106965 2013-01-15 17:58:18 -0800 NodeRareData doesn't need to have a vtable pointer 2013-01-24 11:27:30 -0800 1 1 1 Unclassified WebKit DOM 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P2 Normal --- 106981 1 rniwa rniwa benjamin esprehn hausmann kling koivisto loislo ojan.autocc ossy webkit.review.bot yurys zan oldest_to_newest 807874 0 rniwa 2013-01-15 17:58:18 -0800 We have a vtable pointer in NodeRareData just so that its destructor and reportMemoryUsage can use them. That's silly because the only place wee delete NodeRareData is Node::clearRareData and we already explicitly calls delete. 807885 1 esprehn 2013-01-15 18:09:59 -0800 I have a patch for this from a few months ago but it made the code a little confusing so I had decided against it. I can just upload it now though... You can't just naively rely on us calling delete since deleting ElementRareData from a base class ptr with a non virtual destructor produces undefined behavior. 807891 2 182891 rniwa 2013-01-15 18:20:03 -0800 Created attachment 182891 Removes the vtable pointer 807951 3 182891 esprehn 2013-01-15 19:24:51 -0800 Comment on attachment 182891 Removes the vtable pointer View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review Can we devirtualize createRareData() as well so people can't return subclasses anymore? There's no reason to support generic subclasses of rare data with virtual createRareData() since they wouldn't have their destructors handled properly. > Source/WebCore/dom/ElementRareData.cpp:39 > +struct SameSizeAsElementRareData { Why not just inherit NodeRareData ? 807955 4 rniwa 2013-01-15 19:35:17 -0800 (In reply to comment #3) > (From update of attachment 182891 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review > > Can we devirtualize createRareData() as well so people can't return subclasses anymore? There's no reason to support generic subclasses of rare data with virtual createRareData() since they wouldn't have their destructors handled properly. Let's do that in a separate patch. There are few cleanups we can do there. 808002 5 benjamin 2013-01-15 20:23:23 -0800 > > Source/WebCore/dom/ElementRareData.cpp:39 > > +struct SameSizeAsElementRareData { > > Why not just inherit NodeRareData ? That would defeat the purpose. 808007 6 rniwa 2013-01-15 20:37:33 -0800 (In reply to comment #5) > > > Source/WebCore/dom/ElementRareData.cpp:39 > > > +struct SameSizeAsElementRareData { > > > > Why not just inherit NodeRareData ? > > That would defeat the purpose. It's same thing since I'm including NodeRareData as a member variable. I'm simply ensuring that the size of NodeRareData is not zero, in which case sizeof(SameSizeAsElementRareData) > sizeof(ElementRareData), among other things. 