[U-Boot] Regression caused by "disk: part_efi: fix **pgpt_pte == NULL"

[Stephen Warren]

The commit below causes the command "mmc part" to fail on both Seaboard and
Ventana (and I assume any Tegra platform, and possibly just any platform).
Reverting it solves the problem.

(this is observed in git://git.denx.de/u-boot.git master)

commit deb5ca80275e8cfa74d5680b41204e08a095eca5
Author: Doug Anderson <dianders at chromium.org>
Date:   Wed Oct 19 09:47:31 2011 +0000

    disk: part_efi: fix **pgpt_pte == NULL
    
    Code was setting **pgpt_pte == NULL, which meant that the pointer
    to the gpt_pte would be stored at RAM address 00000000. This 'worked'
    on T20 (SDRAM starts @ 0x00000000), but hung gpt/EFI access on T30
    (SDRAM starts @ 0x80000000).
    
    Signed-off-by: Tom Warren <twarren at nvidia.com>
    Signed-off-by: Doug Anderson <dianders at chromium.org>

Test steps:
mmc dev 1 # Works
mmcinfo   # Works
mmc part  # Hangs, or starts spewing messages such as:

Aos DebugSemiHosting Initialized
Assert on hardware/tegra/core/drivers/nvrm/nvrmkernel/core/ap15/ap15rm_power.c:486
Assert on hardware/tegra/core/drivers/nvrm/nvrmkernel/core/common/nvrm_power_dfs.c:2276
(perhaps it's executing some random address, and I happen to have some
copy of fastboot in RAM from when I was flashing the updated U-Boot?

-- 
nvpublic