Stephen Warren wrote at Sent: Friday, April 29, 2011 7:52 PM:
> To: Grant Likely
> Cc: linaro-dev@lists.linaro.org; linux-te...@vger.kernel.org
> Subject: RE: [PATCH 2/3] ARM: Tegra: dt: Fix machine desc to match other
> boards.
>
> Grant Likely wrote at Friday, April 29, 2011 5:35 PM:
> > On Sat, Apr 23, 2011 at 10:30:03PM -0600, Stephen Warren wrote:
> > > The content of the machine descriptions has be re-organized. Without
> > > fixing
> > > the board-dt.c copy, the system will fail to boot (BUG_ON during timer
> > > initialization, which happens before printk is operational, leading to a
> > > silent hang early during kernel boot). Solve this basiclaly by copying the
> > > existing Harmony version over the copy in board-dt.c
> > >
> > > Signed-off-by: Stephen Warren <swar...@nvidia.com>
> > > ---
> > > arch/arm/mach-tegra/board-dt.c | 11 ++++++-----
> > > 1 files changed, 6 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/arch/arm/mach-tegra/board-dt.c
> > > b/arch/arm/mach-tegra/board-dt.c
> > > index 899311b..7b9d469 100644
> > > --- a/arch/arm/mach-tegra/board-dt.c
> > > +++ b/arch/arm/mach-tegra/board-dt.c
> > > @@ -175,10 +175,11 @@ static const char * tegra_dt_board_compat[] = {
> > > };
> > >
> > > DT_MACHINE_START(TEGRA_DT, "nVidia Tegra (Flattened Device Tree)")
> > > - .fixup = tegra_dt_fixup,
> >
> > I take it that with memory ranges encoded into the device tree this
> > fixup is no longer needed? I'll drop the tegra_dt_fixup() function
> > from the file too then.
>
> I actually tested this all on Seaboard not Harmony, and locally modified
> the tegra-harmony.dts to add to the kernel cmdline mem=/nvmem= options
> identical to what we use for Seaboard + ChromeOS. Now I have things
> working, I should probably remove the kernel cmdline mem=/nvmem= options,
> and validate that the dt memory setup is working correctly. I believe the
> fixup function was only there to cater for broken bootloader setups that
> didn't supply the correct mem=/nvmem= on the kernel cmdline, so this
> should all work OK...
I've now tested on Seaboard with the fixup removed as above, no mem= or
nvmem= kernel cmdline options, and hence just the devicetree memory
definitions. Everything looks like it's fine.
--
nvpublic
_______________________________________________
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev
