Re: [PATCH 2/3] time: replace board_time_init() by plat_clk_setup()

["Franck Bui-Huu" <vagabon.xyz@gmail.com>]

Hi Atsushi,

On 5/5/07, Atsushi Nemoto <anemo@mba.ocn.ne.jp> wrote:
> Though providing plat_clk_setup() for timekeeping code might be a good
> idea, I think your patch break at least those two platforms:
>
> MOMENCO_JAGUAR_ATX: momenco_time_init() assumes tlb_init() was already
> called.  (wire_stupidity_into_tlb() calls local_flush_tlb_all())
Ok, but in this case it seems a bad use of time init hook: the
platform seems to use its time init hook to setup the tlb. That sounds
pretty hackish, isn't it ?

If so, doesn't it mean that we should give the opportunity to
platform's code to modify the tlb mapping properly ? Actually I had
the same problem with mapped kernels and the only way to solve it was
to hack tlb_init()...

BTW, do you know why wire_stupidity_into_tlb() is called twice: one
time in plat_mem_setup() and a second time in momenco_time_init() ?
Note for the former case, tlb_init() hasn't been called yet...

> MOMENCO_OCELOT_G: gt64240_time_init() assumes IRQ subsystem are
> already initialized.
heh ? why not using plat_timer_setup() hook in this case ?

> How about keeping board_time_init pointer as is and adding
> plat_clk_setup only for simple platforms?
Not sure that would force us to duplicate all timekeeping stuff in
time_init() hook because of several hacks. If this is really true,
let's try to clean up some code.

Thanks
--
              Franck