powerpc/kernel: Change the do_syscall_trace_enter() API

The API for calling do_syscall_trace_enter() is currently sensible
enough, it just returns the (modified) syscall number.

However once we enable seccomp filter it will get more complicated. When
seccomp filter runs, the seccomp kernel code (via SECCOMP_RET_ERRNO), or
a ptracer (via SECCOMP_RET_TRACE), may reject the syscall and *may* or may
*not* set a return value in r3.

That means the assembler that calls do_syscall_trace_enter() can not
blindly return ENOSYS, it needs to only return ENOSYS if a return value
has not already been set.

There is no way to implement that logic with the current API. So change
the do_syscall_trace_enter() API to make it deal with the return code
juggling, and the assembler can then just return whatever return code it
is given.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Kees Cook <keescook@chromium.org>

1 files changed, 37 insertions, 11 deletions

diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index f21897b42057..7484221bb3f8 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -1762,26 +1762,42 @@ long arch_ptrace(struct task_struct *child, long request,
 	return ret;
 }
 
-/*
- * We must return the syscall number to actually look up in the table.
- * This can be -1L to skip running any syscall at all.
+/**
+ * do_syscall_trace_enter() - Do syscall tracing on kernel entry.
+ * @regs: the pt_regs of the task to trace (current)
+ *
+ * Performs various types of tracing on syscall entry. This includes seccomp,
+ * ptrace, syscall tracepoints and audit.
+ *
+ * The pt_regs are potentially visible to userspace via ptrace, so their
+ * contents is ABI.
+ *
+ * One or more of the tracers may modify the contents of pt_regs, in particular
+ * to modify arguments or even the syscall number itself.
+ *
+ * It's also possible that a tracer can choose to reject the system call. In
+ * that case this function will return an illegal syscall number, and will put
+ * an appropriate return value in regs->r3.
+ *
+ * Return: the (possibly changed) syscall number.
  */
 long do_syscall_trace_enter(struct pt_regs *regs)
 {
-	long ret = 0;
+	bool abort = false;
 
 	user_exit();
 
 	secure_computing_strict(regs->gpr[0]);
 
-	if (test_thread_flag(TIF_SYSCALL_TRACE) &&
-	    tracehook_report_syscall_entry(regs))
+	if (test_thread_flag(TIF_SYSCALL_TRACE)) {
 		/*
-		 * Tracing decided this syscall should not happen.
-		 * We'll return a bogus call number to get an ENOSYS
-		 * error, but leave the original number in regs->gpr[0].
+		 * The tracer may decide to abort the syscall, if so tracehook
+		 * will return !0. Note that the tracer may also just change
+		 * regs->gpr[0] to an invalid syscall number, that is handled
+		 * below on the exit path.
 		 */
-		ret = -1L;
+		abort = tracehook_report_syscall_entry(regs) != 0;
+	}
 
 	if (unlikely(test_thread_flag(TIF_SYSCALL_TRACEPOINT)))
 		trace_sys_enter(regs, regs->gpr[0]);
@@ -1798,7 +1814,17 @@ long do_syscall_trace_enter(struct pt_regs *regs)
 				    regs->gpr[5] & 0xffffffff,
 				    regs->gpr[6] & 0xffffffff);
 
-	return ret ?: regs->gpr[0];
+	if (abort || regs->gpr[0] >= NR_syscalls) {
+		/*
+		 * If we are aborting explicitly, or if the syscall number is
+		 * now invalid, set the return value to -ENOSYS.
+		 */
+		regs->gpr[3] = -ENOSYS;
+		return -1;
+	}
+
+	/* Return the possibly modified but valid syscall number */
+	return regs->gpr[0];
 }
 
 void do_syscall_trace_leave(struct pt_regs *regs)