| 1 |
On Sun, Jul 1, 2018 at 9:40 PM Andrew Udvare <audvare@×××××.com> wrote: |
| 2 |
> |
| 3 |
> YMMV on what processes will actually work properly after a SIGCONT. If |
| 4 |
> anything a process does is not re-entrant, then you could have very |
| 5 |
> unpredictable things happen including corruption of data. |
| 6 |
> |
| 7 |
|
| 8 |
If a process corrupts data of any significance after being stopped it |
| 9 |
should be considered a bug. |
| 10 |
|
| 11 |
Obviously if the process has sockets open there is a good chance that |
| 12 |
timeouts/etc will happen, and the process needs to handle that. |
| 13 |
However, the same is true if the network goes down/etc, or the system |
| 14 |
is suspended, or even if the system just gets really highly loaded and |
| 15 |
doesn't give the process much time. |
| 16 |
|
| 17 |
As others have pointed out, once the process stops running it is going |
| 18 |
to be a relatively high priority for swapping without having to try to |
| 19 |
do anything else to force things, and it won't come back unless you |
| 20 |
wake it up. |
| 21 |
|
| 22 |
-- |
| 23 |
Rich |
Archives