On Wed, Mar 28, 2012 at 09:56:48AM -0700, Randy Dunlap wrote:

On 03/26/2012 01:16 PM, Greg KH wrote:

quoted

On Mon, Mar 26, 2012 at 04:06:59PM -0400, Alan Stern wrote:

quoted

On Mon, 26 Mar 2012, Greg KH wrote:

quoted

On Mon, Mar 26, 2012 at 09:58:36AM -0700, Randy Dunlap wrote:

quoted

On 03/25/2012 09:28 PM, Stephen Rothwell wrote:

quoted

Hi all,

Reminder: please do not add stuff destined for v3.5 to linux-next
included trees/branches until after v3.4-rc1 has been released.

Changes since 20120323:

on x86_64:

ERROR: "utf16s_to_utf8s" [drivers/usb/core/usbcore.ko] undefined!
ERROR: "usb_speed_string" [drivers/usb/core/usbcore.ko] undefined!

Full randconfig file is attached.

Odd, nothing in the usb core has changed in a while in this area,
although we might have gotten some core dependancies mixed up for random
configurations like this.

This is puzzling.

	utf16s_to_utf8s is defined in fs/nls/nls_base.c,
	which is built according to CONFIG_NLS (in fs/nls/Makefile),
	which is selected by CONFIG_USB (in drivers/usb/Kconfig).

But the config that Randy attached had CONFIG_USB set to M and 
CONFIG_NLS not set.  How is that possible?

	usb_speed_string is defined in drivers/usb/usb-common.c,
	which is built according to CONFIG_USB_COMMON (in
		drivers/usb/Makefile),
	which defaults to Y if CONFIG_USB_SUPPORT and either
		CONFIG_USB or CONFIG_USB_GADGET is set (in 
		drivers/usb/Kconfig).

The problem here was that CONFIG_USB_SUPPORT isn't set but CONFIG_USB
is set to M.  This used not to be possible, since CONFIG_USB was
defined inside an "if USB_SUPPORT" section.  Maybe that got changed and
needs to be fixed (but
https://git.kernel.org/?p=linux/kernel/git/next/linux-next.git thinks
it's still not possible).  Or maybe Randy's tree is a little out of 
date.

Yeah, that is strange, thanks for looking into it.

Randy, any thoughts?

It's weird.  The same config file builds successfully on
linux-next of Mar. 27.

I looked more at linux-next of Mar. 26 and found one problem:
in drivers/gpu/drm/Kconfig, there is a "select USB" that should be
protected by "depends on USB_ARCH_HAS_HCD".  I did that and then
kconfig complained about unmet dependencies in drivers/gpu/drm/udl/Kconfig,
so I added the same "depends on" there.  That led into an awful mess:

drivers/video/Kconfig:36:error: recursive dependency detected!
drivers/video/Kconfig:36:	symbol FB is selected by DRM_KMS_HELPER
drivers/gpu/drm/Kconfig:26:	symbol DRM_KMS_HELPER is selected by DRM_UDL
drivers/gpu/drm/udl/Kconfig:1:	symbol DRM_UDL depends on USB_ARCH_HAS_HCD
drivers/usb/Kconfig:78:	symbol USB_ARCH_HAS_HCD depends on USB_ARCH_HAS_OHCI
drivers/usb/Kconfig:24:	symbol USB_ARCH_HAS_OHCI depends on I2C
drivers/i2c/Kconfig:5:	symbol I2C is selected by FB_DDC
drivers/video/Kconfig:86:	symbol FB_DDC is selected by FB_CYBER2000_DDC
drivers/video/Kconfig:385:	symbol FB_CYBER2000_DDC depends on FB_CYBER2000
drivers/video/Kconfig:373:	symbol FB_CYBER2000 depends on FB

and I don't know what to do with this.

Anyway, I do think that the problem is in drivers/gpu/drm/Kconfig
and drivers/gpu/drm/udl/Kconfig.

Ok, thanks for looking.  I'm worried why DRM would care about USB, but
maybe I don't want to know...

greg k-h