sticker of transparant, white puffy hits Switzerland
Was also included in the 3.9 CD parcel I've received today! Aweseome, now the directors have to look at puffy everytime they enter my door ;) Thanks, guys. -- Stephan A. Rickauer --- Institut f|r Neuroinformatik Tel: +41 44 635 30 50 Universitdt / ETH Z|rich Sek: +41 44 635 30 52 Winterthurerstrasse 190 Fax: +41 44 635 30 53 CH-8057 Z|richWeb: www.ini.ethz.ch RSA public key: https://www.ini.ethz.ch/~stephan/pubkey.asc --- [demime 1.01d removed an attachment of type application/pgp-signature which had a name of signature.asc]
Re: OpenBSD todo list?
On Tue, Apr 11, 2006 at 12:40:45AM -0401, Ray Lai wrote: On Mon, Apr 10, 2006 at 08:44:36PM -0700, Shawn Nock wrote: A quick search of the archive and google didn't turn anything up, so I'll ask here. Is there (if not could there be) a document that describes portions of the tree that particularly need attention? I am looking for a way to contribute and without a little direction the task seems daunting. The FreeBSD folks recently started maintaining such a todo list. It seems to have worked out fairly well for them. I realize that those in a position to put together such a list are also the ones not likely to have the time, but I believe this could prove useful (I assume there are more like me who are new and/or haven't found a comfort zone/focus yet). OpenRCS needs help. Diffs that implement missing functionality, diffs that match GNU RCS behavior in existing functions, and additional regression tests are all welcome. I look forward to your code. There was also an open request for help with m4, not too long ago, on [EMAIL PROTECTED] IIRC, Marc Espie would like traceback functionality. Not sure if someone is already working on it. Joachim
Re: gcc miscompiles ntohs16() inline assembly in OpenBSD 3.8
On Thu, 6 Apr 2006, chefren wrote:
On 04/05/06 02:07, Andrew Pinski wrote:
Actually I bet ntohs16 is violating C aliasing rules.
So getting rid of GCC actually is wrong. Getting rid
of these aliasing violations is the correct way.
-- Pinski
Interesting, how do you figure that?
The relevant definition is:
#define __swap16md(x) ({ \
u_int16_t __swap16md_x = (x); \
\
__asm (rorw $8, %w1 : +r (__swap16md_x)); \
__swap16md_x; \
})
i.e. a block that returns a short value (a compound statement expression
is the gcc terminology, we believe).
Where would aliasing rules enter into this situation?
Thanks for looking at this, if we can help or assist with solving this: please
don't hesitate to ask.
The following diff (backported from gcc PR
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10692) seems to fix the
problem. I hardly know i386 assembly, so please check if the produced
code is correct.
If I see things correctly, this fix is not in either 3.3.5 or 3.3.6.
Debian -stable backported it into their 3.3.5 gcc.
Now the funny thing is that according to the PR, this is a mk68k
specific bug, although the fix is not in target specific code.
-Otto
Index: gcc/reload1.c
===
RCS file: /cvs/src/gnu/usr.bin/gcc/gcc/reload1.c,v
retrieving revision 1.3
diff -u -p -r1.3 reload1.c
--- gcc/reload1.c 25 Dec 2004 00:23:11 - 1.3
+++ gcc/reload1.c 11 Apr 2006 08:48:39 -
@@ -6923,6 +6923,10 @@ do_input_reload (chain, rl, j)
actually no need to store the old value in it. */
if (optimize
+ /* Only attempt this for input reloads; for RELOAD_OTHER we miss
+that there may be multiple uses of the previous output reload.
+Restricting to RELOAD_FOR_INPUT is mostly paranoia. */
+ rl-when_needed == RELOAD_FOR_INPUT
(reload_inherited[j] || reload_override_in[j])
rl-reg_rtx
GET_CODE (rl-reg_rtx) == REG
Re: Limiting userland RAM utilization
On Mon, Apr 10, 2006 at 07:18:02PM -0700, Michael Favinsky wrote: I have a userland process that once in a while goes haywire and starts consuming lots of RAM. While I'm troubleshooting the problem, I need to set up a way to limit this process's RAM consumption, to something along the lines of 200MB. I was looking at using some of the RAM limiting parameters in /etc/login.conf. The three I found most relevant are memorylocked, memoryuse, and vmemoryuse. I'm not sure which one of these is the one I need to tweak. At this point I'm ready to set them all to 200MB. If someone can provide some info on which one of these I should actually be using (or point to somewhere that does) I'd appreciate it. And please, no flaming about how I should use an app that doesn't consume too much RAM. I'm working on that, but I need a short term solution to control this app's RAM consumption without bringing my whole system down. See ulimit, in the manual for whatever shell you are using. It can be used much more elegantly to set per-process limits. Joachim