808013 7 esprehn 2013-01-15 20:45:52 -0800 (In reply to comment #6) > (In reply to comment #5) > > > > Source/WebCore/dom/ElementRareData.cpp:39 > > > > +struct SameSizeAsElementRareData { > > > > > > Why not just inherit NodeRareData ? > > > > That would defeat the purpose. > > It's same thing since I'm including NodeRareData as a member variable. I'm simply ensuring that the size of NodeRareData is not zero, in which case sizeof(SameSizeAsElementRareData) > sizeof(ElementRareData), among other things. I don't understand this comment. Why would this make the struct bigger than the subclass? In general in WebCore we just inherit, ex. SameSizeAsRenderText, ShadowRoot, CSSRule, CSSRuleBase etc. I don't see anyone doing it like this. 808016 8 benjamin 2013-01-15 20:47:08 -0800 > I don't understand this comment. Why would this make the struct bigger than the subclass? In general in WebCore we just inherit, ex. SameSizeAsRenderText, ShadowRoot, CSSRule, CSSRuleBase etc. I don't see anyone doing it like this. I thought you were commenting on SameSizeAsNodeRareData. I should probably read the patch :) 808020 9 rniwa 2013-01-15 20:48:43 -0800 (In reply to comment #7) > I don't understand this comment. Why would this make the struct bigger than the subclass? In general in WebCore we just inherit, ex. SameSizeAsRenderText, ShadowRoot, CSSRule, CSSRuleBase etc. I don't see anyone doing it like this. struct A { }; struct B { A a; int b; } struct C : A { int c; } Then sizeof(B) > sizeof(C) in most implementations. 808023 10 182891 benjamin 2013-01-15 21:04:12 -0800 Comment on attachment 182891 Removes the vtable pointer View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review Looks good to me. I like both Elliott's ideas. > Source/WebCore/dom/ElementRareData.cpp:44 > + void* pointers[7]; > + LayoutSize sizeForResizing; Please move sizeForResizing before pointers. 808024 11 esprehn 2013-01-15 21:04:30 -0800 (In reply to comment #9) > (In reply to comment #7) > > I don't understand this comment. Why would this make the struct bigger than the subclass? In general in WebCore we just inherit, ex. SameSizeAsRenderText, ShadowRoot, CSSRule, CSSRuleBase etc. I don't see anyone doing it like this. > > struct A { }; > struct B { A a; int b; } > struct C : A { int c; } > > Then sizeof(B) > sizeof(C) in most implementations. Oh that's really fascinating! In this case sizeof(A) is actually 1 not 0. Thanks. :) 808026 12 rniwa 2013-01-15 21:09:38 -0800 (In reply to comment #10) > (From update of attachment 182891 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review > > Looks good to me. I like both Elliott's ideas. > > > Source/WebCore/dom/ElementRareData.cpp:44 > > + void* pointers[7]; > > + LayoutSize sizeForResizing; > > Please move sizeForResizing before pointers. I'm gonna move both non-pointer times to before pointers as we discussed in person. 