On 8/2/26 12:55, Wen Yang wrote:
On 7/30/26 13:13, Gabriele Monaco wrote:
Il 29 luglio 2026 18:17:31 UTC, Wen Yang <[email protected]> ha scritto:
+#define rv_get_current() (unlikely(kunit_get_current_test()) ?
rv_get_mock_current() : current)
...
+/*
+ * rv_get_mock_current() is called only if we are running from a
KUnit test.
+ * This can occur from a legitimate RV test or any unrelated test
running when
+ * a real RV monitor is active and triggering events.
+ * We assume the former case is the only one where mock_current is
not NULL and
+ * can occur only sequentially (KUnit doesn't run tests in parallel).
+ * We cannot rely on the test's context because there is no way to
safely
+ * understand from which test we are running and KUnit utilities
require
+ * locking, which is unsafe from NMI or scheduling context.
+ * Note that it is not possible for a real RV monitor to run when
the RV KUnit
+ * tests are running (see rv_set_testing()).
+ */
+static struct task_struct *mock_current;
+
+void rv_mock_current(struct task_struct *tsk)
+{
+ mock_current = tsk;
+}
+EXPORT_SYMBOL_IF_KUNIT(rv_mock_current);
+
+struct task_struct *rv_get_mock_current(void)
+{
+ return mock_current ?: current;
+}
+EXPORT_SYMBOL_GPL(rv_get_mock_current);
#endif
rv_mock_current() uses EXPORT_SYMBOL_IF_KUNIT, but
rv_get_mock_current() uses EXPORT_SYMBOL_GPL. Both are defined inside
the same CONFIG_RV_MONITORS_KUNIT_TEST block, so rv_get_mock_current
should use EXPORT_SYMBOL_IF_KUNIT as well, otherwise it leaks a
test-only symbol into production builds.
With that fixed:
Reviewed-by: Wen Yang <[email protected]>
Thanks for the review.
This was intentional however: rv_get_current() can be called by any
monitor, those don't have to be KUnit.
Since rv_get_current() is a macro also calling rv_get_mock_current()
we need to be able to link that too.
The idea is that a "real" (non-kunit) monitor handler could be run
when interrupting a KUnit test (not an RV one, we make sure of that).
In that case we do call rv_get_mock_current() and return current after
the function call.
rv_mock_current() CANNOT be called outside of the RV KUnit test cases,
it uses a global variable (for problems I tried to explain in the
comment), so should be exported only to KUnit and called directly from
the test case.
Does it make sense to you?
Thanks for the explanation.
That does make sense.
One small thought though -- the kernel already provides
KUNIT_STATIC_STUB_REDIRECT as the idiomatic way to handle this kind of
pattern. It allows a production function to be transparently redirected
to a stub during KUnit tests, with zero overhead when no test is running
(since it uses the same static_branch_unlikely(&kunit_running) path as
kunit_get_current_test()).
This might be a cleaner fit here, since it avoids exporting
rv_mock_current() beyond KUnit and keeps the redirection logic internal
to the test harness.
Just wanted to throw that out there -- what do you think?
I apologize for the noise. After reading the V5 commit notes more
carefully:
* Drop static stub from current to avoid issues with unrelated KUnit
tests
it is clear that KUNIT_STATIC_STUB_REDIRECT was already considered and
deliberately dropped. Kunit_find_resource() may acquires
spin_lock_irqsave, which does not mask NMIs; using the static stub in an
NMI-reachable path would risk deadlock.
This patch is the correct choice, and the comment in rv.c explains
exactly why.
Reviewed-by: Wen Yang <[email protected]>
--
Best wishes,
Wen
--
Best wishes,
Wen
