selftests/bpf: Test packet range of pointers sharing an idgrafted

Add tests where two packet pointers share an id and tightening one
pointer's umax from its var_off would put it less than their constant
distance from the other's umax: with an index & 0x38 capped at 50, the
base pointer keeps umax 50, so the pointer 8 bytes further on must keep
umax 58, even though its known bits allow at most 56.

These refused a valid program or accepted an out-of-bounds access before
the fix:

- check the advanced copy, load through the base: valid, was refused;
- check the base, load the byte at base + 1 through a copy advanced by
  8: was accepted;
- check base + 4, load 4 bytes at base + 2 through base + 8: reads two
  bytes past the checked range, was accepted;
- the same as the second with data_meta pointers checked against data:
  was accepted.

These pass with and without the fix and cover nearby paths:

- subtract an unknown scalar from a checked pointer and load below it
  (the range is kept across a new id);
- reach a load through two paths whose checks cover 8 and 7 bytes after
  the loaded pointer; the second path must not be pruned by the first;
- spill a copy of a pointer, check the pointer, fill the copy and load
  one byte past the checked range: the load is refused, and the copy
  has the range of the check.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20261001145255.855630-2-alexei.starovoitov@gmail.com
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>

1 files changed, 107 insertions, 0 deletions

diff --git a/include/asm-generic/local64.h b/include/asm-generic/local64.h
new file mode 100644
index 000000000..14963a7a6
--- /dev/null
+++ b/include/asm-generic/local64.h
@@ -0,0 +1,107 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_GENERIC_LOCAL64_H
+#define _ASM_GENERIC_LOCAL64_H
+
+#include <linux/percpu.h>
+#include <asm/types.h>
+
+/*
+ * A signed long type for operations which are atomic for a single CPU.
+ * Usually used in combination with per-cpu variables.
+ *
+ * This is the default implementation, which uses atomic64_t.  Which is
+ * rather pointless.  The whole point behind local64_t is that some processors
+ * can perform atomic adds and subtracts in a manner which is atomic wrt IRQs
+ * running on this CPU.  local64_t allows exploitation of such capabilities.
+ */
+
+/* Implement in terms of atomics. */
+
+#if BITS_PER_LONG == 64
+
+#include <asm/local.h>
+
+typedef struct {
+	local_t a;
+} local64_t;
+
+#define LOCAL64_INIT(i)	{ LOCAL_INIT(i) }
+
+#define local64_read(l)		local_read(&(l)->a)
+#define local64_set(l,i)	local_set((&(l)->a),(i))
+#define local64_inc(l)		local_inc(&(l)->a)
+#define local64_dec(l)		local_dec(&(l)->a)
+#define local64_add(i,l)	local_add((i),(&(l)->a))
+#define local64_sub(i,l)	local_sub((i),(&(l)->a))
+
+#define local64_sub_and_test(i, l) local_sub_and_test((i), (&(l)->a))
+#define local64_dec_and_test(l) local_dec_and_test(&(l)->a)
+#define local64_inc_and_test(l) local_inc_and_test(&(l)->a)
+#define local64_add_negative(i, l) local_add_negative((i), (&(l)->a))
+#define local64_add_return(i, l) local_add_return((i), (&(l)->a))
+#define local64_sub_return(i, l) local_sub_return((i), (&(l)->a))
+#define local64_inc_return(l)	local_inc_return(&(l)->a)
+
+static inline s64 local64_cmpxchg(local64_t *l, s64 old, s64 new)
+{
+	return local_cmpxchg(&l->a, old, new);
+}
+
+static inline bool local64_try_cmpxchg(local64_t *l, s64 *old, s64 new)
+{
+	return local_try_cmpxchg(&l->a, (long *)old, new);
+}
+
+#define local64_xchg(l, n)	local_xchg((&(l)->a), (n))
+#define local64_add_unless(l, _a, u) local_add_unless((&(l)->a), (_a), (u))
+#define local64_inc_not_zero(l)	local_inc_not_zero(&(l)->a)
+
+/* Non-atomic variants, ie. preemption disabled and won't be touched
+ * in interrupt, etc.  Some archs can optimize this case well. */
+#define __local64_inc(l)	local64_set((l), local64_read(l) + 1)
+#define __local64_dec(l)	local64_set((l), local64_read(l) - 1)
+#define __local64_add(i,l)	local64_set((l), local64_read(l) + (i))
+#define __local64_sub(i,l)	local64_set((l), local64_read(l) - (i))
+
+#else /* BITS_PER_LONG != 64 */
+
+#include <linux/atomic.h>
+
+/* Don't use typedef: don't want them to be mixed with atomic_t's. */
+typedef struct {
+	atomic64_t a;
+} local64_t;
+
+#define LOCAL64_INIT(i)	{ ATOMIC_LONG_INIT(i) }
+
+#define local64_read(l)		atomic64_read(&(l)->a)
+#define local64_set(l,i)	atomic64_set((&(l)->a),(i))
+#define local64_inc(l)		atomic64_inc(&(l)->a)
+#define local64_dec(l)		atomic64_dec(&(l)->a)
+#define local64_add(i,l)	atomic64_add((i),(&(l)->a))
+#define local64_sub(i,l)	atomic64_sub((i),(&(l)->a))
+
+#define local64_sub_and_test(i, l) atomic64_sub_and_test((i), (&(l)->a))
+#define local64_dec_and_test(l) atomic64_dec_and_test(&(l)->a)
+#define local64_inc_and_test(l) atomic64_inc_and_test(&(l)->a)
+#define local64_add_negative(i, l) atomic64_add_negative((i), (&(l)->a))
+#define local64_add_return(i, l) atomic64_add_return((i), (&(l)->a))
+#define local64_sub_return(i, l) atomic64_sub_return((i), (&(l)->a))
+#define local64_inc_return(l)	atomic64_inc_return(&(l)->a)
+
+#define local64_cmpxchg(l, o, n) atomic64_cmpxchg((&(l)->a), (o), (n))
+#define local64_try_cmpxchg(l, po, n) atomic64_try_cmpxchg((&(l)->a), (po), (n))
+#define local64_xchg(l, n)	atomic64_xchg((&(l)->a), (n))
+#define local64_add_unless(l, _a, u) atomic64_add_unless((&(l)->a), (_a), (u))
+#define local64_inc_not_zero(l)	atomic64_inc_not_zero(&(l)->a)
+
+/* Non-atomic variants, ie. preemption disabled and won't be touched
+ * in interrupt, etc.  Some archs can optimize this case well. */
+#define __local64_inc(l)	local64_set((l), local64_read(l) + 1)
+#define __local64_dec(l)	local64_set((l), local64_read(l) - 1)
+#define __local64_add(i,l)	local64_set((l), local64_read(l) + (i))
+#define __local64_sub(i,l)	local64_set((l), local64_read(l) - (i))
+
+#endif /* BITS_PER_LONG != 64 */
+
+#endif /* _ASM_GENERIC_LOCAL64_H */