808032 13 rniwa 2013-01-15 21:24:20 -0800 Committed r139833: <http://trac.webkit.org/changeset/139833> 808058 14 ossy 2013-01-15 22:37:17 -0800 (In reply to comment #13) > Committed r139833: <http://trac.webkit.org/changeset/139833> It made all tests crash on GTK and on Qt: - http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release/builds/33025 - http://build.webkit.org/builders/Qt%20Linux%20Release/builds/56380 It seems they are GC related crashes. Could you check it please? 808063 15 rniwa 2013-01-15 22:43:02 -0800 (In reply to comment #14) > (In reply to comment #13) > > Committed r139833: <http://trac.webkit.org/changeset/139833> > > It made all tests crash on GTK and on Qt: > - http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release/builds/33025 > - http://build.webkit.org/builders/Qt%20Linux%20Release/builds/56380 > > It seems they are GC related crashes. Could you check it please? Oops, sorry about that.This is strange maybe it doesn't like non-virtual destructor somehow? 808067 16 benjamin 2013-01-15 22:50:29 -0800 Do you have a backtrace with symbols somewhere? 808068 17 ossy 2013-01-15 22:51:36 -0800 and Chromium too: http://build.webkit.org/builders/Chromium%20Mac%20Release%20%28Tests%29/builds/31809 808070 18 ossy 2013-01-15 22:52:10 -0800 (In reply to comment #16) > Do you have a backtrace with symbols somewhere? I'm on it. But unfortunately it doesn't occur in debug mode. 808071 19 benjamin 2013-01-15 22:53:51 -0800 Just a hunch: it looks like it is using libstd delete && libc free instead of the FastAlloc delete and free. 808072 20 rniwa 2013-01-15 22:55:01 -0800 Fix attempted in http://trac.webkit.org/changeset/139838. The problem was that m_rareData is of type NodeRareDataBase even though it's an instance of NodeRareData so we have to explicitly cast it into NodeRareData before deleting it. 808426 21 182891 darin 2013-01-16 09:20:36 -0800 Comment on attachment 182891 Removes the vtable pointer View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review > Source/WebCore/dom/NodeRareData.h:309 > + // This member function is intentionially not virtual to avoid adding a vtable pointer. Typo: intentionially 815654 22 rniwa 2013-01-24 11:27:30 -0800 (In reply to comment #21) > (From update of attachment 182891 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=182891&action=review > > > Source/WebCore/dom/NodeRareData.h:309 > > + // This member function is intentionially not virtual to avoid adding a vtable pointer. > > Typo: intentionially Oops, fixed in http://trac.webkit.org/changeset/140700. 182891 2013-01-15 18:20:03 -0800 2013-01-16 09:20:36 -0800 Removes the vtable pointer bug-106965-20130115181708.patch text/plain 6017 rniwa SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDEzOTgxOCkKKysrIFNvdXJjZS9XZWJDb3JlL0NoYW5n ZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDI5IEBACisyMDEzLTAxLTE1ICBSeW9zdWtl IE5pd2EgIDxybml3YUB3ZWJraXQub3JnPgorCisgICAgICAgIE5vZGVSYXJlRGF0YSBkb2Vzbid0 IG5lZWQgdG8gaGF2ZSBhIHZ0YWJsZSBwb2ludGVyCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJr aXQub3JnL3Nob3dfYnVnLmNnaT9pZD0xMDY5NjUKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JP RFkgKE9PUFMhKS4KKworICAgICAgICBNYWRlIHRoZSBkZXN0cnVjdG9yIGFuZCByZXBvcnRNZW1v cnlVc2FnZSBvZiBOb2RlUmFyZURhdGFCYXNlIGFuZCBOb2RlUmFyZURhdGEKKyAgICAgICAgcmVz cGVjdGl2ZWx5IGFuZCB0aGVpciBzdWJjbGFzc2VzIG5vbi12aXJ0dWFsLgorCisgICAgICAgICog ZG9tL0VsZW1lbnRSYXJlRGF0YS5jcHA6CisgICAgICAgIChTYW1lU2l6ZUFzRWxlbWVudFJhcmVE YXRhKTogQWRkZWQuCisgICAgICAgIChXZWJDb3JlKTogQWRkZWQgYSBjb21waWxlIGFzc2VydGlv biBmb3IgdGhlIHNpemUgb2YgRWxlbWVudFJhcmVEYXRhLgorICAgICAgICAqIGRvbS9FbGVtZW50 UmFyZURhdGEuaDoKKyAgICAgICAgKEVsZW1lbnRSYXJlRGF0YSk6CisgICAgICAgICogZG9tL05v ZGUuY3BwOgorICAgICAgICAoV2ViQ29yZTo6Tm9kZTo6Y2xlYXJSYXJlRGF0YSk6IEV4cGxpY2l0 ZWx5IGRlbGV0ZSBOb2RlUmFyZURhdGEgb3IgRWxlbWVudFJhcmVEYXRhLgorICAgICAgICAoV2Vi Q29yZTo6Tm9kZTo6cmVwb3J0TWVtb3J5VXNhZ2UpOiBFeHBsaWNpdGVseSBhZGQgTm9kZVJhcmVE YXRhIG9yIEVsZW1lbnRSYXJlRGF0YS4KKyAgICAgICAgKiBkb20vTm9kZS5oOgorICAgICAgICAo Tm9kZVJhcmVEYXRhQmFzZSk6IERlbGV0ZWQgdGhlIHZpcnR1YWwgZGVzdHJ1Y3Rvci4KKyAgICAg ICAgKiBkb20vTm9kZVJhcmVEYXRhLmNwcDoKKyAgICAgICAgKFNhbWVTaXplQXNOb2RlUmFyZURh dGEpOiBSZW1vdmVkIG9uZSBwb2ludGVyLgorICAgICAgICAqIGRvbS9Ob2RlUmFyZURhdGEuaDoK KyAgICAgICAgKE5vZGVSYXJlRGF0YSk6IFJlbW92ZWQgdGhlIHZpcnR1YWwgZGVzdHJ1Y3Rvci4K KyAgICAgICAgKE5vZGVSYXJlRGF0YTo6cmVwb3J0TWVtb3J5VXNhZ2UpOiBNYWRlIG5vbi12aXJ0 dWFsLgorCiAyMDEzLTAxLTE1ICBIYXlhdG8gSXRvICA8aGF5YXRvQGNocm9taXVtLm9yZz4KIAog ICAgICAgICBHcm91cCBhbGwgcmVxdWVzdCBwYXJhbWV0ZXJzIHdoaWNoIGFyZSB1c2VkIHRvIG1h dGNoIENTUyBSdWxlcyBpbnRvIGEgcGFyYW1ldGVyIG9iamVjdC4KSW5kZXg6IFNvdXJjZS9XZWJD b3JlL2RvbS9FbGVtZW50UmFyZURhdGEuY3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9XZWJDb3Jl L2RvbS9FbGVtZW50UmFyZURhdGEuY3BwCShyZXZpc2lvbiAxMzk3ODcpCisrKyBTb3VyY2UvV2Vi Q29yZS9kb20vRWxlbWVudFJhcmVEYXRhLmNwcAkod29ya2luZyBjb3B5KQpAQCAtMzYsNiArMzYs MTcgQEAKIAogbmFtZXNwYWNlIFdlYkNvcmUgewogCitzdHJ1Y3QgU2FtZVNpemVBc0VsZW1lbnRS YXJlRGF0YSB7CisgICAgTm9kZVJhcmVEYXRhIG5vZGVEYXRhOworICAgIHNob3J0IGluZGljZXNb Ml07CisgICAgdW5zaWduZWQgYml0ZmllbGRzOworICAgIHZvaWQqIHBvaW50ZXJzWzddOworICAg IExheW91dFNpemUgc2l6ZUZvclJlc2l6aW5nOworICAgIEludFNpemUgc2Nyb2xsT2Zmc2V0Owor fTsKKworQ09NUElMRV9BU1NFUlQoc2l6ZW9mKEVsZW1lbnRSYXJlRGF0YSkgPT0gc2l6ZW9mKFNh bWVTaXplQXNFbGVtZW50UmFyZURhdGEpLCBFbGVtZW50UmFyZURhdGFTaG91bGRTdGF5U21hbGwp OworCiB2b2lkIEVsZW1lbnRSYXJlRGF0YTo6cmVwb3J0TWVtb3J5VXNhZ2UoTWVtb3J5T2JqZWN0 SW5mbyogbWVtb3J5T2JqZWN0SW5mbykgY29uc3QKIHsKICAgICBNZW1vcnlDbGFzc0luZm8gaW5m byhtZW1vcnlPYmplY3RJbmZvLCB0aGlzLCBXZWJDb3JlTWVtb3J5VHlwZXM6OkRPTSk7CkluZGV4 OiBTb3VyY2UvV2ViQ29yZS9kb20vRWxlbWVudFJhcmVEYXRhLmgKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL1dlYkNvcmUvZG9tL0VsZW1lbnRSYXJlRGF0YS5oCShyZXZpc2lvbiAxMzk3ODcpCisrKyBT b3VyY2UvV2ViQ29yZS9kb20vRWxlbWVudFJhcmVEYXRhLmgJKHdvcmtpbmcgY29weSkKQEAgLTM2 LDcgKzM2LDcgQEAgbmFtZXNwYWNlIFdlYkNvcmUgewogY2xhc3MgRWxlbWVudFJhcmVEYXRhIDog cHVibGljIE5vZGVSYXJlRGF0YSB7CiBwdWJsaWM6CiAgICAgRWxlbWVudFJhcmVEYXRhKCk7Ci0g