Repository navigation

A wake to another CPU that is busy is drained at its next preemption point and runs within a bound - #874

Merged

Japabu merged 20 commits into

mainfrom

wt/toyos-busywake

Oct 13, 2026

Merged

A wake to another CPU that is busy is drained at its next preemption point and runs within a bound#874
Japabu merged 20 commits into
mainfrom
wt/toyos-busywake

Commits on Oct 12, 2026
- A wake to a busy CPU is drained at its next preemption point and runs within a bound
  
  Japabu
  and
  claude
  committed
  831fbf1
- A metal row holds a ping-pong beside busy CPUs to a round-trip p99 under 1 ms
  
  Japabu
  and
  claude
  committed
  c633ee4
- The wake_onto_busy row stages its partner beside a spinner, so a kernel that leaves the wake there reds
  
  Japabu
  and
  claude
  committed
  4ba2e63
- issues: the simulator rotates spawn placement unlike the kernel
  
  Japabu
  and
  claude
  committed
  d5165c2
- Merge origin/main (36aff86fc: #885) into wt/toyos-busywake
  
  Japabu
  and
  claude
  committed
  bb21afa
- The steal publishes its whole fair band, and keeps a probe the task it stands on would answer
  
  Japabu
  and
  claude
  committed
  e2808f7
- The wake_onto_busy row times a fresh wake onto the busy CPU every round, and names CPUs by x2APIC id
  
  Japabu
  and
  claude
  committed
  fe3fe76
- issues: the unstaged K3 regression for S4, the sync-wake forward for S3, the threshold-2 staging for S2
  
  Japabu
  and
  claude
  committed
  d6294fd
- Merge origin/main (ab655a115: #871, #881, #861) into wt/toyos-busywake
  
  Japabu
  and
  claude
  committed
  34d7e18
- The wake_onto_busy row keeps a spinner on the busy CPU again when the kernel moves it
  
  Japabu
  and
  claude
  committed
  1dc0e39
- The I14 corpus trace says which path its replay takes
  
  Japabu
  and
  claude
  committed
  17c4045
- issues: the I5 crossings measured again at this head, and the I15 stretch linked
  
  Japabu
  and
  claude
  committed
  a5e2472
- Merge origin/main (1a7464692: #888, #890, #876, #883, #891, #889, #898) into wt/toyos-busywake
  
  Japabu
  and
  claude
  committed
  41261f8
- An ordinary wake posted on the CPU it goes to rings nothing
  
  Japabu
  and
  claude
  committed
  4b70d88
- Merge origin/main (5096551ca: #902, #897) into wt/toyos-busywake
  
  Japabu
  and
  claude
  committed
  fa3a31a
- kernel-sim: the model answers runs_on from its poster alone
  
  Japabu
  and
  claude
  committed
  4118fbb
Commits on Oct 13, 2026
- kernel-sim: drop the test of runs_on's one-line accessor
  
  Japabu
  and
  claude
  committed
  0dcb56e
- issues: a device interrupt's self wake runs its pass on the way out
  
  Japabu
  and
  claude
  committed
  3e17f57
- issues: the kernel brings up at most eight CPUs
  
  Japabu
  and
  claude
  committed
  d398c85
- Merge origin/main (b680a973b: #873) into wt/toyos-busywake
  
  Japabu
  and
  claude
  committed
  0ab013e

Commits

Commits on Oct 12, 2026

Commits on Oct 13, 2026