> Besides avoiding of double assignment of that "length" byte to the buffer, > this move will avoid pollution of the buffer in case of an error (invalid length). This was intendend as a feature not a bug ;) This was the reason I put the data to the buffer at the beginning, so people could see the wrong data length. But yes, it can be argued if it should be logged somewhere instead of being in the buffer. I'll check what other drivers do. Sidenote: It is planned to add SMBus3 features somewhen. Then, there won't be an invalid length anymore because it allows 255 byte transfers.
Attachment:
signature.asc
Description: PGP signature
