Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
12/23/05 00:18
Modified:
  12/23/05 00:19

Read: times


 
#105802 - unkowm state
Responding to: ???'s previous message
Ali Malihi said:
The problem actually is that if for example a slave is suddenly disconnected from master while a read or write cycle from/to it has not been completed, the master goes into an unkowm state which stops it from further read or write from/to other slaves.

So there must be something wrong with your recovery strategy when a slave doesn't respond correctly!

Prime suspects here must be variables that were in use while the slave was present, and are not correctly reset when the master determines that the slave is no longer present; or timers that are allowed to continue, but should've been stopped.

Look very carefully at your code:
  • make sure that all waits for responses have a timeout for "no response" and correctly handle a corrupt response (eg, a glitch when a slave is unplugged just at the "wrong" moment)

  • make sure that absolutely every variable, flag, etc is set to a proper state after a "no response" or "invalid response"

  • make sure that any other running timers are correctly handled


    I'm not familiar with the detail of I2C to be more specific, but this is all common comms protocol stuff!


  • List of 16 messages in thread
    TopicAuthorDate
    Plug & play by I2C ,is it possibe?            01/01/70 00:00      
       You have two problems?            01/01/70 00:00      
          I know            01/01/70 00:00      
             with a "master" with IIC hardware it sho            01/01/70 00:00      
          lot of time not needed !            01/01/70 00:00      
       limited feasibility of I2C PnP            01/01/70 00:00      
          the fact is            01/01/70 00:00      
             Re: the facts is            01/01/70 00:00      
       CAN            01/01/70 00:00      
       unkowm state            01/01/70 00:00      
          I must try            01/01/70 00:00      
             Inappropriate Design Choice?            01/01/70 00:00      
             Hot-swap I2C            01/01/70 00:00      
             a fvew points            01/01/70 00:00      
       ACCESS.bus is PnP I2C            01/01/70 00:00      
       just recalled            01/01/70 00:00      

    Back to Subject List