1 files changed, 223 insertions, 85 deletions

diff --git a/tools/perf/examples/bpf/augmented_raw_syscalls.c b/tools/perf/examples/bpf/augmented_raw_syscalls.c
index f9b2161e1ca4..b80437971d80 100644
--- a/tools/perf/examples/bpf/augmented_raw_syscalls.c
+++ b/tools/perf/examples/bpf/augmented_raw_syscalls.c
@@ -15,17 +15,39 @@
  */
 
 #include <unistd.h>
+#include <linux/limits.h>
+#include <linux/socket.h>
 #include <pid_filter.h>
 
 /* bpf-output associated map */
 bpf_map(__augmented_syscalls__, PERF_EVENT_ARRAY, int, u32, __NR_CPUS__);
 
+/*
+ * string_args_len: one per syscall arg, 0 means not a string or don't copy it,
+ * 		    PATH_MAX for copying everything, any other value to limit
+ * 		    it a la 'strace -s strsize'.
+ */
 struct syscall {
 	bool	enabled;
+	u16	string_args_len[6];
 };
 
 bpf_map(syscalls, ARRAY, int, struct syscall, 512);
 
+/*
+ * What to augment at entry?
+ *
+ * Pointer arg payloads (filenames, etc) passed from userspace to the kernel
+ */
+bpf_map(syscalls_sys_enter, PROG_ARRAY, u32, u32, 512);
+
+/*
+ * What to augment at exit?
+ *
+ * Pointer arg payloads returned from the kernel (struct stat, etc) to userspace.
+ */
+bpf_map(syscalls_sys_exit, PROG_ARRAY, u32, u32, 512);
+
 struct syscall_enter_args {
 	unsigned long long common_tp_fields;
 	long		   syscall_nr;
@@ -38,119 +60,235 @@ struct syscall_exit_args {
 	long		   ret;
 };
 
-struct augmented_filename {
+struct augmented_arg {
 	unsigned int	size;
-	int		reserved;
-	char		value[256];
+	int		err;
+	char		value[PATH_MAX];
 };
 
-#define SYS_OPEN 2
-#define SYS_ACCESS 21
-#define SYS_OPENAT 257
-
 pid_filter(pids_filtered);
 
