Manuel Bouyer <[email protected]> writes: > it occured several times, but I was running a -11 kernel with a -10 > userland. I'm quite confident that this was triggered by firefox. I never > noticed this while I was running a -10 kernel; and I've not seen it since I > updated userland to -11.
I started logging wired memory pages and overnight exited firefox. The data is pretty clear: no firefox running and wired pages are stable; with firefox running they grow, irregularly. It's interesting that you say with netbsd 11 kernel/user you don't see this you added AccelMethod UXA and system/X is stable* * of course firefox is never truly stable