Re: PFlog
On Mon, Apr 10, 2006 at 09:27:53PM +0100, Gaby vanhegan wrote: On 10 Apr 2006, at 17:29, Joachim Schipper wrote: The only problem here is that I'm running 3.6 and pmacct requires libpcap = 0.6, and 0.3 is what I have. I can't do an upgrade at the moment, there's too many variables, but if I were to build libpcap from source, would it clobber the version that's currently installed and break other programs? The OpenBSD libpcap is a pretty heavily hacked version - most should be in it. It appears to be missing the function pcap_open_dead(), so I presume the 3.6 libpcap version is a touch behind the 0.6 version that pmacct requires. Of course, that looks like it's time for a port. ;-) Or just go with pfflowd, or somesuch. I already had a nice little system setup using pmacct to dump data into an SQL db. It would seem that using pfflowd and flowd together could replace that part of the system, and the data analysis part remains the same. The only difference here is that pfflowd would capture traffic at the firewall stage, whereas pmacct captures it directly at the interface. A little more glue required, but it could be made to do the same job. Actually, since the firewall would do most of the packet processing, it's quite likely to be faster, too. Joachim
Re: aliases with carp
Am 10.04.2006 um 18:05 schrieb Simon Slaytor: inet 1.2.3.2 255.255.255.0 1.2.3.255 vhid 1 pass foo carpdev em0 inet alias 1.2.3.6 255.255.255.0 1.2.3.255 Try triming down your alias lines as see if that helps. Might be a shot in the dark but you never know. With carp implemented in FreeBSD 5.4 this doesn't works: Master: ifconfig_carp0=vhid 1 pass secretpassword 192.168.0.2 netmask 255.255.255.128 ifconfig_carp0_alias0=inet 192.168.0.10 netmask 255.255.255.128 ifconfig_carp0_alias1=inet 192.168.0.11 netmask 255.255.255.128 Backup: ifconfig_carp0=vhid 1 advskew 100 pass secretpasswort 192.168.0.3 netmask 255.255.255.128 ifconfig_carp0_alias0=inet 192.168.0.10 netmask 255.255.255.128 ifconfig_carp0_alias1=inet 192.168.0.11 netmask 255.255.255.128 But both servers assumes to be in master-state :-/ Is this a FreeBSD- specific or a generel carp-problem? Regards, Falk Brockerhoff
Re: bgpd, nexthop and dynamically created interfaces
I did find a bug. Not certain wethe rthis is what affects you.
We look at fields from rt_msghdr that the RTM_IFINFO messages do not
have - they use if_msghdr instead. We do abort on rtm-rtm_errno != 0,
but if_msghdr has no errno, so we look at something in the data part
instead. Surprising that this didn't bite us before!
So we must only do these checks for RTM_ADD/CHANGE/DELETE that actually
use rt_msghdr.
ospfd has teh same issue.
Index: kroute.c
===
RCS file: /cvs/src/usr.sbin/bgpd/kroute.c,v
retrieving revision 1.145
diff -u -p -r1.145 kroute.c
--- kroute.c22 Mar 2006 13:30:35 - 1.145
+++ kroute.c11 Apr 2006 11:07:27 -
@@ -2123,21 +2123,23 @@ dispatch_rtmsg(void)
lim = buf + n;
for (next = buf; next lim; next += rtm-rtm_msglen) {
rtm = (struct rt_msghdr *)next;
- sa = (struct sockaddr *)(rtm + 1);
- get_rtaddrs(rtm-rtm_addrs, sa, rti_info);
-
- if (rtm-rtm_pid == kr_state.pid) /* cause by us */
- continue;
-
- if (rtm-rtm_errno) /* failed attempts... */
- continue;
switch (rtm-rtm_type) {
case RTM_ADD:
case RTM_CHANGE:
case RTM_DELETE:
+ sa = (struct sockaddr *)(rtm + 1);
+ get_rtaddrs(rtm-rtm_addrs, sa, rti_info);
+
+ if (rtm-rtm_pid == kr_state.pid) /* cause by us
*/
+ continue;
+
+ if (rtm-rtm_errno) /* failed
attempts... */
+ continue;
+
if (rtm-rtm_flags RTF_LLINFO)/* arp cache */
continue;
+
if (dispatch_rtmsg_addr(rtm, rti_info) == -1)
return (-1);
break;
--
BS Web Services, http://www.bsws.de/
OpenBSD-based Webhosting, Mail Services, Managed Servers, ...
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)
Re: aliases with carp
* Falk Brockerhoff [EMAIL PROTECTED] [2006-04-11 13:01]: With carp implemented in FreeBSD 5.4 this doesn't works: this is not a freebsd ist. Master: ifconfig_carp0=vhid 1 pass secretpassword 192.168.0.2 netmask Backup: ifconfig_carp0=vhid 1 advskew 100 pass secretpasswort 192.168.0.3 But both servers assumes to be in master-state they are, for their specific addresses. They do not match. -- BS Web Services, http://www.bsws.de/ OpenBSD-based Webhosting, Mail Services, Managed Servers, ... Unix is very simple, but it takes a genius to understand the simplicity. (Dennis Ritchie)
Re: bgpd, nexthop and dynamically created interfaces
On 11/04/06, Henning Brauer [EMAIL PROTECTED] wrote:
I did find a bug. Not certain wethe rthis is what affects you.
We look at fields from rt_msghdr that the RTM_IFINFO messages do not
have - they use if_msghdr instead. We do abort on rtm-rtm_errno != 0,
but if_msghdr has no errno, so we look at something in the data part
instead. Surprising that this didn't bite us before!
So we must only do these checks for RTM_ADD/CHANGE/DELETE that actually
use rt_msghdr.
ospfd has teh same issue.
Index: kroute.c
===
RCS file: /cvs/src/usr.sbin/bgpd/kroute.c,v
retrieving revision 1.145
diff -u -p -r1.145 kroute.c
--- kroute.c22 Mar 2006 13:30:35 - 1.145
+++ kroute.c11 Apr 2006 11:07:27 -
@@ -2123,21 +2123,23 @@ dispatch_rtmsg(void)
lim = buf + n;
for (next = buf; next lim; next += rtm-rtm_msglen) {
rtm = (struct rt_msghdr *)next;
- sa = (struct sockaddr *)(rtm + 1);
- get_rtaddrs(rtm-rtm_addrs, sa, rti_info);
-
- if (rtm-rtm_pid == kr_state.pid) /* cause by us */
- continue;
-
- if (rtm-rtm_errno) /* failed
attempts... */
- continue;
switch (rtm-rtm_type) {
case RTM_ADD:
case RTM_CHANGE:
case RTM_DELETE:
+ sa = (struct sockaddr *)(rtm + 1);
+ get_rtaddrs(rtm-rtm_addrs, sa, rti_info);
+
+ if (rtm-rtm_pid == kr_state.pid) /* cause
by us */
+ continue;
+
+ if (rtm-rtm_errno) /* failed
attempts... */
+ continue;
+
if (rtm-rtm_flags RTF_LLINFO)/* arp
cache */
continue;
+
if (dispatch_rtmsg_addr(rtm, rti_info) == -1)
return (-1);
break;
--
BS Web Services, http://www.bsws.de/
OpenBSD-based Webhosting, Mail Services, Managed Servers, ...
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)
I'll apply it tonight and see what it does in my environment.
Oh how productive one can be when the family is out of the country =)
--
Tony Sarendal - [EMAIL PROTECTED]
IP/Unix
-= The scorpion replied,
I couldn't help it, it's my nature =-
Re: bash: home and end keys
On 2006-04-10 23:38:38 +0200, viq wrote: (usually running things in screen - though that doesn't seem to matter much) I just did export TERM=xterm-xfree86 and that didn't help... I use xterm and ssh. The only result of pressing home or end keys is ~ appearing where the cursor is. Best Martin -- http://www.tm.oneiros.de
Re: bash: home and end keys
On Tue, 11 Apr 2006, Martin Schrvder wrote: On 2006-04-10 23:38:38 +0200, viq wrote: (usually running things in screen - though that doesn't seem to matter much) I just did export TERM=xterm-xfree86 and that didn't help... I use xterm and ssh. The only result of pressing home or end keys is ~ appearing where the cursor is. I've no idea why it's working for some people and for others not. But setting the binding explicitly does the trick for me. Not that I use them ever, I'm happy with ^E and ^A, which work out of the box. bind '^[[H'=beginning-of-line bind '^[[F'=end-of-line -Otto
Re: Which Hardware for Firewall
Falk Husemann wrote: Hello misc! We're using OpenBSD on our Hardware since 2003 and have run our Firewall on OpenBSD since that time too (always following -STABLE). Now the box i once built for that purpose has broken down and I'm in need of a replacement I'll assemble myself again. The box serves a squid, a pf (with 2 and 1/2 pages DinA4 ruleset), named and httpd-SSL The hardware will be on the low to medium end (budget 400-600 EURO). Which processor architecture is faster for firewalling purposes? pf runs in kernelspace AFAIK, so will dual-core be useless? AMD64? Pentium 4? I thought about buying 2GB+ of RAM and running parts of the system from RAM (tmp, squid-cache). Is this possible on OpenBSD? A quick google search did not turn up anything. A quick hint would be sufficent, I don't want to get on someones wick, but theres no information about firewall hardware and obsd on the net. Thanks in advance, Falk My home firewall is serving 4-5 machines, 2 of them full-time connected to the internet. And 3 others sometimes are connected. I run apache, dhcpd, named, openvpn, webmin, ifstated, plus some others things. I do have a one and half pages of rules. But the better part, my hardware: CPU: Pentium 133 Mem: 64MB EDO Two 10Mbit ethernet cards: An ep(4) based card and an ne(4) based card. It stays 80% idle most of the time (not counting when i download the last lost episode using torrents and dht :)) So, unless you have a huge amount of traffic, i recommend buying a good machine for yourself and using the one you have for the firewall. My 3 cents, -- Giancarlo Razzolini Linux User 172199 Moleque Sem Conteudo Numero #002 Slackware Current OpenBSD Stable Snike Tecnologia em Informatica 4386 2A6F FFD4 4D5F 5842 6EA0 7ABE BBAB 9C0E 6B85 [demime 1.01d removed an attachment of type application/pgp-signature which had a name of signature.asc]
Re: OpenBSD todo list?
