I thought both HUAWEI and ECI both supported their own native versions of PHyR?
Why is it proving to be such a headache enabling what supposedly already existed? Is this purely a modem problem?
For a start, G.INP is not the same as either of the proprietary PHYR variants. Each DSLAM needs firmware that implements G.INP, irrespective of whether it used to support the proprietary (near-)equivalent.
Because G.INP is different from the proprietary versions, it certainly means that the firmware needs changing, but it likely also requires a change to the DSP coding ... and it could well be that the hardware is not powerful enough to support G.INP bi-directionally (reports that speeds might drop a few Mbps suggest this is true). If the modem is running right on the limits of its capabilities, careful crafting is needed in an upgrade ... and it is likely to take time to get this firmware through BT's acceptance & regression tests. And this could be true at the CPE end and at the DSLAM end.
The second problem seems to be that BT's SIN allows G.INP upstream to be optional in any modem, but they (in an entirely different branch) didn't design DLM to make good choices when G.INP upstream turned out to be unavailable.
DLM was probably designed expecting this possibility to be a rare edge condition, so the engineers didn't pay much attention to it ... and didn't spot the issue arising during trials.