[Please CC: the mailing-list on reply.] Quoting myself: > We usually ignore read errors in chip drivers because they are rare, > but since it happens so frequently with that specific chip, we will > have to do it here. I will try to implement the solutions listed > above, if you want to help me test them. Here is a patch, tested to apply and compile on Linux 2.6.2-rc3. Could you please give it a try? It should solve the rare "-1" readings problem, as well as possible misdetection issues. If the driver is really unable to read a register value, it will silently return the previously known value (or 0 if it doesn't have one yet). Maybe we could log the incident, although I'm fairly certain it will never happen, providing the number of max retries is high enough (I set it to 5 for now). Thanks. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/ -------------- next part -------------- A non-text attachment was scrubbed... Name: linux-2.6.2-rc3-i2c-w83l785ts-retries.diff Type: application/octet-stream Size: 3342 bytes Desc: not available Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20040202/ca50f495/attachment.obj
