Tracking upstream u-boot

[Mike Montour]

This weekend I spent some more time looking at u-boot. I know that there 
have been discussions of using other bootloaders in the future, and I do 
not want to distract the core developers from GTA02 tasks, but I hope 
that the information and patches here will be of some use to the community.

Right now the default bitbake recipe is 
"u-boot-openmoko_1.3.1+git+svn.bb", which uses the upstream "v1.3.2-rc2" 
tag. This recipe also appears to lack the 
"fix-data-abort-from-sd-ombug799.patch" workaround that is required to 
boot from SD.

The first attachment to this email is a patch to create a 
"u-boot-openmoko_1.3.2+git+svn.bb" recipe that uses the upstream "1.3.2" 
version, with the ombug799 patch. I have tested this briefly and it 
looks OK, so if other testers can confirm this I would suggest adding 
this recipe to OE for now (until upstream 1.3.3 comes out).

The second attachment is a patch to create a bitbake recipe that tracks 
the upstream git HEAD (currently v1.3.3-rc3, with a 1.3.3 final release 
scheduled for May 12). Here the situation is not as clean, and 
additional changes are required:

The Openmoko patches do not apply to this tree. The third attachment is 
a set of patches to the Openmoko patches that are enough for the 
do_quilt() stage to succeed, although there are still lots of "offset" 
and "fuzz" warnings that could be cleaned up.

The fourth attachment is for a change for the rtc_get() function from a 
'void' to an 'int' return value. This probably belongs in the "quilt" 
patch series rather than in its current form as an OpenEmbedded patch.

The fifth attachment reverses the upstream commit 
de109d909707e2dfe806be5efc3cdb103b47c8ad, "Makefile: fix parallel 
builds", which was applied after 1.3.3-rc2. I don't know why but this 
change results in a bricked device (at least on my gta01bv4) - on 
power-up, there is no splash-screen and nothing is written to the serial 
console. I do not know whether this is something that should be reported 
to the u-boot development list or whether it is just that some of the 
Openmoko-specific code needs to be updated to deal with the change.

Comments are welcome.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: u-boot_to_1.3.2.patch
Url: http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20080504/194dac60/attachment.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: u-boot_to_head.patch
Url: http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20080504/194dac60/attachment-0001.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: om_patch_changes.patch
Url: http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20080504/194dac60/attachment-0002.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: int_rtc_get.patch
Url: http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20080504/194dac60/attachment-0003.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: revert_git_de109d90.patch
Url: http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20080504/194dac60/attachment-0004.txt