Re: connect(2) failure when initiating multiple connections

Jeremie Courreges-Anglas: > > The pattern is that several ssh processes are spawned in parallel > > that connect to different hosts--actually the amd64*.p package build > > machines. I have observed this both from home a continent away, > > and from amd64.p to amd64-*.p on the same network. > >

connect(2) failure when initiating multiple connections

For at least two weeks, I have observed a strange error on OpenBSD/amd64 -current, across multiple snapshots and rebuilds from source during that time: When several ssh processes are spawned in quick succession, SOMETIMES one of them hangs in connect(2) and will eventually error out because

Re: terminal tab stop incorrect until reset

"Ted Unangst": > This came up long ago, but I've lost the mail, and the problem remains. > > When printing a line of text containing vt100 escapes and tabs, the tab stops > are not set correctly. After running reset, they are. > > Repro: > > Open a new xterm and run printf

Re: SSH_CONFIG(5) error

z...@philomathiclife.com: > >Synopsis: SSH_CONFIG(5) mistakenly claims that the keyword > >"HostbasedKeyTypes" has been replaced with the keyword > >"HostbasedAcceptedAlgorithms". Thanks, fixed now. In the client configuration reader, the option was accidentally misspelled as

Re: disklabel autoinstall arm64 broken

Alexander Bluhm: > http://bluhm.genua.de/regress/results/2021-02-06T16%3A17%3A03Z/setup-ot10.log > The auto-allocated layout for sd0 is: > #size offset fstype [fsize bsize cpg] > i: 2.0G100522880 4.2BSD 2048 16384 1 # /usr/src > > The

Re: apu4 fatal protection fault in supervisor mode [Was: apu4 kernel panic]

On 2020-11-29, Theo Buehler wrote: > Thanks for digging into this. Your APU seems much worse off than mine, > which takes a few weeks before crashing these days, so it's not much use > for bisecting. The APU2 that serves as my home gateway has been running just fine. OpenBSD 6.8-current

Re: ntpd still fails to notice unreachable server

Otto Moerbeek: > This takes the observed issue into account, -snip- This works for my test case of a single peer and the two scenarios of "can't send query" and "port unreachable". I notice that both cases are handled very differently: * Can't send query: ntpd keeps retrying with a fast poll

Re: ntpd still fails to notice unreachable server

Otto Moerbeek: > Currently testing this. For "port unreachable" replies, this caused ntpd to become unsynced, but the peer still remains valid. 1/1 peers valid, clock unsynced peer wt tl st next poll offset delay jitter

ntpd still fails to notice unreachable server

Even after otto@'s commit in -current... If no replies are received for a while due to connectivity issues go into unsynced mode. The existing code to check if we're unsycned is only done on receiving an ntp packet which does not happen if there are connectivity issues. ... ntpd can

arm64: savecore reports error on every boot

On arm64, the default savecore invocation from /etc/rc results in an error every time, e.g.: bootfile: sd0a:/bsd boot device: sd0 root on sd0a (c8405fa6843cfe97.a) swap on sd0b dump on sd0b ... savecore: can't find device 18/0 The error is different each time: Sep 5 16:41:08 kemoauc

realpath "//" -> EISDIR

gmake's regression tests fail the checks for its built-in realpath function, which led me to this: 2537 realpath CALL __realpath(0x7f7bfd6b,0x7f7bf750) 2537 realpath NAMI "//" 2537 realpath RET __realpath -1 errno 21 Is a directory That doesn't seem right. Very recent,

Re: rad(8) fails to advertise new prefix

Florian Obser: > We need to event_set the correct fd :/ This fixes the problem I observed. > OK? ok -- Christian "naddy" Weisgerber na...@mips.inka.de

rad(8) fails to advertise new prefix

My home gateway uses DHCPv6-PD straight out of net/dhcpcd's README. dhcpcd requests prefix delegation from the ISP, assigns subnets to local interfaces, and rad(8) picks up those addresses and advertises the corresponding prefixes on the local networks. I upgraded my home gateway from

Re: suspend broken on Lenovo X1 (5th gen) on -current

lists+m...@ggp2.com: > >Synopsis: suspend not working starting within the past ~3 week snaps I'm typing this on an almost identical machine which also follows -current and has had no problems resuming in recent weeks. > acpi0: tables DSDT FACP UEFI SSDT SSDT HPET APIC MCFG ECDT SSDT BOOT BATB

