On Thu, 5 Oct 2000, Wesley Morgan wrote:
> I'm getting a panic in ufs_extattr_uepm_destroy() because in ffs_vfsops.c
> it is being called (line 788) with ump NULL:
>
> ufs_extattr_uepm_destroy(&ump->um_extattr);
>
> Of course disabling FFS_EXTATTR gets rid of this:)
Hmm. I added these change
On Thu, 14 Sep 2000, Julian Elischer wrote:
> I've never thought of a use for fdescfs...
I used /dev/fd/1 just yesterday for a third-party precompiled binary
that insists on outputting to a file specified on the command line.
Slapping in /dev/fd/1 lets me stuff the command in a pipe chain. Same
On Sat, 9 Sep 2000, Brian Somers wrote:
> > Seigo Tanimura wrote:
> > >
> > > I have been suffering from this problem for almost 2 months. When I
> > > remove a pcmcia ethernet card from my laptop PC, routed(8) announces
> > > updated routing information by multicast, leading to a kernel
> > >
On Fri, 8 Sep 2000, John Baldwin wrote:
> Ok, an update on the dirty buffers on reboot:
>
> If you use the reboot command, you will get dirty buffers. If you use
> 'shutdown -r now' instead, you won't get dirty buffers. Thus, as a
> workaround for now, use the shutdown command to reboot your
On Tue, 5 Sep 2000, Warner Losh wrote:
> In message <[EMAIL PROTECTED]> Wes Peters writes:
> : Ejecting an interface configured up will do that. ifconfig the interface
> : `down' and then `delete' before ejecting it.
>
> At best this is an unsatisfactory workaround. if_detach should cause
> t
As of this afternoon, I seem to be having problems with machines hanging
on incoming TCP connections. As soon as some process attempts to do
useful TCP stuff (incoming telnet, ssh, et al), all process hang. If I
break to ddb, a ps shows that telnetd has no wmesg. If I begin tapping on
various
Yup, there's been chatter on this in a number of forums. I must admit
that I am very please -- the Arla client is great, but Milko has a long
way to go before it is production quality. Having access to the
Transarc/IBM source will presumably greatly facilitate the development of
Arla, and also
te: Tue, 29 Aug 2000 21:49:10 -0700 (PDT)
From: Robert Watson <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED], [EMAIL PROTECTED]
Subject: cvs commit: src/sys/kern kern_event.c kern_ktrace.c kern_proc.c
kern_prot.c kern_resource.c kern_sig.c sys_process.c src/sys/miscfs/pr
stedbsd.org 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Thu Jul 27
13:35:58 PDT 2000
[EMAIL PROTECTED]:/usr/src/sys/compile/DEV1 i386
dev1#
Robert Watson
Research Scientist
NAI Labs at Network Associates
...
/../../contrib/ncurses/ncurses
-I/usr/src/lib/libncurses/../../contrib/ncurses/i
nclude -Wall -DFREE
On Tue, 22 Aug 2000, Walter Belgers wrote:
> Last week I was at USENIX where Niels Provos talked about his
> implementation of encrypted swap in OpenBSD. What is does is encrypting
> all memory that gets swapped out, keeping the encryption keys in memory.
> A test showed that all kinds of intere
"me too"
pcm0: mem 0xfda0-0xfdaf,0xfac0-0xfaff irq
5 at device 0.1 on pci1
pcm0: play interrupt timeout, channel dead
Then EBUSY until the cows come home. I chatted with Cam some this
evening; this problem seems to have come about as a result of driver
restructuring. This chi
I've been using vmware with great success under -STABLE. Take a look at
/usr/ports/emulators/vmware2. It's commercial, but works a lot better
than all the alternatives. I've had decent success with pcemu for
DOS-based programs, although there are a few quirks. I found bochs also
worked pretty
On Fri, 28 Jul 2000, Brian Fundakowski Feldman wrote:
> On Fri, 28 Jul 2000, Robert Watson wrote:
>
> > Nope, it didn't appear to help. When I move the mouse around, it
> > intermitently pauses, perhaps once a second, for a short period of time.
