On Tue Oct 7, 2025 at 11:44 AM CEST, Matthew Brost wrote: > Honestly I'd just lean towards dropping the debug version then. It is > basically useless information in Xe as the number of pending jobs can > dervived from the ring's seqnos which is also printed.
Yes, either do that or provide a dedicated debugfs layer in the scheduler itself. For printing driver specific job information we can easily add a callback.
