OP is using the Seagate FireCuda 530R which according to its dataset consume between 8-9W (8.3 to 8.9W, to be precise) on average during activity. With 6 SSDs at 8W each, the 48W from the SSDs alone would be above the rated 45W output - and that’s not accounting for anything else in the system. Sorry - 5 530R’s is 40W, and then there’s the WD Red SN700 and anything else. Still, those are “average” values - IIRC the M.2 socket is capable of delivering more.
Can you link to the bug/kernel discussion in question here? I see the discussion on the mailing list from 2023/2024: LKML: "Maciej W. Rozycki": [PATCH v9 00/14] pci: Work around ASMedia ASM2824 PCIe link training failures