Re: [PATCHv2 2/3] ARM: drm: Intel FPGA VIP Frame Buffer II drm driver

On Wed, 2017-05-03 at 13:34 -0700, Eric Anholt wrote:

hean.loong.ong@intel.com writes:

...

From: "Ong, Hean Loong" hean.loong.ong@intel.com

Driver for Intel FPGA Video and Image Processing Suite Frame Buffer II. The driver only supports the Intel Arria10 devkit and its variants. This driver can be either loaded staticlly or in modules. The OF device tree binding is located at: Documentation/devicetree/bindings/display/altr,vip-fb2.txt

Signed-off-by: Ong, Hean Loong hean.loong.ong@intel.com

I'm not sure if this driver is going to make sense as-is, if it doesn't actually do display of planes from system memory.  But in case I was wrong, here's my review:

...

diff --git a/drivers/gpu/drm/ivip/intel_vip_conn.c b/drivers/gpu/drm/ivip/intel_vip_conn.c new file mode 100644 index 0000000..499d3b4 --- /dev/null +++ b/drivers/gpu/drm/ivip/intel_vip_conn.c @@ -0,0 +1,96 @@ +/*

• • intel_vip_conn.c -- Intel Video and Image Processing(VIP)
• • Frame Buffer II driver
• • This driver supports the Intel VIP Frame Reader component.
• • More info on the hardware can be found in the Intel Video
• • and Image Processing Suite User Guide at this address
• • http://www.altera.com/literature/ug/ug_vip.pdf.
• • This program is free software; you can redistribute it and/or

modify it

• • under the terms and conditions of the GNU General Public

License,

• • version 2, as published by the Free Software Foundation.
• • This program is distributed in the hope it will be useful, but

WITHOUT

• • ANY WARRANTY; without even the implied warranty of

MERCHANTABILITY or

• • FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public

License for

• • more details.
• • Authors:
• • Ong, Hean-Loong hean.loong.ong@intel.com
• */

+#include <linux/init.h> +#include <linux/kernel.h> +#include <drm/drm_atomic_helper.h> +#include <drm/drm_crtc_helper.h> +#include <drm/drm_encoder_slave.h> +#include <drm/drm_plane_helper.h>

+static enum drm_connector_status +intelvipfb_drm_connector_detect(struct drm_connector *connector, bool force) +{

• return connector_status_connected;

+}

+static void intelvipfb_drm_connector_destroy(struct drm_connector *connector) +{

• drm_connector_unregister(connector);
• drm_connector_cleanup(connector);

+}

+static const struct drm_connector_funcs intelvipfb_drm_connector_funcs = {

• .dpms = drm_helper_connector_dpms,
• .reset = drm_atomic_helper_connector_reset,
• .detect = intelvipfb_drm_connector_detect,
• .fill_modes = drm_helper_probe_single_connector_modes,
• .destroy = intelvipfb_drm_connector_destroy,
• .atomic_duplicate_state =

drm_atomic_helper_connector_duplicate_state,

• .atomic_destroy_state =

drm_atomic_helper_connector_destroy_state, +};

+static int intelvipfb_drm_connector_get_modes(struct drm_connector *connector) +{

• struct drm_device *drm = connector->dev;
• int count;
• count = drm_add_modes_noedid(connector, drm-

...

mode_config.max_width,

• 		     drm->mode_config.max_height);
  

• drm_set_preferred_mode(connector, drm-

...

mode_config.max_width,

• 	       drm->mode_config.max_height);
  

• return count;

+}

You're adding a bunch of modes here, but I don't see anywhere in the driver that you change the resolution being scanned out.

If you can't change resolution, then I'd probably use drm_gtf_mode() or something to generate just the one mode (do you have a vrefresh for it?).  Also, in the simple display pipe's atomic_check, make sure that the mode chosen is the width/height you can support.

I can see the point in the proposed approach.Based on the comment for get_modes in drm_connector_helper_funcs the recommended approach was using drm_add_modes_noedid and drm_set_preferred_mode. Not many drivers uses the drm_gtf_mode directly therefore I am not inclined to this approach.

...

+static const struct drm_connector_helper_funcs +intelvipfb_drm_connector_helper_funcs = {

• .get_modes = intelvipfb_drm_connector_get_modes,

+};

+struct drm_connector * +intelvipfb_conn_setup(struct drm_device *drm) +{

• struct drm_connector *conn;
• int ret;
• conn = devm_kzalloc(drm->dev, sizeof(*conn), GFP_KERNEL);
• if (IS_ERR(conn))

• return NULL;
  

• ret = drm_connector_init(drm, conn,

&intelvipfb_drm_connector_funcs,

• 		 DRM_MODE_CONNECTOR_DisplayPort);
  

Are you actually outputting DisplayPort (https://en.wikipedia.org/wiki/DisplayPort)?

You probably want something else from the DRM_MODE_CONNECTOR list, or maybe just DRM_MODE_CONNECTOR_Unknown.

...

