Re: [PATCH v5 6/6] staging: vc04_services: vchiq: Register devices with a custom bus_type

From: Laurent Pinchart
Date: Fri Jan 20 2023 - 10:19:05 EST


Hi Umang,

On Fri, Jan 20, 2023 at 04:26:00PM +0530, Umang Jain wrote:
> On 1/20/23 7:22 AM, Laurent Pinchart wrote:
> > On Thu, Jan 19, 2023 at 05:25:03PM +0530, Umang Jain wrote:
> >> The devices that the vchiq interface registers(bcm2835-audio,
> >
> > Missing space before '('.
> >
> >> bcm2835-camera) are implemented and exposed by the VC04 firmware.
> >> The device tree describes the VC04 itself with the resources
> >> required to communicate with it through a mailbox interface. However,
> >> the vchiq interface registers these devices as platform devices. This
> >> also means the specific drivers for these devices are also getting
> >
> > Drop one of the two "also".
> >
> >> registered as platform drivers. This is not correct and a blatant
> >> abuse of platform device/driver.
> >>
> >> Replace the platform device/driver model with a standard device driver
> >> model. A custom bus_type, vchiq_bus_type, is created in the vchiq
> >> interface which matches the devices to their specific device drivers
> >> thereby, establishing driver binding. A struct vchiq_device wraps the
> >> struct device for each device being registered on the bus by the vchiq
> >> interface.
> >>
> >> Each device registered will expose a 'name' read-only device attribute
> >> in sysfs (/sys/bus/vchiq-bus/devices). New devices and drivers can be
> >> added by registering on vchiq_bus_type and adding a corresponding
> >> device name entry in the static list of devices, vchiq_devices. There
> >> is currently no way to enumerate the VCHIQ devices that are available
> >> from the firmware.
> >
> > Greg, I don't know if you've followed the conversation in earlier mail
> > threads, so I'll try to summarize it here.
> >
> > There are two layers involved: the VCHIQ layer, which has two clients
> > (audio and MMAL), and the MMAL layer, which has multiple clients
> > (camera, codec, ISP). The reason for this is that audio and mmal are
> > separate hardware, while camera, codec and ISP share some hardware
> > blocks.
> >
> > The VCHIQ layer provides a mailbox API to its clients to communicate
> > with the firmware, and the MMAL layer provides another API implemented
> > on top of the VCHIQ layer. Neither APIs offer a way to discover devices
> > dynamically (that's not a feature implemented by the firmware). We've
> > decided that implementing two buses would be overkill, so Umang went for
> > a single vchiq_bus_type. The only value it provides is to stop abusing
> > platform_device. That's pretty much it.
> >
> > Given the above explanation, do you still think the additional
> > complexity introduced by the vchiq bus type is worth it (it more or less
> > duplicates a small subset of the platform bus type implementation), and
> > are you fine with a single bus type, even if it doesn't exactly match
> > the firmware layers ?
> >
> >> Signed-off-by: Umang Jain <umang.jain@xxxxxxxxxxxxxxxx>
> >> ---
> >> .../vc04_services/bcm2835-audio/bcm2835.c | 19 ++-
> >> .../bcm2835-camera/bcm2835-camera.c | 17 ++-
> >> .../interface/vchiq_arm/vchiq_arm.c | 121 +++++++++++++++---
> >> .../interface/vchiq_arm/vchiq_arm.h | 1 +
> >> 4 files changed, 117 insertions(+), 41 deletions(-)
> >>
> >> diff --git a/drivers/staging/vc04_services/bcm2835-audio/bcm2835.c b/drivers/staging/vc04_services/bcm2835-audio/bcm2835.c
> >> index 00bc898b0189..9f3af84f5d5d 100644
> >> --- a/drivers/staging/vc04_services/bcm2835-audio/bcm2835.c
> >> +++ b/drivers/staging/vc04_services/bcm2835-audio/bcm2835.c
> >> @@ -1,12 +1,11 @@
> >> // SPDX-License-Identifier: GPL-2.0
> >> /* Copyright 2011 Broadcom Corporation. All rights reserved. */
> >>
> >> -#include <linux/platform_device.h>
> >> -
> >> #include <linux/init.h>
> >> #include <linux/slab.h>
> >> #include <linux/module.h>
> >>
> >> +#include "../interface/vchiq_arm/vchiq_arm.h"
> >> #include "bcm2835.h"
> >>
> >> static bool enable_hdmi;
> >> @@ -268,9 +267,8 @@ static int snd_add_child_devices(struct device *device, u32 numchans)
> >> return 0;
> >> }
> >>
