On Nov 6, 2008, at 1:28 PM, Scott Wood wrote:
Becky Bruce wrote:
The memory map on the 8641hpcn is modified to look more like the 85xx boards; this is a step towards a more standardized layout going forward. As part of this change, we now relocate the flash. The regions for some of the mappings were far larger than they needed to be. I have reduced the mappings to match the actual sizes supported by the hardware.
Was this causing any actual problems? It seems to be inviting gratuitous breakage if the device tree isn't updated at the same time, or for any OSes not using the device tree.
Was it causing any problems for anybody running it? Only minor stuff. Was it causing maintenance to be a god-awful nightmare? Definitely. The map for 86xx was a real mess, and really needed to be cleaned up - trying to make sense of the previous code was a real pain, and the whole structure of it was very fragile - any attempt to make a change resulted in random breakage because it was such a house of cards (which I ran into repeatedly while doing the 36b stuff). The decision to change the map was also made for ease of code maintenance - having a similar map to 85xx makes it much easier now to port the 36b changes over to that platform which is the next major task. The boot sequences for the 2 platforms are also a lot more similar now, so I'm hoping that a little bit of pain now will save us a lot more pain in the future.
I agree that there's potential for breakage, but in the case of Linux, which is the vast majority of u-boot users, I think it's minimal. For the non-linux cases, changing an OS to accomodate the new map should be a minor task.
The device tree changes are in hand, and as soon as this goes into the u-boot tree, I will push those to Linux.
We *do* need a comment in the release notes for this revision of u- boot that the map for 8641 has changed so it doesn't catch anyone by surprise.
-B