Christmas in April? ;) A couple requests I recall seeing (*cough* posting *cough*): - enable chroot-ed apps to dump core (this is an easy one) - enable openbsd to run as a para-virtualized Xen guest (this is more involved) Kent Shawn Nock wrote: A quick search of the archive and google didn't turn anything up, so I'll ask here. Is there (if not could there be) a document that describes portions of the tree that particularly need attention? I am looking for a way to contribute and without a little direction the task seems daunting. The FreeBSD folks recently started maintaining such a todo list. It seems to have worked out fairly well for them. I realize that those in a position to put together such a list are also the ones not likely to have the time, but I believe this could prove useful (I assume there are more like me who are new and/or haven't found a comfort zone/focus yet). Cheers, Shawn
Re: OpenBSD todo list?
On Tue, 2006-04-11 at 09:25:55 -0700, Kent Watsen proclaimed... Christmas in April? ;) A couple requests I recall seeing (*cough* posting *cough*): - enable chroot-ed apps to dump core (this is an easy one) - enable openbsd to run as a para-virtualized Xen guest (this is more involved) I see your two requests, and up you the following. IPv6 enabled syslogd(8)
using a 122-key keyboard with OpenBSD
Does anyone here have any experience using TN3270-style keyboards (such as http://www.pckeyboard.com/emulator.html) in OpenBSD? I'm thinking of starting to use them with the machines I build for enhanced functionality (vim would be fantastic with all of those keys, although I feel like an uebernerd for saying so), and am wondering if there're any pre-existing keymaps/terminal entries/etc. What to do in X seems fairly self-explanatory overall, but I figured I'd put a feeler out to see if anyone has anything they'd like to share with me. Regards, --Blair Sadewitz -- What is the practical application of a million galaxies? --Alan W. Watts
Re: ADSL with pppoa (over ATM)
The WAN allocated from the ISP's RADUIS server will be passed through the DLink, via DHCP, to your NIC. If you aren't convinced, put a windows box with a DHCP NIC behind the DLink while in bridge mode, and see it get a routable address. Try this: unplug the telephone wire, reboot the DLink, and your NIC will get a private address (on windows, do an ipconfig /renew). Plug in the telephone wire, and it will get a public one. Job done. Can you explain how to handle the authentication? I missed this point
generic driver inclusion
I'm interested in getting the clcs audio driver included in the generic amd64 kernel, what is necessay to move this process forward? -- Tim Leslie Dept. Of Geography Arizona State University
FAQ 10.11 disk quotas
I'm trying to follow along with FAQ 10.11 and it just doesn't seem to work right for me. $ grep quota /etc/fstab /dev/wd0g /var ffs rw,nodev,nosuid,userquota,groupquota 1 2 $ man edquota | grep -A1 one.second should be imposed. Setting a grace period to one second indicates that no grace period should be granted. $ sudo edquota -t -u Time units may be: days, hours, minutes, or seconds Grace period before enforcing soft limits for users: /var: block grace period: 1 second, file grace period: 0 days $ sudo reboot /* The FAQ mentions quotaon -a, but it seems a reboot is required to create the quota files expected by quotaon */ $ sudo quota clgw Disk quotas for user clgw (uid 1002): Filesystem blocks quota limit grace files quota limit grace /var355240964096 1 0 0 $ cat 700K.txt | mail clgw $ sudo quota clgw Disk quotas for user clgw (uid 1002): Filesystem blocks quota limit grace files quota limit grace /var4256* 40964096 7days 1 0 0 OpenBSD 3.7 (from cdrom). 3.9 has been ordered. Q1) Is there a way to create quota files without reboot? Q2) Why is grace still reported as '7 days' instead of '1 second'? Q3) Why did mail get delivered?
Re: FAQ 10.11 disk quotas
On Tue, 11 Apr 2006, Frank Bax wrote: I'm trying to follow along with FAQ 10.11 and it just doesn't seem to work right for me. $ grep quota /etc/fstab /dev/wd0g /var ffs rw,nodev,nosuid,userquota,groupquota 1 2 $ man edquota | grep -A1 one.second should be imposed. Setting a grace period to one second indicates that no grace period should be granted. $ sudo edquota -t -u Time units may be: days, hours, minutes, or seconds Grace period before enforcing soft limits for users: /var: block grace period: 1 second, file grace period: 0 days $ sudo reboot /* The FAQ mentions quotaon -a, but it seems a reboot is required to create the quota files expected by quotaon */ $ sudo quota clgw Disk quotas for user clgw (uid 1002): Filesystem blocks quota limit grace files quota limit grace /var355240964096 1 0 0 $ cat 700K.txt | mail clgw $ sudo quota clgw Disk quotas for user clgw (uid 1002): Filesystem blocks quota limit grace files quota limit grace /var4256* 40964096 7days 1 0 0 OpenBSD 3.7 (from cdrom). 3.9 has been ordered. Q1) Is there a way to create quota files without reboot? Yes, run quotacheck -a. This should preferably be done before moving to multi-user, or at least on a fs without activity. Q2) Why is grace still reported as '7 days' instead of '1 second'? No idea. Maybe this thas something to do with the fact that the actual delivery got done by root, see below. Q3) Why did mail get delivered? mail.local runs as root, and as such quota checking is not enforced.. I have a diff to solve that, which I'm running on a couple of mail servers. I posted it to tech@ a year ago or so, but due to lack of testing it didn't got committed. Tell me if you're interested. -Otto
Re: OpenBSD todo list?