>
> Robert, how
.
>
> Kazu
>
> >Robert Watson wrote:
> >>
> >> I'm using a Micron P133 box with a PS/2 mouse. Up until this morning, I
> >> was running 4.0-STABLE from a month or two back. I upgraded to
> >> 5.0-CURRENT, and since that time, my mouse has been
I'm using a Micron P133 box with a PS/2 mouse. Up until this morning, I
was running 4.0-STABLE from a month or two back. I upgraded to
5.0-CURRENT, and since that time, my mouse has been responding slowly and
erratically, jumping as it moves, et al. The mouse daemon seems to be
consuming more
On Wed, 5 Jul 2000, John Baldwin wrote:
> The headers will always be installed in the right place in
> /usr/include: Makefile's are editable. As far as kernel
> compiles, symlinks can be created in the work directory as
> one possible solution. For example,
> sys/compile/i386/GENERIC/netinet ->
This is great news -- one of the big hangups in our interop testing at NAI
Labs was the like of IKE on FreeBSD. I notice that right now racoon is a
port -- assuming this interpretation is correct, are their any plans to
integrate racoon as a base system component? As you point out, without
IKE,
On Wed, 7 Jun 2000, Robert Watson wrote:
> That said, I'm a little puzzled as to where securelevel is being defined
> -- a bunch of stuff depends on the variable and yet my test build
> succeeded without it in there. And you go that far also -- far enough to
> boot rather than
At bde's request, I moved kern.suser_permitted to kern_prot.c and
accidentally also trimmed kern.securelevel. I just committed it back into
kern_mib.c. Please let me know if there are further problems.
That said, I'm a little puzzled as to where securelevel is being defined
-- a bunch of stuff
I use a 4.0-RELEASE machine as my build machine, and have been
experiencing intermittent build fails in the perl code for a while. It
does not happen consistently, and when it does, does not give a useful
error message of any sort. I compile with -j, usually between 3 and 6
processes. Here's t
FYI: I committed the addition of the magic number and version information
an hour or two ago. It seems to work fine for me, but please let me know
if you have any problems. A migration tool doesn't seem useful yet, but
is now feasible :-).
In a day or two, I'll send a post to freebsd-fs descri
On Wed, 19 Apr 2000, Garance A Drosihn wrote:
> At 3:41 AM -0400 4/19/00, Robert Watson wrote:
> >I hope not to change the format any further. I've been considering
> >introducing a backing file header version number of some sort, but
> >this is only necessary if
I just committed a change to the extended attribute backing code that
modifies the per-attribute header. The result is that backing files used
and created from now on have a different format, and weird and unfortunate
things will happen with backing files before this change. I doubt anyone
is do
Since it appears to work for me, I'm going to go ahead and commit the
patch before too many other people run into this. Please let me know if
you have further problems and I'll get them fixed up ASAP.
On Sat, 15 Apr 2000, George W. Dinolt wrote:
> Content-Type: text/plain; charset=us-ascii
> C
Yup -- I neglected to update the ext2fs code (which uses UFS stuff) to
include the requisite include files. Please try the attached patch
against src/sys/gnu/ext2fs, and let me know if it works, and I'll go ahead
and commit it. I caught the weird Coda dependancy, but guess I missed
this one.
Not sure if this should go to -current or -stable, since we seem to get a
lot of instant MFC's these days :-). I upgraded a notebook from
4.0-RELEASE to -STABLE last night. After doing so, I noticed that the
middle mouse button emulation in moused seems to be fairly broken -- i.e.,
once it's en
5.0-CURRENT -- was doing a make buildworld -j 2. Sadly, I don't know what
exact date the source was from, as I had just cvsup'd and started
building, but I expect in the last week and a half. I was running with
capabilities patches going, but I wouldn't imagine that it would cause
this particul
I installed 4.0 on a notebook yesterday, using the docking station. As
previously described, I had hardware probing problems without using the
ethernet card in the docking station. Well, sadly, X11 requires an extra
option or two to work when with the docking station, but I figured that
out and
eld
saying, ``Which of these IRQs should I note use'', instead, ``Which should
I use''. Or the like.
