Date: Sat, 5 Dec 2009 16:24:25 +0100 From: Wiktor Niesiobedzki <bsd@w.evip.pl> To: freebsd-geom <freebsd-geom@freebsd.org> Subject: geli freezing 8.0 RELEASE and 7.2-STABLE Message-ID: <2ae8edf30912050724i6f196e53y40ccd06970d59a7a@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hi, I'm currently configuring new box with 8.0-RELEASE. I'm setting it up together with geli. From time to time I notice some small freezes (for 2-5 seconds) of the system. I'm doing a small test. In one shell I'm running: % while true ; do ; sleep 1; date ;done sob 5 gru 2009 16:02:48 CET sob 5 gru 2009 16:02:49 CET sob 5 gru 2009 16:02:50 CET sob 5 gru 2009 16:03:15 CET sob 5 gru 2009 16:03:16 CET sob 5 gru 2009 16:03:17 CET sob 5 gru 2009 16:03:18 CET sob 5 gru 2009 16:03:19 CET (Note the gap between 16:02:50 and 16:03:15) In mean time I'm runnig following in the second terminal: # dd if=/dev/gzero.eli of=/dev/null bs=1M ^C3955+0 records in 3955+0 records out 4147118080 bytes transferred in 23.919693 secs (173376727 bytes/sec) As I've noticed, for the time, when dd is working, top reports, that all CPU is used by system by g_eli[0] gzero process. Until I interrupt the dd process, it is +/- fully busy with encryption. I've setup the geli as following: # kldload geom_zero # geli onetime -s 4096 -l 256 gzero GEOM_ELI: Device gzero.eli created. GEOM_ELI: Encryption: AES-CBC 256 GEOM_ELI: Crypto: hardware # sysctl kern.geom.zero.clear=0 (setting the last one to 1 doesn't make any change). On this system I have VIA Padlock device as crypto hardware. I've tried the same on 7.2-STABLE and got similiar result (though the interrupt signal wasn't delivered to dd process so I need to do a hard reset). Is this a problem within GEOM, or is this related to scheduler? On 8.0-Release I'm still running GENERIC kernel (with SCHED_ULE). Cheers, Wiktor Niesiobedzki
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2ae8edf30912050724i6f196e53y40ccd06970d59a7a>