if we could log the serial start, i think we can gather the info needed to write the script, example from an arris cm802a:
Serial Flash [Bus:0 CS:0] : w25q64bv 8192KB, 2048 sectors each 4KB
partitions[0] = {.name = U-Boot, .offset = 0x00000000,.size = 0x00020000 (128K) }
partitions[1] = {.name = env1, .offset = 0x00020000,.size = 0x00010000 (64K) }
partitions[2] = {.name = env2, .offset = 0x00030000,.size = 0x00010000 (64K) }
partitions[3] = {.name = UBFI1, .offset = 0x00040000,.size = 0x003b0000 (3776K) }
partitions[4] = {.name = UBFI2, .offset = 0x003f0000,.size = 0x003b0000 (3776K) }
partitions[5] = {.name = Kernel, .offset = 0x000416bc,.size = 0x000d0d44 (835K) }
partitions[6] = {.name = RootFileSystem, .offset = 0x00112400,.size = 0x00284400 (2577K) }
partitions[7] = {.name = nvram, .offset = 0x007a0000,.size = 0x00050000 (320K) }
also this:
*** ACTIMAGE = 1, will try to boot UBFI1 stored @0x48040000
## Executing script at 48040000
Bad magic number
*** UBFI1 is corrupted, try UBFI2...
## Executing script at 483f0000
I think it will be pretty easy to change an existing script for a similar model (6120?) I hope that the fact that it bonds more channels than the 6120 doesn't mean that it will not be good for testing. I see that it does have a TI chipset, and it's definitely puma5 based, hopefully this can mean alpha or the like.