On Fri, Jan 09, 2015 at 09:31:08AM +0100, Daniel Lezcano wrote:
> On 01/09/2015 03:09 AM, Paul Walmsley wrote:
> >Hello Daniel
> >
> >On Thu, 8 Jan 2015, Daniel Lezcano wrote:
> >
> >>On 12/09/2014 11:07 PM, Paul Walmsley wrote:
> >>>
> >>>Like several of the other files in drivers/clocksource,
> >>>tegra20_timer.c contains code that can only compile when CONFIG_ARM is
> >>>enabled.  This causes obvious problems when trying to compile this
> >>>code for NVIDIA ARM64-based SoCs, such as Tegra132.  The same timer IP
> >>>blocks exist, so it seems appropriate to provide support for them.
> >>>
> >>>So until we figure out a better way to partition this code, wrap the
> >>>delay_timer and persistent_clock support code with preprocessor tests
> >>>for CONFIG_ARM.
> >>>
> >>>  (The delay_timer code should not be needed at all on
> >>>ARM64 due to the presence of the ARMv8 architected timer.  The
> >>>persistent_clock support code could become important once power
> >>>management modes are implemented that turn off the CPU complex.)
> >>
> >>IIUC, the cpuidle driver is not yet ready, right ?
> >>
> >>If it is the case, this driver is not needed yet, no ?
> >
> >The point of the patch is to allow the hardware drivers selected by
> >CONFIG_ARCH_TEGRA to build for an arm64 kernel, just as they build for
> >32-bit ARM.
> >
> >There's nothing CPUIdle-specific about the patch - that is, this timer can
> >be selected as a clockevent and clocksource provider without the use of
> >CPUIdle - although low-power PM idle is likely to be a primary use-case.
> 
> What I meant is this timer is not needed for the moment.
> 
> >>Perhaps you can rework a bit this driver in the meantime to have a better fix
> >>than disabling the code with macros ?
> >
> >I'm happy to do that, but it would be nice to get the driver compiling
> >first for ARM64 :-)
> >
> >>Otherwise, please try at least to group the code into a minimal set of macros.
> >
> >So, would it be accurate to say that you would prefer a patch that changes
> >more lines of code, but minimizes preprocessor directives, to the current
> >patch?
> 
> Yes at least an attempt to factor out a bit the driver. Those #ifdef are
> like #if 0, which is a quick fix. I am not strongly against this patch, but
> it would be nice to take the opportunity to reorganize it a bit.

How about we do something like the attached patch instead for now. That
avoids any #ifdef'ery and still we don't attempt (and fail) to build the
driver on 64-bit ARM.

With that applied we can incrementally make the changes to untangle the
ARM-specific parts and when the driver can build on 64-bit ARM we simply
select TEGRA_TIMER via Kconfig.

Thierry