[U-Boot] [PATCH 1/4] ARM: Tegra210: Add SoC code/include files for T210

[Tom Warren]

Update WRT gpio.h and hardware.h, below.

> -----Original Message-----
> From: Tom Warren
> Sent: Monday, June 15, 2015 1:05 PM
> To: 'Stephen Warren'; Tom Warren
> Cc: u-boot at lists.denx.de; Stephen Warren
> Subject: RE: [U-Boot] [PATCH 1/4] ARM: Tegra210: Add SoC code/include files
> for T210
> 
> 
> 
> > -----Original Message-----
> > From: Stephen Warren [mailto:swarren at wwwdotorg.org]
> > Sent: Monday, June 15, 2015 10:11 AM
> > To: Tom Warren
> > Cc: u-boot at lists.denx.de; Stephen Warren; Tom Warren
> > Subject: Re: [U-Boot] [PATCH 1/4] ARM: Tegra210: Add SoC code/include
> > files for T210
> >
> > On 06/03/2015 02:35 PM, Tom Warren wrote:
> > > All based off of Tegra124. As a Tegra210 board is brought up, these
> > > may change a bit to match the HW more closely, but probably 90% of
> > > this is identical to T124.
> >
> > Rather than duplicating lots of headers and code, can we share the
> > content with other chips?
> Sure, but I wasn't looking at this patchset as a reworking of all Tegra common
> headers, but an inclusion of T210 support. We can then move to
> common/shared content after this is in, or someone (you?) can do it now
> before I add T210 support, but that'll delay it.
> 
> >
> > > diff --git a/arch/arm/include/asm/arch-tegra210/funcmux.h
> > > b/arch/arm/include/asm/arch-tegra210/funcmux.h
> >
> > We should be able to drop funcmux support completely now that we're
> > programming entire board pinmux tables.
> I'll look into it, but I believe funcmux is only used to get early UART muxes set,
> which is done before the pinmux table is parsed/written.
> 
> >
> > > diff --git a/arch/arm/include/asm/arch-tegra210/gpio.h
> > > b/arch/arm/include/asm/arch-tegra210/gpio.h
> >
> > > +enum gpio_pin {
> > > +	GPIO_PA0 = 0,	/* pin 0 */
> > > +	GPIO_PA1,
> >
> > Given the move to DT, are any of these GPIO_xxx values actually used?
> > I wonder how many other types/defines in the other files are actually
> > used, rather than simply left over from times gone by.
> Again, that's more of a general Tegra cleanup phase then this patchset is
> intended for. I'll take a quick look, but I don't want to get delayed by doing a
> bunch of Tegra cleanup stuff right now.
They're used in the pinmux table (pinmux-config-venice2.h, for example) and some board files (nyan-big.c, cardhu.c, seaboard.c). Can't remove 'em.

> 
> >
> > > diff --git a/arch/arm/include/asm/arch-tegra210/hardware.h
> > > b/arch/arm/include/asm/arch-tegra210/hardware.h
> >
> > Can we drop this file? I don't see a hardware.h in any of the other
> > arch-tegra*/ directories.
> Sure. It's never been used AFAICT.
Can't drop hardware.h. It's expected to be in every build, included from arch/arm/include/asm/hardware.h, created by Wolfgang way back in 2003.

> 
> >
> > > diff --git a/arch/arm/include/asm/arch-tegra210/tegra.h
> > > b/arch/arm/include/asm/arch-tegra210/tegra.h
> >
> > > +#define BCT_ODMDATA_OFFSET	1704	/* offset to ODMDATA word */
> > > +
> > > +#undef NVBOOTINFOTABLE_BCTSIZE
> > > +#undef NVBOOTINFOTABLE_BCTPTR
> > > +#define NVBOOTINFOTABLE_BCTSIZE	0x48	/* BCT size in BIT in
> > IRAM */
> > > +#define NVBOOTINFOTABLE_BCTPTR	0x4C	/* BCT pointer in BIT in
> > IRAM */
> >
> > Have you validated those? I'm pretty sure the BCT and perhaps BIT
> > layout changed in T210, and those values match T124.
> Good point. They have changed, since the BCT structure has changed. I'll
> update them w/real T210 offsets.
> 
> >
> > Have all the clock tables and IDs been updated to match T210? If not,
> > I think we should do that before checking in the code, or it'll be misleading.
> I believe so - I'm using the clock tables from my 'fully working' branch, so they
> should be accurate/jibe with the TRM, but I'll double-check.
> 
> --
> nvpublic