> As strange as it may sound if you consider the age of the code, I > believe that you are right and we should rely on ctl_name instead of > procname. It has gone unnoticed so far most probably because all > hardware monitoring device drivers export all their sysctl entries to > procfs. > > I will fix that in CVS, and send a similar fix to Marcelo Tosatti for > Linux 2.4, although I'm not sure he will accept it. The fix doesn't > apply to 2.6 since we are using sysfs instead of procfs now. > > BTW, for which device are you writing a driver? Bonjour Jean, Je suis tent? de vous r?pondre en fran?ais, mais je r?siste, for other people's sake ! I'm writing a device driver for a Philips PCA9555 chip, which is a 16-bit I/O port with interrupt I2C device. Since our project in for an embedded system, the driver itself it tightly bound to the way the chip is integrated into our hardware. I doubt it is of any interest for anyone outside, considering it relies on a homebrewed CPLD for interrupt management. However, before doing that, I "backported" the LM77 driver found in Linux 2.6 to Linux 2.4. If you are interested in getting it, I will be glad to provide it. Just tell me where to submit it, and maybe give me 1 hour to remove customizations I made for supporting our embedded platform (like interrupts)! The basic functionality (temp read/write) should run fine on any host hardware (PC / Mac / whatever). Thanks for getting back to me. Merci ! Louis-Martin
