52769 2011-01-19 17:02:33 -0800 Crash in DOMWindow::setLocation() due to null m_frame 2011-01-20 16:37:01 -0800 1 1 1 Unclassified WebKit WebCore Misc. 528+ (Nightly build) All All RESOLVED FIXED P2 Normal --- 1 japhet japhet abarth ap commit-queue jschuh oldest_to_newest 336952 0 japhet 2011-01-19 17:02:33 -0800 Original report at http://code.google.com/p/chromium/issues/detail?id=68077 Opening a new window, setting a timeout to close it, and setting a later timeout to set window.location will cause chromium to crash. DOMWindow::setLocation() null checks the Frames for the activeWindow and firstWindow parameters it receives, but it uses its own m_frame without null-checking it. For reasons that aren't clear to me, JSC never calls JSDOMWindow::setLocation() in this example, so this crash doesn't show up in Safari. I think a fix for this still goes in DOMWindow, however, because if DOMWindow is going to use its own m_frame and it isn't guaranteed to be non-null, we should probably be checking it. If others disagree, we can kick this fix into the v8 bindings instead. 336956 1 79525 japhet 2011-01-19 17:11:21 -0800 Created attachment 79525 patch + test 336980 2 79525 abarth 2011-01-19 17:35:49 -0800 Comment on attachment 79525 patch + test View in context: https://bugs.webkit.org/attachment.cgi?id=79525&action=review I don't love the test. Too many timeouts. This test takes 1 whole second to run! There's got to be a better way. For example, you can have the newly opened window postMessage back to the original page to know that it's time to close it. > LayoutTests/fast/dom/Window/Location/set-location-after-close.html:14 > +setTimeout('w.location = \'\'; if (window.layoutTestController) layoutTestController.notifyDone()', 1000); \'\' -> "" 337124 3 ap 2011-01-19 21:58:18 -0800 This test passes in Firefox, right? And how does window.location come back - is the setter really ignored? 337401 4 japhet 2011-01-20 09:42:50 -0800 (In reply to comment #2) > (From update of attachment 79525 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=79525&action=review > > I don't love the test. Too many timeouts. This test takes 1 whole second to run! There's got to be a better way. For example, you can have the newly opened window postMessage back to the original page to know that it's time to close it. > > > LayoutTests/fast/dom/Window/Location/set-location-after-close.html:14 > > +setTimeout('w.location = \'\'; if (window.layoutTestController) layoutTestController.notifyDone()', 1000); > > \'\' -> "" Yeah, I just modified the reduction provided in the Chromium bug to make it a layout test. I'll see if this can be reproed more sanely. (In reply to comment #3) > This test passes in Firefox, right? And how does window.location come back - is the setter really ignored? I'll double-check other browsers on whatever test I end up with. As for the setter being ignored, I set a breakpoint in JSDOMWindow::setLocation() and never saw it break there, so it really does seem that at least the custom binding isn't getting called. I don't know enough about the guts of JSC to understand how this could happen. 