+struct augmented_args_payload {
+       struct syscall_enter_args args;
+       union {
+		struct {
+			struct augmented_arg arg, arg2;
+		};
+		struct sockaddr_storage saddr;
+	};
+};
+
+// We need more tmp space than the BPF stack can give us
+bpf_map(augmented_args_tmp, PERCPU_ARRAY, int, struct augmented_args_payload, 1);
+
+static inline struct augmented_args_payload *augmented_args_payload(void)
+{
+	int key = 0;
+	return bpf_map_lookup_elem(&augmented_args_tmp, &key);
+}
+
+static inline int augmented__output(void *ctx, struct augmented_args_payload *args, int len)
+{
+	/* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */
+	return perf_event_output(ctx, &__augmented_syscalls__, BPF_F_CURRENT_CPU, args, len);
+}
+
+static inline
+unsigned int augmented_arg__read_str(struct augmented_arg *augmented_arg, const void *arg, unsigned int arg_len)
+{
+	unsigned int augmented_len = sizeof(*augmented_arg);
+	int string_len = probe_read_str(&augmented_arg->value, arg_len, arg);
+
+	augmented_arg->size = augmented_arg->err = 0;
+	/*
+	 * probe_read_str may return < 0, e.g. -EFAULT
+	 * So we leave that in the augmented_arg->size that userspace will
+	 */
+	if (string_len > 0) {
+		augmented_len -= sizeof(augmented_arg->value) - string_len;
+		augmented_len &= sizeof(augmented_arg->value) - 1;
+		augmented_arg->size = string_len;
+	} else {
+		/*
+		 * So that username notice the error while still being able
+		 * to skip this augmented arg record
+		 */
+		augmented_arg->err = string_len;
+		augmented_len = offsetof(struct augmented_arg, value);
+	}
+
+	return augmented_len;
+}
+
+SEC("!raw_syscalls:unaugmented")
+int syscall_unaugmented(struct syscall_enter_args *args)
+{
+	return 1;
+}
+
+/*
+ * These will be tail_called from SEC("raw_syscalls:sys_enter"), so will find in
+ * augmented_args_tmp what was read by that raw_syscalls:sys_enter and go
+ * on from there, reading the first syscall arg as a string, i.e. open's
+ * filename.
+ */
+SEC("!syscalls:sys_enter_connect")
+int sys_enter_connect(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *sockaddr_arg = (const void *)args->args[1];
+	unsigned int socklen = args->args[2];
+	unsigned int len = sizeof(augmented_args->args);
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	if (socklen > sizeof(augmented_args->saddr))
+		socklen = sizeof(augmented_args->saddr);
+
+	probe_read(&augmented_args->saddr, socklen, sockaddr_arg);
+
+	return augmented__output(args, augmented_args, len + socklen);
+}
+
+SEC("!syscalls:sys_enter_sendto")
+int sys_enter_sendto(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *sockaddr_arg = (const void *)args->args[4];
+	unsigned int socklen = args->args[5];
+	unsigned int len = sizeof(augmented_args->args);
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	if (socklen > sizeof(augmented_args->saddr))
+		socklen = sizeof(augmented_args->saddr);
+
+	probe_read(&augmented_args->saddr, socklen, sockaddr_arg);
+
+	return augmented__output(args, augmented_args, len + socklen);
+}
+
+SEC("!syscalls:sys_enter_open")
+int sys_enter_open(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *filename_arg = (const void *)args->args[0];
+	unsigned int len = sizeof(augmented_args->args);
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	len += augmented_arg__read_str(&augmented_args->arg, filename_arg, sizeof(augmented_args->arg.value));
+
+	return augmented__output(args, augmented_args, len);
+}
+
+SEC("!syscalls:sys_enter_openat")
+int sys_enter_openat(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *filename_arg = (const void *)args->args[1];
+	unsigned int len = sizeof(augmented_args->args);
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	len += augmented_arg__read_str(&augmented_args->arg, filename_arg, sizeof(augmented_args->arg.value));
+
+	return augmented__output(args, augmented_args, len);
+}
+
+SEC("!syscalls:sys_enter_rename")
+int sys_enter_rename(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *oldpath_arg = (const void *)args->args[0],
+		   *newpath_arg = (const void *)args->args[1];
+	unsigned int len = sizeof(augmented_args->args), oldpath_len;
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	oldpath_len = augmented_arg__read_str(&augmented_args->arg, oldpath_arg, sizeof(augmented_args->arg.value));
+	len += oldpath_len + augmented_arg__read_str((void *)(&augmented_args->arg) + oldpath_len, newpath_arg, sizeof(augmented_args->arg.value));
+
+	return augmented__output(args, augmented_args, len);
+}
+
+SEC("!syscalls:sys_enter_renameat")
+int sys_enter_renameat(struct syscall_enter_args *args)
+{
+	struct augmented_args_payload *augmented_args = augmented_args_payload();
+	const void *oldpath_arg = (const void *)args->args[1],
+		   *newpath_arg = (const void *)args->args[3];
+	unsigned int len = sizeof(augmented_args->args), oldpath_len;
+
+        if (augmented_args == NULL)
+                return 1; /* Failure: don't filter */
+
+	oldpath_len = augmented_arg__read_str(&augmented_args->arg, oldpath_arg, sizeof(augmented_args->arg.value));
+	len += oldpath_len + augmented_arg__read_str((void *)(&augmented_args->arg) + oldpath_len, newpath_arg, sizeof(augmented_args->arg.value));
+
+	return augmented__output(args, augmented_args, len);
+}
+
 SEC("raw_syscalls:sys_enter")
 int sys_enter(struct syscall_enter_args *args)
 {
-	struct {
-		struct syscall_enter_args args;
-		struct augmented_filename filename;
-	} augmented_args;
+	struct augmented_args_payload *augmented_args;
+	/*
+	 * We start len, the amount of data that will be in the perf ring
+	 * buffer, if this is not filtered out by one of pid_filter__has(),
+	 * syscall->enabled, etc, with the non-augmented raw syscall payload,
+	 * i.e. sizeof(augmented_args->args).
+	 *
+	 * We'll add to this as we add augmented syscalls right after that
+	 * initial, non-augmented raw_syscalls:sys_enter payload.
+	 */
+	unsigned int len = sizeof(augmented_args->args);
 	struct syscall *syscall;
-	unsigned int len = sizeof(augmented_args);
-	const void *filename_arg = NULL;
 
 	if (pid_filter__has(&pids_filtered, getpid()))
 		return 0;
 
-	probe_read(&augmented_args.args, sizeof(augmented_args.args), args);
+	augmented_args = augmented_args_payload();
+	if (augmented_args == NULL)
+		return 1;
+
+	probe_read(&augmented_args->args, sizeof(augmented_args->args), args);
 
