DragonFly kernel List (threaded) for 2004-06
[
Date Prev][
Date Next]
[
Thread Prev][
Thread Next]
[
Date Index][
Thread Index]
HEADS UP! DragonFly 1.0-RC1 is going to happen tonight.
I am tagging the tree and doing the final version cleanups and testing
as we speak for RC1. I was hoping to do the release rather then an RC,
but there just hasn't been enough time for testing.
But we probably will have only one RC, and I bet a bunch of things will
come out of the woodwork at USENIX. The release will occur either during
or just after USENIX.
One major issue that people need a heads up on is the ACPICA5 support.
Because ACPICA5 is now built as part of the tree, a /modules/acpi.ko
module will be installed. The LOADER WILL AUTOMATICALLY TRY TO LOAD THIS
MODULE WHEN YOU BOOT! I've decided to make this the default to stay
consistent with FreeBSD, even though I may live to regret that decision :-)
There are two ways to disable ACPI. First, manually, by using menu option
#2 instead of #1 when booting. Second, automatically, by setting
the variable hint.acpi.0.disable=1 in your /boot/loader.conf, which
will cause acpi NOT to be loaded by default (when you don't select
any boot menu options).
If you want to disable ACPI permanently regardless of the menu option you
select, you can put 'unset acpi_load' in your /boot/loader.conf.
-----
Last Minute Changes
The following 11th hour work has gone into the tree:
* acpica5 is now a part of the tree and generates the acpi.ko module.
A huge thank you to YONETANI Tomokazu for doing this work! It isn't
completely finished, but we've run out of time. Since most people
follow are going to treat this release as just another LIVE CD it
shouldn't be a big deal.
* The CD, will dual-console boot. You will see output on both the video
screen and the serial port and you can use either to control boot2
and the loader (but not boot1, yet, unless you compile it specifically
for a serial console). Also, the CD has the getty for ttyd0 enabled
so once booting is complete there should be a login: prompt on the
serial port as well as on the screen.
* I implemented interrupt livelock detection to try to make systems
more debuggable. Interrupts are assumed to livelock when
they pass 50,000 interrupts/sec and will be rate limited, hopefully
preventing a system lockup. These numbers may be a bit low. For
faster machines (> 1GHz) you probably should increase the sysctl
variables (set new ones in /etc/sysctl.conf). The sysctl's are
kern.livelock_limit and kern.livelock_fallback.
------------------
THIS IS THE TIME TO TEST!
This coming week is the time to test! We will be releasing 1.0 in about
a week and while there are going to be issues that we know we won't be
able to fix, such as (probably lots of) ACPI issues, we would like to make
this as good a release as possible!
I'll have the ISO for RC1 up tonight. I will make a separate announcement
when the ISO is ready.
-Matt
Matthew Dillon
<dillon@xxxxxxxxxxxxx>
[
Date Prev][
Date Next]
[
Thread Prev][
Thread Next]
[
Date Index][
Thread Index]