|
4407 | 4407 | reboot_cpu is s[mp]#### with #### being the processor
|
4408 | 4408 | to be used for rebooting.
|
4409 | 4409 |
|
| 4410 | + refperf.holdoff= [KNL] |
| 4411 | + Set test-start holdoff period. The purpose of |
| 4412 | + this parameter is to delay the start of the |
| 4413 | + test until boot completes in order to avoid |
| 4414 | + interference. |
| 4415 | + |
| 4416 | + refperf.loops= [KNL] |
| 4417 | + Set the number of loops over the synchronization |
| 4418 | + primitive under test. Increasing this number |
| 4419 | + reduces noise due to loop start/end overhead, |
| 4420 | + but the default has already reduced the per-pass |
| 4421 | + noise to a handful of picoseconds on ca. 2020 |
| 4422 | + x86 laptops. |
| 4423 | + |
| 4424 | + refperf.nreaders= [KNL] |
| 4425 | + Set number of readers. The default value of -1 |
| 4426 | + selects N, where N is roughly 75% of the number |
| 4427 | + of CPUs. A value of zero is an interesting choice. |
| 4428 | + |
| 4429 | + refperf.nruns= [KNL] |
| 4430 | + Set number of runs, each of which is dumped onto |
| 4431 | + the console log. |
| 4432 | + |
| 4433 | + refperf.readdelay= [KNL] |
| 4434 | + Set the read-side critical-section duration, |
| 4435 | + measured in microseconds. |
| 4436 | + |
| 4437 | + refperf.shutdown= [KNL] |
| 4438 | + Shut down the system at the end of the performance |
| 4439 | + test. This defaults to 1 (shut it down) when |
| 4440 | + rcuperf is built into the kernel and to 0 (leave |
| 4441 | + it running) when rcuperf is built as a module. |
| 4442 | + |
| 4443 | + refperf.verbose= [KNL] |
| 4444 | + Enable additional printk() statements. |
| 4445 | + |
4410 | 4446 | relax_domain_level=
|
4411 | 4447 | [KNL, SMP] Set scheduler's default relax_domain_level.
|
4412 | 4448 | See Documentation/admin-guide/cgroup-v1/cpusets.rst.
|
|
0 commit comments