Merge branch 'destructive bpf_kfuncs'

Artem Savkov says:

====================

eBPF is often used for kernel debugging, and one of the widely used and
powerful debugging techniques is post-mortem debugging with a full memory dump.
Triggering a panic at exactly the right moment allows the user to get such a
dump and thus a better view at the system's state. Right now the only way to
do this in BPF is to signal userspace to trigger kexec/panic. This is
suboptimal as going through userspace requires context changes and adds
significant delays taking system further away from "the right moment". On a
single-cpu system the situation is even worse because BPF program won't even be
able to block the thread of interest.

This patchset tries to solve this problem by allowing properly marked tracing
bpf programs to call crash_kexec() kernel function. The only requirement for
now to run programs calling crash_kexec() or other destructive kfuncs is
CAP_SYS_BOOT capability. When signature checking for bpf programs is available
it is possible that stricter rules will be applied to programs utilizing
destructive kfuncs.

Changes in v5:
 - documentation numbering fixed
 - no more warning on failed kfunc registration

Changes in v4:
 - added description for KF_DESTRUCTIVE flag to documentation

Changes in v3:
 - moved kfunc set registration to kernel/bpf/helpers.c

Changes in v2:
 - BPF_PROG_LOAD flag dropped as it doesn't fully achieve it's aim of
   preventing accidental execution of destructive bpf programs
 - selftest moved to the end of patchset
 - switched to kfunc destructive flag instead of a separate set

Changes from RFC:
 - sysctl knob dropped
 - using crash_kexec() instead of panic()
 - using kfuncs instead of adding a new helper
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Documentation/bpf/kfuncs.rst

@@ -152,6 +152,15 @@ ensure the integrity of the operation being performed on the expected object.
 The KF_SLEEPABLE flag is used for kfuncs that may sleep. Such kfuncs can only
 be called by sleepable BPF programs (BPF_F_SLEEPABLE).
 
+2.4.7 KF_DESTRUCTIVE flag
+--------------------------
+
+The KF_DESTRUCTIVE flag is used to indicate functions calling which is
+destructive to the system. For example such a call can result in system
+rebooting or panicking. Due to this additional restrictions apply to these
+calls. At the moment they only require CAP_SYS_BOOT capability, but more can be
+added later.
+
 2.5 Registering the kfuncs
 --------------------------
 

include/linux/btf.h

@@ -49,7 +49,8 @@
  * for this case.
  */
 #define KF_TRUSTED_ARGS (1 << 4) /* kfunc only takes trusted pointer arguments */
-#define KF_SLEEPABLE   (1 << 5) /* kfunc may sleep */
+#define KF_SLEEPABLE    (1 << 5) /* kfunc may sleep */
+#define KF_DESTRUCTIVE  (1 << 6) /* kfunc performs destructive actions */
 
 struct btf;
 struct btf_member;

kernel/bpf/helpers.c

@@ -1725,3 +1725,21 @@ bpf_base_func_proto(enum bpf_func_id func_id)
 		return NULL;
 	}
 }
+
+BTF_SET8_START(tracing_btf_ids)
+#ifdef CONFIG_KEXEC_CORE
+BTF_ID_FLAGS(func, crash_kexec, KF_DESTRUCTIVE)
+#endif
+BTF_SET8_END(tracing_btf_ids)
+
+static const struct btf_kfunc_id_set tracing_kfunc_set = {
+	.owner = THIS_MODULE,
+	.set   = &tracing_btf_ids,
+};
+
+static int __init kfunc_init(void)
+{
+	return register_btf_kfunc_id_set(BPF_PROG_TYPE_TRACING, &tracing_kfunc_set);
+}
+
+late_initcall(kfunc_init);

kernel/bpf/verifier.c

@@ -7584,6 +7584,11 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 			func_name);
 		return -EACCES;
 	}
+	if (*kfunc_flags & KF_DESTRUCTIVE && !capable(CAP_SYS_BOOT)) {
+		verbose(env, "destructive kfunc calls require CAP_SYS_BOOT capabilities\n");
+		return -EACCES;
+	}
+
 	acq = *kfunc_flags & KF_ACQUIRE;
 
 	/* Check the arguments */

net/bpf/test_run.c

@@ -695,6 +695,10 @@ noinline void bpf_kfunc_call_test_ref(struct prog_test_ref_kfunc *p)
 {
 }
 
+noinline void bpf_kfunc_call_test_destructive(void)
+{
+}
+
 __diag_pop();
 
 ALLOW_ERROR_INJECTION(bpf_modify_return_test, ERRNO);
@@ -719,6 +723,7 @@ BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_pass1)
 BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail1)
 BTF_ID_FLAGS(func, bpf_kfunc_call_test_mem_len_fail2)
 BTF_ID_FLAGS(func, bpf_kfunc_call_test_ref, KF_TRUSTED_ARGS)
+BTF_ID_FLAGS(func, bpf_kfunc_call_test_destructive, KF_DESTRUCTIVE)
 BTF_SET8_END(test_sk_check_kfunc_ids)
 
 static void *bpf_test_init(const union bpf_attr *kattr, u32 user_size,

tools/testing/selftests/bpf/prog_tests/kfunc_call.c

@@ -5,6 +5,9 @@
 #include "kfunc_call_test.lskel.h"
 #include "kfunc_call_test_subprog.skel.h"
 #include "kfunc_call_test_subprog.lskel.h"
+#include "kfunc_call_destructive.skel.h"
+
+#include "cap_helpers.h"
 
 static void test_main(void)
 {
@@ -86,6 +89,36 @@ static void test_subprog_lskel(void)
 	kfunc_call_test_subprog_lskel__destroy(skel);
 }
 
+static int test_destructive_open_and_load(void)
+{
+	struct kfunc_call_destructive *skel;
+	int err;
+
+	skel = kfunc_call_destructive__open();
+	if (!ASSERT_OK_PTR(skel, "prog_open"))
+		return -1;
+
+	err = kfunc_call_destructive__load(skel);
+
+	kfunc_call_destructive__destroy(skel);
+
+	return err;
+}
+
+static void test_destructive(void)
+{
+	__u64 save_caps = 0;
+
+	ASSERT_OK(test_destructive_open_and_load(), "succesful_load");
+
+	if (!ASSERT_OK(cap_disable_effective(1ULL << CAP_SYS_BOOT, &save_caps), "drop_caps"))
+		return;
+
+	ASSERT_EQ(test_destructive_open_and_load(), -13, "no_caps_failure");
+
+	cap_enable_effective(save_caps, NULL);
+}
+
 void test_kfunc_call(void)
 {
 	if (test__start_subtest("main"))
@@ -96,4 +129,7 @@ void test_kfunc_call(void)
 
 	if (test__start_subtest("subprog_lskel"))
 		test_subprog_lskel();
+
+	if (test__start_subtest("destructive"))
+		test_destructive();
 }

tools/testing/selftests/bpf/progs/kfunc_call_destructive.c

@@ -0,0 +1,14 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+
+extern void bpf_kfunc_call_test_destructive(void) __ksym;
+
+SEC("tc")
+int kfunc_destructive_test(void)
+{
+	bpf_kfunc_call_test_destructive();
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";