Re: [PATCH] usb: chipidea: fix RT issue for udc

[Peter Chen <peter.chen@kernel.org>]

On 21-08-10 15:02:28, Jeaho Hwang wrote:
> hw_ep_prime sometimes fails if irq occurs while it rus on RT kernel.
> to prevent local_irq_save should keep the function from irqs.
> 
> I am not sure where is the best to submit this patch, between RT and USB
> community so sending to both. thanks.

Greg, do you have any suggestions about it, the RT kernel schedules the interrupt
handler (top-half) out which causes the USB hardware atomic sequences are broken,
these hardware operations needs to be executed within limited time.

Peter
> 
> Signed-off-by: Jeaho Hwang <jhhwang@rtst.co.kr>
> ---
>  drivers/usb/chipidea/udc.c | 31 +++++++++++++++++++++++++------
>  1 file changed, 25 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/usb/chipidea/udc.c b/drivers/usb/chipidea/udc.c
> index 5f35cdd2cf1d..a90498f17cc4 100644
> --- a/drivers/usb/chipidea/udc.c
> +++ b/drivers/usb/chipidea/udc.c
> @@ -102,6 +102,9 @@ static int hw_ep_flush(struct ci_hdrc *ci, int num, int dir)
>  {
>  	int n = hw_ep_bit(num, dir);
>  
> +    /* From zynq-7000 TRM, It can take a long time
> +     * so irq disable is not a good option for RT
> +     */
>  	do {
>  		/* flush any pending transfer */
>  		hw_write(ci, OP_ENDPTFLUSH, ~0, BIT(n));
> @@ -190,22 +193,32 @@ static int hw_ep_get_halt(struct ci_hdrc *ci, int num, int dir)
>  static int hw_ep_prime(struct ci_hdrc *ci, int num, int dir, int is_ctrl)
>  {
>  	int n = hw_ep_bit(num, dir);
> +	unsigned long flags;
> +	int ret = 0;
>  
>  	/* Synchronize before ep prime */
>  	wmb();
>  
> -	if (is_ctrl && dir == RX && hw_read(ci, OP_ENDPTSETUPSTAT, BIT(num)))
> -		return -EAGAIN;
> +	/* irq affects this routine so irq should be disabled on RT.
> +	 * on standard kernel, irq is already disabled by callers.
> +	 */
> +	local_irq_save(flags);
> +	if (is_ctrl && dir == RX && hw_read(ci, OP_ENDPTSETUPSTAT, BIT(num))) {
> +		ret = -EAGAIN;
> +	goto out;
> +	}
>  
>  	hw_write(ci, OP_ENDPTPRIME, ~0, BIT(n));
>  
>  	while (hw_read(ci, OP_ENDPTPRIME, BIT(n)))
>  		cpu_relax();
>  	if (is_ctrl && dir == RX && hw_read(ci, OP_ENDPTSETUPSTAT, BIT(num)))
> -		return -EAGAIN;
> +		ret = -EAGAIN;
>  
> +out:
> +	local_irq_restore(flags);
>  	/* status shoult be tested according with manual but it doesn't work */
> -	return 0;
> +	return ret;
>  }
>  
>  /**
> -- 
> 2.25.1
> 

-- 

Thanks,
Peter Chen