svn commit: trunk/busybox/libbb

[vda at busybox.net]

Author: vda
Date: 2008-04-01 10:43:03 -0700 (Tue, 01 Apr 2008)
New Revision: 21617

Log:
xfunc_die: resurrect (actually, it's "svn add" being forgotten again)



Added:
   trunk/busybox/libbb/xfunc_die.c


Changeset:
Added: trunk/busybox/libbb/xfunc_die.c
===================================================================
--- trunk/busybox/libbb/xfunc_die.c	                        (rev 0)
+++ trunk/busybox/libbb/xfunc_die.c	2008-04-01 17:43:03 UTC (rev 21617)
@@ -0,0 +1,40 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * Utility routines.
+ *
+ * Copyright (C) 2008 by Denys Vlasenko <vda.linux at googlemail.com>
+ *
+ * Licensed under GPLv2, see file LICENSE in this tarball for details.
+ */
+
+/* Keeping it separate allows to NOT suck in stdio for VERY small applets.
+ * Try building busybox with only "true" enabled... */
+
+#include "libbb.h"
+
+int die_sleep;
+#if ENABLE_FEATURE_PREFER_APPLETS || ENABLE_HUSH
+jmp_buf die_jmp;
+#endif
+
+void xfunc_die(void)
+{
+	if (die_sleep) {
+		if ((ENABLE_FEATURE_PREFER_APPLETS || ENABLE_HUSH)
+		 && die_sleep < 0
+		) {
+			/* Special case. We arrive here if NOFORK applet
+			 * calls xfunc, which then decides to die.
+			 * We don't die, but jump instead back to caller.
+			 * NOFORK applets still cannot carelessly call xfuncs:
+			 * p = xmalloc(10);
+			 * q = xmalloc(10); // BUG! if this dies, we leak p!
+			 */
+			/* -2222 means "zero" (longjmp can't pass 0)
+			 * run_nofork_applet() catches -2222. */
+			longjmp(die_jmp, xfunc_error_retval ? xfunc_error_retval : -2222);
+		}
+		sleep(die_sleep);
+	}
+	exit(xfunc_error_retval);
+}