Re: [PATCH v4 10/18] drm/imagination: Add reset controller support for GPU initialization

[Matt Coster <Matt.Coster@imgtec.com>]

[-- Attachment #1.1: Type: text/plain, Size: 5213 bytes --]

On 28/01/2025 19:48, Michal Wilczynski wrote:
> Certain platforms, such as the T-Head TH1520 and Banana Pi BPI-F3,
> require a controlled GPU reset sequence during the power-up procedure
> to ensure proper initialization. Without this reset, the GPU may remain
> in an undefined state, potentially leading to stability or performance
> issues.

Can you reword this to clarify that _all_ IMG Rogue GPUs have a reset
line that participates in the power-up sequence but some SoCs handle
this in silicon and/or firmware without exposing the reset line
directly (as the currently supported TI SoC does).

> 
> This commit integrates a dedicated reset controller within the
> drm/imagination driver. By doing so, the driver can coordinate the
> necessary reset operations as part of the normal GPU bring-up process,
> improving reliability and ensuring that the hardware is ready for
> operation.
> 
> Signed-off-by: Michal Wilczynski <m.wilczynski@samsung.com>
> ---
>  drivers/gpu/drm/imagination/pvr_device.c | 21 +++++++++++++++++++++
>  drivers/gpu/drm/imagination/pvr_device.h |  9 +++++++++
>  drivers/gpu/drm/imagination/pvr_power.c  | 12 +++++++++++-
>  3 files changed, 41 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/imagination/pvr_device.c b/drivers/gpu/drm/imagination/pvr_device.c
> index 1704c0268589..ef73e95157ee 100644
> --- a/drivers/gpu/drm/imagination/pvr_device.c
> +++ b/drivers/gpu/drm/imagination/pvr_device.c
> @@ -25,6 +25,7 @@
>  #include <linux/interrupt.h>
>  #include <linux/platform_device.h>
>  #include <linux/pm_runtime.h>
> +#include <linux/reset.h>
>  #include <linux/slab.h>
>  #include <linux/stddef.h>
>  #include <linux/types.h>
> @@ -120,6 +121,21 @@ static int pvr_device_clk_init(struct pvr_device *pvr_dev)
>  	return 0;
>  }
>  
> +static int pvr_device_reset_init(struct pvr_device *pvr_dev)
> +{
> +	struct drm_device *drm_dev = from_pvr_device(pvr_dev);
> +	struct reset_control *reset;
> +
> +	reset = devm_reset_control_get_optional_exclusive(drm_dev->dev, NULL);
> +	if (IS_ERR(reset))
> +		return dev_err_probe(drm_dev->dev, PTR_ERR(reset),
> +				     "failed to get gpu reset line\n");
> +
> +	pvr_dev->reset = reset;
> +
> +	return 0;
> +}
> +
>  /**
>   * pvr_device_process_active_queues() - Process all queue related events.
>   * @pvr_dev: PowerVR device to check
> @@ -509,6 +525,11 @@ pvr_device_init(struct pvr_device *pvr_dev)
>  	if (err)
>  		return err;
>  
> +	/* Get the reset line for the GPU */
> +	err = pvr_device_reset_init(pvr_dev);
> +	if (err)
> +		return err;
> +
>  	/* Explicitly power the GPU so we can access control registers before the FW is booted. */
>  	err = pm_runtime_resume_and_get(dev);
>  	if (err)
> diff --git a/drivers/gpu/drm/imagination/pvr_device.h b/drivers/gpu/drm/imagination/pvr_device.h
> index 6d0dfacb677b..f6576c08111c 100644
> --- a/drivers/gpu/drm/imagination/pvr_device.h
> +++ b/drivers/gpu/drm/imagination/pvr_device.h
> @@ -131,6 +131,15 @@ struct pvr_device {
>  	 */
>  	struct clk *mem_clk;
>  
> +	/**
> +	 * @reset: Optional reset line.
> +	 *
> +	 * This may be used on some platforms to provide a reset line that needs to be de-asserted
> +	 * after power-up procedure. It would also need to be asserted after the power-down
> +	 * procedure.
> +	 */
> +	struct reset_control *reset;
> +
>  	/** @irq: IRQ number. */
>  	int irq;
>  
> diff --git a/drivers/gpu/drm/imagination/pvr_power.c b/drivers/gpu/drm/imagination/pvr_power.c
> index ba7816fd28ec..e39460d594bd 100644
> --- a/drivers/gpu/drm/imagination/pvr_power.c
> +++ b/drivers/gpu/drm/imagination/pvr_power.c
> @@ -15,6 +15,7 @@
>  #include <linux/mutex.h>
>  #include <linux/platform_device.h>
>  #include <linux/pm_runtime.h>
> +#include <linux/reset.h>
>  #include <linux/timer.h>
>  #include <linux/types.h>
>  #include <linux/workqueue.h>
> @@ -252,6 +253,8 @@ pvr_power_device_suspend(struct device *dev)
>  	clk_disable_unprepare(pvr_dev->sys_clk);
>  	clk_disable_unprepare(pvr_dev->core_clk);
>  
> +	err = reset_control_assert(pvr_dev->reset);
> +
>  err_drm_dev_exit:
>  	drm_dev_exit(idx);
>  
> @@ -282,16 +285,23 @@ pvr_power_device_resume(struct device *dev)
>  	if (err)
>  		goto err_sys_clk_disable;

This is where I'd expect to see the 32 cycle delay that's currently in
P9 ("reset: thead: Add TH1520 reset controller driver").

If it turns out that delay is required in the reset driver, would you be
opposed to adding it here as well? It's a very small amount of time and
would make this codepath more versatile to future reset controllers.

Cheers,
Matt

>  
> +	err = reset_control_deassert(pvr_dev->reset);
> +	if (err)
> +		goto err_mem_clk_disable;
> +
>  	if (pvr_dev->fw_dev.booted) {
>  		err = pvr_power_fw_enable(pvr_dev);
>  		if (err)
> -			goto err_mem_clk_disable;
> +			goto err_reset_assert;
>  	}
>  
>  	drm_dev_exit(idx);
>  
>  	return 0;
>  
> +err_reset_assert:
> +	reset_control_assert(pvr_dev->reset);
> +
>  err_mem_clk_disable:
>  	clk_disable_unprepare(pvr_dev->mem_clk);
>  

-- 
Matt Coster
E: matt.coster@imgtec.com

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 236 bytes --]