Stefan wrote: > On Mon, 25 Feb 2008, Marty Faltesek wrote: > > > > >> I could need some help and advice to get a working boot/root disk. > >> > >> Technical data: > >> ASUS M2N32 WS Professional / Phenom 9600 / 8GB > >> WD Raptor 150GB > >> Opensolaris NV81 > >> ... > >> > >> This combination didn't work out of the box, but now I'm running a "beta > >> BIOS" and things seem to be OK, as long as I don't access the boot disk to > >> much, like trying to create a lu or install additional packages, then the > >> disk gets "off-line" and unindentified and the system processes that need > >> to > >> access the disk hangs. > >> > >> A fried who have the same MB but with a Athlon X2 processor and is running > >> with Opensolaris NV71 have a not the same problem. > >> > >> When I have been looking for software differences I noticed that on my > >> friends machine with nv71 the disk driver is cmdk and on my machine with > >> nv81 > >> I have nv_sata as the disk driver. > >> > >> One thing I would like to test but don't know if possible and if so how, is > >> to force my nv81 to use the cmdk instaed of nv_sata. Is this doable? > >> > >> Is there any other thing I could test to narrow down the problem to either > >> a > >> software or hardware problem? > >> > >> If there is a software problem any tips on any w/o that I could test? > > > > > > check /var/adm/messages or the msgbuf and if you see the following > > message: > > > > Feb 19 18:27:39 nvsata-2 nv_sata: [ID 517869 kern.warning] WARNING: > > nv_sata inst 0 port 0: excessive interrupt processing. Disabling port > > int_status=1 clear=0 > > > > Then this is the bug you are seeing: > > > > 6662881 nv_sata: systems with MCP55 nVidia SATA controller hang during > > contiguous short writes > > > > If so I have a test binary I could send to you. > > I have seen something like that scrolling away up on my 24-row cosole. But > don't remeber exactly. Later the cosole id displaying a write failure, > just before the machine hangs. Nothing of this seem to make it to any > messages-file. I will try to capture more of the console output.
Yep, I've seen exactly the same behaviour with nv_sata on an ASUS M2NPV-VM mainboard (MCP51 SATA controller), and filed bug 6642154 "nv_sata: excessive interrupt processing": http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6642154 For now, I've worked around this problem by increasing the "NV_MAX_INTR_LOOP" define from 10 to 32, in usr/src/uts/common/sys/sata/adapters/nv_sata/nv_sata.h and installed the recompiled nv_sata module. There have been no more "WARNING: nv_sata inst 1 port 0: excessive interrupt processing. Disabling port int_status=1 clear=0" messages / with system hangs since that change. > For info: > I have installed and tested the nv_stata from nv82, after that I read > about a problem with disable/enable write-cahce (filed for a Hitachi > disk, 6623404), but with no luck. > > I also forgot to mension that I have been forced to disbale, intrd, when I > thought that I had similiar problems as bug 6658565. Which also writes > similar error messages as yours above. Which I can't find on > opensolaris.org, 6662881. Yep, yesterday 6662881 wasn't visible on bugs.opensolaris.org, but today it is. After reading the description text for 6662881 I'm not so sure that this is the same issue as my 6642154. The text visible on the web site does not mention "excessive interrupt processing"; and 6662881 seems to be about some broken hardware hanging the whole OS. > Anyway, I would be more than happy to try your test binary. Just send it > to me or give an URL where I can downlod it. _______________________________________________ driver-discuss mailing list driver-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/driver-discuss
