Re: Shutdown anomaly
On Wednesday 20 February 2008 03:37:29 pm Brent Jones wrote: -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tore Lund Sent: Thursday, 21 February 2008 9:32 a.m. To: Steven Friedrich; freebsd-questions@freebsd.org Subject: Re: Shutdown anomaly Steven Friedrich wrote: iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s I don't know if this is relevant. Anyway, I used to see this error on 7.0-RC1, at any rate when using xdm. I no longer see it on 7.0-RC2. I see it on my RC2 machine. It comes and goes on my machine, and isn't always present on shutdown. Not running xdm, just a plain vanilla box with no X. The previously suggested kernel config option PRINTF_BUFR_SIZE=128 appears to work around the issue on my machine, an HP zd8215us. I'm running 7.0-RC3 (no, really, that's what it sez), built Wed Feb 20. I could RELIABLY reproduce it with single-user mode, mergemaster -p, make installworld, mergemaster, halt, before I added the kernel option. I'm not using xdm. I do use kdm, but it's the FIRST thing I turn off before troubleshooting. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
Robert Huff wrote: Jonathan Chen writes: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. There is no fix. The workaround is to increase the size of the kernel printf() buffer. I don't remember how you do that ... but this is not a new issue - chech the archives for details. I would like to propose a fix. Care to test it? Just apply this patch and add the following line to your /etc/rc.conf file. shutdown_clean_enable=YES --- rc.shutdown 2008-02-20 19:28:45.0 +0100 +++ /etc/rc.shutdown2008-02-20 19:59:03.0 +0100 @@ -43,10 +43,21 @@ PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin export HOME PATH +# The next three lines belong into /etc/defaults/rc.conf. +shutdown_clean_enable=NO # Set to YES to stop all but the first CPU + # core to prevent mixed buffer output + # upon shutdown. + . /etc/rc.subr load_rc_config 'XXX' +# Fall back to single core mode to guarantee clean output. +if checkyesno shutdown_clean_enable; then + bitmask=$(jot -s '' -b1 $(expr $(sysctl -n hw.ncpu) - 1))0 + sysctl machdep.hlt_cpus=$bitmask +fi + # reverse_list list # print the list in reverse order # ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
Dominic Fandrey writes: It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. There is no fix. The workaround is to increase the size of the kernel printf() buffer. I don't remember how you do that ... but this is not a new issue - chech the archives for details. I would like to propose a fix. Care to test it? Me personally? No. Don't have an SMP machine available. Robert Huff ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
Steven Friedrich wrote: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s I don't know if this is relevant. Anyway, I used to see this error on 7.0-RC1, at any rate when using xdm. I no longer see it on 7.0-RC2. -- Tore ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: Shutdown anomaly
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tore Lund Sent: Thursday, 21 February 2008 9:32 a.m. To: Steven Friedrich; freebsd-questions@freebsd.org Subject: Re: Shutdown anomaly Steven Friedrich wrote: iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s I don't know if this is relevant. Anyway, I used to see this error on 7.0-RC1, at any rate when using xdm. I no longer see it on 7.0-RC2. I see it on my RC2 machine. It comes and goes on my machine, and isn't always present on shutdown. Not running xdm, just a plain vanilla box with no X. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
On Wed, Feb 20, 2008 at 08:06:36PM +0100, Dominic Fandrey wrote: Robert Huff wrote: Jonathan Chen writes: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. There is no fix. The workaround is to increase the size of the kernel printf() buffer. I don't remember how you do that ... but this is not a new issue - chech the archives for details. I would like to propose a fix. Care to test it? Just apply this patch and add the following line to your /etc/rc.conf file. shutdown_clean_enable=YES I'm not too keen on the fix, as it only pushes the problem under the carpet just for the sake of nice output on shutdown. There's still tons of mangled output in the logs (esp. if you run a logging firewall). Here's hoping for a kernel-fix sometime in the future. Cheers. -- Jonathan Chen [EMAIL PROTECTED] -- Beer. Now there's a temporary solution. - Homer Simpson ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
On Sat, Feb 16, 2008 at 11:42:52AM -0500, Steven Friedrich wrote: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. -- Jonathan Chen [EMAIL PROTECTED] -- The human mind ordinarily operates at only ten percent of its capacity -- the rest is overhead for the operating system. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
Jonathan Chen writes: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. There is no fix. The workaround is to increase the size of the kernel printf() buffer. I don't remember how you do that ... but this is not a new issue - chech the archives for details. Robert Huff ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Shutdown anomaly
On Sat, Feb 16, 2008 at 05:43:17PM -0500, Robert Huff wrote: Jonathan Chen writes: I am seeing the following messages, which appear to indicate a memory overwrite: Waiting (max 60 seconds) for system processs 'vnlru' to stop...done Waiting (max 60 seconds) for system processs 'bufdaemon' to stop...done a iStyinncgi n(gm adxi s6k0s ,s evcnoonddess) rfeomra isnyisntge.m. .pr0o cess 'syncer' to stop...0 0 done All buffers synced. Uptime: 8m9s It's an interleaved buffer messages on SMP systems. The problem is known, but I haven't heard of a proposed solution yet. There is no fix. The workaround is to increase the size of the kernel printf() buffer. I don't remember how you do that ... but this is not a new issue - chech the archives for details. The PRINTF_BUFR_SIZE=128 workaround doesn't work on the few systems I've seen the issue on; but if you're interested, one way to do it is: include GENERIC ident GENERIC-1 options PRINTF_BUFR_SIZE=128 Cheers -- Jonathan Chen [EMAIL PROTECTED] -- If everything's under control, you're going too slow - Mario Andretti ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]