However, the kernel maintainers turned down the patch because the hardware itself was bizarre, underspecified, and unreliable in common cases.
This is an ongoing story and i expect there'll be a resolution in a few months.
ML discussion thread is here: http://lists.infradead.org/pipermail/linux-nvme/2016-October...
Do you happen to know what does he work/has worked on regarding Linux?
More about him, here: http://www.advogato.org/person/dwmw2/
It sounds like trying to tunnel NVMe through something that is only connected over an AHCI interface.
The kernel patches are refused because the kernel devs expect to maintain the code for as long as there are devices in active use that need it once allowed in.
Thus they demand a proper spec for the device behavior so they can test for breakages down the line. And apparently the patches are already violating AHCI behavior expected by the kernel.
Or at least that is what i got out of reading the short ML thread.