You guys rock! (and other ticket #754 followups)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Amazing.  I report a bug (ticket #754) and go to bed.  By the time I 
wake up, it has been triaged, reproduced on a test machine, fixed, and 
checked into the CVS repository.

You guys rock!

There were three questions in that ticket.  The key one, concerning 
via686a detection, has already been answered.  I wanted to follow up on 
the other two:

(2) Regarding misdetection of an lm80 chip, MDS wrote: "it?s something 
else, perhaps a winbond chip....  Look at your board, perhaps you can 
figure out what it is."

You helped me, so I'm inclined to help you make lm_sensors better by 
trying to answer this question.  When I look at my board, what exactly 
am I looking for?  A chip with the word "winbond" printed somewhere on 
it?  I suppose I could make a list of what's printed on each and every 
chip, then see if one of those is on your list of known sensor chips. 
But it would help if I had a better idea of what I'm looking for.

(3) Regarding the requirement to load i2c-isa in order for via686a to 
work.  I guess that wiring in a forced dependency would only be 
appropriate if i2c-isa is *always* the *only* way that one communicates 
with that chip.  Is it?  Or are there sometimes ways to talk to it on a 
different bus?

Anyway, I did want to note for the record how I established the 
dependency manually.  It's a bit different from the standard approach 
suggested at the end of sensors-detect, so I thought you might find it 
interesting....

The standard thing to do, which is probably what sensors-detect would 
suggest, would be to just "modload i2c-isa" followed by "modload 
via686a".  But that strikes me as inelegant for some reason.  I used a 
different strategy.  In "/etc/modules.conf", I added the line:

	add below via686a i2c-isa

This tells modprobe that the via686a module depends upon the i2c-isa 
module for some underlying functionality.  The "modules.conf" manual 
describes this as "an optimized case of the pre-install and post-remove 
directives."  Now I can just "modload via686a" and the "i2c-isa" module 
is loaded up.  I like this approach because it has a more declarative 
feel.  It just establishes the dependency, and lets someone else decide 
when to load up the sensor module.

In fact, that worked so well that I switched to it as the loading 
strategy for all (both) sensor modules.  Each sensor has a "below" 
dependency on the bus it uses:

	add below eeprom i2c-viapro
	add below via686a i2c-isa

Then at machine boot time, I just "modprobe eeprom" and "modprobe 
via686a".  The sensor drivers will load up their required bus drivers 
automatically.



[Index of Archives]     [Linux Kernel]     [Linux Hardware Monitoring]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux