-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
t #include
without testing on other platforms.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
on't
think we need to create a contrib infrastructure for it.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
re the FP state
from the ucontext.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
hokey.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ot planning on it for 2.0, just getting HAMMER's base
functionality implemented is already pretty challenging. But there's
nothing particularly difficult about the concept.
-Matt
years, it can wind up being a completely different entity than the
original was. Now what do you do?
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
hands and say 'I don't know'.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
The issue between OpenBSD and Linux revolves around Sam Leffler's code
(as in the Sam Leffler of FreeBSDland).
Basically Sam's code is dual-licensed and he has explicitly said that
people can use one license to the exclusion of the other.
OpenBSD then took Sam's code and, per
tead of the scripts under
'localdaemons'.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
he easiest
solution at the moment.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:Great, thanks a lot!
:Btw, Which man page contains doco for all the options?
:
:Petr
man mount_nfs
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
?
:
:Regards.
:Vincent
It's a bug. Try this patch.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Index: md.c
1.10.1 has been rolled! A serious mbuf leak and issues with hardware
packet checksum support have been fixed in this sub-release.
-Matt
ks again.
:
:-- Will
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ysctl with sysctl.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:Damn, I should have caught that. The pkgsrc bootstrap installs a good
mk.conf, so we should remove the mk.conf from the skel. Doing so as we speak.
:
:cheers
: simon
:
:--
:Serve - BSD +++ RENT this banner advert +++ASCII Ribbon /"\
I'm thinking we will want something in the
conf stuff.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
portable you can just avoid using fdisk entirely and disklabel
the unit directly (disklabel -r -w da0s0 auto).
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:Thanks for the hint, but actually I was trying to fdisk a USB stick.
:Kernel says the following when I plug it in:
:
:umass1: ChipsBnk Flash Disk, rev 2.00/1.00, addr 3
:da0 at umass-sim1 bus 1 target 0 lun 0
:da0: Removable Direct Access SCSI-2 device
:da0: 1.000MB/s transfers
:da0: 1010MB (5173
* Also run the size command on the module:
size /modules/if_ipw.ko
That early in the boot there's probably no way to get a kernel core
dump.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
): b_bcount 1024 is not on a sector boundary (ssize 2048)
You are trying to fdisk a CD ? You can't fdisk a CD.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:when inv
See, this is why I just stuck a line in /etc/rc.local. Much easier
then muffing about with a broken RC .
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
orks.
oops. Ok, fixed. I keep meaning to fix the rc script. I'll just
add a manual start in rc.local.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
h dragonflybsd.org
You would have a line like this:
search mydomain.com
What this line does is allow you to shortcut the name you use to
access your machine. That is, you can just go 'ping mymachine'
instead of having to say 'ping mymachine.m
ess up your hard drive, though.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Index: if_re.c
===
RCS file: /cvs/src/sys/dev/ne
the
driver.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
It looks like there is a mbuf leak somewhere in -RELEASE and in -HEAD.
I do not know where it is yet but if anyone has done any recent
networking work (general work, not the wireless stuff), please check
over your commits.
-Matt
ng issues. We probably need to
implement the various posix scheduling methodologies, too.
SMP isn't on my priority list but that shouldn't stop anyone who wants
to work on it.
-Matt
ject these days.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Jeremy did another interview of me in Kernel Trap, here's the URL:
http://kerneltrap.org/node/14116
-Matt
instructions, so you'd better
:put something at that URL -- or duck!
I'm moving on to post-release work, but if someone wants to create a
Wiki page, please be my guest!
-Matt
Matt
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
2d24a36720b37a7e
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
The 1.10 release ISO is now on crater! The official announcement will
be made tomorrow (Monday) after our mirrors have had a chance to get a
copy.
There will be some minor release errata. The pkgsrc issue with the
bootstrap has not been fixed. Also, the installer defaults to
better solution for I/O, for sure.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:Now look how stupid I am. I accidentally exposed my 3128 port to the
:World Wide Web and all creatures from the dark wilderness were trying to
:abuse it. I have restricted it properly now and pflog reports about 100
:connection attempts to port 3128.
:
:Matt, your hints led me to the final soluti
o what the machine is doing,
using netstat -tn, netstat -m, ps, etc. Is squid simply trying to
maintain more open TCP sockets then the system can handle?
-Matt
Matthew
:
:How far do you reckon (percentage) is the project progressed if we take
:distributed as end goal and the source-split as begin?
:
:--
:mph
I'd say we're well over half way there.
-Matt
Matt
:Are any parts of the distributed computing (taking advantage of more than
:one DragonFly computer) available for testing?
:
: Jeremy C. Reed
Nuh uh.
-Matt
Matthew Dillon
d src.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
o use pkgsrc infrastructure as much as we can
because it provides a very easy method to customize our ISO.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ow us to do something like 'make installer dvd release'
or 'make installer dvd quickrel' without having to create a
multiplication of make targets.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
everything, uncompressed and ready to use.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
soon.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
synchronized with the release.
Both the HEAD preview tag and the release slip tag have also been
updated.
No release date has been set yet but this coming weekend is looking
real good now.
-Matt
Matthew Dillon
o elf32_loadfile() is
loaded into the file_format structure). The code you are
referencing is what calls through to that function.
The l_load declration you found is not specifically related.
-Matt
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
a mbuf leak somewhere.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
l the mbufs.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
creates a problem when you tar or cpdup
a directory tree because you wind up breaking the varsym. So for
now I think the only real solution is to modify firefox to understand
varsyms.
