From mboxrd@z Thu Jan 1 00:00:00 1970 From: Karsten de Freese Date: Sun, 20 Nov 2011 23:03:36 +0000 Subject: Re: [lm-sensors] looking for help with W83795ADG Message-Id: <4EC98748.4030805@maxi-dsl.de> List-Id: References: <20101022135607.6996731s0m2gcr48@webmail.mnet-online.de> In-Reply-To: <20101022135607.6996731s0m2gcr48@webmail.mnet-online.de> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: lm-sensors@vger.kernel.org hello Eddi, a less-than-welcome update: at least when the three modules i2c-piix4.ko, i2c-piix4-n36l.ko,=20 w83795.ko are loaded, a restart (sudo shutdown -r now) of my N36L=20 fails, the system does not come up until I do a complete power cycle.=20 I'll have to look more closely at this, once I have physical access to=20 the box again. regards - Karsten Am 20.11.2011 14:38, schrieb Karsten de Freese: > hello Eddi, > > Success! > > Thanks for the detailed steps. This all works as described, at least=20 > as far as the 'sensors' output is concerned. With the respective=20 > modules, I get all the measurements you're listing! (See attached.) > > 'i2cdetect -l' will however not produce any output for whatsoever reason. > > The next thing I'd like to find out is how to load the modules=20 > automatically in a clean / kernel-upgrade-proof way.. > > thanks and regards - > > Karsten > > > Am 18.11.2011 11:09, schrieb Eddi De Pieri: >> I suggest you: >> >> >> mkdir ~/piix4-n36l >> cp=20 >> [yourlinuxtree]/usr/src/linux-source-2.6.32/drivers/i2c/busses/i2c-piix4= .c >> ~/piix4-n36l/ >> cp ~/Downloads/Makefile ~/piix4-n36l/ >> cd ~/piix4-n36l/ >> >> patch -p1< proliant-microserver-n36l.diff >> make >> you will have your new module inside ~/piix4-n36l/ >> >> rmmod i2c-piix4 >> insmod ./i2c-piix4.ko >> insmod ./i2c-piix4-n36l.ko >> >> insmod w83795 (you should already build separately) >> >> sensors >> >> regards Eddi. >> >> I hope an answer from Jean... I'd like if this patch may be inserted >> in mainstream linux with the necessary adjustment.. >> >> On Thu, Nov 17, 2011 at 11:39 PM, Karsten de=20 >> Freese wrote: >>> hello Eddi, >>> >>> I'm ready for two questions ;-) >>> >>> .diff / patch looks simple, and I also found i2c-amd756-s4882.c -=20 >>> however: >>> - the diff file actually describes the patch to be on top of=20 >>> i2c-piix4.c? >>> - if I'd compile the patched file, would that give me a standalone=20 >>> driver? >>> >>> thanks for your help - >>> >>> Karsten >>> >>> On 17.11.2011 22:25, Karsten de Freese wrote: >>>> hi Eddi, >>>> >>>> thanks very much - I had almost forgotten about the topic, but=20 >>>> stumbled >>>> across it again a few days ago, so that's very interesting. >>>> I'm not yet familiar with .diff patches, and may welcome a little=20 >>>> help. >>>> However, let me try to find out on my own first.. >>>> >>>> thanks& regards - >>>> >>>> Karsten >>>> >>>> On 16.11.2011 17:18, Eddi De Pieri wrote: >>>>> Finally I get W83795ADG working on my proliant microserver >>>>> >>>>> The patch is based on i2c-amd756-s4882.c and adapted for sb800=20 >>>>> southbus. >>>>> >>>>> Karsten, please try it too... >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# i2cdetect -l >>>>> i2c-0 smbus SMBus piix4 adapter (SDA0) SMBus >>>>> adapter >>>>> i2c-1 smbus SMBus piix4 adapter (SDA2) SMBus >>>>> adapter >>>>> i2c-2 smbus SMBus piix4 adapter (SDA3) SMBus >>>>> adapter >>>>> i2c-3 smbus SMBus piix4 adapter (SDA4) SMBus >>>>> adapter >>>>> root@proliant:/usr/src/lm-sensors/eddi# >>>>> >>>>> yes SDA1 is reserved... so i can't multiplex it >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# sensors >>>>> k10temp-pci-00c3 >>>>> Adapter: PCI adapter >>>>> temp1: +24.5=B0C (high =3D +70.0=B0C, crit =3D +100.0=B0C) >>>>> >>>>> w83795adg-i2c-1-2f >>>>> Adapter: SMBus piix4 adapter (SDA2) >>>>> in0: +1.02 V (min =3D +0.00 V, max =3D +2.05 V) >>>>> in1: +1.52 V (min =3D +0.00 V, max =3D +2.05 V) >>>>> in2: +1.10 V (min =3D +0.00 V, max =3D +2.05 V) >>>>> in3: +0.89 V (min =3D +0.00 V, max =3D +2.05 V) >>>>> in12: +3.35 V (min =3D +0.00 V, max =3D +6.14 V) >>>>> in13: +3.28 V (min =3D +0.00 V, max =3D +6.14 V) >>>>> fan1: 703 RPM (min =3D 329 RPM) >>>>> temp1: +23.0=B0C (high =3D +109.0=B0C, hyst =3D +109.0=B0C) >>>>> (crit =3D +109.0=B0C, hyst =3D +109.0=B0C) se= nsor =3D >>>>> thermal diode >>>>> temp2: +33.2=B0C (high =3D +105.0=B0C, hyst =3D +105.0=B0C) >>>>> (crit =3D +105.0=B0C, hyst =3D +105.0=B0C) se= nsor =3D >>>>> thermal diode >>>>> temp5: +14.0=B0C (high =3D +39.0=B0C, hyst =3D +39.0=B0C) >>>>> (crit =3D +44.0=B0C, hyst =3D +44.0=B0C) sens= or =3D >>>>> thermistor >>>>> beep_enable:disabled >>>>> >>>>> jc42-i2c-0-18 >>>>> Adapter: SMBus piix4 adapter (SDA0) >>>>> temp1: +20.5=B0C (low =3D +0.0=B0C, high =3D +0.0=B0C) ALA= RM >>>>> (crit =3D +0.0=B0C, hyst =3D +0.0=B0C) ALARM >>>>> >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# i2cdetect -y 0 >>>>> 0 1 2 3 4 5 6 7 8 9 a b c d e f >>>>> 00: -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 10: -- -- -- -- -- -- -- -- UU -- -- -- -- -- -- -- >>>>> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 50: 50 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 70: -- -- -- -- -- -- -- -- >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# i2cdetect -y 1 >>>>> 0 1 2 3 4 5 6 7 8 9 a b c d e f >>>>> 00: -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- UU >>>>> 30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 60: -- 61 -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 70: -- -- -- -- -- -- -- -- >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# i2cdetect -y 2 >>>>> 0 1 2 3 4 5 6 7 8 9 a b c d e f >>>>> 00: -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 40: -- -- -- -- -- -- -- -- -- -- -- -- 4c -- -- -- >>>>> 50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 70: -- -- -- -- -- -- -- -- >>>>> >>>>> root@proliant:/usr/src/lm-sensors/eddi# i2cdetect -y 3 >>>>> 0 1 2 3 4 5 6 7 8 9 a b c d e f >>>>> 00: -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- >>>>> 70: -- -- -- -- -- -- -- -- >>>>> >>>>> pay attention that the msleep seems to be really needed... >>>>> >>>>> Regards, Eddi >>>>> >>>>> On Tue, Aug 16, 2011 at 9:26 AM, Eddi De=20 >>>>> Pieri wrote: >>>>>> Hi Jean... >>>>>> >>>>>> The bios show fan rpm and temperature. Since my own proliant=20 >>>>>> microserver >>>>>> lacks of ipmi card, I doubt the bios give me the results using=20 >>>>>> ipmi call. >>>>>> >>>>>> The HP Proliant Microserver chipset (SB820M) have a multiplexed=20 >>>>>> smbus >>>>>> like your. >>>>>> >>>>>> >>>>>> http://support.amd.com/it/Embedded_TechDocs/47283_sb820m_ds_pub_2.00= .pdf=20 >>>>>> >>>>>> >>>>>> Here is technical document from amd... >>>>>> >>>>>> >>>>>> at p52/53 they say: >>>>>> >>>>>> (1) The SDA1 and SCL1 SMBus interface is dedicated for ASF=20 >>>>>> devices only. >>>>>> It should not be used to >>>>>> connect to any other devices. >>>>>> (2) There are only two SMBus controllers. The SCL1/SDA1 pair is >>>>>> controlled by SMBus controller 1. >>>>>> >>>>>> >>>>>> SCL0/SDA0, SCL2/SDA2, SCL3/SDA3 and SCL4/SDA4 are multiplexed=20 >>>>>> pins that >>>>>> are all controlled by >>>>>> SMBus controller 0, and only 1 pair of those pins can be active=20 >>>>>> at any >>>>>> time. >>>>>> >>>>>> I can't understand how to enable 2, 3 and 4 muxed bus.. >>>>>> >>>>>> >>>>>> If you consolidate a solution for i2c-i801.c I hope we can be=20 >>>>>> replicated >>>>>> for i2c-piix4.c too.. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> Hi Karsten, >>>>>> >>>>>> On Fri, 22 Oct 2010 16:04:10 +0200, karsten at maxi-dsl.de wrote: >>>>>>> hi Jean, >>>>>>>> The w83795 driver uses the same detection mechanism used in >>>>>>>> sensors-detect. So if sensors-detect doesn't see your chip, I=20 >>>>>>>> am not >>>>>>>> surprised that the driver didn't either. >>>>>>>> The full output of sensors-detect is needed to diagnose your=20 >>>>>>>> problem >>>>>>>> further. It could be that your SMBus controller isn't=20 >>>>>>>> supported, or >>>>>>>> maybe you are supposed to access the chip through IPMI. >>>>>>> IPMI is probably the key.. >>>>>>> From the sensors-detect output: >>>>>>> "Warning: the required module ipmisensors is not currently=20 >>>>>>> installed >>>>>>> on your system. If it is built into the kernel then it's OK. >>>>>>> Otherwise, checkhttp://www.lm-sensors.org/wiki/Devices for >>>>>>> driver availability" (full ouput attached) >>>>>>> >>>>>>> I did see the output before, but didn't really understand the >>>>>>> implications (IPMI seems to be an alternative way of accessing the >>>>>>> monitoring chips?) >>>>>> Yes. The strength of IPMI access is that it can be used remotely=20 >>>>>> too, >>>>>> for example from a BMC plugged into your system. >>>>>> >>>>>> >>>>>> >>>>>>> The suggested ipmisensors driver however does not seem to be=20 >>>>>>> available, >>>>>>> right? >>>>>>> (http://lm-sensors.org/wiki/Devices says "(2004-12-12) Port to=20 >>>>>>> Linux >>>>>>> 2.6 in progress by Yani Ioannou") >>>>>> That project is essentially dead. At the time being, the best way to >>>>>> retrieve IPMI-based sensor values is using "ipmitool sensor". The >>>>>> drawback is that this isn't integrated with libsensors, so the=20 >>>>>> usually >>>>>> >>>>>> >>>>>> monitoring applications won't work. >>>>>> >>>>>> If you have a BMC or are otherwise already using IPMI on your=20 >>>>>> machine, >>>>>> just use ipmitool and you're done. >>>>>> >>>>>> If not, then it might be worth trying ipmitool once, and then try >>>>>> >>>>>> >>>>>> sensors-detect again. There is a small chance that, once ipmitool >>>>>> managed to access the monitoring device, sensors-detect will see=20 >>>>>> it. My >>>>>> own mainboard which has a W83795ADG chip has a multiplexed SMBus,=20 >>>>>> and I >>>>>> can only see the W83795ADG in sensors-detect when the multiplexer is >>>>>> >>>>>> >>>>>> set properly. Otherwise I would see the memory modules, which _do_ >>>>>> appear in your sensors-detect output. So maybe your board has a >>>>>> multiplexed SMBus too. >>>>>> >>>>>> If the trick doesn't work, you may want to try the new jc42 driver >>>>>> >>>>>> >>>>>> which can monitor the temperature of some DDR3 memory modules. I >>>>>> suspect your memory modules have such sensors. A more recent=20 >>>>>> version of >>>>>> sensors-detect would confirm that: >>>>>> http://dl.lm-sensors.org/lm-sensors/files/sensors-detect >>>>>> >>>>>> >>>>>> >>>>>> I also see that sensors-detect suggested to use the k10temp driver, >>>>>> which your kernel doesn't have. You can try the standalone flavor of >>>>>> the drivers which I maintain at: >>>>>> http://khali.linux-fr.org/devel/misc/k10temp/ >>>>>> >>>>>> >>>>>> The accuracy of the CPU internal sensors is limited, but that's=20 >>>>>> still >>>>>> better than nothing. >>>>>> >>>>>> --=20 >>>>>> Jean Delvare >>>>>> http://khali.linux-fr.org/wishlist.html >>>>>> >>>>>> >>>> >>> > _______________________________________________ lm-sensors mailing list lm-sensors@lm-sensors.org http://lists.lm-sensors.org/mailman/listinfo/lm-sensors