source:
doc/papers/concurrency/notes/notes.md@
184557e
| Last change on this file since 184557e was 604e76d, checked in by , 8 years ago | |
|---|---|
|
|
| File size: 486 bytes | |
| Line | |
|---|---|
| 1 | Internal scheduling notes. |
| 2 | |
| 3 | Internal scheduling requires a stack or queue to make sense. |
| 4 | We also need a stack of "monitor contexts" to be able to restuore stuff. |
| 5 | |
| 6 | Multi scheduling try 1 |
| 7 | - adding threads to many monitors and synching the monitors |
| 8 | - Too hard |
| 9 | |
| 10 | Multi scheduling try 2 |
| 11 | - using a leader when in a group |
| 12 | - it's hard but doable to manage who is the leader and keep the current context |
| 13 | - basically __monitor_guard_t always saves an restore the leader and current context |
| 14 |
Note:
See TracBrowser
for help on using the repository browser.
