On Fri, 4 May 2012 18:48:56 +0200
Stephen Warren <swarren@xxxxxxxxxxxxx> wrote:
> On 05/04/2012 12:17 AM, Hiroshi Doyu wrote:
> > Stephen Warren wrote at Thu, 3 May 2012 19:41:35 +0200:
> >> On 05/03/2012 10:05 AM, Hiroshi DOYU wrote:
> >>> Tegra AHB Bus conforms to the AMBA Specification (Rev 2.0) Advanced
> >>> High-performance Bus (AHB) architecture.
> ...
> >>> +static int __init tegra_ahb_module_init(void)
> >>> +{
> >>> + return platform_driver_register(&tegra_ahb_driver);
> >>> +}
> >>> +postcore_initcall(tegra_ahb_module_init);
> >>
> >> Can this be a module_init() instead of postcore_initcall()?
> >
> > Since this driver configures prefetch size from AHB client devices,
> > it's better to make this driver available before other AHB client
> > drivers get ready. So "postcore_initcall()" seems to make sense if
> > there's no other better initcall.
>
> I believe this only affects when the driver is registered and has no
> influence over when the device itself is probed.
>
> When booting with board files rather than DT, it was possible to
> register drivers and platform devices early using various initcalls to
> control the order. However, with DT, I believe all the devices are
> instantiated from DT at the same time (well, one by one in whatever
> order as the DT is parsed), so the time when the driver is registered
> isn't relevant.
>
> So, if the other AHB devices really need this AHB driver to initialize
> first, you'd better move all the AHB devices inside the AHB node in DT,
> so that the AHB driver can influence when the children get probed.
> Still, I'd suggest leaving that to a later patch, since everything
> clearly works fine right now without this driver even existing.
Ok, I'll use module_platform_driver(tegra_ahb_driver). These order would be revisited.
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
