> Il 14/10/2020 10:22 Felix Brack <f...@ltec.ch> ha scritto: > > > On 11.10.20 14:13, Dario Binacchi wrote: > > The series was born from the need to manage the PWM backlight of the > > display connected to my beaglebone board. To hit the target, I had to > > develop drivers for PWM management which in turn relied on drivers for > > managing timers and clocks, all developed according to the driver model. > > My intention was to use the SoC-specific API only at strictly necessary > > points in the code. My previous patches for migrating the AM335x display > > driver to the driver model had required the implementation of additional > > functions outside the concerns of the driver, (settings for dividing the > > pixel clock rate, configuring the display DPLL rate, ....) not being > > able to use the API of the related clock drivers. This series shouldn't > > have repeated the same kind of mistake. Furthermore, I also wanted to fix > > that kind of forced choice. Almost everything should have been accessible > > via the driver model API. In the series there are also some patches that > > could be submitted separately, but which I have however inserted to avoid > > applying future patches to incorporate them. > > With this last consideration, I hope I have convincingly justified the > > large number of patches in the series. > > > > The patch enabling address translation into a CPU physical address from > > device-tree even in case of crossing levels with #size-cells = <0>, is > > crucial for the series. The previous implementation was unable to > > perform the address translation required by the am33xx device tree. > > I tried to apply in a conservative way as few changes as possible and > > to verify the execution of all the tests already developed, as well as > > the new ones I added for the new feature. > > > > As recommended by Grygorii Strashko I have removed some patches to > > upload them in new different series: > > - [v2,01/30] clk: remove a redundant header > > - [v2,03/30] arch: sandbox: fix typo in clk.h > > - [v2,20/30] video: backlight: fix pwm's duty cycle calculation > > - [v2,21/30] video: backlight: fix pwm data structure description > > - [v2,22/30] dm: core: improve uclass_get_device_by_phandle_id() description > > - [v2,23/30] gpio: fix gpio_request_by_name() description > > > > Grygorii suggested to remove also other patches that I have kept in the > > series to allow anyone who wants to be able to test it. Moreover after > > the addition of a patch that has largely changed the device tree. The > > patches removed are the one and only ones that do not affect the series > > in any way. I hope that the next version of this series will require > > only minimal changes allowing me to split it according to Grygorii's > > suggestions. > > ---- >
> I must be missing something as this patch series does not apply without > errors to U-Boot master: > Hi Felix, I applied the patch of the series downloaded from patchwork on master and I compiled it without errors. > error: drivers/clk/clk-ti-am3-dpll-x2.c: No such file or directory > error: drivers/clk/clk-ti-am3-dpll.c: No such file or directory > error: drivers/clk/clk-ti-ctrl.c: No such file or directory > error: drivers/clk/clk-ti-divider.c: No such file or directory > error: drivers/clk/clk-ti-gate.c: No such file or directory > error: drivers/clk/clk-ti-mux.c: No such file or directory > error: drivers/clk/clk-ti.c: No such file or directory > error: drivers/clk/clk-ti.h: No such file or directory After applying the patch you should find: ls drivers/clk/ti am3-prcm.c clk-am3-dpll.c clk-am3-dpll-x2.c clk.c clk-ctrl.c clk-divider.c clk-gate.c clk.h clk-mux.c clk-sci.c Kconfig Makefile omap4-cm.c > error: drivers/video/tilcdc-panel.c: No such file or directory > error: drivers/video/tilcdc-panel.h: No such file or directory > error: drivers/video/tilcdc.c: No such file or directory > error: drivers/video/tilcdc.h: No such file or directory > and ls drivers/video/ti am335x-fb.c am335x-fb.h Kconfig Makefile tilcdc.c tilcdc.h tilcdc-panel.c tilcdc-panel.h > Does it probably depend on some other patches? No, the series patch does not depend on other patches > > regards, Felix Best regards Dario
