Lines Matching +full:delay +full:- +full:us

1 // SPDX-License-Identifier: GPL-2.0-only
26 * Any bug related to task migration is likely to be timing-dependent; perform
82 * is in-progress, while a completely different count indicates in migration_worker()
89 * stable, i.e. while changing affinity is in-progress. in migration_worker()
101 * Wait 1-10us before proceeding to the next iteration and more in migration_worker()
102 * specifically, before bumping seq_cnt again. A delay is in migration_worker()
128 * 3. To ensure the read-side makes efficient forward progress, in migration_worker()
129 * e.g. if getcpu() involves a syscall. Stalling the read-side in migration_worker()
131 * to stabilize and less time trying to hit the timing-dependent in migration_worker()
134 * Because any bug in this area is likely to be timing-dependent, in migration_worker()
135 * run with a range of delays at 1us intervals from 1us to 10us in migration_worker()
140 * The original bug can reproduce with a delay up to ~500us on in migration_worker()
141 * x86-64, but starts to require more iterations to reproduce in migration_worker()
142 * as the delay creeps above ~10us, and the average runtime of in migration_worker()
143 * each iteration obviously increases as well. Cap the delay in migration_worker()
144 * at 10us to keep test runtime reasonable while minimizing in migration_worker()
147 * The lower bound for reproducing the bug is likely below 1us, in migration_worker()
148 * e.g. failures occur on x86-64 with nanosleep(0), but at that in migration_worker()
149 * point the overhead of the syscall likely dominates the delay. in migration_worker()
172 min_cpu = -1; in calc_min_max_cpu()
173 max_cpu = -1; in calc_min_max_cpu()
179 if (min_cpu == -1) in calc_min_max_cpu()
225 * count is odd (migration in-progress). in main()
230 * i.e. if a migration is in-progress. in main()
255 * conservative ratio on x86-64, which can do _more_ KVM_RUNs than in main()
256 * migrations given the 1us+ delay in the migration task. in main()