Hi Tom,

On Thu, 30 Aug 2012 17:14:09 -0700, Tom Warren TWarren@nvidia.com wrote:

Seems to be the root Makefile:

ifeq ($(SOC),tegra20) -LIBS += arch/$(ARCH)/cpu/$(SOC)-common/lib$(SOC)-common.o +LIBS-y += arch/$(ARCH)/cpu/$(SOC)-common/lib$(SOC)-common.o endif

fixes it, i.e. ./MAKEALL -s tegra20 w/arm/master branch pullreq-utm passes. This change from Daniel Schwierzeck (Makefile: replace LIBS by LIBS-y) wasn't in the ARM repo I sync'd to before I sent the pull request. I assume it came from ARM/master since then, or maybe from u-boot/master.

Thanks for the analysis.

It *could* be a previous attempt of mine at pulling into ARM some branch where this fix was in and then having to revert to a previous arm/master due to regressions, and your branch was left based on the temporary arm/master with the fix.

The only way to avoid this is to strictly follow U-Boot policies for pull requests: always rebase onto the target branch before sending out a pull request, i.e. pull requests to u-boot-arm/master should always be (re)based on u-boot-arm/master, and my ARM pull requests to u-boot/master should always be (re)based on u-boot/master.

Looks like we'll need a v11 tegra20 SPL patchset, unless Albert is willing to slip-stream this minor fix in.

I'll pick Daniel's commit individually, then locally rebase your branch with the V10 SPL patch set. If regression builds are ok, I'll send out the 'applied' for Daniel's patch and your pull req.

Amicalement,