This branch fixes some odd issues in the i.MX6 clock tree. The i.MX6Q and i.MX6D SoC variants need a workaround for the PFDs and it turned out we have that workaround twice: once in the clk driver and then again in architecture code. When deep probe is enabled the order in which they run may be changed. The difference with both workarounds is that the one in the architecture code leaves the PFD clocks enabled and the one in the clock driver leaves them disabled. That is still not a problem, but our clock driver lacks the hdmi_isfr clock, so that the i.MX hdmi driver works on a dummy clock instead. This works when the architecture workaround runs after the one in the clk driver, but when deep probe is enabled and the order changes then the hdmi driver finds its clock disabled and doesn't work. The issues are resolved in this series: The hdmi_isfr clock is added to the clock tree and one of the duplicated workarounds is removed. There are some other patches I came along while investigating why hdmi doesn't work with deep probe enabled. Sascha Hauer (7): clk: imx6: Add hdmi_isfr clk clk: imx6: fix axi_sels clk: imx6: Fix periph_clk2_sel and periph2_clk2_sel clocks clk: imx6: emi_sel clock muxing clk: imx6: fix cko1_sels clk: imx6: do pfd workaround unconditionally ARM: i.MX: Remove duplicate PFD workaround arch/arm/mach-imx/imx6.c | 44 -------------------------------------- drivers/clk/imx/clk-imx6.c | 37 ++++++++++++++++++++++++-------- 2 files changed, 28 insertions(+), 53 deletions(-) -- 2.30.2 _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox
