[RFC PATCH v2 0/9] rv: Verify generic IRQ context transitions

From: Karl Mehltretter

Date: Sat Nov 07 2026 - 12:15:52 EST


IRQ action tracepoints describe handlers, but not the lifetime of the
enclosing IRQ context. One context may run several actions, and an action
may be dispatched outside hardirq context. PREEMPT_RT may also enter
serving-softirq state without invoking an action. This leaves no generic
event stream from which an RV monitor can check IRQ-context transitions.

This RFC adds tracepoints at the generic hardirq and serving-softirq
boundaries. The irq:irqstate RV monitor checks their order and agreement
with live kernel state. irq:irqexit follows the owner of a
preempt_schedule_irq() invocation across task switches and CPU migration.

The hardirq tracepoints cover paths through irq_enter() or
irq_enter_rcu() and their matching exits. They do not cover direct
__irq_enter() and __irq_exit() users, raw paths, or architecture-specific
entry and return intervals. irqexit covers preempt_schedule_irq(), not
the complete interrupt-return path.

Boundary callbacks can raise softirq work after the pending-work check.
The series wakes ksoftirqd or ktimerd for remaining work. KUnit covers
nesting, monitor synchronization, PREEMPT_RT worker intervals, provenance
and late-work handoff.

Gabriele answered two of the v1 questions [1]:

- The boundaries are now bare DECLARE_TRACE() tracepoints like the
existing RV hooks in include/trace/events/sched.h, so they add no
tracefs ABI.
- irqstate and irqexit stay separate automata because one is per-CPU
and the other per-task. A hardirq or non-RT softirq belongs to the
CPU it interrupts. A preempt_schedule_irq() invocation belongs to
the interrupted task, several tasks can be inside one at the same
time, and it can migrate before it returns. Each RV automaton has a
single monitor type. Both now sit under an irq container, so they
are enabled together by default, and the irqentry/irqexit naming
clash is gone.

Questions still open from v1:

1. Are dedicated scheduler tracepoints the right way to follow
preempt_schedule_irq() across task switches and migration?
2. Is the KUnit-only hook for an empty ksoftirqd invocation
acceptable? Outside a KUnit run, its worker checks take a disabled
static-key path.

Patches 1-5 add the tracepoints, monitors and documentation. Patches 6-9
add KUnit coverage.

The series is based on v7.4-rc1, which contains the v1 prerequisite and
Wen Yang's reactor lockdep fix:

commit feb5c9a2447a ("softirq: Preserve interrupt context on return from interrupt")
commit 39ccedfd9e55 ("rv: Use LD_WAIT_SPIN as the reactor lockdep wait type")

The series is independent of the early-hardirq series [2], which
installs hardirq state before entry tracing. Both change
kernel/softirq.c, so applying one on top of the other needs a small
conflict resolution in __irq_exit_rcu().

Changes since v1:

- Use bare DECLARE_TRACE() tracepoints instead of trace events
(Gabriele)
- Rename the irqentry container to irq (Gabriele)
- Add text schematics of both models to the documentation (Gabriele)
- Document why irqstate and irqexit are separate automata
- Rebase onto v7.4-rc1

Testing on v7.4-rc1 with Clang 23.1.5, GCC 14.2 (x86) and GCC 16.2
(cross), and QEMU 10.0 TCG, with lockdep and the locking selftests:

- Each patch builds with W=1 on x86_64 without new warnings.
RV_MONITORS_KUNIT_TEST=m builds on x86_64 and arm64. kernel-doc
reports no warnings for the changed headers.
- The rv_mon KUnit suite and the reactor suites passed without
failures on:

x86_64 PREEMPT SMP4 and UP, PREEMPT with nohz_full=1-3 and
threadirqs, PREEMPT_RT SMP4, PREEMPT_RT SMP8 with
nohz_full=1-3
arm64 PREEMPT, PREEMPT with nohz_full=1-3 and threadirqs,
PREEMPT_RT, PREEMPT_RT SMP8 with nohz_full=1-3
riscv64 PREEMPT
ARMv7 UP

Each configuration ran 13 to 21 of the 29 rv_mon cases. The others
are skipped because they need a different configuration.
- RV=n kernels boot on x86_64 and arm64.
- On top of the early-hardirq series, the same suites passed on
x86_64 (PREEMPT, PREEMPT_RT), arm64 (PREEMPT, PREEMPT_RT) and
riscv64.

An LLM assisted with the patches and the testing.

v1: https://lore.kernel.org/all/20261022050712.75035-1-kmehltretter@xxxxxxxxx/

[1] https://lore.kernel.org/all/FACA54EC-F846-4051-9ADD-DB46E7665FAD@xxxxxxxxxx/
[2] https://lore.kernel.org/all/20261028033127.82193-1-kmehltretter@xxxxxxxxx/

Karl Mehltretter (9):
tracing/irq: Add generic IRQ context boundary tracepoints
tracing/sched: Add IRQ preemption lifecycle tracepoints
rv: Add a monitor for generic IRQ context transitions
rv: Add a monitor for IRQ-exit scheduling ownership
docs/rv: Document the IRQ monitors
rv: Add KUnit coverage for IRQ context transitions
rv: Add KUnit coverage for IRQ-exit scheduling
rv: Test empty PREEMPT_RT softirq worker intervals
rv: Test IRQ-exit softirq provenance and handoff

Documentation/trace/rv/index.rst | 1 +
Documentation/trace/rv/monitor_irq.rst | 475 ++++
include/linux/interrupt.h | 8 +
include/rv/kunit.h | 34 +-
include/trace/events/irq.h | 56 +
include/trace/events/sched.h | 26 +
kernel/sched/core.c | 6 +
kernel/softirq.c | 324 ++-
kernel/trace/rv/Kconfig | 11 +-
kernel/trace/rv/Makefile | 3 +
kernel/trace/rv/monitors/irq/Kconfig | 13 +
kernel/trace/rv/monitors/irq/irq.c | 35 +
kernel/trace/rv/monitors/irq/irq.h | 3 +
kernel/trace/rv/monitors/irqexit/Kconfig | 14 +
kernel/trace/rv/monitors/irqexit/irqexit.c | 172 ++
kernel/trace/rv/monitors/irqexit/irqexit.h | 85 +
.../trace/rv/monitors/irqexit/irqexit_kunit.c | 540 ++++
.../trace/rv/monitors/irqexit/irqexit_kunit.h | 21 +
.../trace/rv/monitors/irqexit/irqexit_trace.h | 19 +
kernel/trace/rv/monitors/irqstate/Kconfig | 14 +
kernel/trace/rv/monitors/irqstate/irqstate.c | 289 ++
kernel/trace/rv/monitors/irqstate/irqstate.h | 139 +
.../rv/monitors/irqstate/irqstate_kunit.c | 2339 +++++++++++++++++
.../rv/monitors/irqstate/irqstate_kunit.h | 25 +
.../rv/monitors/irqstate/irqstate_trace.h | 15 +
kernel/trace/rv/rv_monitors_test.c | 30 +
kernel/trace/rv/rv_trace.h | 2 +
tools/verification/models/irq/irqexit.dot | 21 +
tools/verification/models/irq/irqstate.dot | 32 +
29 files changed, 4737 insertions(+), 15 deletions(-)
create mode 100644 Documentation/trace/rv/monitor_irq.rst
create mode 100644 kernel/trace/rv/monitors/irq/Kconfig
create mode 100644 kernel/trace/rv/monitors/irq/irq.c
create mode 100644 kernel/trace/rv/monitors/irq/irq.h
create mode 100644 kernel/trace/rv/monitors/irqexit/Kconfig
create mode 100644 kernel/trace/rv/monitors/irqexit/irqexit.c
create mode 100644 kernel/trace/rv/monitors/irqexit/irqexit.h
create mode 100644 kernel/trace/rv/monitors/irqexit/irqexit_kunit.c
create mode 100644 kernel/trace/rv/monitors/irqexit/irqexit_kunit.h
create mode 100644 kernel/trace/rv/monitors/irqexit/irqexit_trace.h
create mode 100644 kernel/trace/rv/monitors/irqstate/Kconfig
create mode 100644 kernel/trace/rv/monitors/irqstate/irqstate.c
create mode 100644 kernel/trace/rv/monitors/irqstate/irqstate.h
create mode 100644 kernel/trace/rv/monitors/irqstate/irqstate_kunit.c
create mode 100644 kernel/trace/rv/monitors/irqstate/irqstate_kunit.h
create mode 100644 kernel/trace/rv/monitors/irqstate/irqstate_trace.h
create mode 100644 tools/verification/models/irq/irqexit.dot
create mode 100644 tools/verification/models/irq/irqstate.dot


base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
--
2.39.5 (Apple Git-154)