Fix On The Way For One Of The Linux 6.19 Regressions: 52.4% Scheduler Regression
The Linux 6.19 kernel has been a bit bumpy in the scheduler department but at least one fix is on the way for addressing fallout.
Linux 6.19 does bring some nice performance improvements overall but during my early testing there were some regressions and I ended up bisecting some of them to the scheduler changes in Linux 6.19.
Intel's Kernel Test Robot did also report a scheduler regression with Schbench using a Phoronix benchmark run:
That regression was spotted on this scheduler commit.
Queued a few days ago into the tip/tip.git's sched/core branch is sched/fair: Fix sched_avg fold to fix that reported 52.4% regression:
That patch will presumably be sent in soon for the ongoing Linux 6.19 cycle.
Linux 6.19 does bring some nice performance improvements overall but during my early testing there were some regressions and I ended up bisecting some of them to the scheduler changes in Linux 6.19.
Intel's Kernel Test Robot did also report a scheduler regression with Schbench using a Phoronix benchmark run:
"kernel test robot noticed a 52.4% regression of pts.schbench.32.usec,_99.9th_latency_percentile"
That regression was spotted on this scheduler commit.
Queued a few days ago into the tip/tip.git's sched/core branch is sched/fair: Fix sched_avg fold to fix that reported 52.4% regression:
"After the robot reported a regression wrt commit: 089d84203ad4 ("sched/fair: Fold the sched_avg update"), Shrikanth noted that two spots missed a factor se_weight()."
That patch will presumably be sent in soon for the ongoing Linux 6.19 cycle.