On Mon, 10 Apr 2006, Shawn Nock wrote: A quick search of the archive and google didn't turn anything up, so I'll ask here. Is there (if not could there be) a document that describes portions of the tree that particularly need attention? I am looking for a way to contribute and without a little direction the task seems daunting. The FreeBSD folks recently started maintaining such a todo list. It seems to have worked out fairly well for them. I realize that those in a position to put together such a list are also the ones not likely to have the time, but I believe this could prove useful (I assume there are more like me who are new and/or haven't found a comfort zone/focus yet). The PR database is one possible starting point. -Otto
Re: generic driver inclusion
Compile your own amd64 kernel with clcs enabled (copy the clcs and audio at clcs lines from i386 GENERIC) Test it, see if it work as well in amd64 mode as it does in i386 mode If so, file a PR or talk to someone who can enable clcs in amd64 GENERIC Tim Leslie [EMAIL PROTECTED] wrote: I'm interested in getting the clcs audio driver included in the generic amd64 kernel, what is necessay to move this process forward? -- Tim Leslie Dept. Of Geography Arizona State University -- The map is not the territory; the word is not the thing defined.
Privilege revocation and privilege separation choice.
I am trying to see the best way or choice in design between privilege revocation and privilege separation. A very simplistic explication of the application is, I have an application that I my putting together that provide network service from a daemon and that daemon gets data from a SQLite file on the server to be send back to the network. Do I understand this properly that may be the separation should be use when part of the software needs to access system resources that are only accessible to root and the revocation should be use in all other cases? Or is the separation should be use all the time anyway and then the part that needs to access files on the system that could be under a low right users like _Apps, and the part that run the daemon for the network part run under nobody or something. Looking at bgpd ntpd, they both start and have part of the design as root and then part under _bgpd or _ntp. Or may be the proper design is always to use separation then, but if any parts doesn't need root access, why should it be, or should it anyway in case of future changes that might need it. I am trying to come to peace with this. Any inside as to how to go at this and be proper in the process? Hope this question make sense.
Re: Privilege revocation and privilege separation choice.
On Tue, 11 Apr 2006 15:50:11 -0400 Daniel Ouellet [EMAIL PROTECTED] wrote: I am trying to see the best way or choice in design between privilege revocation and privilege separation. Its just a question of wether it needs root initially for startup, and then never needs it again (revoke) or if it needs to keep doing stuff as root all the time (seperate). Adam
Re: Privilege revocation and privilege separation choice.
Adam wrote: I am trying to see the best way or choice in design between privilege revocation and privilege separation. Its just a question of wether it needs root initially for startup, and then never needs it again (revoke) or if it needs to keep doing stuff as root all the time (seperate). Thanks for the answer. So, what would be a very good example in the tree of a revoke one then. I love looking and studying the ntpd and bgpd code as it is a very clean and understandable one. Specially ntpd for a small application where separation is needed. That's the best example by far I could find! Anything as good as that for revoke available as well that anyone could suggest for study and example then?
Re: Privilege revocation and privilege separation choice.
On 4/11/06, Daniel Ouellet [EMAIL PROTECTED] wrote: Thanks for the answer. So, what would be a very good example in the tree of a revoke one then. I love looking and studying the ntpd and bgpd code as it is a very clean and understandable one. Specially ntpd for a small application where separation is needed. That's the best example by far I could find! Anything as good as that for revoke available as well that anyone could suggest for study and example then? ping
Re: Privilege revocation and privilege separation choice.
