On Nov 11,  6:44pm, scotte@warped.com (Scott Ellis) wrote:
-- Subject: Re: bin/34979: -current libc causes boot failure

| Christos Zoulas wrote:
| > On Nov 11,  3:58pm, scotte@warped.com (Scott Ellis) wrote:
| > -- Subject: Re: bin/34979: -current libc causes boot failure
| > 
| > | Thanks for the patch, but unfortunately there's no change at all with a 
| > | re-built userland.  Enabling syslogd results in the exact same behavior 
| > | as w/o this patch.
| > | 
| > | What additional steps can I take to help debug?
| > 
| > I did not think so. Can you ktrace it?
| 
| I put three ktraces at 
| http://intrepid.warped.com/~scotte/visible/syslog_ktrace/ , and noted an 
| interesting behavior. The first time I execute syslogd after boot, it 
| produces the output shown in syslogd_Uses100PercentCPU.txt, while using 
| 100% CPU (or close to it).  If I kill syslogd (-9 is the only signal 
| that works), it will restart and work perfectly fine, as shown in 
| syslogd_1.txt and syslogd_2.txt.
| 
| I cannot get the system to hang, as I can when enabling syslogd on boot, 
| just use lots and lots of CPU.  This is the behavior that I saw before 
| in the Oct 24-ish -current (although the syslogd 100% cpu trick happened 
| occasionally, and the initial run at boot didn't seem to cause the problem).
| 
| Let me know what else I can provide.  Thanks!

Try removing /var/run/utmp; there appears to be something wrong with it.

christos