
+			uacce_mm->handle = NULL;
+		}
+	}
+	mutex_unlock(&uacce->mm_lock);
+
+	/* disable sva now since no opened queues */
+	if (uacce->flags & UACCE_DEV_SVA)
+		iommu_dev_disable_feature(uacce->parent, IOMMU_DEV_FEAT_SVA);
+
+	if (uacce->cdev)
+		cdev_device_del(uacce->cdev, &uacce->dev);
+	xa_erase(&uacce_xa, uacce->dev_id);
+	put_device(&uacce->dev);
+}
+EXPORT_SYMBOL_GPL(uacce_remove);
+
+static int __init uacce_init(void)
+{
+	int ret;
+
+	uacce_class = class_create(THIS_MODULE, UACCE_NAME);
+	if (IS_ERR(uacce_class))
+		return PTR_ERR(uacce_class);
+
+	ret = alloc_chrdev_region(&uacce_devt, 0, MINORMASK, UACCE_NAME);
+	if (ret)
+		class_destroy(uacce_class);
+
+	return ret;
+}
+
+static __exit void uacce_exit(void)
+{
+	unregister_chrdev_region(uacce_devt, MINORMASK);
+	class_destroy(uacce_class);
+}
+
+subsys_initcall(uacce_init);
+module_exit(uacce_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Hisilicon Tech. Co., Ltd.");
+MODULE_DESCRIPTION("Accelerator interface for Userland applications");
diff --git a/include/linux/uacce.h b/include/linux/uacce.h
new file mode 100644
index 0000000..904a461
--- /dev/null
+++ b/include/linux/uacce.h
@@ -0,0 +1,161 @@ 
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _LINUX_UACCE_H
+#define _LINUX_UACCE_H
+
+#include <linux/cdev.h>
+#include <uapi/misc/uacce/uacce.h>
+
+#define UACCE_NAME		"uacce"
+#define UACCE_MAX_REGION	2
+#define UACCE_MAX_NAME_SIZE	64
+
+struct uacce_queue;
+struct uacce_device;
+
+/**
+ * struct uacce_qfile_region - structure of queue file region
+ * @type: type of the region
+ */
+struct uacce_qfile_region {
+	enum uacce_qfrt type;
+};
+
+/**
+ * struct uacce_ops - uacce device operations
+ * @get_available_instances:  get available instances left of the device
+ * @get_queue: get a queue from the device
+ * @put_queue: free a queue to the device
+ * @start_queue: make the queue start work after get_queue
+ * @stop_queue: make the queue stop work before put_queue
+ * @is_q_updated: check whether the task is finished
+ * @mmap: mmap addresses of queue to user space
+ * @ioctl: ioctl for user space users of the queue
+ */
+struct uacce_ops {
+	int (*get_available_instances)(struct uacce_device *uacce);
+	int (*get_queue)(struct uacce_device *uacce, unsigned long arg,
+			 struct uacce_queue *q);
+	void (*put_queue)(struct uacce_queue *q);
+	int (*start_queue)(struct uacce_queue *q);
+	void (*stop_queue)(struct uacce_queue *q);
+	int (*is_q_updated)(struct uacce_queue *q);
+	int (*mmap)(struct uacce_queue *q, struct vm_area_struct *vma,
+		    struct uacce_qfile_region *qfr);
+	long (*ioctl)(struct uacce_queue *q, unsigned int cmd,
+		      unsigned long arg);
+};
+
+/**
+ * struct uacce_interface - interface required for uacce_register()
+ * @name: the uacce device name.  Will show up in sysfs
+ * @flags: uacce device attributes
+ * @ops: pointer to the struct uacce_ops
+ */
+struct uacce_interface {
+	char name[UACCE_MAX_NAME_SIZE];
+	unsigned int flags;
+	const struct uacce_ops *ops;
+};
+
+enum uacce_q_state {
+	UACCE_Q_ZOMBIE = 0,
+	UACCE_Q_INIT,
+	UACCE_Q_STARTED,
+};
+
+/**
+ * struct uacce_queue
+ * @uacce: pointer to uacce
+ * @priv: private pointer
+ * @wait: wait queue head
+ * @list: index into uacce_mm
