Re: [PATCH] Revert "clk: imx: fix composite peripheral flags"

[Martin Kepplinger]

On 04.01.21 13:05, Abel Vesa wrote:
> On 20-12-31 17:33:40, Fabio Estevam wrote:
> > Hi Martin,
> >
> > On Thu, Dec 31, 2020 at 11:22 AM Martin Kepplinger
> > <martin.kepplinger@xxxxxxx> wrote:
> > > This reverts commit 936c383673b9e3007432f17140ac62de53d87db9.
> > >
> > > It breaks clock reparenting via devfreq on the imx8mq used in the
> > > Librem 5 phone. When switching dram frequency (which worked before)
> > > the system now hangs after this where the dram_apb clock cannot be
> > > set:
> > >
> > > [  129.391755] imx8m-ddrc-devfreq 3d400000.memory-controller: failed to
> > > set dram_apb parent: -16
> > > [  129.391959] imx8m-ddrc-devfreq 3d400000.memory-controller: ddrc
> > > failed freq switch to 25000000 from 800000000: error -16. now at 25000000
> > > [  129.406133] imx8m-ddrc-devfreq 3d400000.memory-controller: failed to
> > > update frequency from PM QoS (-16)
> >
> > I am wondering whether IMX8MQ_CLK_DRAM_ALT should also be marked as
> > CLK_IS_CRITICAL.
>
> Hmm, the way the DRAM clocks are right registered right now is a real mess.
> The DRAM clocks on i.MX8M are changed in TF-A, but the kernel still needs to
> register them to keep track of the clock tree.
>
> Martin, I already have a patchset waiting to be shipped which doesn't
> only fix the 8MQ, but all the 8M platforms. Unfortunately, I haven't had the time
> to work on that in the last couple of weeks but I intend to switch back to it soon.
>
> Fabio, marking the DRAM clocks as critical will not allow the set_parent to be done,
> as CLK_IS_CRITICAL flag and set_parent do not go together. As of now the devfreq
> tries to reparent to be consistent with TF-A configuration.
>
> My approach here was to make the DRAM clocks read-only. This means adding some
> stuff in the clock core subsystem too.

Hi Abel, thanks a lot for the update. I'm looking forward to seeing your 
patchset.

                              martin