| 1 |
On Thursday 24 July 2008, 17:09, Alan McKinnon wrote: |
| 2 |
|
| 3 |
> > What does it mean a process is "sleeping", technically? |
| 4 |
> |
| 5 |
> It's a misnomer, it means "not running". |
| 6 |
> |
| 7 |
> The cpu gives the illusion of executing many tasks simultaneously. In |
| 8 |
> reality, it is executing them one at a time and very rapidly (many |
| 9 |
> times a second) switching between them. |
| 10 |
> |
| 11 |
> Normally at a given instant in time, one task is running per cpu. The |
| 12 |
> rest are mostly waiting their turn or sleeping. There are various OS |
| 13 |
> strategies for bringing this about - some rely on the task itself to |
| 14 |
> back out after a running for a short while, sometimes the OS kernel |
| 15 |
> enforces it, sometimes you have a combination. If everything is |
| 16 |
> working nicely, the end result is pretty much the same. |
| 17 |
> |
| 18 |
> There's another state worthy of note - blocked. This is when a task is |
| 19 |
> waiting for something else to happen first (most often disk or network |
| 20 |
> I/O) so it won't try and execute till that other thing happens. This |
| 21 |
> is not the same as sleeping. Sleeping is spinning you wheels in idle, |
| 22 |
> blocked is a deliberate stop and sit back and wait. |
| 23 |
|
| 24 |
From what I know, "blocked" is the same as "sleeping", ie waiting for |
| 25 |
something to happen. Tasks that have completed their time slice and are |
| 26 |
forced by the scheduler to stop, are not "sleeping"; they are |
| 27 |
re-inserted in the queue of the runnable processes, and the scheduler |
| 28 |
picks them up again from there when another time slice is assigned to |
| 29 |
them. These processes are in the "runnable" or "ready" state. |
| 30 |
|
| 31 |
But of course I may be wrong, so corrections welcome. |
Archives