tickets needing help

Mark M. Hoffman mhoffman at
Tue Jan 6 06:01:18 CET 2004

* Jean Delvare <khali at> [2004-01-04 21:19:33 +0100]:
> > To really do it properly, it needs to be aware of 2.4 vs. 2.6, due to
> > the fact that the ridiculous PCI device unhiding happens at driver
> > load in the former, at bootup in the latter.
> > 
> > I.e. if we only had 2.6 to worry about, i2c-sis96x would have a single
> > PCI device entry in the table and detection would be nice and generic
> > like all the rest.
> > 
> > Instead, what I should do is yank all the sis96x and sis5595 entries
> > (because that misdetects all 96x) out of that table and write a
> > special- purpose function for detecting them properly.
> > 
> > I'm not a perl wizard (in fact I dislike perl) so I'm inclined to just
> > list both devices and attack the rest later, if ever.
> If you provide the data and explain to me what I should do with them, I
> can do the work for you. I'm not a perl wizard either, but in fact I
> like perl pretty much ;)


in pseudo code; indentation matters...

if kernel is 2.6.x
	if 1039:0016 is present
		use i2c-sis96x
	else if 1039:0008 is present
		use i2c-sis5595
else if kernel is 2.4.x
	if 1039:0008 is present
		if (1039:0645 or 1039:0646 or \
		1039:648 or 1039:0650 or \
		1039:651 or 1039:0735 or \
		1039:745 or 1039:0746) is present
			use i2c-sis645
			use i2c-sis5595
	else if 1039:0016 is present
		use i2c-sis645
	else if 1039:0018 is present
		use i2c-sis645

Also, I think, any SiS device ID which does not appear in the above
function should stay just as it is in the table... because I can't
confirm what IDs are on those systems.


Mark M. Hoffman
mhoffman at

More information about the lm-sensors mailing list