Hi folks, This v3 is mostly about the naming problem - get your paintbrushes ready! Patches ======= o Patch 1 is the meat of the topic - note that it's now in tip/sched/urgent o Patch 2 introduces helpers for the dynamic preempt state o Patches 3-4 make use of said accessors where relevant. Testing ======= Briefly tested the dynamic part on an x86 kernel + QEMU. Compile-tested the kcsan test thingie as a module. Revisions ========= v1: http://lore.kernel.org/r/20211105104035.3112162-1-valentin.schneider@xxxxxxx v1.5: http://lore.kernel.org/r/20211109151057.3489223-1-valentin.schneider@xxxxxxx v2 -> v3 ++++++++ o Turned is_preempt_*() into preempt_model_*() (Frederic) It breaks my rule of "booleans must answer a yes/no question" but is the best I could come with using a "preempt_" prefix o Added preempt_model_preemptible() (Marco) Now used in kcsan_test.c o Dropped powerpc changes Cheers, Valentin Valentin Schneider (4): preempt: Restore preemption model selection configs preempt/dynamic: Introduce preemption model accessors kcsan: Use preemption model accessors ftrace: Use preemption model accessors for trace header printout include/linux/kernel.h | 2 +- include/linux/sched.h | 41 ++++++++++++++++++++++++++++++++++++++ include/linux/vermagic.h | 2 +- init/Makefile | 2 +- kernel/Kconfig.preempt | 42 +++++++++++++++++++-------------------- kernel/kcsan/kcsan_test.c | 5 +++-- kernel/sched/core.c | 18 ++++++++++++++--- kernel/trace/trace.c | 14 ++++--------- 8 files changed, 87 insertions(+), 39 deletions(-) -- 2.25.1
