Merge branch 'sleepable uprobe support'

Delyan Kratunov says:

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

This series implements support for sleepable uprobe programs.
Key work is in patches 2 and 3, the rest is plumbing and tests.

The main observation is that the only obstacle in the way of sleepable uprobe
programs is not the uprobe infrastructure, which already runs in a user-like
context, but the rcu usage around bpf_prog_array.

Details are in patch 2 but the tl;dr is that we chain trace_tasks and normal rcu
grace periods when releasing to array to accommodate users of either rcu type.
This introduces latency for non-sleepable users (kprobe, tp) but that's deemed
acceptable, given recent benchmarks by Andrii [1]. We're a couple of orders of
magnitude under the rate of bpf_prog_array churn that would raise flags (~1MM/s per Paul).

  [1]: https://lore.kernel.org/bpf/CAEf4BzbpjN6ca7D9KOTiFPOoBYkciYvTz0UJNp5c-_3ptm=Mrg@mail.gmail.com/

v3 -> v4:
 * Fix kdoc and inline issues
 * Rebase

v2 -> v3:
 * Inline uprobe_call_bpf into trace_uprobe.c, it's just a bpf_prog_run_array_sleepable call now.
 * Do not disable preemption for uprobe non-sleepable programs.
 * Add acks.

v1 -> v2:
 * Fix lockdep annotations in bpf_prog_run_array_sleepable
 * Chain rcu grace periods only for perf_event-attached programs. This limits
   the additional latency on the free path to use cases where we know it won't
   be a problem.
 * Add tests calling helpers only available in sleepable programs.
 * Remove kprobe.s support from libbpf.
====================

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

3 files changed, 112 insertions, 2 deletions

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index d989b0a17a89..49e359cd34df 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -9177,8 +9177,10 @@ static const struct bpf_sec_def section_defs[] = {
 	SEC_DEF("sk_reuseport",		SK_REUSEPORT, BPF_SK_REUSEPORT_SELECT, SEC_ATTACHABLE | SEC_SLOPPY_PFX),
 	SEC_DEF("kprobe+",		KPROBE,	0, SEC_NONE, attach_kprobe),
 	SEC_DEF("uprobe+",		KPROBE,	0, SEC_NONE, attach_uprobe),
+	SEC_DEF("uprobe.s+",		KPROBE,	0, SEC_SLEEPABLE, attach_uprobe),
 	SEC_DEF("kretprobe+",		KPROBE, 0, SEC_NONE, attach_kprobe),
 	SEC_DEF("uretprobe+",		KPROBE, 0, SEC_NONE, attach_uprobe),
+	SEC_DEF("uretprobe.s+",		KPROBE, 0, SEC_SLEEPABLE, attach_uprobe),
 	SEC_DEF("kprobe.multi+",	KPROBE,	BPF_TRACE_KPROBE_MULTI, SEC_NONE, attach_kprobe_multi),
 	SEC_DEF("kretprobe.multi+",	KPROBE,	BPF_TRACE_KPROBE_MULTI, SEC_NONE, attach_kprobe_multi),
 	SEC_DEF("usdt+",		KPROBE,	0, SEC_NONE, attach_usdt),
@@ -11571,7 +11573,8 @@ static int attach_uprobe(const struct bpf_program *prog, long cookie, struct bpf
 		break;
 	case 3:
 	case 4:
-		opts.retprobe = strcmp(probe_type, "uretprobe") == 0;
+		opts.retprobe = strcmp(probe_type, "uretprobe") == 0 ||
+				strcmp(probe_type, "uretprobe.s") == 0;
 		if (opts.retprobe && offset != 0) {
 			pr_warn("prog '%s': uretprobes do not support offset specification\n",
 				prog->name);
diff --git a/tools/testing/selftests/bpf/prog_tests/attach_probe.c b/tools/testing/selftests/bpf/prog_tests/attach_probe.c
index 08c0601b3e84..0b899d2d8ea7 100644
--- a/tools/testing/selftests/bpf/prog_tests/attach_probe.c
+++ b/tools/testing/selftests/bpf/prog_tests/attach_probe.c
@@ -17,6 +17,14 @@ static void trigger_func2(void)
 	asm volatile ("");
 }
 
