Hi, On Thu, Jun 13, 2013 at 10:37:23AM +0200, Andrzej Pietrasiewicz wrote: > Hello Felipe, Hello Greg, > > I've seen that you pulled a bunch of my patches. That's nice. > > However, Felipe was unable to apply one of them due to some whitespace issues, > so he dropped it. > > While dropping the said patch does not break anything, it makes the usb > functions configurable through configfs unavailable to be built > stand-alone. They still can be built but only as a side effect of building > some old gadgets which depend on the functions of interest, e.g. ecm, > ecm subset, eem, rndis will be built as a side effect of building g_ether.ko. > > In Kconfig the problem described here manifests itself by some USB_CONFIGFS_* > entries being dependent on nonexistent USB_CONFIGFS. In practice they do not > appear in menuconfig, so they can't be selected. > > This isn't terribly bad, but the goal of configfs integration is to allow > composing the gadgets of functions of choice at runtime, provided that > the functions themselves are compiled. Which, in turn, stops the > proliferation of new g_xxxx.ko gadgets and, ultimately, allows > dropping the g_xxxx.ko modules at all. And this is in contradiction > to not being able to build only f_xxxx.ko modules without building some > g_xxxx.ko gadgets. > > The series I'm sending does the following: > > 1) fix separate building of configfs-enabled functions > 2) add a piece of documentation about configfs-based usb gadgets > 3) add some Documentation/ABI/* related to the newly added configfs entries > > I kindly ask you to consider the series; either as a whole, or individual > patches. The patches in the series do not depend on each other. Verified by > checkpatch. Greg, if you want to apply them directly I'm fine with it: Acked-by: Felipe Balbi <balbi@xxxxxx> Also ok to wait for v3.12 merge window as nothing is *really* broken. -- balbi
Attachment:
signature.asc
Description: Digital signature
