Details
Diff Detail
- Repository
- rS FreeBSD src repository - subversion
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| sys/sys/gtaskqueue.h | ||
|---|---|---|
| 95 ↗ | (On Diff #32147) | This one is a bit more complicated. Here you would remove the EARLY_AP_STARTUP #ifdef's and only use the second case (with separate SI_SUB_TASKQ and SI_SUB_SMP SYSINITs). |
The LinuxKPI bits look good to me. Please address John Baldwins comments before committing.
Remove EARLY_AP_STARTUP gtaskqueue code, as it does not work when run before
APs are started. Now gtaskqueue is SI_SUB_TASKQ, which precedes SI_SUB_SMP.
Boots fine on my laptop (whereas it did not boot fine before removing the
EARLY_AP_STARTUP code).
| sys/sys/gtaskqueue.h | ||
|---|---|---|
| 93 ↗ | (On Diff #32192) | Follow-up question: is it ok that this is SI_ORDER_FIRST, while taskqueues themselves are only being initialized SI_SUB_SECOND? I guess I'm not sure what, if any, relationship gtaskqueue / taskqgroup have with ordinary taskqueues. |
| sys/sys/gtaskqueue.h | ||
|---|---|---|
| 93 ↗ | (On Diff #32192) | It matches the previous behavior when all these SYSINITs were SI_SUB_NET_IF so it should be ok. (At some point the SYSINIT's recently changed to use SI_SUB_TASKQ should probably have their orders adjusted so taskqueues are created at SI_SUB_FIRST instead of SI_SUB_SECOND, and the linuxkpi ones can use SI_SUB_SECOND instead of SI_SUB_THIRD, but that isn't critical / urgent) |
| sys/sys/gtaskqueue.h | ||
|---|---|---|
| 93 ↗ | (On Diff #32192) | That makes sense, thanks. Can you mark the review approved, or do you have more feedback? |