89673 2012-06-21 10:09:59 -0700 showModalDialog fix creates risk of never returning from RunLoop::performWork, potentially blocking other event sources 2012-06-21 11:54:43 -0700 1 1 1 Unclassified WebKit WebCore Misc. 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 1 beidson beidson darin jonlee levin+threading webkit.review.bot oldest_to_newest 654644 0 beidson 2012-06-21 10:09:59 -0700 showModalDialog fix creates risk of never returning from RunLoop::performWork, potentially starving other event sources The fix in question was for bug 89590 and can be seen at http://trac.webkit.org/changeset/120879 The scenario is as follows: -RunLoop::performWork() is invoked and there is 1 function on the queue -Handling that one function results in 1 more function being placed on the queue -Handling that one results in yet another being placed on the queue -Ad infinitum Therefore that invocation of performWork() never returns and all other event sources are blocked. Before r120879 this could never happen because we only ever did the amount of work that was in the queue when performWork() was invoked. Any additional work that was enqueued would have to wait until the next time performWork() was invoked which would be after other event sources had a chance to spin. We can approximate that same behavior with a simple count. Patch is coming up. In radar as <rdar://problem/11718988> 654675 1 148832 beidson 2012-06-21 10:36:38 -0700 Created attachment 148832 Patch v1 - Proposed fix This patch implements a rule where we only perform the number of functions in the queue at the time performWork was called. Additional functions will wait until the next time performWork is called. This very closely approximates our behavior before http://trac.webkit.org/changeset/120879. I originally re-wrote the loop to be a for loop but that required acquiring the lock before the for loop *just* to get the function count. By far the common case in RunLoop::performWork is that it is handling 1 function. So the for-loop solution would've required acquiring and releasing the lock twice just to perform one function. Having this simple branch inside the loop seems much more efficient than having that addition lock acquisition in the common case of 1 function, but I'm open to input on this. 654677 2 webkit.review.bot 2012-06-21 10:39:52 -0700 Attachment 148832 did not pass style-queue: Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Source/WebCore/ChangeLog', u'Source/WebCor..." exit_code: 1 Source/WebCore/platform/RunLoop.cpp:82: Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons. [readability/comparison_to_zero] [5] Total errors found: 1 in 2 files If any of these errors are false positives, please file a bug against check-webkit-style. 