On Tue, 14 Mar 2000, Robert Watson wrote:
>
> Yesterday I spent a fair amount of time attempting to get a 4.0 snapshot
> to install on my notebook (Dell Latitude CPi), whi
Sounds good to me, as long as it runs :-).
BTW, ran into another nit from the 02/13 snapshot. I installed the
X-kern-developer distribution, discovered X11 didn't work, so went back
into sysinstall to install X11 stuff. I selected some combination of X11
components, and chose releng3.freebsd.o
It looks like the X11 associated with the snapshots on current.freebsd.org
is still broken:
/usr/libexec/ld-elf.so.1: Share object "libXThrStub.so.6" not found
Installing -current boxes for testing and development would be a lot
easier if this worked. :-) Especially leading up to releases wh
Yesterday I spent a fair amount of time attempting to get a 4.0 snapshot
to install on my notebook (Dell Latitude CPi), which until now has been
happily running 3.3-PAO. Sadly, it seems not to like my ethernet card.
When installing, sysinstall provides three IRQ exclude options before
initializ
Just a quicky. Since we now support IPv6 by default, you can imagine
environments where you have an interface and IPv4 is not in use. My
tcpdump is from a rather elderly 4.0, so this may be fixed already, but:
# tcpdump -eni bridge0
tcpdump: WARNING: bridge0: no IPv4 address assigned
tcp
On Thu, 9 Mar 2000, Boris Staeblow wrote:
> On Wed, Mar 08, 2000 at 11:15:23PM +0100, Luigi Rizzo wrote:
>
> > >
> > > Is it possible that bridging is broken in -current and -stable?
> >
> > no, but the "de" driver on bridging is now unsupported and i could not
> > find the time to make it wor
On Sun, 27 Feb 2000, Kris Kennaway wrote:
> On Sun, 27 Feb 2000, Robert Watson wrote:
>
> > path, we should consider moving a few of the things in /sbin and /usr/sbin
> > into /bin and /usr/bin respectively. For example, md5 is sufficiently
> > useful for all use
I've posted about this previously, and am still hoping for a useful answer
:-). I have a box with an ISA 3Com 3C509 Etherlink III card in it, but
the ie0 and fe0 probes now come before the ep0 probe in the boot sequence.
If those two probes are enabled, as they are on the same port (0x300) they
While trying to explain to someone how to run ping, I noticed that by
default, we exclude /sbin and /usr/sbin (etc) from the normal user path.
There are various arguments as to whether that is a good idea or not
(especially given that ping and md5 are both in /sbin...?), but my current
observatio
I have a somewhat old 486 box I'm using as a dual-homed host. It has two
SMC EtherEZ ISA cards in it, and works fine under 3.4-STABLE. The cards
are on irregular ports/IRQs/et al.
My problem is that in the kernel config, it is only possible to hard code
ed0's settings, but that ed1 is probed (
On Fri, 18 Feb 2000, Kris Kennaway wrote:
> All of the ports which explicitly depend on openssl should be working on
> all supported versions of FreeBSD, modulo screwups :) Jim Bloom has been
> putting a lot of work into getting these working - I have a couple of
> patches to commit, but they mos
Kris,
I was pointed to you for questions regarding whether or not certain ports
would be working udner 4.0-RELEASE -- specifically, OpenSSH and related
applications which depend on SSL/RSA. Do we plan to provide a consistent
and documented way for users of FreeBSD to go from the RSA-disabled ba
On Fri, 18 Feb 2000, Kris Kennaway wrote:
> On Fri, 18 Feb 2000, Robert Watson wrote:
>
> > > What value of -j? I've built fine with -j4 and -j8, but this says that the
> > > depend target hadn't run before the all target did. Did you 'make depend'
On Fri, 18 Feb 2000, Kris Kennaway wrote:
> On Fri, 18 Feb 2000, Robert Watson wrote:
>
> > A further data point: I was building with a -j, so maybe there's a
> > dependency mixup with the beforedepend target? If I manually cd to the
> > directory and do a make
Another technique that could be used, and gets discussed occasionally on
-security, is passing authentication information via ancillary data
transfer on UNIX domain sockets. You could limit the effectiveness of DOS
attacks by rate limiting per-uid, for example.