+/* attach point for byname sleepable uprobe */
+static void trigger_func3(void)
+{
+	asm volatile ("");
+}
+
+static char test_data[] = "test_data";
+
 void test_attach_probe(void)
 {
 	DECLARE_LIBBPF_OPTS(bpf_uprobe_opts, uprobe_opts);
@@ -49,9 +57,17 @@ void test_attach_probe(void)
 	if (!ASSERT_GE(ref_ctr_offset, 0, "ref_ctr_offset"))
 		return;
 
-	skel = test_attach_probe__open_and_load();
+	skel = test_attach_probe__open();
 	if (!ASSERT_OK_PTR(skel, "skel_open"))
 		return;
+
+	/* sleepable kprobe test case needs flags set before loading */
+	if (!ASSERT_OK(bpf_program__set_flags(skel->progs.handle_kprobe_sleepable,
+		BPF_F_SLEEPABLE), "kprobe_sleepable_flags"))
+		goto cleanup;
+
+	if (!ASSERT_OK(test_attach_probe__load(skel), "skel_load"))
+		goto cleanup;
 	if (!ASSERT_OK_PTR(skel->bss, "check_bss"))
 		goto cleanup;
 
@@ -151,6 +167,30 @@ void test_attach_probe(void)
 	if (!ASSERT_OK_PTR(skel->links.handle_uretprobe_byname2, "attach_uretprobe_byname2"))
 		goto cleanup;
 
+	/* sleepable kprobes should not attach successfully */
+	skel->links.handle_kprobe_sleepable = bpf_program__attach(skel->progs.handle_kprobe_sleepable);
+	if (!ASSERT_ERR_PTR(skel->links.handle_kprobe_sleepable, "attach_kprobe_sleepable"))
+		goto cleanup;
+
+	/* test sleepable uprobe and uretprobe variants */
+	skel->links.handle_uprobe_byname3_sleepable = bpf_program__attach(skel->progs.handle_uprobe_byname3_sleepable);
+	if (!ASSERT_OK_PTR(skel->links.handle_uprobe_byname3_sleepable, "attach_uprobe_byname3_sleepable"))
+		goto cleanup;
+
+	skel->links.handle_uprobe_byname3 = bpf_program__attach(skel->progs.handle_uprobe_byname3);
+	if (!ASSERT_OK_PTR(skel->links.handle_uprobe_byname3, "attach_uprobe_byname3"))
+		goto cleanup;
+
+	skel->links.handle_uretprobe_byname3_sleepable = bpf_program__attach(skel->progs.handle_uretprobe_byname3_sleepable);
+	if (!ASSERT_OK_PTR(skel->links.handle_uretprobe_byname3_sleepable, "attach_uretprobe_byname3_sleepable"))
+		goto cleanup;
+
+	skel->links.handle_uretprobe_byname3 = bpf_program__attach(skel->progs.handle_uretprobe_byname3);
+	if (!ASSERT_OK_PTR(skel->links.handle_uretprobe_byname3, "attach_uretprobe_byname3"))
+		goto cleanup;
+
+	skel->bss->user_ptr = test_data;
+
 	/* trigger & validate kprobe && kretprobe */
 	usleep(1);
 
@@ -164,6 +204,9 @@ void test_attach_probe(void)
 	/* trigger & validate uprobe attached by name */
 	trigger_func2();
 
