> On 01.06.2016, at 21:05, Eric Anholt <e...@anholt.net> wrote: > > I figured out another critical clock (patch 3), but didn't use the > CLK_IS_CRITICAL flag since I want to just protect whatever clock > happens to be the parent (there are #ifdefs in the firmware indicating > that they've experimented with using different clocks as the parent). > > I think these fixes are all suitable for 4.7. > > Eric Anholt (4): > clk: bcm2835: Mark the VPU clock as critical > clk: bcm2835: Mark GPIO clocks enabled at boot as critical > clk: bcm2835: Mark the CM SDRAM clock's parent as critical > clk: bcm2835: Skip PLLC clocks when deciding on a new clock parent > > drivers/clk/bcm/clk-bcm2835.c | 63 +++++++++++++++++++++++++++++++++++++++++-- > 1 file changed, 61 insertions(+), 2 deletions(-)
Whole series: Acked-by: Martin Sperl <ker...@martin.sperl.org> Note that these patches are also seeing more testing downstream in 4.7 and there have been no hiccups seen either. Clock selection is working as expected for I2S as well.