654678 3 beidson 2012-06-21 10:40:08 -0700 (In reply to comment #1) > Created an attachment (id=148832) [details] > By far the common case in RunLoop::performWork is that it is handling 1 function. So the for-loop solution would've required acquiring and releasing the lock twice just to perform one function. > > Having this simple branch inside the loop seems much more efficient than having that addition lock acquisition in the common case of 1 function, but I'm open to input on this. I thought of a better way. New patch coming. 654687 4 148837 beidson 2012-06-21 10:48:04 -0700 Created attachment 148837 Patch v2 - Avoid taking the lock twice for the common case *and* avoid the branch comparison in Patch v1 654698 5 148837 darin 2012-06-21 11:01:07 -0700 Comment on attachment 148837 Patch v2 - Avoid taking the lock twice for the common case *and* avoid the branch comparison in Patch v1 View in context: https://bugs.webkit.org/attachment.cgi?id=148837&action=review I know Anders will review too. Looks fine to me, thought. > Source/WebCore/platform/RunLoop.cpp:72 > + // we guarantee that we will occassionally return from the run loop and other event sources will have Typo: occasionally has only one s 654715 6 148837 jonlee 2012-06-21 11:14:38 -0700 Comment on attachment 148837 Patch v2 - Avoid taking the lock twice for the common case *and* avoid the branch comparison in Patch v1 View in context: https://bugs.webkit.org/attachment.cgi?id=148837&action=review > Source/WebCore/platform/RunLoop.cpp:90 > + for (size_t functionsHandled = 1; functionsHandled < functionsToHandle; functionsHandled++) { pre, not post, plz 2 fix 654723 7 beidson 2012-06-21 11:21:17 -0700 (In reply to comment #6) > (From update of attachment 148837 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=148837&action=review > > > Source/WebCore/platform/RunLoop.cpp:90 > > + for (size_t functionsHandled = 1; functionsHandled < functionsToHandle; functionsHandled++) { > > pre, not post, plz 2 fix Definitely fixed locally. Waiting to land until after Anders has also taken a look 654765 8 beidson 2012-06-21 11:54:43 -0700 http://trac.webkit.org/changeset/120954 148832 2012-06-21 10:36:38 -0700 2012-06-21 10:48:04 -0700 Patch v1 - Proposed fix patch.txt text/plain 3409 beidson ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZyBiL1NvdXJjZS9XZWJDb3JlL0No YW5nZUxvZwppbmRleCA3MGFlMjAwLi41ODM4ZmQ4IDEwMDY0NAotLS0gYS9Tb3VyY2UvV2ViQ29y ZS9DaGFuZ2VMb2cKKysrIGIvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMjEg QEAKKzIwMTItMDYtMjEgIEJyYWR5IEVpZHNvbiAgPGJlaWRzb25AYXBwbGUuY29tPgorCisgICAg ICAgIDxyZGFyOi8vcHJvYmxlbS8xMTcxODk4OD4gYW5kIGh0dHBzOi8vYnVncy53ZWJraXQub3Jn L3Nob3dfYnVnLmNnaT9pZD04OTY3MworICAgICAgICBzaG93TW9kYWxEaWFsb2cgZml4IGNyZWF0 ZXMgcmlzayBvZiBuZXZlciByZXR1cm5pbmcgZnJvbSBSdW5Mb29wOjpwZXJmb3JtV29yaywgcG90 ZW50aWFsbHkgYmxvY2tpbmcgb3RoZXIgZXZlbnQgc291cmNlcworCisgICAgICAgIEluIGNhc2Ug aGFuZGxpbmcgYSBmdW5jdGlvbiBvbiB0aGUgcXVldWUgcGxhY2VzIGFkZGl0aW9uYWwgZnVuY3Rp b25zIG9uIHRoZSBxdWV1ZSwgd2Ugc2hvdWxkCisgICAgICAgIGxpbWl0IHRoZSBudW1iZXIgb2Yg ZnVuY3Rpb25zIGVhY2ggaW52b2NhdGlvbiBvZiBwZXJmb3JtV29yaygpIHBlcmZvcm1zIHNvIGl0 IGNhbiByZXR1cm4gYW5kCisgICAgICAgIG90aGVyIGV2ZW50IHNvdXJjZXMgaGF2ZSBhIGNoYW5j ZSB0byBzcGluLgorCisgICAgICAgIFRoZSBzaG93TW9kYWxEaWFsb2cgZml4IGluIHF1ZXN0aW9u IGlzIGh0dHA6Ly90cmFjLndlYmtpdC5vcmcvY2hhbmdlc2V0LzEyMDg3OQorCisgICAgICAgIFJl dmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAgICAgICogcGxhdGZvcm0vUnVuTG9vcC5j cHA6CisgICAgICAgIChXZWJDb3JlOjpSdW5Mb29wOjpwZXJmb3JtV29yayk6IElmIHRoZXJlIGFy ZSBvbmx5IE4gZnVuY3Rpb25zIGluIHRoZSBxdWV1ZSB3aGVuIHBlcmZvcm1Xb3JrIGlzIGNhbGxl ZCwKKyAgICAgICAgICBvbmx5IGhhbmRsZSB1cCB0byBOIGZ1bmN0aW9ucyBiZWZvcmUgcmV0dXJu aW5nLiBBbnkgYWRkaXRpb25hbCBmdW5jdGlvbnMgd2lsbCBiZSBoYW5kbGVkIHRoZSBuZXh0IHRp bWUKKyAgICAgICAgICB0aGUgcnVubG9vcCBzcGlucy4KKwogMjAxMi0wNi0yMSAgVGFpanUgVHN1 aWtpICA8dHppa0BjaHJvbWl1bS5vcmc+CiAKICAgICAgICAgV2ViIEluc3BlY3RvcjogVW5pZnkg RmlsZVN5c3RlbSBjYWxsYmFja3MKZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3Jt L1J1bkxvb3AuY3BwIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVuTG9vcC5jcHAKaW5kZXgg ZGE3Mzg5ZC4uNTE2MGNkNSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVu TG9vcC5jcHAKKysrIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVuTG9vcC5jcHAKQEAgLTU3 LDE1ICs1NywzNCBAQCBSdW5Mb29wKiBSdW5Mb29wOjptYWluKCkKIAogdm9pZCBSdW5Mb29wOjpw ZXJmb3JtV29yaygpCiB7CisgICAgLy8gT25lIHBvc3NpYmxlIHNjZW5hcmlvIHdoZW4gaGFuZGxp bmcgdGhlIGZ1bmN0aW9uIHF1ZXVlIGlzIGFzIGZvbGxvd3M6CisgICAgLy8gLSBSdW5Mb29wOjpw ZXJmb3JtV29yaygpIGlzIGludm9rZWQgYW5kIHRoZXJlIGlzIDEgZnVuY3Rpb24gb24gdGhlIHF1 ZXVlCisgICAgLy8gLSBIYW5kbGluZyB0aGF0IG9uZSBmdW5jdGlvbiByZXN1bHRzIGluIDEgbW9y ZSBmdW5jdGlvbiBiZWluZyBwbGFjZWQgb24gdGhlIHF1ZXVlCisgICAgLy8gLSBIYW5kbGluZyB0 aGF0IG9uZSByZXN1bHRzIGluIHlldCBhbm90aGVyIGJlaW5nIHBsYWNlZCBvbiB0aGUgcXVldWUK KyAgICAvLyAtIEFuZCBzbyBvbgorICAgIC8vCisgICAgLy8gSW4gdGhpcyBzaXR1YXRpb24sIG9u ZSBpbnZvY2F0aW9uIG9mIHBlcmZvcm1Xb3JrKCkgbmV2ZXIgcmV0dXJucyBhbmQgYWxsIG90aGVy IGV2ZW50IHNvdXJjZXMgYXJlIGJsb2NrZWQuCisgICAgLy8gQnkgb25seSBoYW5kbGluZyB1cCB0 byB0aGUgbnVtYmVyIG9mIGZ1bmN0aW9ucyB0aGF0IHdlcmUgaW4gdGhlIHF1ZXVlIHdoZW4gcGVy Zm9ybVdvcmsoKSBpcyBjYWxsZWQKKyAgICAvLyB3ZSBndWFyYW50ZWUgdGhhdCB3ZSB3aWxsIG9j Y2Fzc2lvbmFsbHkgcmV0dXJuIGZyb20gdGhlIHJ1biBsb29wIGFuZCBvdGhlciBldmVudCBzb3Vy Y2VzIHdpbGwgaGF2ZQorICAgIC8vIGEgY2hhbmNlIHRvIHNwaW4uCisKICAgICBGdW5jdGlvbjx2 b2lkKCk+IGZ1bmN0aW9uOworICAgIHNpemVfdCBmdW5jdGlvbnNUb0hhbmRsZSA9IDA7CisgICAg c2l6ZV90IGZ1bmN0aW9uc0hhbmRsZWQgPSAwOwogICAgIAotICAgIHdoaWxlICh0cnVlKSB7Cisg ICAgZG8gewogICAgICAgICAvLyBJdCBpcyBpbXBvcnRhbnQgdG8gaGFuZGxlIHRoZSBmdW5jdGlv bnMgaW4gdGhlIHF1ZXVlIG9uZSBhdCBhIHRpbWUgYmVjYXVzZSB3aGlsZSBpbnNpZGUgb25lIG9m IHRoZXNlCiAgICAgICAgIC8vIGZ1bmN0aW9ucyB3ZSBtaWdodCByZS1lbnRlciBSdW5Mb29wOjpw ZXJmb3JtV29yaygpIGFuZCB3ZSBuZWVkIHRvIGJlIGFibGUgdG8gcGljayB1cCB3aGVyZSB3ZSBs ZWZ0IG9mZi4KICAgICAgICAgLy8gU2VlIGh0dHA6Ly93ZWJraXQub3JnL2IvODk1OTAgZm9yIG1v cmUgZGlzY3Vzc2lvbi4KIAogICAgICAgICB7CiAgICAgICAgICAgICBNdXRleExvY2tlciBsb2Nr ZXIobV9mdW5jdGlvblF1ZXVlTG9jayk7CisgICAgICAgICAgICBpZiAoZnVuY3Rpb25zVG9IYW5k bGUgPT0gMCkKKyAgICAgICAgICAgICAgICBmdW5jdGlvbnNUb0hhbmRsZSA9IG1fZnVuY3Rpb25R dWV1ZS5zaXplKCk7CisKKyAgICAgICAgICAgIC8vIEV2ZW4gaWYgd2Ugc3RhcnQgb2ZmIHdpdGgg TiBmdW5jdGlvbnMgdG8gaGFuZGxlIGFuZCB3ZSd2ZSBvbmx5IGhhbmRsZWQgbGVzcyB0aGFuIE4g ZnVuY3Rpb25zLCB0aGUgcXVldWUKKyAgICAgICAgICAgIC8vIHN0aWxsIG1pZ2h0IGJlIGVtcHR5 IGJlY2F1c2UgdGhvc2UgZnVuY3Rpb25zIG1pZ2h0IGhhdmUgYmVlbiBoYW5kbGVkIGluIGFuIGlu bmVyIFJ1bkxvb3A6OnBlcmZvcm1Xb3JrKCkuCisgICAgICAgICAgICAvLyBJbiB0aGF0IGNhc2Ug d2Ugc2hvdWxkIGJhaWwgaGVyZS4KICAgICAgICAgICAgIGlmIChtX2Z1bmN0aW9uUXVldWUuaXNF bXB0eSgpKQogICAgICAgICAgICAgICAgIGJyZWFrOwogCkBAIC03Myw3ICs5Miw4IEBAIHZvaWQg UnVuTG9vcDo6cGVyZm9ybVdvcmsoKQogICAgICAgICB9CiAgICAgICAgIAogICAgICAgICBmdW5j dGlvbigpOwotICAgIH0KKyAgICAgICAgKytmdW5jdGlvbnNIYW5kbGVkOworICAgIH0gd2hpbGUg KGZ1bmN0aW9uc0hhbmRsZWQgPCBmdW5jdGlvbnNUb0hhbmRsZSk7CiB9CiAKIHZvaWQgUnVuTG9v cDo6ZGlzcGF0Y2goY29uc3QgRnVuY3Rpb248dm9pZCgpPiYgZnVuY3Rpb24pCg== 148837 2012-06-21 10:48:04 -0700 2012-06-21 11:14:38 -0700 Patch v2 - Avoid taking the lock twice for the common case *and* avoid the branch comparison in Patch v1 patch.txt text/plain 3656 beidson ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZyBiL1NvdXJjZS9XZWJDb3JlL0No YW5nZUxvZwppbmRleCA3MGFlMjAwLi41ODM4ZmQ4IDEwMDY0NAotLS0gYS9Tb3VyY2UvV2ViQ29y ZS9DaGFuZ2VMb2cKKysrIGIvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMjEg QEAKKzIwMTItMDYtMjEgIEJyYWR5IEVpZHNvbiAgPGJlaWRzb25AYXBwbGUuY29tPgorCisgICAg ICAgIDxyZGFyOi8vcHJvYmxlbS8xMTcxODk4OD4gYW5kIGh0dHBzOi8vYnVncy53ZWJraXQub3Jn L3Nob3dfYnVnLmNnaT9pZD04OTY3MworICAgICAgICBzaG93TW9kYWxEaWFsb2cgZml4IGNyZWF0 ZXMgcmlzayBvZiBuZXZlciByZXR1cm5pbmcgZnJvbSBSdW5Mb29wOjpwZXJmb3JtV29yaywgcG90 ZW50aWFsbHkgYmxvY2tpbmcgb3RoZXIgZXZlbnQgc291cmNlcworCisgICAgICAgIEluIGNhc2Ug aGFuZGxpbmcgYSBmdW5jdGlvbiBvbiB0aGUgcXVldWUgcGxhY2VzIGFkZGl0aW9uYWwgZnVuY3Rp b25zIG9uIHRoZSBxdWV1ZSwgd2Ugc2hvdWxkCisgICAgICAgIGxpbWl0IHRoZSBudW1iZXIgb2Yg ZnVuY3Rpb25zIGVhY2ggaW52b2NhdGlvbiBvZiBwZXJmb3JtV29yaygpIHBlcmZvcm1zIHNvIGl0 IGNhbiByZXR1cm4gYW5kCisgICAgICAgIG90aGVyIGV2ZW50IHNvdXJjZXMgaGF2ZSBhIGNoYW5j ZSB0byBzcGluLgorCisgICAgICAgIFRoZSBzaG93TW9kYWxEaWFsb2cgZml4IGluIHF1ZXN0aW9u IGlzIGh0dHA6Ly90cmFjLndlYmtpdC5vcmcvY2hhbmdlc2V0LzEyMDg3OQorCisgICAgICAgIFJl dmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAgICAgICogcGxhdGZvcm0vUnVuTG9vcC5j cHA6CisgICAgICAgIChXZWJDb3JlOjpSdW5Mb29wOjpwZXJmb3JtV29yayk6IElmIHRoZXJlIGFy ZSBvbmx5IE4gZnVuY3Rpb25zIGluIHRoZSBxdWV1ZSB3aGVuIHBlcmZvcm1Xb3JrIGlzIGNhbGxl ZCwKKyAgICAgICAgICBvbmx5IGhhbmRsZSB1cCB0byBOIGZ1bmN0aW9ucyBiZWZvcmUgcmV0dXJu aW5nLiBBbnkgYWRkaXRpb25hbCBmdW5jdGlvbnMgd2lsbCBiZSBoYW5kbGVkIHRoZSBuZXh0IHRp bWUKKyAgICAgICAgICB0aGUgcnVubG9vcCBzcGlucy4KKwogMjAxMi0wNi0yMSAgVGFpanUgVHN1 aWtpICA8dHppa0BjaHJvbWl1bS5vcmc+CiAKICAgICAgICAgV2ViIEluc3BlY3RvcjogVW5pZnkg RmlsZVN5c3RlbSBjYWxsYmFja3MKZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3Jt L1J1bkxvb3AuY3BwIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVuTG9vcC5jcHAKaW5kZXgg ZGE3Mzg5ZC4uNmRhZGVmNSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVu TG9vcC5jcHAKKysrIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vUnVuTG9vcC5jcHAKQEAgLTU3 LDE1ICs1Nyw0MyBAQCBSdW5Mb29wKiBSdW5Mb29wOjptYWluKCkKIAogdm9pZCBSdW5Mb29wOjpw ZXJmb3JtV29yaygpCiB7CisgICAgLy8gSXQgaXMgaW1wb3J0YW50IHRvIGhhbmRsZSB0aGUgZnVu Y3Rpb25zIGluIHRoZSBxdWV1ZSBvbmUgYXQgYSB0aW1lIGJlY2F1c2Ugd2hpbGUgaW5zaWRlIG9u ZSBvZiB0aGVzZQorICAgIC8vIGZ1bmN0aW9ucyB3ZSBtaWdodCByZS1lbnRlciBSdW5Mb29wOjpw ZXJmb3JtV29yaygpIGFuZCB3ZSBuZWVkIHRvIGJlIGFibGUgdG8gcGljayB1cCB3aGVyZSB3ZSBs ZWZ0IG9mZi4KKyAgICAvLyBTZWUgaHR0cDovL3dlYmtpdC5vcmcvYi84OTU5MCBmb3IgbW9yZSBk aXNjdXNzaW9uLgorCisgICAgLy8gT25lIHBvc3NpYmxlIHNjZW5hcmlvIHdoZW4gaGFuZGxpbmcg