> In the patch: > > http://jdelvare.net1.nerim.net/sensors/lm_sensors-CVS-separate-mallocs.diff > > to break the kmalloc's out, why did you remove the > > - default : > - printk("lm85: Internal error, invalid kind (%d)!", kind); > - err = -EFAULT ; > - goto ERROR1; > > > clause from the kind detection switch statement in the ADM1026 and > LM85 drivers? Shouldn't we trap any invalid "kind" values? After taking a closer look at the code it seemed obvious that this case just couldn't happen. The negative "kind" case is handled by the "if (kind <= 0)" block right above, and all the other values are handled in the swicth/case block. So why would we try to catch an error that, by construction, cannot happen? Thanks for reviewing my patch, BTW. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
