With regards to question 3. If I were force to turn off interrupts to get
reliable performance are there any rules of thumb that should be followed
to cause minimal impact to the OS?
Or better yet? Is there a reasonable worse case scenario as to how long my
code will be out of control?
Oliver
At 11:34 PM 7/13/99 -0700, you wrote:
>At 3:50 PM -0700 1999/07/13, Oliver King-Smith wrote:
>>I have three questions about the Palm OS.
>>
>>1) If I am running in my own code and I am not calling any Palm OS API
>>points (code is compiled C++ in this case) I have seen evidence of the Palm
>>OS taking control for a part of a second. Is this true? Can the Palm OS
>>take control while your code is running even if you are not making an API
>>call.
>
>Yes. The Palm OS uses a preemptive multitasking kernel.
>
>
>>2) If the Palm OS can take control (and this is not unreasonable think to
>>do) do I have any guarantees about the longest time the Palm OS code can
>>take during one of these interrupts?
>
>Nope. It's not a real-time OS.
>
>
>>3) Assume 1 is true, if I turn off interrupts, does this prevent the Palm
>>OS from taking control, and what are the adverse consequences of doing
>>this? For example if I turn of interrupts for 250ms will that disturb
the OS.
>
>This will disturb the OS (so don't do it).
>
>Regards,
>
>Jim Schram
>3Com/Palm Computing
>Partner Engineering
>