dGhlIGZ1bmN0aW9uIHF1ZXVlIGlzIGFzIGZvbGxvd3M6CisgICAgLy8gLSBSdW5Mb29wOjpwZXJm b3JtV29yaygpIGlzIGludm9rZWQgYW5kIHRoZXJlIGlzIDEgZnVuY3Rpb24gb24gdGhlIHF1ZXVl CisgICAgLy8gLSBIYW5kbGluZyB0aGF0IG9uZSBmdW5jdGlvbiByZXN1bHRzIGluIDEgbW9yZSBm dW5jdGlvbiBiZWluZyBwbGFjZWQgb24gdGhlIHF1ZXVlCisgICAgLy8gLSBIYW5kbGluZyB0aGF0 IG9uZSByZXN1bHRzIGluIHlldCBhbm90aGVyIGJlaW5nIHBsYWNlZCBvbiB0aGUgcXVldWUKKyAg ICAvLyAtIEFuZCBzbyBvbgorICAgIC8vCisgICAgLy8gSW4gdGhpcyBzaXR1YXRpb24sIG9uZSBp bnZvY2F0aW9uIG9mIHBlcmZvcm1Xb3JrKCkgbmV2ZXIgcmV0dXJucyBhbmQgYWxsIG90aGVyIGV2 ZW50IHNvdXJjZXMgYXJlIGJsb2NrZWQuCisgICAgLy8gQnkgb25seSBoYW5kbGluZyB1cCB0byB0 aGUgbnVtYmVyIG9mIGZ1bmN0aW9ucyB0aGF0IHdlcmUgaW4gdGhlIHF1ZXVlIHdoZW4gcGVyZm9y bVdvcmsoKSBpcyBjYWxsZWQKKyAgICAvLyB3ZSBndWFyYW50ZWUgdGhhdCB3ZSB3aWxsIG9jY2Fz c2lvbmFsbHkgcmV0dXJuIGZyb20gdGhlIHJ1biBsb29wIGFuZCBvdGhlciBldmVudCBzb3VyY2Vz IHdpbGwgaGF2ZQorICAgIC8vIGEgY2hhbmNlIHRvIHNwaW4uCisKICAgICBGdW5jdGlvbjx2b2lk KCk+IGZ1bmN0aW9uOwotICAgIAotICAgIHdoaWxlICh0cnVlKSB7Ci0gICAgICAgIC8vIEl0IGlz IGltcG9ydGFudCB0byBoYW5kbGUgdGhlIGZ1bmN0aW9ucyBpbiB0aGUgcXVldWUgb25lIGF0IGEg dGltZSBiZWNhdXNlIHdoaWxlIGluc2lkZSBvbmUgb2YgdGhlc2UKLSAgICAgICAgLy8gZnVuY3Rp b25zIHdlIG1pZ2h0IHJlLWVudGVyIFJ1bkxvb3A6OnBlcmZvcm1Xb3JrKCkgYW5kIHdlIG5lZWQg dG8gYmUgYWJsZSB0byBwaWNrIHVwIHdoZXJlIHdlIGxlZnQgb2ZmLgotICAgICAgICAvLyBTZWUg aHR0cDovL3dlYmtpdC5vcmcvYi84OTU5MCBmb3IgbW9yZSBkaXNjdXNzaW9uLgorICAgIHNpemVf dCBmdW5jdGlvbnNUb0hhbmRsZSA9IDA7CisKKyAgICB7CisgICAgICAgIE11dGV4TG9ja2VyIGxv Y2tlcihtX2Z1bmN0aW9uUXVldWVMb2NrKTsKKyAgICAgICAgZnVuY3Rpb25zVG9IYW5kbGUgPSBt X2Z1bmN0aW9uUXVldWUuc2l6ZSgpOworCisgICAgICAgIGlmIChtX2Z1bmN0aW9uUXVldWUuaXNF bXB0eSgpKQorICAgICAgICAgICAgcmV0dXJuOwogCisgICAgICAgIGZ1bmN0aW9uID0gbV9mdW5j dGlvblF1ZXVlLnRha2VGaXJzdCgpOworICAgIH0KKworICAgIGZ1bmN0aW9uKCk7CisKKyAgICBm b3IgKHNpemVfdCBmdW5jdGlvbnNIYW5kbGVkID0gMTsgZnVuY3Rpb25zSGFuZGxlZCA8IGZ1bmN0 aW9uc1RvSGFuZGxlOyBmdW5jdGlvbnNIYW5kbGVkKyspIHsKICAgICAgICAgewogICAgICAgICAg ICAgTXV0ZXhMb2NrZXIgbG9ja2VyKG1fZnVuY3Rpb25RdWV1ZUxvY2spOworCisgICAgICAgICAg ICAvLyBFdmVuIGlmIHdlIHN0YXJ0IG9mZiB3aXRoIE4gZnVuY3Rpb25zIHRvIGhhbmRsZSBhbmQg d2UndmUgb25seSBoYW5kbGVkIGxlc3MgdGhhbiBOIGZ1bmN0aW9ucywgdGhlIHF1ZXVlCisgICAg ICAgICAgICAvLyBzdGlsbCBtaWdodCBiZSBlbXB0eSBiZWNhdXNlIHRob3NlIGZ1bmN0aW9ucyBt aWdodCBoYXZlIGJlZW4gaGFuZGxlZCBpbiBhbiBpbm5lciBSdW5Mb29wOjpwZXJmb3JtV29yaygp LgorICAgICAgICAgICAgLy8gSW4gdGhhdCBjYXNlIHdlIHNob3VsZCBiYWlsIGhlcmUuCiAgICAg ICAgICAgICBpZiAobV9mdW5jdGlvblF1ZXVlLmlzRW1wdHkoKSkKICAgICAgICAgICAgICAgICBi cmVhazsKIAo=