337511 5 79634 japhet 2011-01-20 12:33:07 -0800 Created attachment 79634 Test using postMessage instead of setTimeout Thanks for the postMessage suggestion, Adam. :) This test behaves the same way as the old test on existing Chromium builds, and works correctly in FF as well. 337522 6 ap 2011-01-20 12:43:38 -0800 Did you check what Firefox returns as window.location value after setting? 337535 7 japhet 2011-01-20 13:01:21 -0800 (In reply to comment #6) > Did you check what Firefox returns as window.location value after setting? It appears FF throws an error trying to set window.location. window.location is inaccessible (for getting or setting) as soon as the window.close() call returns. 337600 8 ap 2011-01-20 14:19:11 -0800 Between the different behavior in Firefox and unexplained behavior in Safari, it's not entirely clear why the patch implements the right thing. 337659 9 abarth 2011-01-20 15:39:00 -0800 (In reply to comment #8) > Between the different behavior in Firefox and unexplained behavior in Safari, it's not entirely clear why the patch implements the right thing. Generally speaking, the behavior for Frame-less objects is very inconsistent between browsers. This patch follows our usual approach. At some point in the future, we might want to make an effort to converge browser behavior in this area. We tried discussing this issue a year or so ago, and folks had particular reasons for liking their individual approaches and not overly much incentive to align. In any case, that's a topic for another day. 337660 10 79634 abarth 2011-01-20 15:40:01 -0800 Comment on attachment 79634 Test using postMessage instead of setTimeout View in context: https://bugs.webkit.org/attachment.cgi?id=79634&action=review > LayoutTests/fast/dom/Window/Location/resources/set-location-after-close-new-window.html:2 > +<body onunload="opener.postMessage('closed', '*');"> I wonder if this is racy. It's probably ok. 337665 11 ap 2011-01-20 15:43:55 -0800 > We tried discussing this issue a year or so ago, Do you at least have a link to that discussion? This bug is not in a very good form now, I'm not sure why you had to hurry with r+/cq+. 337677 12 abarth 2011-01-20 15:55:32 -0800 (In reply to comment #11) > > We tried discussing this issue a year or so ago, > > Do you at least have a link to that discussion? It was either in the TC39 working group or the HTML working group. I don't recall off-hand. Maciej didn't want to implement what IE is doing because he was worried about adding a branch to every JS function call. Someone else didn't want to implement what we do because it lets you execute JavaScript in a strange Frame-less state that they were worried would lead to security