UIO Driver Info

619 views
Skip to first unread message

Nick Burkitt

unread,
Jul 1, 2016, 11:00:34 PM7/1/16
to krtkl-sni...@googlegroups.com
And a final note before the long weekend...

UIO Magic

The snickerdoodle prebuilt Linux image includes the uio_pdrv_genirq driver built into the kernel. This allows you talk to memory-mapped hardware, and even handle interrupts, from user-space code.

I've installed vsftpd on my snickerdoodle to facilitate transferring files.

Install package

# apt-get install vsftpd

Edit /etc/vsftpd.conf, and uncomment this line.

#write_enable=YES

Restart service

# service vsftpd restart

Using the UIO Driver

Add a description of your device to the devicetree source. One quick and dirty way is just to decompile the dtb file from the boot image provided by krtkl, then modify and recompile it. The reader is encouraged to post better ways to do it.

There's no man page for dtc, but you can use

$ dtc --help

to see the syntax of the commands.

Add Your Device to the Devicetree BLOB

Reverse-compile the devicetree BLOB.


$ dtc -I dtb -O dts -o reversed.dts devicetree.dtb

Edit (a copy of, so you won't have to do it again) the reversed devicetree source.

$ cp reversed.dts uio.dts

$ vi uio.dts

Add your devicetree node. This is what mine looked like. The reverse-compiled source has lost its symbols, such as that for the interrupt parent device, so look to see what other nodes are using. Of course if you know what you're doing, you may have a better idea.

counters@0x43C00000 {

    compatible = "nix-uio";      // your unique id

    reg = < 0x43C00000 0x1000 >; // device address and size

    interrupts = <0 57 1>;       // interrupt number

    interrupt-parent = <0x3>;    // see what the other nodes use

};

Recompile the modified devicetree source. I set DTS_SOURCE for convenience.

$ export DTS_SOURCE=~/git/Snickerdoodle/snickerdoodle-linux/arch/arm/boot/dts

$ dtc -I dts -i $DTS_SOURCE -O dtb -o uio.dtb uio.dts

Modify the Boot Configuration

Modify uEnv.txt, which lives in snickerdoodle's BOOT partition, by changing the value of the "bootargs" parameter to include the string "uio_pdrv_genirq.of_id=my-uio" (excluding quotation marks).

You can do this on the board (as root).

Mount the BOOT partition.

# mkdir /BOOT

# mount /dev/mmcblk0p1 /BOOT

Edit /BOOT/uEnv.txt. Add the blue part to the definition of bootargs, of course using the "compatible" string from your devicetree.

bootargs=console=ttyPS0,115200 root=/dev/mmcblk0p2 rw rootwait earlyprintk uio_pdrv_genirq.of_id=nix-uio

Copy your custom .dtb file to the BOOT partition of your snickerdoodle SD card.

Read this post by Bush to see how to manage bistreams and their associated .dtb files.

Unmount /BOOT

# sync

# umount /BOOT

Now reboot, and check for a device named /dev/uioN, where N is an integer, typically 0.

Simple C++ user space driver code is attached as an example. This version doesn't use interrupts. 

Some Links, most of which are rather out-of-date.

http://fpga.org/2013/05/28/how-to-design-and-access-a-memory-mapped-device-part-one/

http://fpga.org/2013/05/28/how-to-design-and-access-a-memory-mapped-device-part-two/

http://alvarom.com/2014/12/17/linux-user-space-drivers-with-interrupts/

https://yurovsky.github.io/2014/10/10/linux-uio-gpio-interrupt/

 

UioDriver.zip

weath...@krtkl.com

unread,
Jul 1, 2016, 11:17:31 PM7/1/16
to snickerdoodle forum
Hi Nick, this is great stuff.  I think it is something especially the example code that we will want to add to the snickerdoodle book although we may rewrite it in C.

The device that you are accessing in the FPGA via UIO, what does it do?  I saw the offsets declared but I was just curious so that I could understand.

It is one of our objectives to get a very very simple end to end example together showing C->UIO->Linux->FPGA->Simple Compute/I/O on snickerdoodle documented.

    interrupts = < 57 >;         // interrupt number

    interrupt-parent = <0x3>;    // see what the other nodes use

};

Recompile the modified devicetree source. I set DTS_SOURCE for convenience.

$ export DTS_SOURCE=~/git/Snickerdoodle/snickerdoodle-linux/arch/arm/boot/dts

$ dtc -I dts -i $DTS_SOURCE -O dtb -o uio.dtb uio.dts

Nick Burkitt

unread,
Jul 5, 2016, 12:24:30 PM7/5/16
to snickerdoodle forum
Hi Jamil.

Feel free to use anything that you feel will help the cause. The device on the other side of the driver does...absolutely nothing. It's my first FPGA IP (I'm the software guy), and I'm quite proud of it, but it's basically just four 32-bit R/W registers.

chrisnr...@gmail.com

unread,
Jul 28, 2016, 2:04:40 PM7/28/16
to snickerdoodle forum

Just to add in case anyone wants a C example of a UIO device driver. Below(tried to attach but did not see option in browser) is a UIO driver I used on another Zynq board for another project that interfaces with a FPGA hardware step generator component.



Something to keep in mind if using multiple UIO devices. The order devices are mapped by Linux(uio0, uio1,etc) will change between boots, so whatever code you have that attaches to the uio device needs to be able to check that it is attaching to the correct device.


Hope this is helpful to someone.

Thanks,
Chris



/*

*

*

* Userspace I/O platform driver

*

* Copyright (C) 2008 Magnus Damm

*

* Based on uio_pdrv.c by Uwe Kleine-Koenig,

* Copyright (C) 2008 by Digi International Inc.

* All rights reserved.

*

* This program is free software; you can redistribute it and/or modify it

* under the terms of the GNU General Public License version 2 as published by

* the Free Software Foundation.

*/

#include <linux/platform_device.h>

#include <linux/uio_driver.h>

#include <linux/spinlock.h>

#include <linux/bitops.h>

#include <linux/module.h>

#include <linux/interrupt.h>

#include <linux/stringify.h>

#include <linux/pm_runtime.h>

#include <linux/slab.h>

#include <linux/of.h>

#include <linux/of_platform.h>

#include <linux/of_address.h>

#define DRIVER_NAME "mk_basic_step"

struct uio_pdrv_genirq_platdata {

struct uio_info *uioinfo;

spinlock_t lock;

unsigned long flags;

struct platform_device *pdev;

};

/* Bits in uio_pdrv_genirq_platdata.flags */

enum {

UIO_IRQ_DISABLED = 1,

};

 

static int uio_pdrv_genirq_open(struct uio_info *info, struct inode *inode)

{

struct uio_pdrv_genirq_platdata *priv = info->priv;

/* Wait until the Runtime PM code has woken up the device */

pm_runtime_get_sync(&priv->pdev->dev);

return 0;

}

static int uio_pdrv_genirq_release(struct uio_info *info, struct inode *inode)

{

struct uio_pdrv_genirq_platdata *priv = info->priv;

/* Tell the Runtime PM code that the device has become idle */

pm_runtime_put_sync(&priv->pdev->dev);

return 0;

}

static irqreturn_t uio_pdrv_genirq_handler(int irq, struct uio_info *dev_info)

{

struct uio_pdrv_genirq_platdata *priv = dev_info->priv;

/* Just disable the interrupt in the interrupt controller, and

* remember the state so we can allow user space to enable it later.

*/

if (!test_and_set_bit(0, &priv->flags))

disable_irq_nosync(irq);

return IRQ_HANDLED;

}

static int uio_pdrv_genirq_irqcontrol(struct uio_info *dev_info, s32 irq_on)

{

struct uio_pdrv_genirq_platdata *priv = dev_info->priv;

unsigned long flags;

/* Allow user space to enable and disable the interrupt

* in the interrupt controller, but keep track of the

* state to prevent per-irq depth damage.

*

* Serialize this operation to support multiple tasks.

