In message <[EMAIL PROTECTED]>, Bryan Liesner writes: >> >> > Revision 1.20 / (download) - annotate - [select for diffs], Mon Mar 10 23:41:41 >> > 2003 UTC (5 days, 5 hours ago) by phk >> > Branch: MAIN >> > CVS Tags: HEAD >> > Changes since 1.19: +5 -0 lines >> > Diff to previous 1.19 (colored) >> > >> > If we run out of consumers while orphaning them, and the provider's geom >> > is withering, destroy the provider when done. >> > >> > This was exposed by the recent change to geom_dev's orphaning logic >> >> If I reverted it back to a previous version (1.19) then the machine booted OK. >> >> BTW, I also found that adding INVARIANTS options into the kernel can prevent this >> problem as well. >> >> Regards, > >I have reverted back to rev 1.19 and all seems to be running OK. I >have /dev/null, /dev/stderr, /dev/apm, and /dev/mixer back. >When the faulty kernel _did_ boot (after about a million retries to >coax a core dump), these devices were missing at boot, or would >disappear shortly after. > >Thanks. >I think Poul-Henning will have enough information to go with now...
You guys _way_ overestimate my abilities here. Right now I have a hard time imagining what geom's eventhandling for "withering" geoms can possibly have to do with any non-geom device node being present in /dev or not, and even more so how INVARIANTS could affect that. The only influence I can imagine is one of timing... I don't think I'll stand a chance on this one until I can reproduce it on one of my machines :-( One thing I'd like you to try is to remove any trace of USB from your systems. USB does some ugly VOP_REVOKES which I am not happy about, and I would like to exclude them from the list of suspects. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
