Hi Stephen,
On Thu, Feb 24, 2022 at 02:54:20PM -0800, Stephen Boyd wrote:
> Quoting Daniel Latypov (2022-02-23 14:50:59)
> > On Wed, Feb 23, 2022 at 2:56 AM Maxime Ripard <maxime@xxxxxxxxxx> wrote:
> > Incremental coverage for 3/9 files in --diff_file
> > Total incremental: 99.29% coverage (281/283 lines)
> > drivers/clk/clk.c: 84.62% coverage (11/13 lines)
> > drivers/clk/clk_test.c: 100.00% coverage (269/269 lines)
> > include/linux/clk.h: 100.00% coverage (1/1 lines)
> >
> > Missing lines are drivers/clk/clk.c:2397-8, i.e. this part of the diff:
> > + if (ret) {
> > + /* rollback the changes */
> > + clk->min_rate = old_min; <- 2397
> > + clk->max_rate = old_max; <- 2398
> >
> > These are from before and were just moved around.
>
> We could trigger a failure in the provider when the rate is set, and
> then we could call round_rate() again and make sure the boundaries from
> before are maintained.
I tried to do that, and it turns out we can't, since we ignore the
set_rate return code:
https://elixir.bootlin.com/linux/latest/source/drivers/clk/clk.c#L2107
We could make determine_rate fail, but then clk_round_rate would fail as
well and wouldn't allow us to test whether the boundaries are still in
place.
Maxime
Attachment:
signature.asc
Description: PGP signature