bugs. > This bug is not in a very good form now, I'm not sure why you had to hurry with r+/cq+. This patch is adding a null check like hundreds of other null checks we have in the code. Certainly not crashing is better behavior than crashing. We can file a follow-up bug about harmonizing our Frame-less execution behavior with other browsers if you like, but that's going to take a lot longer than we want to wait before fixing this crash. 337684 13 ap 2011-01-20 16:05:11 -0800 What I'm worried about is that we didn't capture information about what happens in IE, or even in Safari in this bug. The description says: > For reasons that aren't clear to me, JSC never calls JSDOMWindow::setLocation() in this example Maybe there is protection against this that's just simply broken in v8? It's not clear that the fix should be so local. 337686 14 79634 commit-queue 2011-01-20 16:05:43 -0800 Comment on attachment 79634 Test using postMessage instead of setTimeout Clearing flags on attachment: 79634 Committed r76303: <http://trac.webkit.org/changeset/76303> 337687 15 commit-queue 2011-01-20 16:05:48 -0800 All reviewed patches have been landed. Closing bug. 337703 16 abarth 2011-01-20 16:37:01 -0800 > Maybe there is protection against this that's just simply broken in v8? It's not clear that the fix should be so local. Generally speaking, every method on DOMWindow that can be called by JavaScript needs to be prepared to be called with a null frame. We commonly forget because the case doesn't occur very often, but that's one of the things I was trying to convey in this diagram: http://webkit.org/coding/major-objects.html 79525 2011-01-19 17:11:21 -0800 2011-01-20 12:33:07 -0800 patch + test dw.txt text/plain 3313 japhet SW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24t YWZ0ZXItY2xvc2UtZXhwZWN0ZWQudHh0Cj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91dFRlc3RzL2Zhc3Qv ZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UtZXhwZWN0ZWQudHh0 CShyZXZpc2lvbiAwKQorKysgTGF5b3V0VGVzdHMvZmFzdC9kb20vV2luZG93L0xvY2F0aW9uL3Nl dC1sb2NhdGlvbi1hZnRlci1jbG9zZS1leHBlY3RlZC50eHQJKHJldmlzaW9uIDApCkBAIC0wLDAg KzEgQEAKK1RoaXMgdGVzdHMgb3BlbnMgYSBuZXcgd2luZG93LCBzZXRzIGEgdGltZW91dCB0byBj bG9zZSBpdCwgdGhlbiBzZXRzIGEgdGltZW91dCB0byBuYXZpZ2F0ZSBpdCB2aWEgd2luZG93Lmxv Y2F0aW9uLiBXZSBwYXNzIGlmIHdlIGRvbid0IGNyYXNoIHdoZW4gc2V0dGluZyB0aGUgbG9jYXRp b24gb2YgdGhlIGFscmVhZHkgY2xvc2VkIHdpbmRvdy4KSW5kZXg6IExheW91dFRlc3RzL2Zhc3Qv ZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UuaHRtbAo9PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09Ci0tLSBMYXlvdXRUZXN0cy9mYXN0L2RvbS9XaW5kb3cvTG9jYXRpb24vc2V0LWxvY2F0aW9u LWFmdGVyLWNsb3NlLmh0bWwJKHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9mYXN0L2RvbS9X aW5kb3cvTG9jYXRpb24vc2V0LWxvY2F0aW9uLWFmdGVyLWNsb3NlLmh0bWwJKHJldmlzaW9uIDAp CkBAIC0wLDAgKzEsMTcgQEAKKzxodG1sPgorPGJvZHk+CitUaGlzIHRlc3RzIG9wZW5zIGEgbmV3 IHdpbmRvdywgc2V0cyBhIHRpbWVvdXQgdG8gY2xvc2UgaXQsIHRoZW4gc2V0cyBhIHRpbWVvdXQg dG8gbmF2aWdhdGUgaXQgdmlhIHdpbmRvdy5sb2NhdGlvbi4KK1dlIHBhc3MgaWYgd2UgZG9uJ3Qg Y3Jhc2ggd2hlbiBzZXR0aW5nIHRoZSBsb2NhdGlvbiBvZiB0aGUgYWxyZWFkeSBjbG9zZWQgd2lu ZG93LgorPHNjcmlwdD4KK2lmICh3aW5kb3cubGF5b3V0VGVzdENvbnRyb2xsZXIpIHsKKyAgIGxh eW91dFRlc3RDb250cm9sbGVyLmR1bXBBc1RleHQoKTsKKyAgIGxheW91dFRlc3RDb250cm9sbGVy LnNldENhbk9wZW5XaW5kb3dzKCk7CisgICBsYXlvdXRUZXN0Q29udHJvbGxlci53YWl0VW50aWxE b25lKCk7Cit9CisKK3cgPSBvcGVuKCdodHRwOi8vJyk7CitzZXRUaW1lb3V0KCd3LmNsb3NlKCkn LCA1MDApOworc2V0VGltZW91dCgndy5sb2NhdGlvbiA9IFwnXCc7IGlmICh3aW5kb3cubGF5b3V0 VGVzdENvbnRyb2xsZXIpIGxheW91dFRlc3RDb250cm9sbGVyLm5vdGlmeURvbmUoKScsIDEwMDAp OworPC9zY3JpcHQ+Cis8L2JvZHk+Cis8L2h0bWw+CkluZGV4OiBMYXlvdXRUZXN0cy9DaGFuZ2VM b2cKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvQ2hhbmdlTG9nCShyZXZpc2lvbiA3NjE4MikK KysrIExheW91dFRlc3RzL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDEyIEBA CisyMDExLTAxLTE5ICBOYXRlIENoYXBpbiAgPGphcGhldEBjaHJvbWl1bS5vcmc+CisKKyAgICAg ICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAgICAgVGVzdCBmb3IgaHR0cHM6 Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTUyNzY5LgorCisgICAgICAgICogZmFz dC9kb20vV2luZG93L0xvY2F0aW9uL3NldC1sb2NhdGlvbi1hZnRlci1jbG9zZS1leHBlY3RlZC50 eHQ6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRp b24tYWZ0ZXItY2xvc2UuaHRtbDogQWRkZWQuCisKIDIwMTEtMDEtMTkgIE9saXZlciBIdW50ICA8 b2xpdmVyQGFwcGxlLmNvbT4KIAogICAgICAgICBSZXZpZXdlZCBieSBHYXZpbiBCYXJyYWNsb3Vn aC4KSW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2Uv V2ViQ29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDc2MTgyKQorKysgU291cmNlL1dlYkNvcmUvQ2hh bmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEsMTcgQEAKKzIwMTEtMDEtMTkgIE5hdGUg Q2hhcGluICA8amFwaGV0QGNocm9taXVtLm9yZz4KKworICAgICAgICBSZXZpZXdlZCBieSBOT0JP RFkgKE9PUFMhKS4KKworICAgICAgICBOdWxsLWNoZWNrIG1fZnJhbWUgaW4gRE9NV2luZG93Ojpz ZXRMb2NhdGlvbigpLCBzaW5jZSBpdCdzCisgICAgICAgIHBvc3NpYmxlIHRvIHJlYWNoIHRoaXMg cG9pbnQgd2l0aG91dCBpdCBoYXZpbmcgYmVlbiBjaGVja2VkCisgICAgICAgIGFscmVhZHkuCisg ICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD01Mjc2OQorCisg ICAgICAgIFRlc3Q6IGZhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXIt Y2xvc2UuaHRtbAorCisgICAgICAgICogcGFnZS9ET01XaW5kb3cuY3BwOgorICAgICAgICAoV2Vi Q29yZTo6RE9NV2luZG93OjpzZXRMb2NhdGlvbik6CisKIDIwMTEtMDEtMTIgIE1hcnRpbiBSb2Jp bnNvbiAgPG1yb2JpbnNvbkBpZ2FsaWEuY29tPgogCiAgICAgICAgIFJldmlld2VkIGJ5IEd1c3Rh dm8gTm9yb25oYSBTaWx2YS4KSW5kZXg6IFNvdXJjZS9XZWJDb3JlL3BhZ2UvRE9NV2luZG93LmNw cAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2ViQ29yZS9wYWdlL0RPTVdpbmRvdy5jcHAJKHJldmlz aW9uIDc2MTM3KQorKysgU291cmNlL1dlYkNvcmUvcGFnZS9ET01XaW5kb3cuY3BwCSh3b3JraW5n IGNvcHkpCkBAIC0xNTk4LDYgKzE1OTgsOSBAQAogCiB2b2lkIERPTVdpbmRvdzo6c2V0TG9jYXRp b24oY29uc3QgU3RyaW5nJiB1cmxTdHJpbmcsIERPTVdpbmRvdyogYWN0aXZlV2luZG93LCBET01X aW5kb3cqIGZpcnN0V2luZG93LCBTZXRMb2NhdGlvbkxvY2tpbmcgbG9ja2luZykKIHsKKyAgICBp ZiAoIW1fZnJhbWUpCisgICAgICAgIHJldHVybjsKKwogICAgIEZyYW1lKiBhY3RpdmVGcmFtZSA9 