Re: [PATCH 10/11] ARM: DRA7: hwmod: Add data for McASP1/2/4/5/6/7/8

From: Paul Walmsley
Date: Tue Mar 01 2016 - 12:01:31 EST


Hi Péter,

On Tue, 1 Mar 2016, Peter Ujfalusi wrote:

> Hi Paul,
>
> On 03/01/2016 11:11 AM, Paul Walmsley wrote:
> > Hi Péter
> >
> > A few questions:
> >
> > On Thu, 25 Feb 2016, Peter Ujfalusi wrote:
> >
> >> Add missing data for all McASP ports for the dra7 family
> >>
> >> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@xxxxxx>
> >
> > 1. The patch doesn't set the HWMOD_OPT_CLKS_NEEDED flag for McASP1 and 2,
> > but does set it for McASP4-8. Could you please confirm that this is
> > intentional, and if so, why?
>
> All should have the HWMOD_OPT_CLKS_NEEDED as both fclk and ahclkx is treated
> as functional clock and needs to be available in order to be able to access
> McASP registers.
> Sorry, I can only test McASP3 and somehow I overlooked this when copy-pasting
> the data.

OK

> > 2. The patch sets HWMOD_SWSUP_SIDLE for McASP1 and 2, but doesn't set it
> > for McASP4-8. Could you please confirm that this is intentional, and if
> > so, why? The descriptions of the MODULEMODE fields in SPRUHZ6 look
> > identical.
>
> I need to confirm this, but all McASP should have the same set of flags.

OK. Looking at McASP3 data this morning, they probably shouldn't need
HWMOD_SWSUP_SIDLE, but probably all need

.modulemode = MODULEMODE_SWCTRL,

> > 3. Can McASP1,2,3 bus-master onto the L3? If so, then there should be
> > "dra7xx_mcasp1__l3_main_1"-style links to indicate this.
>
> I need to check this, but I don't think McASP1,2,3 can be bus-master onto L3.

OK. When you get back, maybe doublecheck this - it looks to me from
SPRUHZ6 that McASP1-3 have built-in DMA controllers.

> I can resend the series next week as I'm out of office this week.

That's fine. It's most likely v4.7 material at this point.


- Paul