While 4.18.0-rc3 kernel can kdump with 'pdbg -p0 -c1 -t0 sreset', in 4.14.49 there isn't any response.
As Nick comments:
the problem is that earlier kernels did not look at SRR1 wake
reason for power save wakeups, and just relied on the interrupt
triggering again when interrupts were enabled. SRESET interrupts
do not re-trigger like this so Linux does not respond to it properly.
These patches are a starting point: 771d4304d07f0 and 78adf6c214f04
While 4.18.0-rc3 kernel can kdump with 'pdbg -p0 -c1 -t0 sreset', in 4.14.49 there isn't any response.
As Nick comments:
Here is output from 4.18.0-rc3: