adjtimex 1.13 hangs my machine
I downloaded, built and installed adjtimex-1.13. I stopped xntpd, waited a while for the time to drift and ran adjtimex --compare as is mentioned in the documentation. My machine hung immediately. I had to force reboot and then learn about restoring the journal on my two EXT3 filesystems (system wouldn't boot). I rebooted, stopped xntpd again, waited a while and ran adjtimex --compare again with the same result (though learning how to restore the journal was easier this time ;). There was no information in the adjtimex documentation that I could see that said anything about needing a certain kernel version. So, should I try an older adjtimex package (1.12) or should I upgrade to the 2.4.16 kernel on SuSE's FTP server and try that with adjtimex 1.13? I am running a stock SuSE 7.3 system with kernel 2.4.12. Thanks, Ron
Ron,
OUCH, adjtimex should not be kernel specific, and if it compiled I don't
see any good reason for a crash, but given that they are to add adjtimex,
I'd just go with the ntp again, that seemed to be working, and given the
greater understanding of a previous method in the thread, It sounds that it
would eventually calm down the number of lookups after it figured out the
drift amount also.
Dave
On 2/17/02 2:28 PM, "Ron McCall"
I downloaded, built and installed adjtimex-1.13. I stopped xntpd, waited a while for the time to drift and ran adjtimex --compare as is mentioned in the documentation. My machine hung immediately. I had to force reboot and then learn about restoring the journal on my two EXT3 filesystems (system wouldn't boot). I rebooted, stopped xntpd again, waited a while and ran adjtimex --compare again with the same result (though learning how to restore the journal was easier this time ;).
There was no information in the adjtimex documentation that I could see that said anything about needing a certain kernel version. So, should I try an older adjtimex package (1.12) or should I upgrade to the 2.4.16 kernel on SuSE's FTP server and try that with adjtimex 1.13? I am running a stock SuSE 7.3 system with kernel 2.4.12.
Thanks, Ron
I did a little debugging and found that adjtimex was hanging while doing a read of one byte from /dev/port (trying to read the hardware clock). The code looked awfully x86 specific to me (unless /dev/port provides an x86 style CMOS interface on PPC) so I downloaded a Mandrake binary RPM (version 1.12) thinking that maybe they patched it for PPC. It hung my machine as well. I also downloaded a YellowDog source RPM (version 1.11) but the code was pretty much the same so I didn't bother trying it out. Now, if this code is correct for PPC, wouldn't it imply that there is something different with these new Macs? NTP is working like a charm and is still making about 6 updates per hour on the average. On Sun, Feb 17, 2002 at 08:19:41PM -0700, Dave Gomez wrote:
Ron, OUCH, adjtimex should not be kernel specific, and if it compiled I don't see any good reason for a crash, but given that they are to add adjtimex, I'd just go with the ntp again, that seemed to be working, and given the greater understanding of a previous method in the thread, It sounds that it would eventually calm down the number of lookups after it figured out the drift amount also.
Dave
participants (2)
-
Dave Gomez
-
Ron McCall