* Keerthy <j-keerthy@xxxxxx> [190905 03:54]:
> On 04/09/19 9:52 PM, Tony Lindgren wrote:
> > * Grygorii Strashko <grygorii.strashko@xxxxxx> [190904 11:39]:
> > > Module should also enabled here.
> >
> > You are right, good catch. Maybe this in addition to the clocks is
> > why some dra7 boards fail to enable cpsw depending on the bootloader?
> >
> > > > @@ -1677,10 +1679,10 @@ static int sysc_init_module(struct sysc *ddata)
> > > > if (manage_clocks)
> > > > sysc_disable_main_clocks(ddata);
> > > > err_opt_clocks:
> > > > - if (manage_clocks) {
> > > > + if (manage_clocks)
> > > > sysc_disable_opt_clocks(ddata);
> > > > - sysc_clkdm_allow_idle(ddata);
> > > > - }
> > > > +
> > > > + sysc_clkdm_allow_idle(ddata);
> > >
> > > clkdm doesn't have counters while clock do, so if module is in HW_AUTO
> > > and clkdm in HW_AUTO - the module can go IDLE between this point and ti_sysc_idle() call.
> > >
> > > Errate i877 required
> > > CM_GMAC_CLKSTCTRL[1:0] CLKTRCTRL = 0x2:SW_WKUP.
> > > to be set at boot time and never changed.
> > >
> > > and
> > > "In addition to programming SW_WKUP(0x2) on CM_GMAC_CLKSTCTRL, SW should
> > > also program modulemode field as ENABLED(0x2) on CM_GMAC_GMAC_CLKCTRL
> > > register."
> >
> > OK makes sense now thanks. I've dropped that change and added a comment
> > there. Updated patch below again to test.
>
> Tony,
>
> Thanks. The updated patch fixes the NFS boot issue.
OK good to hear and thanks for testing it. I sent out a proper patch
with a description as " [PATCH] bus: ti-sysc: Fix clock handling for
no-idle quirks", please review one more time and I'll apply it to
fixes.
Regards,
Tony