• if (ret < 0) {

• dev_err(drm->dev, "failed to initialize drm
  

connector\n");

• ret = -ENOMEM;
  

• goto error_connector_cleanup;
  

• }
• drm_connector_helper_add(conn,

&intelvipfb_drm_connector_helper_funcs);

• return conn;

+error_connector_cleanup:

• drm_connector_cleanup(conn);
• return NULL;

+} diff --git a/drivers/gpu/drm/ivip/intel_vip_core.c b/drivers/gpu/drm/ivip/intel_vip_core.c new file mode 100644 index 0000000..4e83343 --- /dev/null +++ b/drivers/gpu/drm/ivip/intel_vip_core.c @@ -0,0 +1,171 @@ +/*

• • intel_vip_core.c -- Intel Video and Image Processing(VIP)
• • Frame Buffer II driver
• • This driver supports the Intel VIP Frame Reader component.
• • More info on the hardware can be found in the Intel Video
• • and Image Processing Suite User Guide at this address
• • http://www.altera.com/literature/ug/ug_vip.pdf.
• • This program is free software; you can redistribute it and/or

modify it

• • under the terms and conditions of the GNU General Public

License,

• • version 2, as published by the Free Software Foundation.
• • This program is distributed in the hope it will be useful, but

WITHOUT

• • ANY WARRANTY; without even the implied warranty of

MERCHANTABILITY or

• • FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public

License for

• • more details.
• • Authors:
• • Ong, Hean-Loong hean.loong.ong@intel.com
• */

+#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h>

+#include <drm/drmP.h> +#include <drm/drm_atomic.h> +#include <drm/drm_atomic_helper.h> +#include <drm/drm_crtc_helper.h> +#include <drm/drm_fb_helper.h> +#include <drm/drm_fb_cma_helper.h> +#include <drm/drm_gem_cma_helper.h> +#include <drm/drm_plane_helper.h> +#include <drm/drm_simple_kms_helper.h>

+#include "intel_vip_drv.h"

+static const u32 fbpriv_formats[] = {

• DRM_FORMAT_XRGB8888,
• DRM_FORMAT_RGB565

+};

You're registering that you support this set of formats, but I don't see anything programming the hardware differently based on the format of the plane.

...

+static void intelvipfb_start_hw(void __iomem *base, resource_size_t addr) +{

• /*
• * The frameinfo variable has to correspond to the size of

the VIP Suite

• * Frame Reader register 7 which will determine the

maximum size used

• * in this frameinfo
• */
• u32 frameinfo;
• frameinfo =

• readl(base + INTELVIPFB_FRAME_READER) &
  

0x00ffffff;

• writel(frameinfo, base + INTELVIPFB_FRAME_INFO);
• writel(addr, base + INTELVIPFB_FRAME_START);
• /* Finally set the control register to 1 to start

streaming */

• writel(1, base + INTELVIPFB_CONTROL);

+}

The addr you're passing in here is from dev->mode_config.fb_base, which is this weird sideband value from drm_fbdev_cma.  It's the wrong value to use if anything else uses the KMS interfaces to change the plane -- you should be using drm_fb_cma_get_gem_addr(drm_simple_display_pipe->plane->state, 0) instead.

The function drm_fb_cma_get_gem_addr(drm_simple_display_pipe, drm_simple_display_pipe->plane->state, 0) is causing an exception in git tag next-20170526. I have confirmed that the pipe is okay. What other configuration should I be looking into before using this function

...

+static void intelvipfb_disable_hw(void __iomem *base) +{

• /* set the control register to 0 to stop streaming */
• writel(0, base + INTELVIPFB_CONTROL);

+}

These two functions should be be called from fbpriv_funcs.enable() and .disable(), not device load/unload.

I agree to this. However in my recent test (git tag next-20170526) it does not seem to be triggered when the driver is loaded

...

+static const struct drm_mode_config_funcs intelvipfb_mode_config_funcs = {

• .fb_create = drm_fb_cma_create,
• .atomic_check = drm_atomic_helper_check,
• .atomic_commit = drm_atomic_helper_commit,

+};

+static struct drm_mode_config_helper_funcs intelvipfb_mode_config_helpers = {

• .atomic_commit_tail = drm_atomic_helper_commit_tail,

+};

+static void intelvipfb_setup_mode_config(struct drm_device *drm) +{

• drm_mode_config_init(drm);
• drm->mode_config.funcs = &intelvipfb_mode_config_funcs;
• drm->mode_config.helper_private =

&intelvipfb_mode_config_helpers; +}

+static int intelvipfb_pipe_prepare_fb(struct drm_simple_display_pipe *pipe,

• 		      struct drm_plane_state
  

*plane_state) +{

• return drm_fb_cma_prepare_fb(&pipe->plane, plane_state);

+}

+static struct drm_simple_display_pipe_funcs fbpriv_funcs = {

• .prepare_fb = intelvipfb_pipe_prepare_fb,

+};

+int intelvipfb_probe(struct device *dev, void __iomem *base) +{

• int retval;
• struct drm_device *drm;
• struct intelvipfb_priv *fbpriv = dev_get_drvdata(dev);
• struct drm_connector *connector;
• dev_set_drvdata(dev, fbpriv);
• drm = fbpriv->drm;
• intelvipfb_setup_mode_config(drm);
• connector = intelvipfb_conn_setup(drm);
• if (!connector) {

• dev_err(drm->dev, "Connector setup failed\n");
  

• goto err_mode_config;
  

• }
• retval = drm_simple_display_pipe_init(drm, &fbpriv->pipe,

• 			      &fbpriv_funcs,
  

• 			      fbpriv_formats,
  

• 			      ARRAY_SIZE(fbpriv_fo
  

rmats),

• 			      connector);
  

• if (retval < 0) {

• dev_err(drm->dev, "Cannot setup simple display
  

pipe\n");

• goto err_mode_config;
  

• }
• fbpriv->fbcma = drm_fbdev_cma_init(drm, PREF_BPP,

• 			   drm-
  

...

mode_config.num_connector);

• if (!fbpriv->fbcma) {

• fbpriv->fbcma = NULL;
  

• dev_err(drm->dev, "Failed to init FB CMA area\n");
  

• goto err_mode_config;
  

• }

On failure, drm_fbdev_cma_init() returns an ERR_PTR, not NULL.

Also, you're passing this PREF_BPP value here, ignoring the altr,bits-per-symbol property.  That seems wrong.