On Sat, Feb 07, 2009, Bob Perry wrote: > The issues with accessing the micron via the UART is a difficult one. > What do you do if it does not respond to the write? The reads are > handled as they are important. UART writes, as with the nature of the > device, are FIFO and generally fire and forget. Some systems this is > not an issue as the same commands are generally fired again and can be > picked up next time. These are not. Some commands, user types, are > issued the once and are expected to succeed. I will make some changes > to the UART write logic to support this.
I understand UART write are FIFO-ed, but up to a certain limit. I think you should handle write failures by bubbling this error up the call stack until you hit a user requested action, or an automatic action; if this is from the command line, return an error exit code, if this is from the config, log an error. Also, I think there are sequence of writes and reads where it would be best to cancel the whole sequence on the first error rather than blindly continuing the IO (which might be interpreted differently IIUC). Probably such safety checks are worth the extra hundred bytes of program code? Thanks a lot for your comments! -- Loïc Minier -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