Re: "at " doesn't work as documented

Todd C. Miller: > The manual is correct. Below is one way to fix this, though the > logic probably belongs in tod() itself. I think tod() already has the required logic, but it doesn't handle the case where the time is only past by minutes. Index: parsetime.c

"at " doesn't work as documented

OpenBSD 6.4-current (GENERIC.MP) #2: Thu Feb 14 21:19:17 CET 2019 na...@ariolc.mips.inka.de:/sys/arch/amd64/compile/GENERIC.MP The at(1) man page says: at allows some moderately complex timespec specifications. It accepts times of the form HHMM or HH:MM to run a job at a specific

Re: Some change you did in SSL libraries lately...

Walter Alejandro Iglesias: > I usually encrypt my private files with this command: > > $ openssl aes-256-cbc -in file -salt -out file.enc > > and decrypt them with this another command: > > $ openssl aes-256-cbc -d -in file.enc -out file > > After upgrading to the latest snapshot I cannot

Re: Dell T20 with Pentium G3220 needs "disable inteldrm"

On 2018-11-01, Jonathan Gray wrote: > Ah yes, that has been seen before when trying to run i386 on a recent > intel machine not seen with amd64. I ran into that when running i386 on > an x230 (ivy bridge) with 8gb physram as well. Doesn't occur with machines > that use inteldrm with intagp

Re: Dell T20 with Pentium G3220 needs "disable inteldrm"

Jonathan Gray: > These machines appear to have a serial port so you should be able to get a > proper trace with the console on serial. Right. What do you need? Here's the ddb trace after the uvm_fault: OpenBSD 6.4-current (GENERIC.MP) #979: Tue Oct 30 13:11:06 MDT 2018

Dell T20 with Pentium G3220 needs "disable inteldrm"

>Synopsis: Dell T20 with Pentium G3220 needs "disable inteldrm" >Category: kernel >Environment: System : OpenBSD 6.4 Details : OpenBSD 6.4-current (GENERIC.MP) #979: Tue Oct 30 13:11:06 MDT 2018

File::Copy bug

Not sure who to report this to... Perl's File::Copy::move has a bug or limitation that can cause problems with autoconf, which uses File::Copy. File::Copy::move tries to rename() the file, but necessarily falls back to copy and unlink if source and destination are on different file systems. It

ntpd claims "clock synced" even if servers go away

Once ntpd has successfully talked to the configured servers and considers the clock as _synced_, it will never mark the clock _unsynced_ if the servers go away. Here's a corresponding bug report for OpenNTPD on FreeBSD... https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=221282 ... but it's

Re: v6 autoconf broken in bsd.rd

On 2017-07-22, Florian Obser wrote: > reenabling debug output on the ramdisk I see this: > > engine_dispatch_frontend: IMSG_PROPOSAL_ACK: 1 - 25467 > configure_dfr: 1 > engine_dispatch_frontend: IMSG_PROPOSAL_ACK: 2 - 25467 > configure_address: 1 > engine_dispatch_frontend:

v6 autoconf broken in bsd.rd

i386 snapshot, Build date: 1500601432 - Fri Jul 21 01:43:52 UTC 2017 I tried to run an install with... ---> IPv6 address for em0? (or 'autoconf' or 'none') [none] autoconf ... DNS nameservers? (IP address list or 'none') [none] fddd:28ee:243::1 <--- ... but there

Re: multicast reception fails for certain groups on interfaces with TI ThunderLAN chip

Nick Briggs: > The code in if_tl.c, tl_calchash() is wrong in that it doesn't > compensate for signed char type so that the XOR operations to generate the > multicast hash index are polluted by sign extension if the MSbit of the 1st > and 4th or 2nd and 5th bytes of the multicast

Re: WPA passphrase 63 character with ampersand fails in OpenBSD 6

On 2016-11-14, Stefan Sperling wrote: > Don't use characters in your wpa key which are special to the shell. > That's just asking for trouble. > > I would recommend: > > pkg_add pwgen > pwgen -s 63 Personally, I find "openssl rand -base64" convenient for generating

Re: IPv6/NDP/IPsec breakage in -current

Alexander Bluhm: > I also see issues with IPv6 and NDP, but no IPsec involved. There > are several other threads on bugs@ about broken IPv6. > > It seems that sending neighbor solicitation retries for expired ND > entries does not work. The diff below helps in my case, although > it is only a

