heh. I understand typos happen, but it's still funny.
My early UNIX systems were on Western Electric 3B20S systems and UNISYS 1100-64 and 1100-92 mainframes. Later, a suite of 80 UNISYS 7000 (Power 6/32) machines and HP 9000 minis. I moved on to my current employer where I did system admin on SunOS and HP-UX systems as well kernel level device drivers for the HP-UX 7.0 systems built on the 68030 including full integration of the Mentat System V STREAMS package, Spider X25 L2/L3, HP 9.0 multi-LUN SCSI drivers and a tunnel driver between the BSD TCP/IP stack the the Mentat STREAMS stack to encapsulate TCP/IP over X.25 wrapped in SCSI frames to an external radio.
Part of the process of bringing in the UNISYS 7000 machines was fixing a dreadful attempt at an X.25 L2/L3 in their MPCC interface board and kernel drivers. In the process we found defective DMA interface chips between memory and the synchronous serial input from our Amdahl network. I found tons of defects in the X.25 L2 as well. After going live, I detected a defect in the CPU scheduler and passed the improved version to CCI in Irvine. The impact at their site was amusing. Under the old scheduler, the machine was nearly unusable by the time the 3rd user logged in. After the update they had new problem. All 32 dial-in lines were in use and the machine was still very responsive. That scheduler was rolled out to the Bell Operating Companies, but called the "Bellcore Scheduler" to have some corporate "gravitas". My reward was another promotion.