IGFjdGl2ZVdpbmRvdy0+ZnJhbWUoKTsKICAgICBpZiAoIWFjdGl2ZUZyYW1lKQogICAgICAgICBy ZXR1cm47Cg== 79634 2011-01-20 12:33:07 -0800 2011-01-20 16:05:43 -0800 Test using postMessage instead of setTimeout dw2.txt text/plain 4255 japhet SW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24t YWZ0ZXItY2xvc2UtZXhwZWN0ZWQudHh0Cj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91dFRlc3RzL2Zhc3Qv ZG9tL1dpbmRvdy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UtZXhwZWN0ZWQudHh0 CShyZXZpc2lvbiAwKQorKysgTGF5b3V0VGVzdHMvZmFzdC9kb20vV2luZG93L0xvY2F0aW9uL3Nl dC1sb2NhdGlvbi1hZnRlci1jbG9zZS1leHBlY3RlZC50eHQJKHJldmlzaW9uIDApCkBAIC0wLDAg KzEsMiBAQAorbWFpbiBmcmFtZSAtIGhhcyAxIG9udW5sb2FkIGhhbmRsZXIocykKK1RoaXMgdGVz dHMgb3BlbnMgYSBuZXcgd2luZG93LCBzZXRzIGEgdGltZW91dCB0byBjbG9zZSBpdCwgdGhlbiBz ZXRzIGEgdGltZW91dCB0byBuYXZpZ2F0ZSBpdCB2aWEgd2luZG93LmxvY2F0aW9uLiBXZSBwYXNz IGlmIHdlIGRvbid0IGNyYXNoIHdoZW4gc2V0dGluZyB0aGUgbG9jYXRpb24gb2YgdGhlIGFscmVh ZHkgY2xvc2VkIHdpbmRvdy4KSW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL1dpbmRvdy9Mb2Nh dGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UuaHRtbAo9PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRU ZXN0cy9mYXN0L2RvbS9XaW5kb3cvTG9jYXRpb24vc2V0LWxvY2F0aW9uLWFmdGVyLWNsb3NlLmh0 bWwJKHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9mYXN0L2RvbS9XaW5kb3cvTG9jYXRpb24v c2V0LWxvY2F0aW9uLWFmdGVyLWNsb3NlLmh0bWwJKHJldmlzaW9uIDApCkBAIC0wLDAgKzEsMjcg QEAKKzxodG1sPgorPGJvZHk+CitUaGlzIHRlc3RzIG9wZW5zIGEgbmV3IHdpbmRvdywgc2V0cyBh IHRpbWVvdXQgdG8gY2xvc2UgaXQsIHRoZW4gc2V0cyBhIHRpbWVvdXQgdG8gbmF2aWdhdGUgaXQg dmlhIHdpbmRvdy5sb2NhdGlvbi4KK1dlIHBhc3MgaWYgd2UgZG9uJ3QgY3Jhc2ggd2hlbiBzZXR0 aW5nIHRoZSBsb2NhdGlvbiBvZiB0aGUgYWxyZWFkeSBjbG9zZWQgd2luZG93LgorPHNjcmlwdD4K K2lmICh3aW5kb3cubGF5b3V0VGVzdENvbnRyb2xsZXIpIHsKKyAgICBsYXlvdXRUZXN0Q29udHJv bGxlci5kdW1wQXNUZXh0KCk7CisgICAgbGF5b3V0VGVzdENvbnRyb2xsZXIuc2V0Q2FuT3Blbldp bmRvd3MoKTsKKyAgICBsYXlvdXRUZXN0Q29udHJvbGxlci53YWl0VW50aWxEb25lKCk7Cit9CisK K3ZhciB3ID0gd2luZG93Lm9wZW4oJ3Jlc291cmNlcy9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2Ut bmV3LXdpbmRvdy5odG1sJyk7CisKK3dpbmRvdy5hZGRFdmVudExpc3RlbmVyKCJtZXNzYWdlIiwg cHJvY2Vzc01lc3NhZ2UsIGZhbHNlKTsKKworZnVuY3Rpb24gcHJvY2Vzc01lc3NhZ2UoZXZlbnQp IHsKKyAgICBpZiAoZXZlbnQuZGF0YSA9PSAib3BlbmVkIikKKyAgICAgICAgdy5jbG9zZSgpOwor ICAgIGVsc2UgaWYgKGV2ZW50LmRhdGEgPT0gImNsb3NlZCIpIHsKKyAgICAgICAgdy5sb2NhdGlv biA9ICcnOworICAgICAgICBpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKQorICAgICAg ICAgICAgbGF5b3V0VGVzdENvbnRyb2xsZXIubm90aWZ5RG9uZSgpOworICAgIH0KK30KKzwvc2Ny aXB0PgorPC9ib2R5PgorPC9odG1sPgpJbmRleDogTGF5b3V0VGVzdHMvZmFzdC9kb20vV2luZG93 L0xvY2F0aW9uL3Jlc291cmNlcy9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UtbmV3LXdpbmRvdy5o dG1sCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT0KLS0tIExheW91dFRlc3RzL2Zhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9y ZXNvdXJjZXMvc2V0LWxvY2F0aW9uLWFmdGVyLWNsb3NlLW5ldy13aW5kb3cuaHRtbAkocmV2aXNp