It should be noted that both the
A further data point: I was building with a -j, so maybe there's a
dependency mixup with the beforedepend target? If I manually cd to the
directory and do a make beforedepend, things continue naturally from
there.
On Fri, 18 Feb 2000, Robert Watson wrote:
>
> Was wondering if anyo
Was wondering if anyone else was seeing this one. I'm following the
jail(8) build directions on the 02-14 snapshot. I've had the build
directions work previously, so I'm a little puzzled. During the make all
phase, I get the errors below. I cvsup'd this morning, but still get the
same results
On Thu, 6 Jan 100, Darren Reed wrote:
> For what it's worth, I think releasing 4.0 *without* IPv6 support
> is a mistake. Why ? Because in < 12 months FreeBSD 5.0 will be
> released *with* IPv6 support (I'd count IPv6 as being a big enough
> change to signify a major release number change). If
This is not a comment on your code, which I have not inspected yet, but
instead on the idea of the optimization. This is probably not a serious
objection, but keep in mind that this optimization does not produce
identical behavior in some case. For example, imagine that the user has a
number o
On Tue, 14 Dec 1999, Peter Wemm wrote:
> The RZ1000 is *dangerous*! We are doing no favours by making it run.. :-/
> IMHO It is better to loose the user by not playing ball than to corrupt
> their data or run unreliably and make them hate us for it.
>
> http://www.faqs.org/faqs/pc-hardware-faq/
; and then the audit tag slides forward.
>
> Not to interrupt in the middle of this discussion but you might
> want to check with robert watson before you guys get too deep here since
> he is working on a FUNDED Posix.1e implementation for FreeBSD. And has
> already posted some EARLY MA
On Sat, 13 Nov 1999, Alexander Leidinger wrote:
> On 12 Nov, Robert Watson wrote:
>
> >> >> >> (102) netchild@ttyp2 > grep cat /etc/rc.conf.local
> >> >> >> spppconfig_isp0="`cat /etc/isdn/connect.parameters`"
> >> >
On Wed, 10 Nov 1999, Alexander Leidinger wrote:
> On 11 Nov, Daniel C. Sobral wrote:
> >> >> (102) netchild@ttyp2 > grep cat /etc/rc.conf.local
> >> >> spppconfig_isp0="`cat /etc/isdn/connect.parameters`"
> >^^^
> > Calling programs from any of the rc.conf files is conside
Love to know why my freebsd-arch subscription disappeared, although the
rest appeared to stick around. I just resubscribed, but was subscribed
before (but not sure about until when) as
[EMAIL PROTECTED]
What is strange is that if that was bouncing, I would have expected, say,
my -current subs
As I understand it, RTP is only a framework for a protocol, and not
actually a protocol. I.e., they define some headers, but no delivery
semantics. There are some RFC's that talk about how to put various
video/audio formats on top, but I think they have explicitely avoided
talking about things li
So will bitkeeper provide a nice interface for migrating code from an
existing and well-established CVS repository to whatever they use?
I'm quite happy to allow them to test bitkeeper in a production
environment before using it in one myself, needless to say. :)
On Sat, 1 May 1999, Matthew Jacob
On 1 May 1999, Dag-Erling Smorgrav wrote:
> Soren Schmidt writes:
> > [...]
> > There are lots of better cards out there, but stating that the 3c509
> > doesn't work, is totally out in the dark...
>
> Well, excuse me for breathing. I'll just take your word, and dismiss
> the problems I've had wi
I suspect not related to the Matt NFS stuff, because I see this same
ICMP/Dummynet problem (see my post recently, independent of PHK's post) on
code from before egcs. I just noticed this morning while testing some
network protocol simulations.
