Hi all, After reading the following post on LKML: http://marc.theaimsgroup.com/?l=linux-kernel&m=107377669427411&w=2 I did some research about how ISA and I2C are supposed to interact through the different Kconfig files in Linux 2.6. Here are my conclusions, on which I'd like to have feedback and comments. #1 In arch/i386/Kconfig, it is suggested that ISA busses are old and deprecated, and that new machines don't have one. This is wrong. Recent motherboards *do* have an ISA bus. They do not have ISA *slots* which is quite different. I think that this information should be added to CONFIG_ISA's help (and possibly backported to Linux 2.4). #2 The Elektor is an ISA card (at least according to i2c/busses/Kconfig), so I'd expect it to depend on ISA. It doesn't, nor does it in Linux 2.4. Should be fixed? #3 The ELV and Velleman adapters are said to depend on ISA, while I don't see any reason why they would. They are parallel port adapters. The dependancy doesn't exit in our CVS repository nor in Linux 2.4. #4 The i2c-isa driver depends on ISA. This is a problem because most people won't have ISA support selected, but are likely to need i2c-isa. This is the problem described in the post on LKML those URL is given above. I think I remember seeing a way to do reverse-dependancies in Kconfigs (i.e. if B is selected, then A is automatically selected) but cannot find an example of that anymore. This is done with the "select" directive if I'm not mistaking. Thus I'd suggest to use it in this case. Another point raised here is wether i2c-isa really does depend on ISA. Looking through the code, it doesn't, at least not directly. OTOH i2c-sensor does as soon as any adapter uses the isa algorithm (which for some reason seems to be included in i2c-sensors instead of being in i2c-isa -- I guess there is a reason for that but I did not look deep enough into the code to tell what it is). As long as only i2c-isa does use that algorithm it doesn't make much different I guess, still its not very clean. No proposal here. After listening to your remarks and comments, I could submit a patch to fix whatever needs to be. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
