qcom_gcc: migrate to use sleep_clk directly

gcc_sleep_clk_src does not exist, just use sleep_clk directly.
I believe this is what feeds them anyway!

This node isn't in the Linux clock driver (drive

qcom_gcc: migrate to use sleep_clk directly

gcc_sleep_clk_src does not exist, just use sleep_clk directly.
I believe this is what feeds them anyway!

This node isn't in the Linux clock driver (drivers/clk/qcom) ipq401x
clock tree, but there are sleep_clk_src nodes in the tree in
ipq50xx, ipq54xx, ipq60xx, ipq8xxxx, ipq9xxxx, msm8xxx, etc.

Differential Revision: https://reviews.freebsd.org/D49686

show more ...

qcom_gcc: begin refactoring sys/dev/qcom_gcc to support multiple chipsets

Although the driver structure is almost supportive of multiple
chipsets, there's a lot of subtle hard coded IPQ4018 assumpti

qcom_gcc: begin refactoring sys/dev/qcom_gcc to support multiple chipsets

Although the driver structure is almost supportive of multiple
chipsets, there's a lot of subtle hard coded IPQ4018 assumptions
here.

This is a partial refactor of the driver in order to have a single
qcom_gcc driver that will eventually support multiple chipsets.

* rename qcom_gcc_ipq4018 -> qcom_gcc
* remove the ipq4018 specific naming from things
* create a table to drive probe/attach, with a chipset id to
use during attach
* migrate the clock register accessors to not be ipq4018 specific
* migrate the reset register accessors to not be ipq4018 specific

Note this won't compile (yet) for an arm64 kernel because there's
a hard-coded clock tree for an earlier 64 bit MSM part in
sys/arm64/qualcomm/qcom_gcc.c . That will need to be rolled into this
driver.

Differential Revision: https://reviews.freebsd.org/D49683

show more ...