-	syscall = bpf_map_lookup_elem(&syscalls, &augmented_args.args.syscall_nr);
-	if (syscall == NULL || !syscall->enabled)
-		return 0;
 	/*
-	 * Yonghong and Edward Cree sayz:
-	 *
-	 * https://www.spinics.net/lists/netdev/msg531645.html
-	 *
-	 * >>   R0=inv(id=0) R1=inv2 R6=ctx(id=0,off=0,imm=0) R7=inv64 R10=fp0,call_-1
-	 * >> 10: (bf) r1 = r6
-	 * >> 11: (07) r1 += 16
-	 * >> 12: (05) goto pc+2
-	 * >> 15: (79) r3 = *(u64 *)(r1 +0)
-	 * >> dereference of modified ctx ptr R1 off=16 disallowed
-	 * > Aha, we at least got a different error message this time.
-	 * > And indeed llvm has done that optimisation, rather than the more obvious
-	 * > 11: r3 = *(u64 *)(r1 +16)
-	 * > because it wants to have lots of reads share a single insn.  You may be able
-	 * > to defeat that optimisation by adding compiler barriers, idk.  Maybe someone
-	 * > with llvm knowledge can figure out how to stop it (ideally, llvm would know
-	 * > when it's generating for bpf backend and not do that).  -O0?  ¯\_(ツ)_/¯
-	 *
-	 * The optimization mostly likes below:
-	 *
-	 *	br1:
-	 * 	...
-	 *	r1 += 16
-	 *	goto merge
-	 *	br2:
-	 *	...
-	 *	r1 += 20
-	 *	goto merge
-	 *	merge:
-	 *	*(u64 *)(r1 + 0)
-	 *
-	 * The compiler tries to merge common loads. There is no easy way to
-	 * stop this compiler optimization without turning off a lot of other
-	 * optimizations. The easiest way is to add barriers:
-	 *
-	 * 	 __asm__ __volatile__("": : :"memory")
-	 *
-	 * 	 after the ctx memory access to prevent their down stream merging.
+	 * Jump to syscall specific augmenter, even if the default one,
+	 * "!raw_syscalls:unaugmented" that will just return 1 to return the
+	 * unagmented tracepoint payload.
 	 */
-	switch (augmented_args.args.syscall_nr) {
-	case SYS_ACCESS:
-	case SYS_OPEN:	 filename_arg = (const void *)args->args[0];
-			__asm__ __volatile__("": : :"memory");
-			 break;
-	case SYS_OPENAT: filename_arg = (const void *)args->args[1];
-			 break;
-	}
+	bpf_tail_call(args, &syscalls_sys_enter, augmented_args->args.syscall_nr);
 
-	if (filename_arg != NULL) {
-		augmented_args.filename.reserved = 0;
-		augmented_args.filename.size = probe_read_str(&augmented_args.filename.value,
-							      sizeof(augmented_args.filename.value),
-							      filename_arg);
-		if (augmented_args.filename.size < sizeof(augmented_args.filename.value)) {
-			len -= sizeof(augmented_args.filename.value) - augmented_args.filename.size;
-			len &= sizeof(augmented_args.filename.value) - 1;
-		}
-	} else {
-		len = sizeof(augmented_args.args);
-	}
-
-	/* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */
-	return perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, &augmented_args, len);
+	// If not found on the PROG_ARRAY syscalls map, then we're filtering it:
+	return 0;
 }
 
 SEC("raw_syscalls:sys_exit")
 int sys_exit(struct syscall_exit_args *args)
 {
 	struct syscall_exit_args exit_args;
-	struct syscall *syscall;
 
 	if (pid_filter__has(&pids_filtered, getpid()))
 		return 0;
 
 	probe_read(&exit_args, sizeof(exit_args), args);
-
-	syscall = bpf_map_lookup_elem(&syscalls, &exit_args.syscall_nr);
-	if (syscall == NULL || !syscall->enabled)
-		return 0;
-
-	return 1;
+	/*
+	 * Jump to syscall specific return augmenter, even if the default one,
+	 * "!raw_syscalls:unaugmented" that will just return 1 to return the
+	 * unagmented tracepoint payload.
+	 */
+	bpf_tail_call(args, &syscalls_sys_exit, exit_args.syscall_nr);
+	/*
+	 * If not found on the PROG_ARRAY syscalls map, then we're filtering it:
+	 */
+	return 0;
 }
 
 license(GPL);