From mboxrd@z Thu Jan 1 00:00:00 1970 From: Harald Judt Date: Wed, 21 Dec 2011 22:43:47 +0000 Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c Message-Id: <4EF26123.10005@gmx.at> List-Id: References: <20110410160433.GA22075@localhost> In-Reply-To: <20110410160433.GA22075@localhost> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lm-sensors@vger.kernel.org Am 21.12.2011 21:55, schrieb Ira W. Snyder: > On Wed, Dec 21, 2011 at 06:55:31PM +0100, Ian Dobson wrote: >> Hi, >> -------------------------------------------------- >> From: "Guenter Roeck" >> Sent: Wednesday, December 21, 2011 6:01 PM >> To: "Harald Judt" >> Cc: >> Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c >> onASRock Extreme4 >> >>> Hi, >>> On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote: >>>> Hi, >>>> >>>>> On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote: >>>>>> Hi, >>>>>> >>>>>> -------------------------------------------------- >>>>>> From: "Mike Campin" >>>>>> Sent: Sunday, April 10, 2011 6:04 PM >>>>>> To: >>>>>> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c >>>>>> onASRock >>>>>> Extreme4 >>>>>> >>>>>>> Hi, >>>>>>> >>>>>>> Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 >>>>>>> motherboard. >>>>>>> It due to the NCT6776 global configuration register 0x1c and 0x24 >>>>>>> values. >>>>>>> The fans work if I modify these registers prior to loading the >>>>>>> w83627ehf >>>>>>> driver. >>>>>>> >>>>>>> My question is where are these registers initialized? Should I ask >>>>>>> ASRock >>>>>>> to fix the BIOS? >>>>>>> >>>>>>> isadump -y -k 0x87,0x87 0x2e 0x2f >>>>>>> >>>>>>> 0 1 2 3 4 5 6 7 8 9 a b c d e f >>>>>>> 00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff >>>>>>> 10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff >>>>>>> 20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01 >>>>>>> 30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff >>>>>>> >>>>>>> isaset -y -f 0x2e 0x87 >>>>>>> isaset -y -f 0x2e 0x87 >>>>>>> isaset -y 0x2e 0x2f 0x1c 0x3 >>>>>>> isaset -y 0x2e 0x2f 0x24 0x1c >>>>>>> >>>>>>> Thanks, Mike >>>>>>> >>>>>>> -- >>>>>>> Mike Campin >>>>>>> lm_sensors at ootsa.homelinux.net >>>>>>> >>>>>>> _______________________________________________ >>>>>>> lm-sensors mailing list >>>>>>> lm-sensors at lm-sensors.org >>>>>>> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors >>>>>> >>>>>> ASRock designed the board, so they should setup the hardware >>>>>> monitoring chip >>>>>> correctly. Allowing users to change the chip configuration is a really >>>>>> bad >>>>>> idea, as many of the pins can have different functionality/electrical >>>>>> spcifications depending on the config. Setting up a pin incorrectly >>>>>> could >>>>>> blow the chip (Configuring a pin so that it's a source that's >>>>>> connected to >>>>>> an unprotected/limited sink for example). >>>>>> >>>>>> So ASRock should fix the BIOS. >>>>> >>>>> Seconded. But before flaming Asrock, two things worth checking: >>>>> * Availability of a BIOS update fixing the issue. >>>>> * Options in the BIOS to enable/disable the monitoring of specific >>>>> fans. I can imagine that the BIOS skips the configuration steps for >>>>> fans for which monitoring was disabled (although that would probably >>>>> mean more code than just doing it unconditionally...) >>>>> >>>>> -- >>>>> Jean Delvare >>>> >>>> I own an Asrock P67 Extreme4 too and have the same problem. It can be >>>> solved by following Mike Campin's instructions, which I successfully >>>> incorporated into /etc/init.d/lm_sensors. There is no BIOS available >>>> that fixes this issue, but I noticed that SpeedFan for Windows >>>> (www.almico.com/speedfan.php) seems to do something similar, otherwise >>>> it would show only 2 fans too. So maybe we could get that into >>>> lm_sensors, even if the BIOS is at fault here? >>>> >>> See above ... worst case this can result in defective hardware if >>> misconfigured. >>> >>> We could possibly introduce some kind of module parameter, but I don't >>> feel >>> too comfortable with that - wrong values may still blow up the hardware. >>> Any opinions, anyone ? >>> >>> Guenter >>> >>> _______________________________________________ >>> lm-sensors mailing list >>> lm-sensors@lm-sensors.org >>> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors >> >> I wouldn't recommend a module parameter, leave the driver as it is. >> If there is a hack (using isaset) just use that. It's too easy for a user to >> set a module parameter hoping that it'll fix some random problem, and if the >> module kills the hardware, then the module is too blame. I understand and agree with that. >> Asrock should fix the BIOS, that's the correct option. Unfortunately, the correct option is not always the most likely to happen. > I wonder if you could set these register values only on this specific > motherboard by checking the DMI table to determine the motherboard > vendor and model number. This is what I currently do using dmidecode in /etc/init.d/lm_sensors, because I too fear to forget about this and accidently wreck a new mainboard I'm going to buy in the future: # We need to do some initialization before loading the module, # otherwise only 2 of the 5 fans will be shown init_module_w83627ehf() { modprobe -q -r w83627ehf # make sure to do this only for the right mainboard if /usr/sbin/dmidecode | grep -i 'Base Board Information' -A2 -B1 | grep -q 'P67 Extreme4' \ && /usr/sbin/dmidecode | grep -i 'BIOS Information' -A3 | grep -q 'Version: L1.61' && \ ! lsmod | grep w83627ehf; then isaset -y -f 0x2e 0x87 2>/dev/null isaset -y -f 0x2e 0x87 2>/dev/null isaset -y 0x2e 0x2f 0x1c 0x3 2>/dev/null isaset -y 0x2e 0x2f 0x24 0x1c 2>/dev/null fi } Actually, checking the BIOS version might make no difference, but since I can change it very easily and know when I have to do so, I decided to include it. I have asked the author of Speedfan how he did it, maybe he will respond. His approach seems to work and be safe. Somehow I don't think it will be very different from this solution though. > At least then you can work around a broken BIOS until the vendor > provides a fix. > > Ira Harald -- `Experience is the best teacher.' _______________________________________________ lm-sensors mailing list lm-sensors@lm-sensors.org http://lists.lm-sensors.org/mailman/listinfo/lm-sensors