> >> -static int snd_bcm2835_alsa_probe(struct platform_device *pdev)
> >> +static int snd_bcm2835_alsa_probe(struct device *dev)
> >> {
> >> - struct device *dev = &pdev->dev;
> >> int err;
> >>
> >> if (num_channels <= 0 || num_channels > MAX_SUBSTREAMS) {
> >> @@ -292,30 +290,29 @@ static int snd_bcm2835_alsa_probe(struct platform_device *pdev)
> >>
> >> #ifdef CONFIG_PM
> >>
> >> -static int snd_bcm2835_alsa_suspend(struct platform_device *pdev,
> >> +static int snd_bcm2835_alsa_suspend(struct device *pdev,
> >> pm_message_t state)
> >> {
> >> return 0;
> >> }
> >>
> >> -static int snd_bcm2835_alsa_resume(struct platform_device *pdev)
> >> +static int snd_bcm2835_alsa_resume(struct device *pdev)
> >> {
> >> return 0;
> >> }
> >>
> >> #endif
> >>
> >> -static struct platform_driver bcm2835_alsa_driver = {
> >> +static struct device_driver bcm2835_alsa_driver = {
> >> .probe = snd_bcm2835_alsa_probe,
> >> #ifdef CONFIG_PM
> >> .suspend = snd_bcm2835_alsa_suspend,
> >> .resume = snd_bcm2835_alsa_resume,
> >> #endif
> >> - .driver = {
> >> - .name = "bcm2835_audio",
> >> - },
> >> + .name = "bcm2835_audio",
> >> + .bus = &vchiq_bus_type,
> >> };
> >> -module_platform_driver(bcm2835_alsa_driver);
> >> +module_driver(bcm2835_alsa_driver, driver_register, driver_unregister);
> >
> > Shouldn't you create a struct vchiq_device that wraps struct device, a
> > struct vchiq_driver that wraps struct device_driver, and a
> > module_vchiq_driver() macro ? It shouldn't be up to individual drivers
> > to deal with the plumbing.
> >
> >>
> >> MODULE_AUTHOR("Dom Cobley");
> >> MODULE_DESCRIPTION("Alsa driver for BCM2835 chip");
> >> diff --git a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c
> >> index 4f81765912ea..199a49f9ec1e 100644
> >> --- a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c
> >> +++ b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c
> >> @@ -24,8 +24,8 @@
> >> #include <media/v4l2-event.h>
> >> #include <media/v4l2-common.h>
> >> #include <linux/delay.h>
> >> -#include <linux/platform_device.h>
> >>
> >> +#include "../interface/vchiq_arm/vchiq_arm.h"
> >> #include "../vchiq-mmal/mmal-common.h"
> >> #include "../vchiq-mmal/mmal-encodings.h"
> >> #include "../vchiq-mmal/mmal-vchiq.h"
> >> @@ -1841,7 +1841,7 @@ static struct v4l2_format default_v4l2_format = {
> >> .fmt.pix.sizeimage = 1024 * 768,
> >> };
> >>
> >> -static int bcm2835_mmal_probe(struct platform_device *pdev)
> >> +static int bcm2835_mmal_probe(struct device *device)
> >> {
> >> int ret;
> >> struct bcm2835_mmal_dev *dev;
> >> @@ -1896,7 +1896,7 @@ static int bcm2835_mmal_probe(struct platform_device *pdev)
> >> &camera_instance);
> >> ret = v4l2_device_register(NULL, &dev->v4l2_dev);
> >> if (ret) {
> >> - dev_err(&pdev->dev, "%s: could not register V4L2 device: %d\n",
> >> + dev_err(device, "%s: could not register V4L2 device: %d\n",
> >> __func__, ret);
> >> goto free_dev;
> >> }
> >> @@ -1976,7 +1976,7 @@ static int bcm2835_mmal_probe(struct platform_device *pdev)
> >> return ret;
> >> }
> >>
> >> -static int bcm2835_mmal_remove(struct platform_device *pdev)
> >> +static int bcm2835_mmal_remove(struct device *device)
> >> {
> >> int camera;
> >> struct vchiq_mmal_instance *instance = gdev[0]->instance;
> >> @@ -1990,15 +1990,14 @@ static int bcm2835_mmal_remove(struct platform_device *pdev)
> >> return 0;
> >> }
> >>
> >> -static struct platform_driver bcm2835_camera_driver = {
> >> +static struct device_driver bcm2835_camera_driver = {
> >> + .name = "bcm2835-camera",
> >> .probe = bcm2835_mmal_probe,
> >> .remove = bcm2835_mmal_remove,
> >> - .driver = {
> >> - .name = "bcm2835-camera",
> >> - },
> >> + .bus = &vchiq_bus_type,
> >> };
> >>
> >> -module_platform_driver(bcm2835_camera_driver)
> >> +module_driver(bcm2835_camera_driver, driver_register, driver_unregister)
> >>
> >> MODULE_DESCRIPTION("Broadcom 2835 MMAL video capture");
> >> MODULE_AUTHOR("Vincent Sanders");
> >> diff --git a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> >> index 22de23f3af02..86c8e5df7cf6 100644
> >> --- a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> >> +++ b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> >> @@ -12,6 +12,8 @@
