Date: Fri, 17 Oct 2003 12:43:59 +0200 From: "Willie Viljoen" <will@unfoldings.net> To: <john@johnrshannon.com>, <freebsd-hardware@freebsd.org> Subject: Re: Repetitive filesystem corruption Message-ID: <009101c3949b$93eb9f00$0a00a8c0@arista> References: <200310170421.55829.john@johnrshannon.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On ATA drivers, this is commonly caused by a controller that is reporting data to be written to drive, but is actually write caching it in RAM and writing it later creating a backlog which mostly isn't noticed. When the system is shutting down, the controller will suddely report this backlog to the system and keep the OS waiting until it can clear its own backlog, before allowing the OS to write further data. The numbers are you seeing means that two buffers of data cached by the OS can not be written because the controller is refusing to write them (or accept them into its own cache, which will mean it gets lost when the system goes down) I'm not sure if SCSI devices should be doing this though. On ATA devices, it can be easily fixed by setting the correct sysctl to 0. I think it is hw.ata.ata_wc or hw.ata.wc, check ata(4) manpage. This should be done in /boot/loader.conf, sysctl.conf is too late This disables hardware write caching on these controllers. There should be a similar setting for SCSI devices that do this. You might also want to look at bus speed. I have had one system where the hard disk controller would not work at 133MHz FSB speed. The PCI clock was FSB/4, meaning 33.34MHz or thereabouts. If I ran the same system at 124MHz/4, meaning + - 31MHz, the same controller would work perfectly. Have you carried this controller over from earlier hardware, and is your system overclocked? Will ----- Original Message ----- From: "John R. Shannon" <john@johnrshannon.com> To: <freebsd-hardware@freebsd.org> Sent: Friday, October 17, 2003 12:21 PM Subject: Repetitive filesystem corruption FreeBSD 5.1-RELEASE-p10 CPU: Intel(R) Pentium(R) 4 CPU 2.80GHz (2793.02-MHz 686-class CPU) I'm experiencing a problem on one computer with the filesystem corrupting repetitively; another system here with the same software configuration but older hardware is working fine. When the system shuts down, I get the message: syncing disks, buffers remaining... 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 giving up on 2 buffers If I reboot into single user mode and run fsck, it reports corruption most of the time. I have verified the media using the verify command in the SCSI controllers BIOS. Any help would be appreciated. Dmesg output attached. -- John R. Shannon john@johnrshannon.com ---------------------------------------------------------------------------- ---- > _______________________________________________ > freebsd-hardware@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-hardware > To unsubscribe, send any mail to "freebsd-hardware-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?009101c3949b$93eb9f00$0a00a8c0>