sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

2 files changed, 196 insertions, 0 deletions

diff --git a/fs/tests/binfmt_elf_kunit.c b/fs/tests/binfmt_elf_kunit.c
new file mode 100644
index 000000000..11d734fec
--- /dev/null
+++ b/fs/tests/binfmt_elf_kunit.c
@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <kunit/test.h>
+
+static void total_mapping_size_test(struct kunit *test)
+{
+	struct elf_phdr empty[] = {
+		{ .p_type = PT_LOAD, .p_vaddr = 0, .p_memsz = 0, },
+		{ .p_type = PT_INTERP, .p_vaddr = 10, .p_memsz = 999999, },
+	};
+	/*
+	 * readelf -lW /bin/mount | grep '^  .*0x0' | awk '{print "\t\t{ .p_type = PT_" \
+	 *				$1 ", .p_vaddr = " $3 ", .p_memsz = " $6 ", },"}'
+	 */
+	struct elf_phdr mount[] = {
+		{ .p_type = PT_PHDR, .p_vaddr = 0x00000040, .p_memsz = 0x0002d8, },
+		{ .p_type = PT_INTERP, .p_vaddr = 0x00000318, .p_memsz = 0x00001c, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x00000000, .p_memsz = 0x0033a8, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x00004000, .p_memsz = 0x005c91, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x0000a000, .p_memsz = 0x0022f8, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x0000d330, .p_memsz = 0x000d40, },
+		{ .p_type = PT_DYNAMIC, .p_vaddr = 0x0000d928, .p_memsz = 0x000200, },
+		{ .p_type = PT_NOTE, .p_vaddr = 0x00000338, .p_memsz = 0x000030, },
+		{ .p_type = PT_NOTE, .p_vaddr = 0x00000368, .p_memsz = 0x000044, },
+		{ .p_type = PT_GNU_PROPERTY, .p_vaddr = 0x00000338, .p_memsz = 0x000030, },
+		{ .p_type = PT_GNU_EH_FRAME, .p_vaddr = 0x0000b490, .p_memsz = 0x0001ec, },
+		{ .p_type = PT_GNU_STACK, .p_vaddr = 0x00000000, .p_memsz = 0x000000, },
+		{ .p_type = PT_GNU_RELRO, .p_vaddr = 0x0000d330, .p_memsz = 0x000cd0, },
+	};
+	size_t mount_size = 0xE070;
+	/* https://lore.kernel.org/linux-fsdevel/YfF18Dy85mCntXrx@fractal.localdomain */
+	struct elf_phdr unordered[] = {
+		{ .p_type = PT_LOAD, .p_vaddr = 0x00000000, .p_memsz = 0x0033a8, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x0000d330, .p_memsz = 0x000d40, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x00004000, .p_memsz = 0x005c91, },
+		{ .p_type = PT_LOAD, .p_vaddr = 0x0000a000, .p_memsz = 0x0022f8, },
+	};
+
+	/* No headers, no size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(NULL, 0), 0);
+	KUNIT_EXPECT_EQ(test, total_mapping_size(empty, 0), 0);
+	/* Empty headers, no size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(empty, 1), 0);
+	/* No PT_LOAD headers, no size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(&empty[1], 1), 0);
+	/* Empty PT_LOAD and non-PT_LOAD headers, no size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(empty, 2), 0);
+
+	/* Normal set of PT_LOADS, and expected size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(mount, ARRAY_SIZE(mount)), mount_size);
+	/* Unordered PT_LOADs result in same size. */
+	KUNIT_EXPECT_EQ(test, total_mapping_size(unordered, ARRAY_SIZE(unordered)), mount_size);
+}
+
+static struct kunit_case binfmt_elf_test_cases[] = {
+	KUNIT_CASE(total_mapping_size_test),
+	{},
+};
+
+static struct kunit_suite binfmt_elf_test_suite = {
+	.name = KBUILD_MODNAME,
+	.test_cases = binfmt_elf_test_cases,
+};
+
+kunit_test_suite(binfmt_elf_test_suite);
diff --git a/fs/tests/exec_kunit.c b/fs/tests/exec_kunit.c
new file mode 100644
index 000000000..1c32cac09
--- /dev/null
+++ b/fs/tests/exec_kunit.c
@@ -0,0 +1,132 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <kunit/test.h>
+
+struct bprm_stack_limits_result {
+	struct linux_binprm bprm;
+	int expected_rc;
+	unsigned long expected_argmin;
+};
+
+static const struct bprm_stack_limits_result bprm_stack_limits_results[] = {
+	/* Negative argc/envc counts produce -E2BIG */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = INT_MIN, .envc = INT_MIN }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = 5, .envc = -1 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = -1, .envc = 10 }, .expected_rc = -E2BIG },
+	/* The max value of argc or envc is MAX_ARG_STRINGS. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = INT_MAX, .envc = INT_MAX }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = MAX_ARG_STRINGS, .envc = MAX_ARG_STRINGS }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = 0, .envc = MAX_ARG_STRINGS }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = MAX_ARG_STRINGS, .envc = 0 }, .expected_rc = -E2BIG },
+	/*
+	 * On 32-bit system these argc and envc counts, while likely impossible
+	 * to represent within the associated TASK_SIZE, could overflow the
+	 * limit calculation, and bypass the ptr_size <= limit check.
+	 */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = 0x20000001, .envc = 0x20000001 }, .expected_rc = -E2BIG },
+#ifdef CONFIG_MMU
+	/* Make sure a pathological bprm->p doesn't cause an overflow. */
+	{ { .p = sizeof(void *), .rlim_stack.rlim_cur = ULONG_MAX,
+	    .argc = 10, .envc = 10 }, .expected_rc = -E2BIG },
+#endif
+	/*
+	 * 0 rlim_stack will get raised to ARG_MAX. With 1 string pointer,
+	 * we should see p - ARG_MAX + sizeof(void *).
+	 */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = 1, .envc = 0 }, .expected_argmin = ULONG_MAX - ARG_MAX + sizeof(void *)},
+	/* Validate that argc is always raised to a minimum of 1. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = 0, .envc = 0 }, .expected_argmin = ULONG_MAX - ARG_MAX + sizeof(void *)},
+	/*
+	 * 0 rlim_stack will get raised to ARG_MAX. With pointers filling ARG_MAX,
+	 * we should see -E2BIG. (Note argc is always raised to at least 1.)
+	 */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = ARG_MAX / sizeof(void *), .envc = 0 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = 0, .envc = ARG_MAX / sizeof(void *) - 1 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = ARG_MAX / sizeof(void *) + 1, .envc = 0 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = 0, .envc = ARG_MAX / sizeof(void *) }, .expected_rc = -E2BIG },
+	/* And with one less, we see space for exactly 1 pointer. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = (ARG_MAX / sizeof(void *)) - 1, .envc = 0 },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 0,
+	    .argc = 0, .envc = (ARG_MAX / sizeof(void *)) - 2, },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	/* If we raise rlim_stack / 4 to exactly ARG_MAX, nothing changes. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = ARG_MAX / sizeof(void *), .envc = 0 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = 0, .envc = ARG_MAX / sizeof(void *) - 1 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = ARG_MAX / sizeof(void *) + 1, .envc = 0 }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = 0, .envc = ARG_MAX / sizeof(void *) }, .expected_rc = -E2BIG },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = (ARG_MAX / sizeof(void *)) - 1, .envc = 0 },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = ARG_MAX * 4,
+	    .argc = 0, .envc = (ARG_MAX / sizeof(void *)) - 2, },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	/* But raising it another pointer * 4 will provide space for 1 more pointer. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = (ARG_MAX + sizeof(void *)) * 4,
+	    .argc = ARG_MAX / sizeof(void *), .envc = 0 },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = (ARG_MAX + sizeof(void *)) * 4,
+	    .argc = 0, .envc = ARG_MAX / sizeof(void *) - 1 },
+	  .expected_argmin = ULONG_MAX - sizeof(void *) },
+	/* Raising rlim_stack / 4 to _STK_LIM / 4 * 3 will see more space. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * (_STK_LIM / 4 * 3),
+	    .argc = 0, .envc = 0 },
+	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
+	/* But raising it any further will see no increase. */
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * (_STK_LIM / 4 * 3 + sizeof(void *)),
+	    .argc = 0, .envc = 0 },
+	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
+	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * _STK_LIM,
+	    .argc = 0, .envc = 0 },
+	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
+};
+
+static void exec_test_bprm_stack_limits(struct kunit *test)
+{
+	/* Double-check the constants. */
+	KUNIT_EXPECT_EQ(test, _STK_LIM, SZ_8M);
+	KUNIT_EXPECT_EQ(test, ARG_MAX, 32 * SZ_4K);
+	KUNIT_EXPECT_EQ(test, MAX_ARG_STRINGS, 0x7FFFFFFF);
+
+	for (int i = 0; i < ARRAY_SIZE(bprm_stack_limits_results); i++) {
+		const struct bprm_stack_limits_result *result = &bprm_stack_limits_results[i];
+		struct linux_binprm bprm = result->bprm;
+		int rc;
+
+		rc = bprm_stack_limits(&bprm);
+		KUNIT_EXPECT_EQ_MSG(test, rc, result->expected_rc, "on loop %d", i);
+#ifdef CONFIG_MMU
+		KUNIT_EXPECT_EQ_MSG(test, bprm.argmin, result->expected_argmin, "on loop %d", i);
+#endif
+	}
+}
+
+static struct kunit_case exec_test_cases[] = {
+	KUNIT_CASE(exec_test_bprm_stack_limits),
+	{},
+};
+
+static struct kunit_suite exec_test_suite = {
+	.name = "exec",
+	.test_cases = exec_test_cases,
+};
+
+kunit_test_suite(exec_test_suite);