Re: [PATCH v6] usb: Reduce the 'SET_ADDRESS' request timeout with a new quirk

[Sergey Shtylyov]

Hello!

   Please don't post the patches as a reply to the other thread, start a new
thread with a new patch version (I thought others would tell you that but nobody
has so far).
   And how about changing the wording of the subject to s/th like below?

usb: new quirk to reduce the SET_ADDRESS request timeout

On 10/26/23 1:15 PM, Hardik Gajjar wrote:

> This patch introduces a new USB quirk,
> USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT, which modifies the timeout value
> for the 'SET_ADDRESS' request. The standard timeout for USB request/command

   The upper case is enough of the emphasis, I don't think the apostrophes
are needed arounnd SET_ADDRESS...

[...]

> Signed-off-by: Hardik Gajjar <hgajjar@xxxxxxxxxxxxxx>
> ---
[...]
> Changes since version 5:
> 	- Changed the terminology in USB core driver files from 'command' to 'request'
> 	  as it is more commonly used. 
> 	  It's important to note that USB specifications indicate these terms are interchangeable.

   Didn't know that... tried to find the proof in the USB specs but haven't
managed to do it...

> 	  For example, USB spec 3.2, section 9.2.6.1, uses the term 'command' in its text
> 	  "USB sets an upper limit of 5 seconds for any command to be processed. "

   Hm, indeed; and this wording is even inherited from USB 1.1...

[...]

> diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> index 3c54b218301c..98db92af2cce 100644
> --- a/drivers/usb/core/hub.c
> +++ b/drivers/usb/core/hub.c
> @@ -54,6 +54,19 @@
>  #define USB_TP_TRANSMISSION_DELAY_MAX	65535	/* ns */
>  #define USB_PING_RESPONSE_TIME		400	/* ns */
>  
> +/*
> + * USB 3.2 spec, section 9.2.6.1
> + * USB sets an upper limit of 5000 ms for any command/request
> + * to be processed.
> + */
> +#define USB_DEFAULT_REQUEST_TIMEOUT_MS	5000 /* ms */
> +
> +/*
> + * The SET_ADDRESS request timeout will be 500 ms when
> + * USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT enable.
> + */
> +#define USB_SHORT_SET_ADDRESS_REQ_TIMEOUT_MS	500  /* ms */

   I don'ts see the _MS-like suffixes in the other timeout #define's there...

[...]
> diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
> index e1b1b64a0723..d856c4717ca9 100644
> --- a/drivers/usb/host/xhci.c
> +++ b/drivers/usb/host/xhci.c
> @@ -3997,12 +3997,18 @@ int xhci_alloc_dev(struct usb_hcd *hcd, struct usb_device *udev)
>  	return 0;
>  }
>  
> -/*
> - * Issue an Address Device command and optionally send a corresponding
> - * SetAddress request to the device.
> +/**
> + * xhci_setup_device - issues an Address Device command to assign a unique
> + *			USB bus address.
> + * @hcd: USB host controller data structure.
> + * @udev: USB dev structure representing the connected device.
> + * @setup: Enum specifying setup mode: address only or with context.
> + * @timeout_ms: Max wait time (ms) for the command operation to complete.
> + *
> + * Return: 0 if successful; otherwise, negative error code.

   I still think the above change should be a separate follow-up (or even
a preceding) patch...

[...]

MBR, Sergey