wim wauters wrote:
> Actually, better to aim Theo at YouTube and their ilk so they stop using
> Flash in the first place.
+1
Flash is a pox. Youtube, and any other flash site, should go over to
open standards. Ogg Theora comes to mind there.
Back to the topic of 4.6 and the extra time for VFS development, I've
been able to fit some use of -current into my regular routine sometimes
and think I might try to work out how to be more active in testing
patches. That means being able to collect traces. Maybe a best
practices for the casual -current user can be posted to Undeadly to help
bring more people closer to the bleeding edge of testing.
At Slackathon, Theo using setting kern.splassert to level 2 to collect
traces to test on more configurations. That seems a bit harder on some
hardware, since using a serial line is the familiar way to collect
traces and a lot of common hardware lacks serial out. What is the best
way to collect traces from hardware that lacks a built-in serial output?
Josh Grosse mentioned last year that the ddb output can be found in the
dmesg saved in the core dump, using the -M and -N options of dmesg(8).
The core dumps are saved in /var/crash, if things go well.
crash(8) mentions a little about how to retrieve information from the
dump after reboot.
e.g.
ddb> boot crash
. . .
ps -N /var/crash/bsd.0 -M /var/crash/bsd.0.core -O paddr
dmesg -N /var/crash/bsd.1
Or systctl ddb.panic=0 might be used to reboot automatically. Again, a
best practices would be useful.
Regards,
-Lars