> >> #include <linux/cdev.h>
> >> #include <linux/fs.h>
> >> #include <linux/device.h>
> >> +#include <linux/device/bus.h>
> >> +#include <linux/string.h>
> >> #include <linux/mm.h>
> >> #include <linux/highmem.h>
> >> #include <linux/pagemap.h>
> >> @@ -65,9 +67,6 @@ int vchiq_susp_log_level = VCHIQ_LOG_ERROR;
> >> DEFINE_SPINLOCK(msg_queue_spinlock);
> >> struct vchiq_state g_state;
> >>
> >> -static struct platform_device *bcm2835_camera;
> >> -static struct platform_device *bcm2835_audio;
> >> -
> >> struct vchiq_drvdata {
> >> const unsigned int cache_line_size;
> >> struct rpi_firmware *fw;
> >> @@ -132,6 +131,51 @@ struct vchiq_pagelist_info {
> >> unsigned int scatterlist_mapped;
> >> };
> >>
> >> +struct vchiq_device {
> >> + const char *name;
> >> + struct device dev;
> >> +};
> >
> > Ah there we go :-) Move this structure to a header file that drivers can
> > include. I'd name it vchiq_device.h. The code below should go to
> > vchiq_device.c.
> >
> > I would also move the dev field first.
> >
> >> +
> >> +static ssize_t vchiq_dev_show(struct device *dev,
> >> + struct device_attribute *attr, char *buf)
> >> +{
> >> + struct vchiq_device *device = container_of(dev, struct vchiq_device, dev);
> >> +
> >> + return sprintf(buf, "%s", device->name);
> >> +}
> >> +
> >> +static DEVICE_ATTR_RO(vchiq_dev);
> >> +
> >> +static struct attribute *vchiq_dev_attrs[] = {
> >> + &dev_attr_vchiq_dev.attr,
> >> + NULL
> >> +};
> >> +
> >> +ATTRIBUTE_GROUPS(vchiq_dev);
> >> +
> >> +static const struct device_type vchiq_device_type = {
> >> + .groups = vchiq_dev_groups
> >> +};
> >> +
> >> +static int vchiq_bus_type_match(struct device *dev, struct device_driver *drv)
> >> +{
> >> + if (dev->bus == &vchiq_bus_type &&
> >> + strcmp(dev_name(dev), drv->name) == 0)
> >> + return 1;
> >> + return 0;
> >> +}
> >> +
> >> +struct bus_type vchiq_bus_type = {
> >> + .name = "vchiq-bus",
> >> + .match = vchiq_bus_type_match,
> >> +};
> >> +EXPORT_SYMBOL(vchiq_bus_type);
> >
> > EXPORT_SYMBOL_GPL ?
> >
> >> +
> >> +static const char *const vchiq_devices[] = {
> >> + "bcm2835_audio",
> >> + "bcm2835-camera",
> >> +};
> >
> > This however should stay in this file.
> >
> >> +
> >> static void __iomem *g_regs;
> >> /* This value is the size of the L2 cache lines as understood by the
> >> * VPU firmware, which determines the required alignment of the
> >> @@ -1763,26 +1807,52 @@ static const struct of_device_id vchiq_of_match[] = {
> >> };
> >> MODULE_DEVICE_TABLE(of, vchiq_of_match);
> >>
> >> -static struct platform_device *
> >> +static void
> >> +vchiq_release_device(struct device *dev)
> >> +{
> >> + struct vchiq_device *device;
> >> +
> >> + device = container_of(dev, struct vchiq_device, dev);
> >> + kfree(device);
> >> +}
> >> +
> >> +static int
> >> vchiq_register_child(struct platform_device *pdev, const char *name)
> >
> > Pass a struct device * for the first argument, you don't need a platform
> > device. I'd also name the function vchiq_register_device, and rename the
> > pdev parameter to parent.
> >
> >> {
> >> - struct platform_device_info pdevinfo;
> >> - struct platform_device *child;
> >> + struct vchiq_device *device = NULL;
> >> + int ret;
> >>
> >> - memset(&pdevinfo, 0, sizeof(pdevinfo));
> >> + device = kzalloc(sizeof(*device), GFP_KERNEL);
> >> + if (!device)
> >> + return -ENOMEM;
> >>
> >> - pdevinfo.parent = &pdev->dev;
> >> - pdevinfo.name = name;
> >> - pdevinfo.id = PLATFORM_DEVID_NONE;
> >> - pdevinfo.dma_mask = DMA_BIT_MASK(32);
> >> + device->name = name;
> >> + device->dev.init_name = name;
> >> + device->dev.parent = &pdev->dev;
> >> + device->dev.bus = &vchiq_bus_type;
> >> + device->dev.type = &vchiq_device_type;
> >> + device->dev.release = vchiq_release_device;
> >> +
> >> + ret = dma_set_mask_and_coherent(&device->dev, DMA_BIT_MASK(32));
> >
> > Do vchiq devices perform DMA ?
>
> As far as I can tell, the devices right now (in the mainline) don't use
> DMA allocation. So it's not relevant right now.
>
> But looking at RPi's BSP - the vc-sm-cma driver does the DMA. So
> probably we want to dma_set_mask_and_coherent() for that only?

