blkdiscard hangs on md raid 10 on PCIe SSDs

I have a running Linux 4.19 amd64 system with two PCIe SSDs that I'd like to clear, test, reformat and put in use again. All of the data has already been moved away. I've - tried to trim/discard blocks from the SSD devices /dev/sdX with hdparm, but it said that the devices "do not support SSD TRIM". - tested the random R/W speed with different block sizes - successfully - created md RAID 10 (far 2) /dev/md/ssd taking care of chunk sizes / alignment - just to test, created an ext4 FS over the RAID and it showed that it was discarding the device blocks, which did take some time (as I would expect). - Encouraged by that I've unmounted the FS and then issued blkdiscard -v /dev/md/ssd but it did not produce any output and is blocked (state D) since more than 12 hours. Reading from /dev/md/ssd is still possible, writing few blocks to the start was/is possible, but following writes are blocked (D+) since hours. Killing blkdiscard, with TERM or KILL does apparently nothing. Stopping the RAID is not possible, mdadm reports that it is busy - in use by the blkdiscard process. Are there any options to resolve that short of rebooting the system? (Rebooting is costly because I'd drive to the site, just in case...)