*/

spin_lock_irqsave(&priv->lock, flags);

if (irq_on) {

if (test_and_clear_bit(0, &priv->flags))

enable_irq(dev_info->irq);

} else {

if (!test_and_set_bit(0, &priv->flags))

disable_irq(dev_info->irq);

}

spin_unlock_irqrestore(&priv->lock, flags);

return 0;

}

static int uio_pdrv_mk_basic_step_io_probe(struct platform_device *pdev)

{

struct uio_info *uioinfo = pdev->dev.platform_data;

struct uio_pdrv_genirq_platdata *priv;

struct uio_mem *uiomem;

int ret = -EINVAL;

int i;

if (pdev->dev.of_node) {

int irq;

/* alloc uioinfo for one device */

uioinfo = kzalloc(sizeof(*uioinfo), GFP_KERNEL);

if (!uioinfo) {

ret = -ENOMEM;

dev_err(&pdev->dev, "unable to kmalloc\n");

goto bad2;

}

uioinfo->name = pdev->dev.of_node->name;

uioinfo->version = "devicetree";

/* Multiple IRQs are not supported */

irq = platform_get_irq(pdev, 0);

if (irq == -ENXIO)

uioinfo->irq = UIO_IRQ_NONE;

else

uioinfo->irq = irq;

}

if (!uioinfo || !uioinfo->name || !uioinfo->version) {

dev_err(&pdev->dev, "missing platform_data\n");

goto bad0;

}

if (uioinfo->handler || uioinfo->irqcontrol ||

uioinfo->irq_flags & IRQF_SHARED) {

dev_err(&pdev->dev, "interrupt configuration error\n");

goto bad0;

}

priv = kzalloc(sizeof(*priv), GFP_KERNEL);

if (!priv) {

ret = -ENOMEM;

dev_err(&pdev->dev, "unable to kmalloc\n");

goto bad0;

}

priv->uioinfo = uioinfo;

spin_lock_init(&priv->lock);

priv->flags = 0; /* interrupt is enabled to begin with */

priv->pdev = pdev;

if (!uioinfo->irq) {

ret = platform_get_irq(pdev, 0);

if (ret < 0) {

dev_err(&pdev->dev, "setting: uioinfo->irq = UIO_IRQ_NONE \n");

// goto bad0;

uioinfo->irq = UIO_IRQ_NONE;

}

// uioinfo->irq = ret;

}

uiomem = &uioinfo->mem[0];

for (i = 0; i < pdev->num_resources; ++i) {

struct resource *r = &pdev->resource[i];

if (r->flags != IORESOURCE_MEM)

continue;

if (uiomem >= &uioinfo->mem[MAX_UIO_MAPS]) {

dev_warn(&pdev->dev, "device has more than "

__stringify(MAX_UIO_MAPS)

" I/O memory resources.\n");

break;

}

uiomem->memtype = UIO_MEM_PHYS;

uiomem->addr = r->start;

uiomem->size = resource_size(r);

uiomem->name = r->name;

++uiomem;

}

while (uiomem < &uioinfo->mem[MAX_UIO_MAPS]) {

uiomem->size = 0;

++uiomem;

}

/* This driver requires no hardware specific kernel code to handle

* interrupts. Instead, the interrupt handler simply disables the

* interrupt in the interrupt controller. User space is responsible

* for performing hardware specific acknowledge and re-enabling of

* the interrupt in the interrupt controller.

*

* Interrupt sharing is not supported.

*/

uioinfo->handler = uio_pdrv_genirq_handler;

uioinfo->irqcontrol = uio_pdrv_genirq_irqcontrol;

uioinfo->open = uio_pdrv_genirq_open;

uioinfo->release = uio_pdrv_genirq_release;

uioinfo->priv = priv;

/* Enable Runtime PM for this device:

* The device starts in suspended state to allow the hardware to be

* turned off by default. The Runtime PM bus code should power on the

* hardware and enable clocks at open().

*/

pm_runtime_enable(&pdev->dev);

ret = uio_register_device(&pdev->dev, priv->uioinfo);

if (ret) {

dev_err(&pdev->dev, "unable to register uio device\n");

goto bad1;

}

platform_set_drvdata(pdev, priv);

return 0;

bad1:

kfree(priv);

pm_runtime_disable(&pdev->dev);

bad0:

/* kfree uioinfo for OF */

if (pdev->dev.of_node)

kfree(uioinfo);

bad2:

return ret;

}

static int uio_pdrv_mk_basic_step_io_remove(struct platform_device *pdev)

{

struct uio_pdrv_genirq_platdata *priv = platform_get_drvdata(pdev);

uio_unregister_device(priv->uioinfo);

pm_runtime_disable(&pdev->dev);

priv->uioinfo->handler = NULL;

priv->uioinfo->irqcontrol = NULL;

/* kfree uioinfo for OF */

if (pdev->dev.of_node)

kfree(priv->uioinfo);

kfree(priv);

return 0;

}

static int uio_pdrv_genirq_runtime_nop(struct device *dev)

{

/* Runtime PM callback shared between ->runtime_suspend()

* and ->runtime_resume(). Simply returns success.

*

* In this driver pm_runtime_get_sync() and pm_runtime_put_sync()

* are used at open() and release() time. This allows the

* Runtime PM code to turn off power to the device while the

* device is unused, ie before open() and after release().

*

* This Runtime PM callback does not need to save or restore

* any registers since user space is responsbile for hardware

* register reinitialization after open().

*/

return 0;

}

static const struct dev_pm_ops uio_pdrv_genirq_dev_pm_ops = {

.runtime_suspend = uio_pdrv_genirq_runtime_nop,

.runtime_resume = uio_pdrv_genirq_runtime_nop,

};

#ifdef CONFIG_OF

static const struct of_device_id uio_of_genirq_match[] = {

{ .compatible = "xlnx,basic-stepper-1.3", },

{ }

};

MODULE_DEVICE_TABLE(of, uio_of_genirq_match);

#else

# define uio_of_genirq_match NULL

#endif

static struct platform_driver mk_basic_step = {

.probe = uio_pdrv_mk_basic_step_io_probe,

.remove = uio_pdrv_mk_basic_step_io_remove,

.driver = {

.name = DRIVER_NAME,

.owner = THIS_MODULE,

.pm = &uio_pdrv_genirq_dev_pm_ops,

.of_match_table = uio_of_genirq_match,

},

};

module_platform_driver(mk_basic_step);

MODULE_AUTHOR("Chris Rodriguez");

MODULE_DESCRIPTION("Stepper Userspace I/O platform driver with generic IRQ handling developed for Machinekit use");

MODULE_LICENSE("GPL v2");

MODULE_ALIAS("platform:" DRIVER_NAME);

Nick Burkitt

unread,
Aug 1, 2016, 3:31:50 PM8/1/16
to snickerdoodle forum
Having finally gotten around to dealing with interrupts, I found an error in my devicetree source. The interrupt definition requires three cells, not just the one I had used. The first cell is the interrupt type (shared vs. per-processor); the second is the interrupt number, and the third is the trigger/level flags. Documented here.


On Friday, July 1, 2016 at 8:00:34 PM UTC-7, Nick Burkitt wrote:

Nick Burkitt

unread,
Aug 1, 2016, 4:08:03 PM8/1/16
to snickerdoodle forum, chrisnr...@gmail.com
Chris - it looks as if you've duplicated the uio_pdrv_genirq driver source, which pretty much defeats the purpose. The generic UIO driver is, well, generic - it's usable as-is for any memory-mapped device. Just set up the devicetree - no kernel code is required.
I create six devices that way and they are always created in the same order in which they appear in the devicetree source.

Chris Rodriguez

unread,
Aug 1, 2016, 4:38:51 PM8/1/16
to Nick Burkitt, snickerdoodle forum
Nick,
Your correct. I was working in a ver. 3 kernel and the "generic" UIO device had some issues back then(ex. only allowed 1 instance to be created). I guess it has since been fixed in the newer kernels, cool.

Maybe they fixed the uio device creation also then. Did you create 6 instances of the same thing or multiple types. I had multiple stepgens, encoders and rategens when I had the issue. When I would reboot uio0 might be a stepgen or sometimes a rategen or encoder.

Thanks,
Chris

Nick Burkitt

unread,
Aug 1, 2016, 7:52:47 PM8/1/16
to snickerdoodle forum, google.1...@spamgourmet.com, chrisnr...@gmail.com
Hi Chris.

I saw some discussions of the single-instance bug, too. I double-checked the code, though, and it's fixed in 4.4. I'm creating six devices with the same memory range, but different names and interrupt IDs, since UIO only allows one interrupt per device.
 If the UIO driver version you were using wouldn't support more than one instance, it certainly wasn't tested for deterministic device creation order. It's not unreasonable to assume that that behavior was fixed when the single-instance bug was. I'd definitely call it a bug. :-)

-Nick
Reply all
Reply to author
Forward
0 new messages