Re: [PATCH v2 0/4] clk_phase error caching problems

[Jerome Brunet]

On Thu 06 Feb 2020 at 00:27, Stephen Boyd <sboyd@xxxxxxxxxx> wrote:

> This patch series is a follow up to[1] which I sent out a few months
> ago. We no longer cache the clk phase if it's an error value, so that
> things like debugfs don't return us nonsense values for the phase.
>
> Futhermore, the last patch fixes up the locking so that debugfs code
> can avoid doing a recursive prepare lock because we know what we're
> doing in that case. While we get some more functions, we avoid taking
> the lock again.
>
> Changes from v1:
>  * A pile of new patches
>  * Rebased to clk-next
>  * New patch to bail out of registration if getting the phase fails

Series look good - comment on patch 1 is just a nitpick

Acked-by: Jerome Brunet <jbrunet@xxxxxxxxxxxx>

>
> Cc: Douglas Anderson <dianders@xxxxxxxxxxxx>
> Cc: Heiko Stuebner <heiko@xxxxxxxxx>
> Cc: Jerome Brunet <jbrunet@xxxxxxxxxxxx>
>
> Stephen Boyd (4):
>   clk: Don't cache errors from clk_ops::get_phase()
>   clk: Use 'parent' to shorten lines in __clk_core_init()
>   clk: Move rate and accuracy recalc to mostly consumer APIs
>   clk: Bail out when calculating phase fails during clk registration
>
>  drivers/clk/clk.c | 119 +++++++++++++++++++++++++++-------------------
>  1 file changed, 70 insertions(+), 49 deletions(-)
>
> [1] https://lkml.kernel.org/r/20191001174439.182435-1-sboyd@xxxxxxxxxx
>
> base-commit: 5df867145f8adad9e5cdf9d67db1fbc0f71351e9