Ted Unangst wrote: On 4/11/06, Daniel Ouellet [EMAIL PROTECTED] wrote: Thanks for the answer. So, what would be a very good example in the tree of a revoke one then. I love looking and studying the ntpd and bgpd code as it is a very clean and understandable one. Specially ntpd for a small application where separation is needed. That's the best example by far I could find! Anything as good as that for revoke available as well that anyone could suggest for study and example then? ping Thank you! Daniel
Re: gcc miscompiles ntohs16() inline assembly in OpenBSD 3.8
On Tue, 11 Apr 2006, chefren wrote: On 04/11/06 11:39, Otto Moerbeek wrote: On Thu, 6 Apr 2006, chefren wrote: On 04/05/06 02:07, Andrew Pinski wrote: Actually I bet ntohs16 is violating C aliasing rules. Interesting, how do you figure that? The following diff (backported from gcc PR http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10692) seems to fix the problem. I hardly know i386 assembly, so please check if the produced code is correct. Cool! We patched compiled gcc and inspected the new assembly output, it indeed fixes the bug. The bug is an over-enthusiastic optimization in gcc, which removes a store temporary value to stack without full justification. How did you manage to find that Bugzilla PR based on this problem description? It does not seem trivial! I knew debian gcc 3.3.5 doesn't contain the bug. So I diffed debian's gcc against our gcc, and went through the changelog. That gave me a list of candidate diffs; I ordered the list based on a gut feeling. Then I worked trough the diffs to test them. I was lucky, the second diff I tried was the right one. Only later I saw the diff was marked m68k. If I see things correctly, this fix is not in either 3.3.5 or 3.3.6. Debian -stable backported it into their 3.3.5 gcc. Yes. The Bugzilla PR indicates that someone applies the fix to gcc 3.4 and beyond, the Debian GCC maintainers suggest applying it to gcc 3.3, but the story ends there. The fix was never applied to the gcc 3.3 branch. (See line 6936 in http://gcc.gnu.org/viewcvs/branches/gcc-3_3-branch/gcc/reload1.c?view=markup ) We notice that the upcoming OpenBSD 3.9 still uses gcc 3.3.5. Is the gcc 3.3 branch still a supported product for the gcc people, and this a fix that slipped through the cracks in their usual maintenance process? I have no idea if the 3.3 branch is actively maintained by the gcc people. Or does OpenBSD 3.9 use a slightly long-in-the-tooth gcc? In that case, how is the process of back porting all bug fixes made to gcc 3.4 and onward arranged for OpenBSD? We use 3.3.5, because it suits us. I believe that the 3.4 branch has some changes that make it difficult to include our local changes like propolice. Also, 3.4 is even slower and more memory hungry than 3.3. As for bug fixes, we tend to do that on a as needed. Now the funny thing is that according to the PR, this is a mk68k specific bug, although the fix is not in target specific code. Yes, the originator hit it with mk68k, we hit it with i686, but we see nothing obvious about the bug that makes clear that it can't be hit with i386 for example. Thanks for enlightening us. I think the bug is misclassified by the gcc people. I'm currently building on several platforms to see if there are any regressions. -Otto
Re: Strange rxvt/xterm behavior
On Mon, Apr 10, 2006 at 10:54:50PM -0300, Gustavo Rios wrote: Dear friends, i am running two xterm. When i run from any of them the w command i got this: $ tty w /dev/ttyp0 10:47PM up 8 mins, 2 users, load averages: 0.88, 0.50, 0.23 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 w griosp1 :0.0 10:46PM 0 -sh $ $ tty w /dev/ttyp1 10:48PM up 8 mins, 2 users, load averages: 0.75, 0.49, 0.23 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp1 :0.0 10:46PM 0 w $ Now, from xterm terminal running over ttyp1, i start a new terminal, by means of rxvt. And the output from the same previous command execution is totally different: $ tty w /dev/ttyp0 10:50PM up 10 mins, 2 users, load averages: 0.34, 0.44, 0.25 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 w griosp2 :0.0 10:48PM 0 -sh $ $ rxvt [1] 28321 $ tty w /dev/ttyp1 10:49PM up 9 mins, 2 users, load averages: 0.56, 0.49, 0.26 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp2 :0.0 10:48PM 0 -sh $ $ tty w /dev/ttyp2 10:49PM up 10 mins, 2 users, load averages: 0.44, 0.47, 0.25 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp2 :0.0 10:48PM 0 w $ What i could not understand is: ttyp1 terminal was associated with a xterm terminal. From that terminal, after starting a rxvt process, that association was broken. Although i am still running xterm on ttyp1, it is not shown on any of the three screen output. What i was expecting it was to have an entry for the same login (grios) on ttyp1. Anyhow, it does not make any sense to me. Thanks a lot for your time and cooperation. My .Xdefaults has the following relevant parts: *VT100.loginShell: true *VT100.scrollBar: off *VT100*colorULMode: on *VT100*underLine: off *VT100*colorBDMode: on *VT100*colorUL: red *VT100*colorBD: magenta *VT100*font: -misc-*-*-*-normal--*-*-*-*-c-*-*-* *VT100*background: black *VT100*cursorBlink: true *VT100*foreground: yellow *VT100*cursorColor: green *VT100*cursorOffTime: 200 *VT100*cursorOnTime: 200 *VT100.pointerColor: red *VT100.pointerColorBackground: black XTerm.termName: xterm Rxvt.loginShell: true Rxvt.scrollBar: off Rxvt*colorUL: red Rxvt*colorBD: magenta Rxvt*font: -misc-*-*-*-normal--*-*-*-*-c-*-*-* Rxvt*background: black Rxvt*foreground: darkgrey Rxvt*cursorColor: yellow Rxvt.pointerColor: red Search the rxvt(1) page for 'utmp'. Essentially, unlike xterm, rxvt does not write to utmp by default. use +ut/utmpInhibit: false to change this behaviour, if desired. Joachim
spamd not logging to /var/log/spamd
I think I just need a second pair of eyes because I'm obviously missing something. I've just installed a new firewall, and i'm trying to get spamd to log to /var/log/spamd. It *does* log to /var/log/daemon though, and the greylisting daemon is working fine. fire:/var/log#ls -al spamd -rw-r- 1 root wheel 0 Apr 5 16:05 spamd ---/var/log/daemon--- Apr 11 15:33:29 fire spamd[8627]: 218.38.56.27: connected (7/6), lists: korea Apr 11 15:33:34 fire spamd[8627]: 80.72.152.151: connected (8/6) ---My modifications to syslog.conf--- !spamd daemon.err;daemon.warn;daemon.info /var/log/spamd ---My modifications to newsyslog.conf--- /var/log/spamd 640 30100 * Z
[solved] spamd not logging to /var/log/spamd
the problem was here: ---My modifications to syslog.conf--- !spamd daemon.err;daemon.warn;daemon.info /var/log/spamd When I started syslog with syslogd -d I saw this error: syslogd: unknown priority name info /var/log/spamd I double checked and between daemon.info and /var/log/spamd I had spaces. I changed the spaces to tab chars, restarted syslog, and now all is well. --Bryan
Re: spamd not logging to /var/log/spamd
On 4/12/06, Bryan Irvine [EMAIL PROTECTED] wrote: I've just installed a new firewall, and i'm trying to get spamd to log to /var/log/spamd. Have you SIGHUP'ed the syslogd process? It should re-read its configuration file at that point, using your new configuration. !spamd daemon.err;daemon.warn;daemon.info /var/log/spamd Also, if you want spamd to only log to /var/log/spamd, try !!spamd in /etc/syslog.conf. See syslog.conf(5) for more information. Cheers, Rogier -- If you don't know where you're going, any road will get you there.
