Re: Regression due to 59cf44575456 ("USB: core: Fix oversight in SuperSpeed initialization")

[Alan Stern <stern@rowland.harvard.edu>]

On Mon, Apr 22, 2024 at 07:33:21PM +0200, Linux regression tracking (Thorsten Leemhuis) wrote:
> Hi, Thorsten here, the Linux kernel's regression tracker. Top-posting
> for once, to make this easily accessible to everyone.
> 
> Is anyone still working on fixing below regression? From here it looks
> stalled, but I might have missed something.

I've been waiting to hear back from Oliver or Takashi.  A revised patch 
taking my comments into account would be welcome; it should be a very 
small change (just one or two lines of code).

Alan Stern

> Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
> --
> Everything you wanna know about Linux kernel regression tracking:
> https://linux-regtracking.leemhuis.info/about/#tldr
> If I did something stupid, please tell me, as explained on that page.
> 
> #regzbot poke
> 
> On 09.04.24 16:56, Alan Stern wrote:
> > On Tue, Apr 09, 2024 at 03:49:01PM +0200, Oliver Neukum wrote:
> >> Hi,
> >>
> >> with the following device:
> >>
> >> Bus 002 Device 002: ID fb5d:0001 BHYVE HID Tablet
> >> Device Descriptor:
> >>   bLength                18
> >>   bDescriptorType         1
> >>   bcdUSB               3.00
> >>   bDeviceClass            0
> >>   bDeviceSubClass         0
> >>   bDeviceProtocol         0
> >>   bMaxPacketSize0         8
> > 
> > A USB-3 device, running at SuperSpeed with its bMaxPacketSize0 set to 8 
> > instead of 9?  Presumably this thing never received a USB certification.  
> > Does the packaging use the USB logo?
> > 
> >>   idVendor           0xfb5d
> >>   idProduct          0x0001
> >>   bcdDevice            0.00
> >>   iManufacturer           1 BHYVE
> >>   iProduct                2 HID Tablet
> >>   iSerial                 3 01
> >>   bNumConfigurations      1
> > 
> > Why on earth would an HID tablet need to run at SuperSpeed?
> > 
> >> Binary Object Store Descriptor:
> >>   bLength                 5
> >>   bDescriptorType        15
> >>   wTotalLength       0x000f
> >>   bNumDeviceCaps          1
> >>   SuperSpeed USB Device Capability:
> >>     bLength                10
> >>     bDescriptorType        16
> >>     bDevCapabilityType      3
> >>     bmAttributes         0x00
> >>     wSpeedsSupported   0x0008
> >>       Device can operate at SuperSpeed (5Gbps)
> >>     bFunctionalitySupport   3
> >>       Lowest fully-functional device speed is SuperSpeed (5Gbps)
> >>     bU1DevExitLat          10 micro seconds
> >>     bU2DevExitLat          32 micro seconds
> > 
> > A tablet not capable of running at any speed below 5 Gbps?
> > 
> >> we are getting a regression on enumeration. It used to work with the
> >> code prior to your patch. Takashi is proposing the attached fixed.
> >> It looks like we are a bit too restrictive and should just try.
> >>
> >> 	Regards
> >> 		Oliver
> > 
> >> From: Takashi Iwai <tiwai@suse.de>
> >> Subject: [PATCH] USB: hub: Workaround for buggy max packet size with super
> >>  speed
> >> Patch-mainline: Not yet, testing
> >> References: bsc#1220569
> >>
> >> Signed-off-by: Takashi Iwai <tiwai@suse.de>
> >>
> >> ---
> >>  drivers/usb/core/hub.c | 13 ++++++++++---
> >>  1 file changed, 10 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> >> index e38a4124f610..64193effc456 100644
> >> --- a/drivers/usb/core/hub.c
> >> +++ b/drivers/usb/core/hub.c
> >> @@ -4830,7 +4830,7 @@ hub_port_init(struct usb_hub *hub, struct usb_device *udev, int port1,
> >>  	const char		*driver_name;
> >>  	bool			do_new_scheme;
> >>  	const bool		initial = !dev_descr;
> >> -	int			maxp0;
> >> +	int			maxp0, ep0_maxp;
> >>  	struct usb_device_descriptor	*buf, *descr;
> >>  
> >>  	buf = kmalloc(GET_DESCRIPTOR_BUFSIZE, GFP_NOIO);
> >> @@ -5070,7 +5070,8 @@ hub_port_init(struct usb_hub *hub, struct usb_device *udev, int port1,
> >>  		else
> >>  			i = 0;		/* Invalid */
> >>  	}
> >> -	if (usb_endpoint_maxp(&udev->ep0.desc) == i) {
> >> +	ep0_maxp = usb_endpoint_maxp(&udev->ep0.desc);
> >> +	if (ep0_maxp == i) {
> > 
> > This variable looks like it was left over from earlier testing.  It's 
> > not really needed.
> > 
> >>  		;	/* Initial ep0 maxpacket guess is right */
> >>  	} else if ((udev->speed == USB_SPEED_FULL ||
> >>  				udev->speed == USB_SPEED_HIGH) &&
> >> @@ -5082,9 +5083,15 @@ hub_port_init(struct usb_hub *hub, struct usb_device *udev, int port1,
> >>  			dev_warn(&udev->dev, "Using ep0 maxpacket: %d\n", i);
> >>  		udev->ep0.desc.wMaxPacketSize = cpu_to_le16(i);
> >>  		usb_ep0_reinit(udev);
> >> +	} else if (udev->speed >= USB_SPEED_SUPER && initial) {
> >> +		/* FIXME: should be more restrictive? */
> >> +		/* Initial guess is wrong; use the descriptor's value */
> >> +		dev_warn(&udev->dev, "Using ep0 maxpacket: %d\n", i);
> >> +		udev->ep0.desc.wMaxPacketSize = cpu_to_le16(i);
> >> +		usb_ep0_reinit(udev);
> > 
> > This could be merged with the previous case fairly easily.
> > 
> >>  	} else {
> >>  		/* Initial guess is wrong and descriptor's value is invalid */
> >> -		dev_err(&udev->dev, "Invalid ep0 maxpacket: %d\n", maxp0);
> >> +		dev_err(&udev->dev, "Invalid ep0 maxpacket: %d, expected %d\n", maxp0, ep0_maxp);
> > 
> > This also looks like a remnant from earlier testing.
> > 
> > Alan Stern
> > 
> >>  		retval = -EMSGSIZE;
> >>  		goto fail;
> >>  	}
> >> -- 
> >> 2.35.3
> >>
> > 
>