On Fri, 30 Apr 1999, Alfred Perlstein wrote:
> On F
I have a -current install from just before egcs and am using DummyNet to
experiment with network protocols. I've been applying the rules to the
loopback device so as to prevent other interference. It seems to work
well, except that ICMP seems not to be working for me. At one point in
the past,
Singed shorts are often the result of sitting on a stove during the
summer.
We also recently saw a "hanging root device to..." in a boot on one of our
machines.
:-)
Robert N Watson
rob...@fledge.watson.org http://www.watson.org/~robert/
PGP key fingerprint: AF B5 5F FF A6 4A
We upgraded a crash machine from 3.1-RELEASE to 4.0-CURRENT from just
before the EGCS switch was pulled. The machine is a Pentium 166 MMX
overdrive. Prior to the upgrade, it correctly probed the Kensington KNE
2100 (something like that) with the lnc driver as being at 0x300 irq 5
drq 6.
The wo
On Sat, 10 Apr 1999, Kevin Day wrote:
> On the shell servers I run, we've got 200-300 users running tasks.
> Occasionally, through intent or misconfiguration, a user either forkbombs,
> or gets a large number of processes running sucking lots of cpu.
>
> I'd like to see an option that makes all t
On Mon, 12 Apr 1999, N wrote:
> Hi,
>
> > I don't use the FreeBSD patched version, as I use the version with the
> > KerberosIV patches (unfortunately the FreeBSD port doesn't do that, but I
> > don't have time just now to make it do that :-). It seems to put the IP
> > address into the wtmp corr
On Mon, 12 Apr 1999, Brian Somers wrote:
> [.]
> > I got sick of seing "invalid hostname" in my wtmps a while ago on my 2.x
> > machines. That is an exceptionally useless piece of behavior, if you ask
> > me. Sshd writes out IPs and I find that to be much more consistent (and
> > useful).
>
On Sat, 10 Apr 1999, Rahul Dhesi wrote:
> For some years I have been using patched utilities under SunOS to show
> full host names in the output from the 'who', 'finger', and 'last'
> commands. (Traditional UNIXes truncate host names to about 16
> characters.)
>
> I have been thinking of patchin
On Mon, 5 Apr 1999, Terry Lambert wrote:
> After you said this, I found it so hard to believe that I had
> to go look.
:-)
> All I can say is, well I'll be damned; you could knock me over
> with a feather, and that doesn't happen often.
>
> I'm sure SVR4 and UnixWare is not like this; I had to
On Mon, 5 Apr 1999, Terry Lambert wrote:
> > What I would like to do is have a child process read
> > from an inherited file descriptor without affecting the
> > parent process's access to the descriptor (for example,
> > the offset it reads from using a normal read() or readv()).
>
> This should
Terry,
In the BUGS section of the 4.0-CURRENT aio_read man page, the following
comment is made:
The value of iocb->aio_offset is ignored.
Is this actually the case, and what would be required to fix it? Does
this comment imply that reads always occur at the current file offset from
the
On Fri, 19 Mar 1999, Andrew McNaughton wrote:
> > Dmitry Valdov wrote:
> > > I think that there is only one way to fix it - it's to disable making
> > > *hard*links to directory with mode 1777.