+ * @uacce_mm: the corresponding mm
+ * @qfrs: pointer of qfr regions
+ * @state: queue state machine
+ */
+struct uacce_queue {
+	struct uacce_device *uacce;
+	void *priv;
+	wait_queue_head_t wait;
+	struct list_head list;
+	struct uacce_mm *uacce_mm;
+	struct uacce_qfile_region *qfrs[UACCE_MAX_REGION];
+	enum uacce_q_state state;
+};
+
+/**
+ * struct uacce_device
+ * @algs: supported algorithms
+ * @api_ver: api version
+ * @ops: pointer to the struct uacce_ops
+ * @qf_pg_num: page numbers of the queue file regions
+ * @parent: pointer to the parent device
+ * @is_vf: whether virtual function
+ * @flags: uacce attributes
+ * @dev_id: id of the uacce device
+ * @cdev: cdev of the uacce
+ * @dev: dev of the uacce
+ * @priv: private pointer of the uacce
+ * @mm_list: list head of uacce_mm->list
+ * @mm_lock: lock for mm_list
+ */
+struct uacce_device {
+	const char *algs;
+	const char *api_ver;
+	const struct uacce_ops *ops;
+	unsigned long qf_pg_num[UACCE_MAX_REGION];
+	struct device *parent;
+	bool is_vf;
+	u32 flags;
+	u32 dev_id;
+	struct cdev *cdev;
+	struct device dev;
+	void *priv;
+	struct list_head mm_list;
+	struct mutex mm_lock;
+};
+
+/**
+ * struct uacce_mm - keep track of queues bound to a process
+ * @list: index into uacce_device
+ * @queues: list of queues
+ * @mm: the mm struct
+ * @lock: protects the list of queues
+ * @pasid: pasid of the uacce_mm
+ * @handle: iommu_sva handle return from iommu_sva_bind_device
+ */
+struct uacce_mm {
+	struct list_head list;
+	struct list_head queues;
+	struct mm_struct *mm;
+	struct mutex lock;
+	int pasid;
+	struct iommu_sva *handle;
+};
+
+#if IS_ENABLED(CONFIG_UACCE)
+
+struct uacce_device *uacce_alloc(struct device *parent,
+				 struct uacce_interface *interface);
+int uacce_register(struct uacce_device *uacce);
+void uacce_remove(struct uacce_device *uacce);
+
+#else /* CONFIG_UACCE */
+
+static inline
+struct uacce_device *uacce_alloc(struct device *parent,
+				 struct uacce_interface *interface)
+{
+	return ERR_PTR(-ENODEV);
+}
+
+static inline int uacce_register(struct uacce_device *uacce)
+{
+	return -EINVAL;
+}
+
+static inline void uacce_remove(struct uacce_device *uacce) {}
+
+#endif /* CONFIG_UACCE */
+
+#endif /* _LINUX_UACCE_H */
diff --git a/include/uapi/misc/uacce/uacce.h b/include/uapi/misc/uacce/uacce.h
new file mode 100644
index 0000000..cc71856
--- /dev/null
+++ b/include/uapi/misc/uacce/uacce.h
@@ -0,0 +1,38 @@ 
+/* SPDX-License-Identifier: GPL-2.0+ WITH Linux-syscall-note */
+#ifndef _UAPIUUACCE_H
+#define _UAPIUUACCE_H
+
+#include <linux/types.h>
+#include <linux/ioctl.h>
+
+/*
+ * UACCE_CMD_START_Q: Start queue
+ */
+#define UACCE_CMD_START_Q	_IO('W', 0)
+
+/*
+ * UACCE_CMD_PUT_Q:
+ * User actively stop queue and free queue resource immediately
+ * Optimization method since close fd may delay
+ */
+#define UACCE_CMD_PUT_Q		_IO('W', 1)
+
+/*
+ * UACCE Device flags:
+ * UACCE_DEV_SVA: Shared Virtual Addresses
+ *		  Support PASID
+ *		  Support device page faults (PCI PRI or SMMU Stall)
+ */
+#define UACCE_DEV_SVA		BIT(0)
+
+/**
+ * enum uacce_qfrt: queue file region type
+ * @UACCE_QFRT_MMIO: device mmio region
+ * @UACCE_QFRT_DUS: device user share region
+ */
+enum uacce_qfrt {
+	UACCE_QFRT_MMIO = 0,
+	UACCE_QFRT_DUS = 1,
+};
+
+#endif