all: add syz_clone() and syz_clone3() pseudo calls

As was pointed out in #2921, the current approach of limiting the number
of pids per process does not work on all Linux-based kernels.

We could just treat fork, clone and clone3 in a special way (e.g. exit
on a zero return). However, in that case we also need to sanitize the
arguments for clone and clone3 - if CLONE_VM is passed and stack is 0,
the forked child processes (threads) will become nearly unindentifiable
and will corrupt syz-executor's memory. While we could sanitize clone's
arguments, we cannot do so for clone3 - nothing can guarantee that they
will not be changed concurrently.

Instead of calling those syscalls directly, introduce a special pseudo
syscall syz_clone3. It copies and sanitizes the arguments and then
executes clone3 (or fork, if we're on an older kernel) in such a way so
as to prevent fork bombs from happening.

Also introduce syz_clone() to still be able to fuzz it on older systems.

1 files changed, 59 insertions, 0 deletions

diff --git a/executor/common_linux.h b/executor/common_linux.h
index 479c229dc..520c13506 100644
--- a/executor/common_linux.h
+++ b/executor/common_linux.h
@@ -5215,3 +5215,62 @@ static long syz_80211_join_ibss(volatile long a0, volatile long a1, volatile lon
 }
 
 #endif
+
+#if SYZ_EXECUTOR || __NR_syz_clone || __NR_syz_clone3
+#if SYZ_EXECUTOR
+// The slowdown multiplier is already taken into account.
+#define USLEEP_FORKED_CHILD (3 * syscall_timeout_ms * 1000)
+#else
+#define USLEEP_FORKED_CHILD (3 * /*{{{BASE_CALL_TIMEOUT_MS}}}*/ *1000)
+#endif
+
+static long handle_clone_ret(long ret)
+{
+	if (ret != 0)
+		return ret;
+	// Exit if we're in the child process - not all kernels provide the proper means
+	// to prevent fork-bombs.
+	// But first sleep for some time. This will hopefully foster IPC fuzzing.
+	usleep(USLEEP_FORKED_CHILD);
+	// Note that exit_group is a bad choice here because if we created just a thread, then
+	// the whole process will be killed. A plain exit will work fine in any case.
+	syscall(__NR_exit, 0);
+	while (1) {
+	}
+}
+#endif
+
+#if SYZ_EXECUTOR || __NR_syz_clone
+// syz_clone is mostly needed on kernels which do not suport clone3.
+static long syz_clone(volatile long flags, volatile long stack, volatile long stack_len,
+		      volatile long ptid, volatile long ctid, volatile long tls)
+{
+	// ABI requires 16-byte stack alignment.
+	long sp = (stack + stack_len) & ~15;
+	// Clear the CLONE_VM flag. Otherwise it'll very likely corrupt syz-executor.
+	long ret = (long)syscall(__NR_clone, flags & ~CLONE_VM, sp, ptid, ctid, tls);
+	return handle_clone_ret(ret);
+}
+#endif
+
+#if SYZ_EXECUTOR || __NR_syz_clone3
+#include <linux/sched.h>
+#include <sched.h>
+
+#define MAX_CLONE_ARGS_BYTES 256
+static long syz_clone3(volatile long a0, volatile long a1)
+{
+	unsigned long copy_size = a1;
+	if (copy_size < sizeof(uint64) || copy_size > MAX_CLONE_ARGS_BYTES)
+		return -1;
+	// The structure may have different sizes on different kernel versions, so copy it as raw bytes.
+	char clone_args[MAX_CLONE_ARGS_BYTES];
+	memcpy(&clone_args, (void*)a0, copy_size);
+
+	// As in syz_clone, clear the CLONE_VM flag. Flags are in the first 8-byte integer field.
+	uint64* flags = (uint64*)&clone_args;
+	*flags &= ~CLONE_VM;
+	return handle_clone_ret((long)syscall(__NR_clone3, &clone_args, copy_size));
+}
+
+#endif