I wonder if even that is required. The vc-sm-cma driver uses a platform
device creates as a child of the vchiq, which already has its DMA mask
set to 32-bit.

We can figure this out later when we'll deal with the vc-sm-cma driver.

> >> + if (ret < 0) {
> >> + vchiq_release_device(&device->dev);
> >> + return ret;
> >> + }
> >>
> >> - child = platform_device_register_full(&pdevinfo);
> >> - if (IS_ERR(child)) {
> >> - dev_warn(&pdev->dev, "%s not registered\n", name);
> >> - child = NULL;
> >> + ret = device_register(&device->dev);
> >> + if (ret) {
> >> + put_device(&device->dev);
> >> + return -EINVAL;
> >> }
> >>
> >> - return child;
> >> + return 0;
> >> +}
> >> +
> >> +static int
> >> +vchiq_unregister_child(struct device *dev, void *data)
> >> +{
> >> + device_unregister(dev);
> >> + return 0;
> >> }
> >>
> >> static int vchiq_probe(struct platform_device *pdev)
> >> @@ -1790,7 +1860,7 @@ static int vchiq_probe(struct platform_device *pdev)
> >> struct device_node *fw_node;
> >> const struct of_device_id *of_id;
> >> struct vchiq_drvdata *drvdata;
> >> - int err;
> >> + int i, err;
> >
> > i can be an unsigned int.
> >
> >>
> >> of_id = of_match_node(vchiq_of_match, pdev->dev.of_node);
> >> drvdata = (struct vchiq_drvdata *)of_id->data;
> >> @@ -1832,8 +1902,12 @@ static int vchiq_probe(struct platform_device *pdev)
> >> goto error_exit;
> >> }
> >>
> >> - bcm2835_camera = vchiq_register_child(pdev, "bcm2835-camera");
> >> - bcm2835_audio = vchiq_register_child(pdev, "bcm2835_audio");
> >> + for (i = 0; i < ARRAY_SIZE(vchiq_devices); i++) {
> >> + err = vchiq_register_child(pdev, vchiq_devices[i]);
> >> + if (!err)
> >> + dev_err(&pdev->dev, "Failed to register %s vchiq device\n",
> >> + vchiq_devices[i]);
> >> + }
> >>
> >> return 0;
> >>
> >> @@ -1845,8 +1919,8 @@ static int vchiq_probe(struct platform_device *pdev)
> >>
> >> static int vchiq_remove(struct platform_device *pdev)
> >> {
> >> - platform_device_unregister(bcm2835_audio);
> >> - platform_device_unregister(bcm2835_camera);
> >> + bus_for_each_dev(&vchiq_bus_type, NULL, NULL, vchiq_unregister_child);
> >> +
> >> vchiq_debugfs_deinit();
> >> vchiq_deregister_chrdev();
> >>
> >> @@ -1866,6 +1940,10 @@ static int __init vchiq_driver_init(void)
> >> {
> >> int ret;
> >>
> >> + ret = bus_register(&vchiq_bus_type);
> >> + if (ret)
> >> + pr_err("Failed to register %s\n", vchiq_bus_type.name);
> >
> > This should be a fatal error, you should return an error value.
> >
> >> +
> >> ret = platform_driver_register(&vchiq_driver);
> >> if (ret)
> >> pr_err("Failed to register vchiq driver\n");
> >> @@ -1876,6 +1954,7 @@ module_init(vchiq_driver_init);
> >>
> >> static void __exit vchiq_driver_exit(void)
> >> {
> >> + bus_unregister(&vchiq_bus_type);
> >> platform_driver_unregister(&vchiq_driver);
> >> }
> >> module_exit(vchiq_driver_exit);
> >> diff --git a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.h b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.h
> >> index 2fb31f9b527f..98c3af32774a 100644
> >> --- a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.h
> >> +++ b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.h
> >> @@ -81,6 +81,7 @@ extern int vchiq_susp_log_level;
> >>
> >> extern spinlock_t msg_queue_spinlock;
> >> extern struct vchiq_state g_state;
> >> +extern struct bus_type vchiq_bus_type;
> >>
> >> extern struct vchiq_state *
> >> vchiq_get_state(void);

--
Regards,

Laurent Pinchart