Date: Sat, 19 Nov 2005 16:33:42 -0500 From: Doug Lee <dgl@dlee.org> To: Kris Kennaway <kris@obsecurity.org> Cc: freebsd-questions@freebsd.org Subject: Re: deadlock caused by idprio? Message-ID: <20051119213342.GB1042@kirk.dlee.org> In-Reply-To: <20051119210903.GC21199@xor.obsecurity.org> References: <20051119183236.GA1042@kirk.dlee.org> <20051119210903.GC21199@xor.obsecurity.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Any hidden hazards in rtprio then? I plan to use rtprio when recording sound, so i/o bursts etc. won't cause things to be missed in the recording. Thanks much for the idprio heads-up. And I do hope, sometime, to jump from 4 to 6 directly--though I also plan to buy a new machine for that. I currently run 4.10 on a P166. Nice to have support for old hardware, and amazing that all I miss in using it are speed of MySQL and mail searches and sound-handling performance; but it is time... On Sat, Nov 19, 2005 at 04:09:03PM -0500, Kris Kennaway wrote: On Sat, Nov 19, 2005 at 01:32:36PM -0500, Doug Lee wrote: > I just ran a MySQL lookup process (written in Perl) as root prefixed > with "idprio 1." I expected it to take a while, but not several > minutes. After a while I decided to abort it, so I typed ^c in its > `screen' window. From then on (either from the ^c point or the idprio > run, I know not which), I could not create any new processes, nor > could I kill the running task. Any attempt to do either would hang > indefinitely. I could end processes and work within existing > processes as long as they didn't try to create new ones. I entered > the debugger (I use the alt method of <cr>~^b) and typed, among other > things, "show lockedvnodes" and got one vnode which said "... with 22 > pending," and this count went up by 1 each time I tried creating a new > process. Sadly, I forgot to snapshot that screen, so I can't quote > the rest of that entry. I remember it said VDIR and type > something+VOBJECT, but I don't remember what the something was. > Unable to retrieve my system, I finally typed "panic" in the debugger > so at least the disks would sync. Other than "giving up on 4 > buffers," that went fine. > > Any ideas what this is, and whether it's a bug? I thought idprio was > harmless as far as affecting other processes. No, it's known to cause deadlocks. I don't know if this is still the case on 6.0. Kris -- Doug Lee dgl@dlee.org SSB + BART Group doug@bartsite.com http://www.bartsite.com "It is not the mountain in the distance which makes you want to stop walking; but the grain of sand in your shoe." --Anon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051119213342.GB1042>