Re: spamd not logging to /var/log/spamd
At 06:42 PM 4/11/06, Bryan Irvine wrote: I think I just need a second pair of eyes because I'm obviously missing something. I've just installed a new firewall, and i'm trying to get spamd to log to /var/log/spamd. Did you 'touch' the file? You need to create the file yourself.
Re: spamd not logging to /var/log/spamd
On Tue, Apr 11, 2006 at 03:42:09PM -0700, Bryan Irvine wrote: ---My modifications to syslog.conf--- !spamd daemon.err;daemon.warn;daemon.info /var/log/spamd when you: $ sed -ne '/spamd/l' /etc/syslog.conf do you have !spamd\n$ daemon.err;daemon.warn;daemon.info /var/log/spamd\n$ or !spamd\n$ daemon.err;daemon.warn;daemon.info\t\t\t/var/log/spamd\n$ last sentence in first paragraph of manpage went under my radar for years and continually bit me in the ass -- jared [ openbsd 3.9-current GENERIC ( mar 15 ) // i386 ]
Re: Strange rxvt/xterm behavior
Excuse, but this is already accomplished by my .Xdefaults file. On 4/11/06, Joachim Schipper [EMAIL PROTECTED] wrote: On Mon, Apr 10, 2006 at 10:54:50PM -0300, Gustavo Rios wrote: Dear friends, i am running two xterm. When i run from any of them the w command i got this: $ tty w /dev/ttyp0 10:47PM up 8 mins, 2 users, load averages: 0.88, 0.50, 0.23 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 w griosp1 :0.0 10:46PM 0 -sh $ $ tty w /dev/ttyp1 10:48PM up 8 mins, 2 users, load averages: 0.75, 0.49, 0.23 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp1 :0.0 10:46PM 0 w $ Now, from xterm terminal running over ttyp1, i start a new terminal, by means of rxvt. And the output from the same previous command execution is totally different: $ tty w /dev/ttyp0 10:50PM up 10 mins, 2 users, load averages: 0.34, 0.44, 0.25 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 w griosp2 :0.0 10:48PM 0 -sh $ $ rxvt [1] 28321 $ tty w /dev/ttyp1 10:49PM up 9 mins, 2 users, load averages: 0.56, 0.49, 0.26 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp2 :0.0 10:48PM 0 -sh $ $ tty w /dev/ttyp2 10:49PM up 10 mins, 2 users, load averages: 0.44, 0.47, 0.25 USERTTY FROM LOGIN@ IDLE WHAT griosp0 :0.0 10:45PM 0 -sh griosp2 :0.0 10:48PM 0 w $ What i could not understand is: ttyp1 terminal was associated with a xterm terminal. From that terminal, after starting a rxvt process, that association was broken. Although i am still running xterm on ttyp1, it is not shown on any of the three screen output. What i was expecting it was to have an entry for the same login (grios) on ttyp1. Anyhow, it does not make any sense to me. Thanks a lot for your time and cooperation. My .Xdefaults has the following relevant parts: *VT100.loginShell: true *VT100.scrollBar: off *VT100*colorULMode: on *VT100*underLine: off *VT100*colorBDMode: on *VT100*colorUL: red *VT100*colorBD: magenta *VT100*font: -misc-*-*-*-normal--*-*-*-*-c-*-*-* *VT100*background: black *VT100*cursorBlink: true *VT100*foreground: yellow *VT100*cursorColor: green *VT100*cursorOffTime: 200 *VT100*cursorOnTime: 200 *VT100.pointerColor: red *VT100.pointerColorBackground: black XTerm.termName: xterm Rxvt.loginShell: true Rxvt.scrollBar: off Rxvt*colorUL: red Rxvt*colorBD: magenta Rxvt*font: -misc-*-*-*-normal--*-*-*-*-c-*-*-* Rxvt*background: black Rxvt*foreground: darkgrey Rxvt*cursorColor: yellow Rxvt.pointerColor: red Search the rxvt(1) page for 'utmp'. Essentially, unlike xterm, rxvt does not write to utmp by default. use +ut/utmpInhibit: false to change this behaviour, if desired. Joachim
Installing X after OpenBSD 3.8 installation
Hi, understand that there are options to select xbase, game etcs during OpenBSD installation. Can I install these options, (particularly X) post-install same as the standard install, and not for Ports or other methods? I would not want to re-install the system unless necessary. Appreciate any help. Thanks. Regards Andrew Ng -- Andrew Ng [EMAIL PROTECTED] -- http://www.fastmail.fm - Same, same, but different
Re: Installing X after OpenBSD 3.8 installation
Andrew Ng wrote: Hi, understand that there are options to select xbase, game etcs during OpenBSD installation. Can I install these options, (particularly X) post-install same as the standard install, and not for Ports or other methods? I would not want to re-install the system unless necessary. Appreciate any help. Thanks. localmirror/faq/faq4.html#AddFileSet -- http://rlworkman.net
Installing X after OpenBSD 3.8 installation