b24gMCkKKysrIExheW91dFRlc3RzL2Zhc3QvZG9tL1dpbmRvdy9Mb2NhdGlvbi9yZXNvdXJjZXMv c2V0LWxvY2F0aW9uLWFmdGVyLWNsb3NlLW5ldy13aW5kb3cuaHRtbAkocmV2aXNpb24gMCkKQEAg LTAsMCArMSw3IEBACis8aHRtbD4KKzxib2R5IG9udW5sb2FkPSJvcGVuZXIucG9zdE1lc3NhZ2Uo J2Nsb3NlZCcsICcqJyk7Ij4KKzxzY3JpcHQ+CitvcGVuZXIucG9zdE1lc3NhZ2UoIm9wZW5lZCIs ICIqIik7Cis8L3NjcmlwdD4KKzwvYm9keT4KKzwvaHRtbD4KSW5kZXg6IExheW91dFRlc3RzL0No YW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9DaGFuZ2VMb2cJKHJldmlzaW9uIDc2 MjY4KQorKysgTGF5b3V0VGVzdHMvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEs MTMgQEAKKzIwMTEtMDEtMjAgIE5hdGUgQ2hhcGluICA8amFwaGV0QGNocm9taXVtLm9yZz4KKwor ICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBUZXN0IGZvciBo dHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9NTI3NjkuCisKKyAgICAgICAg KiBmYXN0L2RvbS9XaW5kb3cvTG9jYXRpb24vcmVzb3VyY2VzL3NldC1sb2NhdGlvbi1hZnRlci1j bG9zZS1uZXctd2luZG93Lmh0bWw6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL1dpbmRvdy9M b2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UtZXhwZWN0ZWQudHh0OiBBZGRlZC4KKyAg ICAgICAgKiBmYXN0L2RvbS9XaW5kb3cvTG9jYXRpb24vc2V0LWxvY2F0aW9uLWFmdGVyLWNsb3Nl Lmh0bWw6IEFkZGVkLgorCiAyMDExLTAxLTIwICBQaGlsaXBwZSBOb3JtYW5kICA8cG5vcm1hbmRA aWdhbGlhLmNvbT4KIAogICAgICAgICBVbnJldmlld2VkLCByZW1vdmUgR1RLIHJlc3VsdCBmb3Ig d2luZG93LW9uZXJyb3IyIGFzIGl0J3MgdGhlIHNhbWUKSW5kZXg6IFNvdXJjZS9XZWJDb3JlL0No YW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cJKHJldmlzaW9u IDc2MTgyKQorKysgU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0x LDMgKzEsMTcgQEAKKzIwMTEtMDEtMTkgIE5hdGUgQ2hhcGluICA8amFwaGV0QGNocm9taXVtLm9y Zz4KKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBOdWxs LWNoZWNrIG1fZnJhbWUgaW4gRE9NV2luZG93OjpzZXRMb2NhdGlvbigpLCBzaW5jZSBpdCdzCisg ICAgICAgIHBvc3NpYmxlIHRvIHJlYWNoIHRoaXMgcG9pbnQgd2l0aG91dCBpdCBoYXZpbmcgYmVl biBjaGVja2VkCisgICAgICAgIGFscmVhZHkuCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQu b3JnL3Nob3dfYnVnLmNnaT9pZD01Mjc2OQorCisgICAgICAgIFRlc3Q6IGZhc3QvZG9tL1dpbmRv dy9Mb2NhdGlvbi9zZXQtbG9jYXRpb24tYWZ0ZXItY2xvc2UuaHRtbAorCisgICAgICAgICogcGFn ZS9ET01XaW5kb3cuY3BwOgorICAgICAgICAoV2ViQ29yZTo6RE9NV2luZG93OjpzZXRMb2NhdGlv bik6CisKIDIwMTEtMDEtMTIgIE1hcnRpbiBSb2JpbnNvbiAgPG1yb2JpbnNvbkBpZ2FsaWEuY29t PgogCiAgICAgICAgIFJldmlld2VkIGJ5IEd1c3Rhdm8gTm9yb25oYSBTaWx2YS4KSW5kZXg6IFNv dXJjZS9XZWJDb3JlL3BhZ2UvRE9NV2luZG93LmNwcAo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9wYWdlL0RPTVdpbmRvdy5jcHAJKHJldmlzaW9uIDc2MTM3KQorKysgU291cmNlL1dlYkNv cmUvcGFnZS9ET01XaW5kb3cuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC0xNTk4LDYgKzE1OTgsOSBA QAogCiB2b2lkIERPTVdpbmRvdzo6c2V0TG9jYXRpb24oY29uc3QgU3RyaW5nJiB1cmxTdHJpbmcs IERPTVdpbmRvdyogYWN0aXZlV2luZG93LCBET01XaW5kb3cqIGZpcnN0V2luZG93LCBTZXRMb2Nh dGlvbkxvY2tpbmcgbG9ja2luZykKIHsKKyAgICBpZiAoIW1fZnJhbWUpCisgICAgICAgIHJldHVy bjsKKwogICAgIEZyYW1lKiBhY3RpdmVGcmFtZSA9IGFjdGl2ZVdpbmRvdy0+ZnJhbWUoKTsKICAg ICBpZiAoIWFjdGl2ZUZyYW1lKQogICAgICAgICByZXR1cm47Cg==