Re: IPv6/NDP/IPsec breakage in -current

Mike Belopuhov: > It's also not clear what's wrong with those broken NS/ND > packets that you receive. Oct 12 17:30:10 bardioc /bsd: nd6_na_input: ND packet from non-neighbor Oct 12 17:30:12 bardioc last message repeated 2 times Oct 12 17:30:15 bardioc /bsd: nd6_ns_input: NS packet from

Re: IPv6/NDP/IPsec breakage in -current

On 2016-10-09, Christian Weisgerber <na...@mips.inka.de> wrote: > Found by bisection. The culprit is this commit: > > > CVSROOT:/cvs > Module name:src > Changes by: mar...@cvs.openb

Re: IPv6/NDP/IPsec breakage in -current

On 2016-10-06, Christian Weisgerber <na...@mips.inka.de> wrote: > Something is very broken at the intersection of IPv6, NDP, and IPsec > in -current. Found by bisection. The culprit is this commit: CVSROOT:

Re: thunderbird segfault: mmap W^X violation

Peter N. M. Hansteen: > >Description: Running post-6.0 amd64 snapshots, thunderbird segfaults at > >startup: That's a combination of three things: (1) The Mozilla code has W^X violations. (2) The ports for the various Mozilla programs fail to label the binaries with the wxneeded marker.

getsockname() change or bug?

I updated my home gateway to -current as of yesterday, and the SixXS tunnel client can no longer set up a tunnel. The problem appears to be that getsockname() now returns 0.0.0.0 as the local IPv4 address used for sending UDP heartbeat packets to the remote endpoint. That seems wrong. 41467

Re: ifconfig media doesn't show inet6 addresses

On 2016-03-28, Theo de Raadt wrote: >> Providing the 'media' parameter to ifconfig only shows ipv4 >> addresses and not ipv6. > > This decision was made almost 15+ years and still stands today. > > If you want to see v6 addresses, use -a. > > Most people who use

Re: ACPI thinkpad brightness regression

Mark Kettenis: > Does the following diff work? No. (Thinkpad X230.) > Index: acpithinkpad.c > === > RCS file: /cvs/src/sys/dev/acpi/acpithinkpad.c,v > retrieving revision 1.51 > diff -u -p -r1.51 acpithinkpad.c > ---

Re: ACPI thinkpad brightness regression

On 2016-02-09, Martin Pieuchot wrote: > Since brightness support has been added to acpithinkpad(4) I can easily > trigger a regression on my x220: > > - Switching to the first virtual console just after xdm starts using the > Ctrl+Alt+F1 key combination turns my screen dark.

Re: tail -f plus Ctrl-4 --> core dump

On 2015-11-28, Stuart Henderson wrote: >> running tail -f on a file and then pressing Ctrl-4 while it waits for data >> results in a crash: > > Somewhere (terminal emulator?) this is converted to Ctrl-\ which defaults > to doing 'quit'. On the DEC VT220 and good emulations

sh: parse error for case inside $()

sh(1) fails to correctly parse a "case" conditional inside a $(...) command substitution: syntax error: `case' unmatched Example script below: #!/bin/sh echo $( case $1 in hello) echo

RCS is confused if latest revision is empty file

System: 5.8-current/amd64 RCS gets confused if the latest revision is an empty file: checking out any other revision will only produce empty files. Here's an example: $ rm -f foo foo,v $ echo hello world > foo $ ci -i -t-bla -l foo foo,v <-- foo initial revision: 1.1 done $ echo bonjour le

Re: i386-current bsd.rd fails on net6501

Mark Kettenis: > > ioapic0 at mainbus0: apid 0 pa 0xfec0, version 20, 24 pins > > uvm_fault(0xd0b6c5e0, 0xd0d7e000, 0, 4) -> d > > kernel: page fault trap, code=0 > > Stopped at __kernel_bss_end+0x130c40: cmpl$0x49435024,%eax > > ddb> > > no trace?

Re: i386-current bsd.rd fails on net6501

Mike Larkin: > That address and instruction seem bogus. When did it last work? > > This seems awfully similar to the other bug we fixed in Calgary a couple > months back; And that may very well be the last time I tried i386 on that machine. -- Christian "naddy" Weisgerber

i386-current bsd.rd fails on net6501

When I try to boot bsd.rd from the Sep 7 i386 snapshot on my Soekris net6501, it blows up. boot> boot bsd.rd booting hd0a:bsd.rd: 3176316+1310736+2057216+0+425984 [72+247696+238008]=0x71eb2c entry point at 0x2000d4 [7205c766, 3404, 24448b12, e568a304] Copyright (c) 1982, 1986, 1989, 1991,

Re: i386-current bsd.rd fails on net6501

Mark Kettenis: > Does the bsd kernel from the same snapshot blow up as well? Yes. booting hd0a:bsd.i386: 7520308+2015344+189444+0+1069056 [72+411072+405023]=0xb155c4 entry point at 0x2000d4 [7205c766, 3404, 24448b12, de60a304] [ using 816580 bytes of bsd ELF symbol table ] Copyright (c)

Re: install57.iso does not recognize a burned image of itself in an attached CD-R drive

On 2015-05-19, James Hartley jjhart...@gmail.com wrote: Let's install the sets! Location of sets? (cd0 cd1 disk http or 'done') [cd0] cd1 No filesystems found on cd1. My first guess would be that there are no /dev/cd1[ac] device nodes on the install kernel's file system. -- Christian naddy

Re: install57.iso does not recognize a burned image of itself in an attached CD-R drive

Christian Weisgerber: Let's install the sets! Location of sets? (cd0 cd1 disk http or 'done') [cd0] cd1 No filesystems found on cd1. My first guess would be that there are no /dev/cd1[ac] device nodes on the install kernel's file system. I don't have a second CD drive, but I'm now

Re: cvs error while checking out openbsd sources

Thomas Schweikle: + export CVSROOT=anon...@openbsd.cs.fau.de/cvs + CVSROOT=anon...@openbsd.cs.fau.de/cvs + export CVS_RSH=/usr/bin/ssh_2022 + CVS_RSH=/usr/bin/ssh_2022 + mkdir -p /var/lib/jenkins/jobs/openbsd_openbsd-5-7/workspace + cd /var/lib/jenkins/jobs/openbsd_openbsd-5-7/workspace +

Thinkpad X230 resets at pckbd0 attach

Occasionally, when I power on my Thinkpad X230, it resets during kernel autoconf: white-on-blue kernel messages - boom! Thinkpad logo. It subsequently boots fine. Today I finally thought to save the dmesg, and it shows that the first boot was cut short at or after pckbd0 attaching. OpenBSD

Re: Race condition in boot scripts when using DHCP (with wireless)

Lars Engblom lars.engb...@kimitotelefon.fi wrote: I have been checking, the bg option is not mention in fstab(5), mount(8) and mount_nfs(8). Please update the manual pages together with the faq. At this moment, there is no way anyone could figure it out by themselves. -b in mount_nfs(8) is

Re: Race condition in boot scripts when using DHCP (with wireless)

Kenneth R Westerback: -b in mount_nfs(8) is hard to miss. But I think the mapping from command line options to fstab(5)/-o option names should be documented. At least I can't find such verbiage. You can use -b etc in fstab(5). We need to make up our mind which variant is preferred and

Re: umount -tnfs -a does not unmount all NFS filesystems

Todd C. Miller: Switching the processing order in getmntname() fixes the behavior. Can anybody think of something that would be broken by this? I can't see any reason not do to this. I'm sure you were trying for a minimal diff but doesn't it make more sense to make the loop more like

Re: ntpd fails to adjust very large deltas

Otto Moerbeek o...@drijf.net wrote: ntpd is never going to be able to correct these large offets within reasonable time. It was designed to adjust clocks that have offsets of a couple of minutes. Seeing the messages in syslog clear indicates you have some work to do. So i like them to stay.

Re: security vulnerability - NTP not authenticated

adrelanos adrela...@riseup.net wrote: I read http://www.openbsd.org/faq/faq6.html#OpenNTPD and also found this nowhere else. NTP is not authenticated in OpenBSD by default, right? Right. OpenBSD's ntpd implements the publically specified protocol, SNTPv4 (RFC2030), which simply doesn't

Soekris net6501, wrong hw.{cpuspeed,setperf}

After boot, hw.setperf and .cpuspeed are wrong on the Soekris net6501-70. hw.cpuspeed claims the box is running at 1600 MHz when in fact it starts up at 600 MHz. # sysctl hw.{setperf,cpuspeed} hw.setperf=100 hw.cpuspeed=1600 # md5 -ttt MD5 time trial. Processing 100 1-byte blocks...