AWS Engineer Reports PostgreSQL Performance Halved By Linux 7.0, But A Fix May Not Be Easy
19 points by radim
19 points by radim
The fix here is to make PostgreSQL make use of rseq slice extension: https://lkml.kernel.org/r/20251215155615.870031952@linutronix.de
The lkml thread is far more informative than the Phoronix summary, especially the messages from Andres Freund.
https://lore.kernel.org/lkml/yr3inlzesdb45n6i6lpbimwr7b25kqkn37qzlvvzgad5hfd7ut@xv4cihno76wu/
He wrote on the orange site
But it's an absurd workload in an insane configuration. Not using huge pages hurts way more than the regression due to PREEMPT_LAZY does.
With what we know so far, I expect that there are just about no real world workloads that aren't already completely falling over that will be affected.
I had the impression that rseq has very difficult requirements that are hard to satisfy if you aren’t libc. Maybe the slice extension makes rseq easier to use?