Most recent kernel where this bug did *NOT* occur: 2.6.19-rc5-mm2 Distribution: Fedora Devel Hardware Environment: - Silicon Image, Inc. SiI 3114 [SATALink/SATARaid] Serial ATA Controller (rev 02) - nVidia Corporation CK804 Serial ATA Controller (rev f3) - nVidia Corporation CK804 IDE (rev f2) Software Environment: hddtemp-0.3-0.9.beta15.fc6 Problem Description: Starting with 2.6.19-rc6-mm1 hddtemp can not monitor disk temperatures anymore Steps to reproduce: # /usr/sbin/hddtemp /dev/sda /dev/sda: : S.M.A.R.T. not available [root@rousalka nim]# /usr/sbin/hddtemp /dev/sdb /dev/sdb: : S.M.A.R.T. not available [root@rousalka nim]# /usr/sbin/hddtemp /dev/sdc /dev/sdc: : S.M.A.R.T. not available [root@rousalka nim]# /usr/sbin/hddtemp /dev/sdd /dev/sdd: : S.M.A.R.T. not available
kernel built with patch in attachment #9540 [details] from bug #7538
Created attachment 9623 [details] dmesg
Created attachment 9624 [details] kernel config
BTW two of the four disks hang from SiI 3114 so the sata_nv ADMA are not responsible for hddtemp problems
I see this too, in 2.6.20-rc1. I'm fairly sure the problem is caused in mainline by 0e75f9063f5c55fb0b0b546a7c356f8ec186825e ("support larger block pc requests") and it isn't fixed by 77d172ce ("[PATCH] fix SG_IO bio leak"). I think SATA is off the hook. Works in 2.6.19, but not in 2.6.19+0e75f9063f5c55fb0b0b546a7c356f8ec186825e.
I can reproduce this now, am looking at fixing it.
Created attachment 9883 [details] Patch for hddtemp hddtemp includes fragile and wrong sense checks, remove those.
Also see: http://lkml.org/lkml/2006/12/19/131 Closing this bug.
Thanks for looking at this, and happy new year to all