[PATCH V3] clk: imx: Fix reparenting of UARTs not associated with sdout

[Abel Vesa]

On 21-01-18 08:00:43, Adam Ford wrote:
> On Mon, Jan 18, 2021 at 6:52 AM Abel Vesa <abel.vesa at nxp.com> wrote:
> >
> > On 21-01-15 12:29:08, Adam Ford wrote:
> >
> > ...
> >
> > > diff --git a/drivers/clk/imx/clk-imx25.c b/drivers/clk/imx/clk-imx25.c
> > > index a66cabfbf94f..66192fe0a898 100644
> > > --- a/drivers/clk/imx/clk-imx25.c
> > > +++ b/drivers/clk/imx/clk-imx25.c
> > > @@ -73,16 +73,6 @@ enum mx25_clks {
> > >
> > >  static struct clk *clk[clk_max];
> > >
> > > -static struct clk ** const uart_clks[] __initconst = {
> > > -     &clk[uart_ipg_per],
> > > -     &clk[uart1_ipg],
> > > -     &clk[uart2_ipg],
> > > -     &clk[uart3_ipg],
> > > -     &clk[uart4_ipg],
> > > -     &clk[uart5_ipg],
> > > -     NULL
> > > -};
> > > -
> >
> > I'm assuming there is another patch that updates the dts files. Right ?
> 
> I have only been able to test this on an i.MX8M Mini.  I need to set
> the parent clock of the i.MX8M Mini to an 80 MHz clock in order to run
> the UART at 4Mbps.   With this patch, I can stop enabling the all the
> UART clocks early and allow the clock parent configuration to occur.
> From what I can tell, the remaining clocks should get activated as
> they are needed, because I was able to use Bluetooth connected to
> UART1 running at 4MBps using a 80MHz clock source with this patch, and
> the clk_summary shows the clock is running at the proper speed.
> Without this patch, the UART fails to re-parent, so I'm stuck at lower
> speeds and that means choppy Bluetooth audio.
> 
> The Kernel that NXP hosts on Code Aurora that they use for Yocto
> attempts scan through stdout to only enable those clocks [1].  I
> attempted to push it upstream, but it was rejected [2].  Sascha
> suggested creating an array which could be filled when the clocks are
> enabled and that array would be used to deactivate the clocks at
> shutdown.  That's what I attempted to do here.
> 
> I don't have older imx boards to know if their device trees are
> configured in such a way without modifications to the device tree or
> not, but since it appears to work for NXP in [2], I assumed it would
> work here.
> 
> [1] - https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fsource.codeaurora.org%2Fexternal%2Fimx%2Flinux-imx%2Fcommit%2Fdrivers%2Fclk%2Fimx%2Fclk.c%3Fh%3Dimx_5.4.47_2.2.0%26id%3D754ae82cc55b7445545fc2f092a70e0f490e9c1b&data=04%7C01%7Cabel.vesa%40nxp.com%7Cf8922e76fa85485b86c508d8bbb97c47%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637465752633257569%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=bVmPaM1nN7Sm%2BISVvlIBoWYozfJE96fHpw431IEuggk%3D&reserved=0
> [2] - https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpatchwork.kernel.org%2Fproject%2Flinux-arm-kernel%2Fpatch%2F20201229145130.2680442-1-aford173%40gmail.com%2F&data=04%7C01%7Cabel.vesa%40nxp.com%7Cf8922e76fa85485b86c508d8bbb97c47%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637465752633257569%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=226HwbeVhZUW%2FJ3hsCSaVIxghOsPBH9EWeF8vFxaTWE%3D&reserved=0
> 
> >
> > TBH, I'm against the idea of having to call consumer API from a clock provider driver.
> > I'm still investigating a way of moving the uart clock control calls in drivers/serial/imx,
> > where they belong.
> 
> That makes sense.
> 

Just a thought. The uart clock used for console remains on from u-boot,
so maybe it's enough to just add the CLK_IGNORE_UNUSED flag to all the
uart root clocks and remove the prepare/enable calls for uart clocks 
for good. I don't really have a way to test it right now, but maybe
you could give it a try.