* Grygorii Strashko <grygorii.strashko@xxxxxx> [160621 10:15]:
> On 06/21/2016 07:50 PM, Tony Lindgren wrote:
> > * Grygorii Strashko <grygorii.strashko@xxxxxx> [160621 05:21]:
> >> On 06/21/2016 02:56 PM, Tony Lindgren wrote:
> >>> So are these safe to apply in whatever order?
> >>>
> >>
> >> Patch 1 should go first.
> >>
> >> but do you agree with comment I've added in patch 1:
> >>>>>
> >>
> >> RFC: because it will break DT ABI, mdio will not work on devices with
> >> old DTs and new kernel which still do have [ti,hwmods = "davinci_mdio"]
> >> property defined in DT:
> >> platform 48485000.mdio: Cannot lookup hwmod 'davinci_mdio'
> >> davinci_mdio 48485000.mdio: _od_fail_runtime_resume: FIXME: missing
> >> hwmod/omap_dev info
> >> davinci_mdio: probe of 48485000.mdio failed with error -5
> >
> > Well the ti,hwmods entry we can keep and remove later. How
> > about parse it manually in the driver and just print a warning
> > if no clocks are found and ti,hwmods is set?
> >
> > I think that's the best we can do here. With the warning people
> > know at least what to do.
> >
>
> Yeh. But it will fail not only from the driver :(
> First of all, it will fail because of
> commit f5c33b070de3fdb8ad995b767ad0e3487cf0d242
> Author: Nishanth Menon <nm@xxxxxx>
> Date: Tue Dec 3 19:39:13 2013 -0600
>
> ARM: OMAP2+: omap_device: add fail hook for runtime_pm when bad data is detected
>
> ^ which will cause all PM runtime calls to fail if there is ti,hwmods prop in DT,
> but no hwmod in code.
>
> I invented few ways to fix above problem:
> - introduce list of obsolete hwmods - any hwmods listed in this per SoC list
> will be ignored by hwmod and omap_device frameworks;
> - implement mach_desc->dt_fixup() and patch DT on the fly by removing
> known deprecated props;
> - Or just ignore...
OK I see where the RFC comes from then.. Sounds like we need to
think about this a bit more for the safe set of patches.
> In general, "fck" problem can be solved by adding clk_dev alias (if it's still
> possible/acceptable), like:
> DT_CLK(NULL, "dss_deshdcp_clk", "dss_deshdcp_clk"),
> + DT_CLK("48485000.mdio", "fck", "dpll_gmac_ck"),
> { .node_name = NULL },
Seems like that's a good place to start before disabling the
bogus hwmod for this module.
Regards,
Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
