i2c-dev lines in modprobe.conf now trigger WARNING

[ticket]

Tyan S2466 mobo. Somewhere between kernels 2.6.19.3 and 2.6.24.7 a major change in i2c took place, such that 2 lines that were needed in modprobe.conf before are now ignored. Older kernel's modprobe.conf:

alias char-major-89 i2c-dev
options w83781d force_w83782d=0,0x2d force_subclients=0,0x2d,0x4c,0x4d ignore=0,0x2c init=0
alias eth0 3c59x
blacklist audio
blacklist snd-usb-audio

Newer kernel's:

alias char-major-89
i2c-dev options w83781d force_w83782d=0,0x2d force_subclients=0,0x2d,0x48,0x49
alias eth0 3c59x
install ide-controller /sbin/modprobe amd74xx; /bin/true
alias pci:v000010B7d00009200sv000010F1sd00002466bc02sc00i00 3c59x
blacklist audio
blacklist snd-usb-audio

On the newer kernel there are warnings in /var/log/messages:

Nov 18 11:11:17 monkey01 modprobe: WARNING: /etc/modprobe.conf line 1: ignoring bad line starting with 'alias' 
Nov 18 11:11:17 monkey01 modprobe: WARNING: /etc/modprobe.conf line 2: ignoring bad line starting with 'i2c-dev'

and dmesg shows: i2c-adapter i2c-0: Client creation failed at 0x2d (-16)

The modules are slightly different too. New:

i2c_amd756              6340  0 
i2c_core               19360  3 eeprom,w83781d,i2c_amd756

Old:

i2c_isa                 5696  2 w83627hf,w83781d
i2c_amd756              7620  0 
i2c_core               19328  5 w83627hf,eeprom,w83781d,i2c_isa,i2c_amd756

The "alias char-marjor-89" was from the original sensors.conf from Tyan. The "i2c-dev options" line was put in place to force selection of certain devices (which otherwise were a bit random, so that sensors did not come up the same each time), and to eliminate a warning at startup in previous kernels:

w83781d 0-002c: Subclient 0 registration at address 0x49 failed.

But now seems to be enough (or rather, since it is ignored), this happens:

i2c-adapter i2c-0: Client creation failed at 0x2d (-16)

Suggestions?

Thanks

[ticket]

Replying to ticket:

Sorry, forgot to put may name on the original, this is by David Mathog

Other info, At the end of dmesg at startup it says:

w83781d 0-002c: The W83627HF chip is better supported by the w83627hf driver, support will be dropped from the w83781d driver soon
w83781d 0-002d: Subclient 0 registration at address 0x49 failed.
i2c-adapter i2c-0: Client creation failed at 0x2d (-16)
w83627hf: Found W83627HF chip at 0xc00
w83627hf: Found W83627HF chip at 0xc00

That implies that the proper fix to this is to add some sort of W83627HF configuration. Can somebody point me at the relevant documentation? The problem is that only the w83782d device gives the CPU temp, which is the key measurement for me among all of these. Here is the output of sensors:

w83782d-i2c-0-2d
Adapter: SMBus AMD768 adapter at 80e0
VCore 1:   +1.63 V  (min =  +0.00 V, max =  +0.00 V)       ALARM  
VCore 2:   +1.65 V  (min =  +0.00 V, max =  +0.00 V)       ALARM  
V AGP:     +3.42 V  (min =  +3.14 V, max =  +3.47 V)              
+5V:       +4.92 V  (min =  +4.76 V, max =  +5.24 V)              
V DDR:     +4.68 V  (min =  +4.50 V, max =  +5.47 V)              
+3.3VSB:   +3.33 V  (min =  +3.14 V, max =  +3.47 V)              
Ch Fan 1:    0 RPM  (min = 1585 RPM, div = 2)              ALARM  
Ch Fan 2: 4239 RPM  (min = 1585 RPM, div = 2)                     
Ch Fan 3: 4377 RPM  (min = 1585 RPM, div = 2)                     
VRM 2:       +27°C  (high =    +0°C, hyst =   +32°C)   sensor = transistor           
CPU1 Temp: +25.5°C  (high =   +80°C, hyst =   +75°C)   sensor = transistor           
CPU0 Temp: +26.5°C  (high =   +80°C, hyst =   +75°C)   sensor = transistor           
vid1:     +0.000 V  (VRM Version 9.0)
alarms:   
beep_enable:
          Sound alarm enabled

w83627hf-isa-0c00
Adapter: ISA adapter
VCore 1:   +1.63 V  (min =  +1.57 V, max =  +1.73 V)              
VCore 2:   +1.63 V  (min =  +1.57 V, max =  +1.73 V)              
+3.3V:     +3.39 V  (min =  +3.14 V, max =  +3.47 V)              
+12V:     +11.60 V  (min = +10.83 V, max = +13.21 V)              
-12V:     -12.28 V  (min = -13.18 V, max = -10.80 V)              
CPU1 Fan:    0 RPM  (min = 7031 RPM, div = 2)              ALARM  
CPU0 Fan: 4687 RPM  (min = 3729 RPM, div = 2)                     
Ch Fan 4:    0 RPM  (min = 3214 RPM, div = 2)              ALARM  
VRM 1:       +30°C  (high =   +77°C, hyst =  -125°C)   sensor = transistor           
AGP:       +30.5°C  (high =   +80°C, hyst =   +75°C)   sensor = transistor           
DDR:       +27.5°C  (high =   +80°C, hyst =   +75°C)   sensor = transistor           
vid2:     +1.650 V  (VRM Version 9.0)
alarms:   Chassis intrusion detection                      ALARM
beep_enable:
          Sound alarm enabled

Thanks.

[khali]

Your /etc/sensors.conf file is so obviously corrupted that it's not even fun. This has nothing to do with lm-sensors and nothing to do with a "major i2c change". You somehow broke your config file, so just fix it and be done with it:

alias char-major-89
i2c-dev options w83781d force_w83782d=0,0x2d force_subclients=0,0x2d,0x48,0x49

should clearly be:

alias char-major-89 i2c-dev
options w83781d force_w83782d=0,0x2d force_subclients=0,0x2d,0x48,0x49

Or maybe you even want to copy the w83781d options back from the original file. If they were working, I see no reason why you would want to change them (even though I'm not sure why you need them in the first place, but let's assume the Tyan folks know what they're doing.)

[khali]

Answer received on the mailing list, for the record:

"You were right: somewhere along the line /etc/modprobe.conf was mangled. Editing the problem lines to match those on the older system resolved these issues."

Closing accordingly.