Hi, understand that there are options to select xbase, game etcs during OpenBSD installation. Can I install these options, (particularly X) post-install same as the standard install, and not for Ports or other methods? I would not want to re-install the system unless necessary. Appreciate any help. Thanks. Regards Andrew Ng -- Andrew Ng [EMAIL PROTECTED] -- http://www.fastmail.fm - Send your email first class
Re: Installing X after OpenBSD 3.8 installation
On 4/12/06, Andrew Ng [EMAIL PROTECTED] wrote: Hi, understand that there are options to select xbase, game etcs during OpenBSD installation. Can I install these options, (particularly X) post-install same as the standard install, and not for Ports or other methods? I would not want to re-install the system unless necessary. Appreciate any help. Thanks. Since X is just contained in a .tgz file, just mount the CD (or whatever other install media you used) and do something like: $su #cd / #tar -zxvf /path/to/install/sets/x* #exit $startx
heads up about filesystem troubles
As part of the work to integrate FFS2 support, there have been several changes to the filesystem code. Unfortunately, there are cases where it may cause data corruption. The problematic change was altering the format of the on disk superblock. FFS2 uses a slightly different superblock than FFS1 used. As of two weeks ago, the kernel would automatically upgrade the superblock to the new format, setting a flag that it had done so. Several fields in the block became redundant. The userland utilities were not updated to work with this new format, but instead would remove the flag, meaning the kernel would again read the old fields and upgrade. The problem is that fsck was not taught to remove the flag until several days after the kernel was taught to upgrade. Meaning booting a new kernel but running old fsck could result in superblock corruption because they were operating on different structures. The window of danger was booting a kernel from any time _after_ 2 weeks ago and running a fsck from any time _before_ 4 days ago. If you have booted a new kernel, do not use the old fsck. I have backed out the new superblock changes. The next snapshot will not upgrade the superblock and will simply use the old format. You are encouraged to move away from any snapshot installed in the last 2 weeks.
Re: Installing X after OpenBSD 3.8 installation
tar -zxpf permissions are important -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Nick Guenther Sent: 12 April 2006 04:21 To: OpenBSD-Misc Subject: Re: Installing X after OpenBSD 3.8 installation On 4/12/06, Andrew Ng [EMAIL PROTECTED] wrote: Hi, understand that there are options to select xbase, game etcs during OpenBSD installation. Can I install these options, (particularly X) post-install same as the standard install, and not for Ports or other methods? I would not want to re-install the system unless necessary. Appreciate any help. Thanks. Since X is just contained in a .tgz file, just mount the CD (or whatever other install media you used) and do something like: $su #cd / #tar -zxvf /path/to/install/sets/x* #exit $startx
Re: OpenBSD todo list?
On 4/10/06, Shawn Nock [EMAIL PROTECTED] wrote: A quick search of the archive and google didn't turn anything up, so I'll ask here. rewrite units. it can convert euros to dollars at an awesome rate of 94 cents per euro, but can't convert temperature. Is there (if not could there be) a document that describes portions of the tree that particularly need attention? I am looking for a way to contribute and without a little direction the task seems daunting. The FreeBSD folks recently started maintaining such a todo list. It seems to have worked out fairly well for them. I realize that those in a position to put together such a list are also the ones not likely to have the time, but I believe this could prove useful (I assume there are more like me who are new and/or haven't found a comfort zone/focus yet). Cheers, Shawn -- Shawn Nock (OpenPGP: 0xEF9B08E7) Broadcast Engineer; KUAT Communications Group University of Arizona nock 'at ' arizona 'dot' edu
Re: Installing X after OpenBSD 3.8 installation
Since X is just contained in a .tgz file, just mount the CD (or whatever other install media you used) and do something like: $su #cd / #tar -zxvf /path/to/install/sets/x* #exit $startx Don't forget the 'p' flag in there, when dealing with install sets: tar zxvpf /path/to/install/sets/x* Benny -- God help us all if cats had thumbs. -- Me, 2006
