Hi Rafael,
These patches parse the newly added opp-supported-hw/opp-<prop>-<name> bindings.
Rebased over: Latest pm/linux-next Tested-on: Exynos 5250, dual core A15.
V2->V3: - Stephen had an interesting point to make for V2, as the patches created the dev_opp first and then updated it within the OPP-list lock. But a simultaneous reader of the dev_opp list need a rcu way of handling this. - The big difference, why its not a problem here, is that the platform needs to call this routine before the OPPs are parsed from DT and the opp-list will be empty. And so no-readers. - It wasn't guaranteed earlier, but now we have installed few WARN_ON() to make sure about that there are no OPPs while these routines are called. - It should be pretty safe to get these merged now. - Lee's work depends on this to be merged, http://marc.info/?l=linux-kernel&m=144958553611215&w=2
V1->V2: - Fixed locking - NUL terminate strings instead of sprintf - Remove NULL checkers for the routines - constify 'versions' - s/EINVAL/EBUSY - updated comments over routines - Use of_property_read_u32_index() instead of allocating arrays - remove dev_opp for failures
Viresh Kumar (2): PM / OPP: Parse 'opp-supported-hw' binding PM / OPP: Parse 'opp-<prop>-<name>' bindings
drivers/base/power/opp/core.c | 313 ++++++++++++++++++++++++++++++++++++++++-- drivers/base/power/opp/opp.h | 7 + include/linux/pm_opp.h | 22 +++ 3 files changed, 327 insertions(+), 15 deletions(-)