Date: Sun, 20 Mar 2005 09:32:32 +0100 From: Anthony Atkielski <atkielski.anthony@wanadoo.fr> To: freebsd-questions@freebsd.org Subject: Re: Serious issue with SATA disks again Message-ID: <88453814.20050320093232@wanadoo.fr> In-Reply-To: <20050319112231.GA35477@Grumpy.DynDNS.org> References: <583197724.20050319103813@wanadoo.fr> <20050319112231.GA35477@Grumpy.DynDNS.org>
next in thread | previous in thread | raw e-mail | index | archive | help
David Kelly writes: > Its impossible to _prove_ the software is _not_ at fault just as its > impossible to prove the hardware is not at fault. When software works > for others but not on your hardware then one can only conclude there is > _something_ about your hardware. It doesn't work for others. I found lots of messages complaining about this on various platforms, but no explanations. > With seemingly random timeouts such as you are seeing I would suspect > the SATA cable. SATA runs gigabits/sec and could be very sensitive. Try > a different cable from another source. I don't want suspicions, I want answers. Who generates the message, and exactly what does it mean? I see the string in ata-queue.c, and references in a couple of other modules, but as usual, there are no comments at all, so there's no way to figure out what's going on. > Also run the HD manufacturer's test utility. I don't think Western Digital has one (?). If it does, where can I find it? > "smartctl" from ports was also quite useful at reading the error log > maintained by the HD firmware. Interesting reading, such as my drive > temperature was 35, lifetime max/min was 19/45 (Celsius). I tried running the offline self-test, but it didn't seem to do anything. > It means the driver asked the HD to fill a buffer, but it didn't > complete the task within alloted time. Either the drive didn't begin, or > data was lost and fell short. Or there's a bug in the code. > A few years ago one of my then-new machines could not write a floppy in > FreeBSD but could in NT4. Tried lots of things, also got the attention > of the floppy driver maintainer. A few weeks later got the idea to > "Reset to Defaults" in the BIOS. Then reset the few specific things I > needed back the way they were. Magic. There was something undocumented > being set by BIOS at boot that didn't bother NT. Or that NT was programmed to handle (i.e., a better driver in NT than in FreeBSD). > More recently, in 5.2.1, I had no problems with a parallel ATA drive > with Hyperthreading enabled on my P4. No problems running sysinstall to > prep the new SATA drives. But the SATA drives locked the kernel solid > moments after first use. Disabled HT and all was fine. Something about > HT and the new Geom framework used for SATA (but not for PATA, at least > then) didn't work. Or something about the way FreeBSD handled this situation contained a bug. -- Anthony
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?88453814.20050320093232>