-Matt
Matt
rapper for this anyway so I went
ahead and wrote one. The pps rate check code is kinda hokey.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Index: ker
There are two outstanding issues which I think have to be fixed
before we can release.
The first is that loading EHCI (usb) at boot time causes USB hard
drives to fail to detect.
The second is that 'vinum start' often winds up corrupting the whole
kernel in a fairly scary
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ave no idea.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
and amd64_elf_obj into
our boot loader and seeing if that fixes the problem.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
We're going to have to delay the 1.10 release for another week due
to the NATA fix that just went in and a few other hangers-on. It needs
more testing.
I will be branching the source tree for the release today or tomorrow
though.
you get, if any, when you load them with the boot loader.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
e device and attaches the driver.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:http://lists.freebsd.org/pipermail/freebsd-current/2003-July/006889.html
:
:Annoucning DragonFly BSD!
:Matthew Dillon dillon at apollo.backplane.com
:Wed Jul 16 12:41:34 PDT 2003
Happy Birthday to us!
-Matt
ally solves the same problem that SSI
does, but I also firmly believe that doing SSI properly requires complete
integration into the kernel to really be effective.
-Matt
Very nice. Justin, you gonna throw it into cvs so we don't lose it?
-Matt
www. Nah, I'll make it 195000.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:i know what 1.10 is mean. but just not compatible with the fact that
:the numerical value of the new version number 1.10 is less than those
:old version number, 1.8, 1.6, 2.0 is not what we want.
Yup. But I'm afraid it is the best choice out of a host of bad choices.
;ll use 195000 for
DragonFly_version.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Matthew Dillon
<[EMAIL PROTECTED]>
:Binary packages are available from
:http://www.pkgsrc-box.org/packages/2007Q2/DragonFly-1.8/All
:and the stable link been adjusted accordingly.
:
:Joerg
Good timing!
-Matt
Matthew Dillon
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ly associate
the descriptor used to restore the checkpoint instead of trying to
lookup the original checkpoint file.
I also got rid of the debugging kprintfs to the console on checkpoint
restore.
-Matt
ting
a program to force it to be written out in full (thus removing the
dependancy on earlier checkpoints).
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
bly but the one in HEAD will compile under RELEASE.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:Matthew Dillon <[EMAIL PROTECTED]> writes:
:> You can monitor system calls made by programs with ktrace.
:
:Ok, but what if I want to watch access to a particular file system
:(e.g. a mounted cdrom) by *all* system processes? :)
:--
:Dave Hayes - Consultant - Altadena CA, USA
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
race.
you can monitor *MODIFYING* operations to filesystems with the
journaling support, e.g.:
mountctl -a -o fd=1 /tmp:monitor | jscan -d
Hmm. mountctl -l doesn't work. I'll go fix that.
-Matt
use a 64 bit off_t at the beginning, even though
at that time the BSDs already *HAD* 64 bit off_t's.
Actually, I think Microsoft is to blame too, but at least they had an
excuse 20 years ago.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
talk to it via ucom0.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
el32.h and sys/disklabel.h
will become a generic abstraction layer. The original struct disklabel
will become struct disklabel32, etc.
All of the disentangling of the disklabel is almost done!
-Matt
Matt
a again until you
have it right.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
#!/bin/csh
#
# A file like this is typically copied to /usr/local/etc/rconfig/auto.sh on
# the rconfig server and the rconfig demon is
luck.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:/home on leaf seems to be full.
:
:Best Regards,
:sephe
Time to clean up, everyone!
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
do a make ugprade ? If not, do it. Maybe you don't have
the new MAKEDEV.
ls -la the devices you are trying to use, before and after.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
advance,
:Joe Holden
Yes, what you did will work, once you fix the devices in /dev.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
I will note one thing on remote sessions... if you are running through
NAT (network address translation), which is a very typical function of
today's cable modems, the NAT translation will often timeout if the
link stays idle too long. This will cause the link to die when you
c
I think the last time I used screen was 20 years ago. I just leave
all my xterms open. Sometimes I have upwards of 30 windows open across
four virtual screens in X. When people were describing the NATA bugs
I had an xterm open in an unsaved vi for over a week with all my notes
. Timer 1 basically either works, or the machine hardlocks
when the kernel tries to configure it at boot. Since your machine
seems to be working, I'd say everything is good now.
-Matt
disable certain features.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
with the timer. Is your clock running
at normal speed or is it sped up too? (e.g. run 'date' a few times
from a shell).
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
he explantion of the cause and effect would be highly appreciated.
:
:Cheers
:kmb
That is ultra odd. What did you upgrade from? An earlier release or
a less recent HEAD?
-Matt
M
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
ically linked. I really doubt init is gonna be happy with
that.
Get rid of the special make variables. Just compile up a normal world.
/bin is only 6MB and /sbin is only 16MB anyway. That's nothing in
today's world.
-Matt
interrupts still need it, too.
We're looking for volunteers to help lock up subsystems. The main
one that is just about ready is the network stack.
-Matt
Matthew Dillon
Matthew Dillon
<[EMAIL PROTECTED]>
t the same.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
A large chunk of the kernel still runs under the big giant
lock, including the light weight processes that libthread_xu
uses, so something like mysql is going to hit a lot of BGL
contention.
You may be able to get DragonFly to run on the machines you
were having problems wi
As people may have noticed from the commits I'm working on separating
out the all the hardcoded references to struct disklabel. I'm
separating out all the disk management layering code into distinct
pieces (slice code, disklabel, partition management).
I might make a mistake,
801 - 900 of 1411 matches
Mail list logo