+	/* trigger & validate sleepable uprobe attached by name */
+	trigger_func3();
+
 	ASSERT_EQ(skel->bss->kprobe_res, 1, "check_kprobe_res");
 	ASSERT_EQ(skel->bss->kprobe2_res, 11, "check_kprobe_auto_res");
 	ASSERT_EQ(skel->bss->kretprobe_res, 2, "check_kretprobe_res");
@@ -174,6 +217,10 @@ void test_attach_probe(void)
 	ASSERT_EQ(skel->bss->uretprobe_byname_res, 6, "check_uretprobe_byname_res");
 	ASSERT_EQ(skel->bss->uprobe_byname2_res, 7, "check_uprobe_byname2_res");
 	ASSERT_EQ(skel->bss->uretprobe_byname2_res, 8, "check_uretprobe_byname2_res");
+	ASSERT_EQ(skel->bss->uprobe_byname3_sleepable_res, 9, "check_uprobe_byname3_sleepable_res");
+	ASSERT_EQ(skel->bss->uprobe_byname3_res, 10, "check_uprobe_byname3_res");
+	ASSERT_EQ(skel->bss->uretprobe_byname3_sleepable_res, 11, "check_uretprobe_byname3_sleepable_res");
+	ASSERT_EQ(skel->bss->uretprobe_byname3_res, 12, "check_uretprobe_byname3_res");
 
 cleanup:
 	test_attach_probe__destroy(skel);
diff --git a/tools/testing/selftests/bpf/progs/test_attach_probe.c b/tools/testing/selftests/bpf/progs/test_attach_probe.c
index ce9acf4db8d2..f1c88ad368ef 100644
--- a/tools/testing/selftests/bpf/progs/test_attach_probe.c
+++ b/tools/testing/selftests/bpf/progs/test_attach_probe.c
@@ -5,6 +5,7 @@
 #include <linux/bpf.h>
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_tracing.h>
+#include <stdbool.h>
 #include "bpf_misc.h"
 
 int kprobe_res = 0;
@@ -17,6 +18,11 @@ int uprobe_byname_res = 0;
 int uretprobe_byname_res = 0;
 int uprobe_byname2_res = 0;
 int uretprobe_byname2_res = 0;
+int uprobe_byname3_sleepable_res = 0;
+int uprobe_byname3_res = 0;
+int uretprobe_byname3_sleepable_res = 0;
+int uretprobe_byname3_res = 0;
+void *user_ptr = 0;
 
 SEC("kprobe")
 int handle_kprobe(struct pt_regs *ctx)
@@ -32,6 +38,17 @@ int BPF_KPROBE(handle_kprobe_auto)
 	return 0;
 }
 
+/**
+ * This program will be manually made sleepable on the userspace side
+ * and should thus be unattachable.
+ */
+SEC("kprobe/" SYS_PREFIX "sys_nanosleep")
+int handle_kprobe_sleepable(struct pt_regs *ctx)
+{
+	kprobe_res = 2;
+	return 0;
+}
+
 SEC("kretprobe")
 int handle_kretprobe(struct pt_regs *ctx)
 {
@@ -93,4 +110,47 @@ int handle_uretprobe_byname2(struct pt_regs *ctx)
 	return 0;
 }
 
+static __always_inline bool verify_sleepable_user_copy(void)
+{
+	char data[9];
+
+	bpf_copy_from_user(data, sizeof(data), user_ptr);
+	return bpf_strncmp(data, sizeof(data), "test_data") == 0;
+}
+
+SEC("uprobe.s//proc/self/exe:trigger_func3")
+int handle_uprobe_byname3_sleepable(struct pt_regs *ctx)
+{
+	if (verify_sleepable_user_copy())
+		uprobe_byname3_sleepable_res = 9;
+	return 0;
+}
+
+/**
+ * same target as the uprobe.s above to force sleepable and non-sleepable
+ * programs in the same bpf_prog_array
+ */
+SEC("uprobe//proc/self/exe:trigger_func3")
+int handle_uprobe_byname3(struct pt_regs *ctx)
+{
+	uprobe_byname3_res = 10;
+	return 0;
+}
+
+SEC("uretprobe.s//proc/self/exe:trigger_func3")
+int handle_uretprobe_byname3_sleepable(struct pt_regs *ctx)
+{
+	if (verify_sleepable_user_copy())
+		uretprobe_byname3_sleepable_res = 11;
+	return 0;
+}
+
+SEC("uretprobe//proc/self/exe:trigger_func3")
+int handle_uretprobe_byname3(struct pt_regs *ctx)
+{
+	uretprobe_byname3_res = 12;
+	return 0;
+}
+
+
 char _license[] SEC("license") = "GPL";