>
> I don't use quotas, and don't know a great deal about how they operate,
> but I think there's anoth
On Wed, 10 Mar 1999 sth...@nethelp.no wrote:
> > Uh, no. Invariants are for developers who want to make sure their code
> > is correct. There is no reason why an end user would want to build a
> > kernel with invariants enabled. Invariants will *not* increase data
> > safety. If they have any effe
So this is actually just a general response to the whole thing--one of the
things I actually dislike about rc.conf is its flexibility: the user can
put anything script-wise they like into it. My temptation would be to
reduce the flexibility: to have a simple name:value configuration file
(with app
So actually, I have a question about this. How is the syscall glue
generated, and when. Pretty much all the userland libraries call syscalls
using symbols of the same name rather than the syscall() wrapper,
presumably for performance and feasiability reasons (especially with
pipe()). When is th
I get the following message from the zp0 driver on boot:
zp0 XXX: driver didn't set ifq_maxlen
lo0 XXX: driver didn't set ifq_maxlen
I assume they should be setting it? :-)
Robert N Watson
rob...@fledge.watson.org http://www.watson.org/~robert/
PGP key fingerprint: 03 01 DD 8E
On Wed, 3 Mar 1999, SXren Schmidt wrote:
> It seems Daniel O'Connor wrote:
> >
> > On 02-Mar-99 SXren Schmidt wrote:
> > > Its in the works, together with the tagged queuing some of the
> > > newer drives supports.
> > Wow! :)
> >
> > Is there any chance od adding the ability to 'wire' devices
Looks like it may actually have been triggered by a bug in my kernel
auditing code, which turned out to be doing stuff when I didn't think it
should have. Sorry about that :-).
On Mon, 22 Feb 1999, Robert Watson wrote:
> kernel code cvsup'd this morning, I belie
kernel code cvsup'd this morning, I believe.
FreeBSD sleipnir.watson.org 4.0-CURRENT FreeBSD 4.0-CURRENT #20: Mon Feb
22 11:53:53 EST 1999
rob...@sleipnir.watson.org:/usr/src/sys/compile/SLEIPNIR_AUDIT i386
panic(f01f13a7) at Debugger+0x37
worklist_remove(f0bd56a0) at worklist_remove+0x2a
free_di
I've been unable to buildworld for a day or so now, much to my
frustration. I've tried recvsuping my /home/ncvs, as well as rechecking
out, etc. I get the following:
cd /usr/src/lib/libskey; /usr/obj/usr/src/tmp/usr/bin/make all;
/usr/obj/usr/src
/tmp/usr/bin/make -B install
cc -pipe -DPERMIT_C
On Mon, 1 Feb 1999, Matthew Dillon wrote:
> :So this probably works for non-root users on files like /dev/zero that can
> :produce as much data as you might be interested in, suggesting a fun
> :denial of service attack for the bored and/or insane.
>
> Presumably the datasize limit can be use
On Mon, 1 Feb 1999, Matthew Dillon wrote:
> Uh. Mm.. Hmm :-)
>
> i = read(fd, &size, sizeof(size));
> ... malloc(bufsize * sizeof(char))
> i = read(fd, buf, bufsize);
>
> When you are reading /dev/mem, 'size' can turn out to be anything.
> You are t
. My kernel code is a few days old on the 4.0-CURRENT branch.
Source: auditd.c (just disable auditing stuff to make it compile). Invoke
as auditd /dev/mem.
/*-
* Copyright (c) 1999 Robert Watson
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
On Wed, 27 Jan 1999, Garrett Wollman wrote:
> < said:
>
> > It's not clear to me, when thinking of introducing a new file (say, for
> > auditing support :), what I should name it. Would it be kern_audit.c or
> > sys_audit.c?
>
> Depends on what it is auditing. If it only auditing the basic I/
I was wondering what the naming scheme for files in /usr/src/sys/kern was
:). There seem to be several sorts of files there--
bus_if.mdevice_.m imgact_*.c
inflate.c init_*.ckern_*.c
link_*.cmake*.{pl,sh} md5c.c
On Tue, 26 Jan 1999, RT wrote:
> I highly doubt that I'll ever use FORTRAN directly or indirectly. If it's
> not used by a vast majority, it should be optional...
So the problem seems to be that 'included in the system' is a problem
because the system gets unwieldy in terms of junk a lot of peo
On Wed, 20 Jan 1999, Mike Smith wrote:
> > I was just pointing out that having things in subdirectories
> > is better than having a zillion files piled into a single directory.
>
> I'm torn between agreeing that it's tidier and disagreeing on the
> grounds that it's much more of a pain to admini
601 - 682 of 682 matches
Mail list logo
