1 /* ***** BEGIN LICENSE BLOCK ***** 2 * Version: NPL 1.1/GPL 2.0/LGPL 2.1 3 * 4 * The contents of this file are subject to the Netscape Public License 5 * Version 1.1 (the "License"); you may not use this file except in 6 * compliance with the License. You may obtain a copy of the License at 7 * http://www.mozilla.org/NPL/ 8 * 9 * Software distributed under the License is distributed on an "AS IS" basis, 10 * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License 11 * for the specific language governing rights and limitations under the 12 * License. 13 * 14 * The Original Code is JavaScript Engine testing utilities. 15 * 16 * The Initial Developer of the Original Code is Netscape Communications Corp. 17 * Portions created by the Initial Developer are Copyright (C) 2002 18 * the Initial Developer. All Rights Reserved. 19 * 20 * Contributor(s): zen-parse (at) gmx.net, pschwartau (at) netscape.com 21 * 22 * Alternatively, the contents of this file may be used under the terms of 23 * either the GNU General Public License Version 2 or later (the "GPL"), or 24 * the GNU Lesser General Public License Version 2.1 or later (the "LGPL"), 25 * in which case the provisions of the GPL or the LGPL are applicable instead 26 * of those above. If you wish to allow use of your version of this file only 27 * under the terms of either the GPL or the LGPL, and not to allow others to 28 * use your version of this file under the terms of the NPL, indicate your 29 * decision by deleting the provisions above and replace them with the notice 30 * and other provisions required by the GPL or the LGPL. If you do not delete 31 * the provisions above, a recipient may use your version of this file under 32 * the terms of any one of the NPL, the GPL or the LGPL. 33 * 34 * ***** END LICENSE BLOCK ***** 35 * 36 * 37 * Date: 16 July 2002 38 * SUMMARY: Testing that Array.sort() doesn't crash on very large arrays 39 * See http://bugzilla.mozilla.org/show_bug.cgi?id=157652 40 * 41 * How large can a JavaScript array be? 42 * ECMA-262 Ed.3 Final, Section 15.4.2.2 : new Array(len) 43 * 44 * This states that |len| must be a a uint32 (unsigned 32-bit integer). 45 * Note the UBound for uint32's is 2^32 -1 = 0xFFFFFFFF = 4,294,967,295. 46 * 47 * Check: 48 * js> var arr = new Array(0xFFFFFFFF) 49 * js> arr.length 50 * 4294967295 51 * 52 * js> var arr = new Array(0x100000000) 53 * RangeError: invalid array length 54 * 55 * 56 * We'll try the largest possible array first, then a couple others. 57 * We're just testing that we don't crash on Array.sort(). 58 * 59 * Try to be good about memory by nulling each array variable after it is 60 * used. This will tell the garbage collector the memory is no longer needed. 61 * 62 * As of 2002-08-13, the JS shell runs out of memory no matter what we do, 63 * when trying to sort such large arrays. 64 * 65 * We only want to test that we don't CRASH on the sort. So it will be OK 66 * if we get the JS "out of memory" error. Note this terminates the test 67 * with exit code 3. Therefore we put 68 * 69 * |expectExitCode(3);| 70 * 71 * The only problem will arise if the JS shell ever DOES have enough memory 72 * to do the sort. Then this test will terminate with the normal exit code 0 73 * and fail. 74 * 75 * Right now, I can't see any other way to do this, because "out of memory" 76 * is not a catchable error: it cannot be trapped with try...catch. 77 * 78 * 79 * FURTHER HEADACHE: Rhino can't seem to handle the largest array: it hangs. 80 * So we skip this case in Rhino. Here is correspondence with Igor Bukanov. 81 * He explains that Rhino isn't actually hanging; it's doing the huge sort: 82 * 83 * Philip Schwartau wrote: 84 * 85 * > Hi, 86 * > 87 * > I'm getting a graceful OOM message on trying to sort certain large 88 * > arrays. But if the array is too big, Rhino simply hangs. Note that ECMA 89 * > allows array lengths to be anything less than Math.pow(2,32), so the 90 * > arrays I'm sorting are legal. 91 * > 92 * > Note below, I'm getting an instantaneous OOM error on arr.sort() for LEN 93 * > = Math.pow(2, 30). So shouldn't I also get one for every LEN between 94 * > that and Math.pow(2, 32)? For some reason, I start to hang with 100% CPU 95 * > as LEN hits, say, Math.pow(2, 31) and higher. SpiderMonkey gives OOM 96 * > messages for all of these. Should I file a bug on this? 97 * 98 * Igor Bukanov wrote: 99 * 100 * This is due to different sorting algorithm Rhino uses when sorting 101 * arrays with length > Integer.MAX_VALUE. If length can fit Java int, 102 * Rhino first copies internal spare array to a temporary buffer, and then 103 * sorts it, otherwise it sorts array directly. In case of very spare 104 * arrays, that Array(big_number) generates, it is rather inefficient and 105 * generates OutOfMemory if length fits int. It may be worth in your case 106 * to optimize sorting to take into account array spareness, but then it 107 * would be a good idea to file a bug about ineficient sorting of spare 108 * arrays both in case of Rhino and SpiderMonkey as SM always uses a 109 * temporary buffer. 110 * 111 */ 112 //----------------------------------------------------------------------------- 113 var bug = 157652; 114 var summary = "Testing that Array.sort() doesn't crash on very large arrays"; 115 116 printBugNumber(bug); 117 printStatus(summary); 118 119 // JSC doesn't run out of memory, so we don't expect an abnormal exit code 120 //expectExitCode(3); 121 var IN_RHINO = inRhino(); 122 123 if (!IN_RHINO) 124 { 125 var a1=Array(0xFFFFFFFF); 126 a1.sort(); 127 a1 = null; 128 } 129 130 var a2 = Array(0x40000000); 131 a2.sort(); 132 a2=null; 133 134 var a3=Array(0x10000000/4); 135 a3.sort(); 136 a3=null; 137
