On Sat, 7 Nov 2009, David Miller wrote: > From: Thomas Gleixner <tglx@xxxxxxxxxxxxx> > Date: Sat, 07 Nov 2009 18:58:13 -0000 > > > of_set_property_mutex is taken inside the devtree_lock write locked > > region which triggers the might_sleep check. The mutex protects the > > call to prom_setprop() which is not necessary as the code is the only > > caller of prom_setprop() and already serialized by devtree_lock. The > > mutex is nowhere else used despite being exported. So it can be > > removed safely. > > > > Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx> > > Please see commit 2481d76615d5e15340ccfb0243fe8779766dfc6e to see why > this mutex really is necessary, regardless of the locking done by the > caller(s) of of_set_property(). And how it will thus need to be used > in the future. > > The long and short of it is that the firmware uses I2C accesses to > write the property values on some systems, and therefore the sparc64 > I2C bus drivers will need this mutex to coordinate with callers of > prom_set_property(). Those I2C drivers aren't merged yet, but I > definitely plan to get them in soon :-) Then you need to fix the problem that you lock the mutex inside the preempt disabled region under devtree_lock. :) Thanks, tglx -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