ICAgdmlydHVhbCB+RWxlbWVudFJhcmVEYXRhKCk7CisgICAgfkVsZW1lbnRSYXJlRGF0YSgpOwog CiAgICAgdm9pZCBzZXRQc2V1ZG9FbGVtZW50KFBzZXVkb0lkLCBQYXNzUmVmUHRyPFBzZXVkb0Vs ZW1lbnQ+KTsKICAgICBQc2V1ZG9FbGVtZW50KiBwc2V1ZG9FbGVtZW50KFBzZXVkb0lkKSBjb25z dDsKQEAgLTk0LDcgKzk0LDggQEAgcHVibGljOgogICAgIHVuc2lnbmVkIGNoaWxkSW5kZXgoKSBj b25zdCB7IHJldHVybiBtX2NoaWxkSW5kZXg7IH0KICAgICB2b2lkIHNldENoaWxkSW5kZXgodW5z aWduZWQgaW5kZXgpIHsgbV9jaGlsZEluZGV4ID0gaW5kZXg7IH0KIAotICAgIHZpcnR1YWwgdm9p ZCByZXBvcnRNZW1vcnlVc2FnZShNZW1vcnlPYmplY3RJbmZvKikgY29uc3QgT1ZFUlJJREU7Cisg ICAgLy8gTWFudWFsbHkgY2FsbGVkIGJ5IE5vZGU6OnJlcG9ydE1lbW9yeVVzYWdlLgorICAgIHZv aWQgcmVwb3J0TWVtb3J5VXNhZ2UoTWVtb3J5T2JqZWN0SW5mbyopIGNvbnN0OwogCiAgICAgRWxl bWVudFNoYWRvdyogc2hhZG93KCkgY29uc3QgeyByZXR1cm4gbV9zaGFkb3cuZ2V0KCk7IH0KICAg ICB2b2lkIHNldFNoYWRvdyhQYXNzT3duUHRyPEVsZW1lbnRTaGFkb3c+IHNoYWRvdykgeyBtX3No YWRvdyA9IHNoYWRvdzsgfQpJbmRleDogU291cmNlL1dlYkNvcmUvZG9tL05vZGUuY3BwCj09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT0KLS0tIFNvdXJjZS9XZWJDb3JlL2RvbS9Ob2RlLmNwcAkocmV2aXNpb24gMTM5Nzg3KQor KysgU291cmNlL1dlYkNvcmUvZG9tL05vZGUuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC00NzEsNyAr NDcxLDEwIEBAIHZvaWQgTm9kZTo6Y2xlYXJSYXJlRGF0YSgpCiAgICAgQVNTRVJUKCF0cmFuc2ll bnRNdXRhdGlvbk9ic2VydmVyUmVnaXN0cnkoKSB8fCB0cmFuc2llbnRNdXRhdGlvbk9ic2VydmVy UmVnaXN0cnkoKS0+aXNFbXB0eSgpKTsKIAogICAgIFJlbmRlck9iamVjdCogcmVuZGVyZXIgPSBt X2RhdGEubV9yYXJlRGF0YS0+cmVuZGVyZXIoKTsKLSAgICBkZWxldGUgbV9kYXRhLm1fcmFyZURh dGE7CisgICAgaWYgKGlzRWxlbWVudE5vZGUoKSkKKyAgICAgICAgZGVsZXRlIHN0YXRpY19jYXN0 PEVsZW1lbnRSYXJlRGF0YSo+KG1fZGF0YS5tX3JhcmVEYXRhKTsKKyAgICBlbHNlCisgICAgICAg IGRlbGV0ZSBtX2RhdGEubV9yYXJlRGF0YTsKICAgICBtX2RhdGEubV9yZW5kZXJlciA9IHJlbmRl cmVyOwogICAgIGNsZWFyRmxhZyhIYXNSYXJlRGF0YUZsYWcpOwogfQpAQCAtMjU3OCw4ICsyNTgx LDEyIEBAIHZvaWQgTm9kZTo6cmVwb3J0TWVtb3J5VXNhZ2UoTWVtb3J5T2JqZWMKICAgICBpbmZv LmFkZE1lbWJlcihtX25leHQpOwogICAgIGluZm8uYWRkTWVtYmVyKG1fcHJldmlvdXMpOwogICAg IGluZm8uYWRkTWVtYmVyKHRoaXMtPnJlbmRlcmVyKCkpOwotICAgIGlmIChoYXNSYXJlRGF0YSgp KQotICAgICAgICBpbmZvLmFkZE1lbWJlcihyYXJlRGF0YSgpKTsKKyAgICBpZiAoaGFzUmFyZURh dGEoKSkgeworICAgICAgICBpZiAoaXNFbGVtZW50Tm9kZSgpKQorICAgICAgICAgICAgaW5mby5h ZGRNZW1iZXIoc3RhdGljX2Nhc3Q8RWxlbWVudFJhcmVEYXRhKj4ocmFyZURhdGEoKSkpOworICAg ICAgICBlbHNlCisgICAgICAgICAgICBpbmZvLmFkZE1lbWJlcihyYXJlRGF0YSgpKTsKKyAgICB9 CiB9CiAKIHZvaWQgTm9kZTo6dGV4dFJlY3RzKFZlY3RvcjxJbnRSZWN0PiYgcmVjdHMpIGNvbnN0 CkluZGV4OiBTb3VyY2UvV2ViQ29yZS9kb20vTm9kZS5oCj09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9X ZWJDb3JlL2RvbS9Ob2RlLmgJKHJldmlzaW9uIDEzOTc4NykKKysrIFNvdXJjZS9XZWJDb3JlL2Rv bS9Ob2RlLmgJKHdvcmtpbmcgY29weSkKQEAgLTExNSw5ICsxMTUsOSBAQCBwdWJsaWM6CiAgICAg UmVuZGVyT2JqZWN0KiByZW5kZXJlcigpIGNvbnN0IHsgcmV0dXJuIG1fcmVuZGVyZXI7IH0KICAg ICB2b2lkIHNldFJlbmRlcmVyKFJlbmRlck9iamVjdCogcmVuZGVyZXIpIHsgbV9yZW5kZXJlciA9 IHJlbmRlcmVyOyB9CiAKLSAgICB2aXJ0dWFsIH5Ob2RlUmFyZURhdGFCYXNlKCkgeyB9CiBwcm90 ZWN0ZWQ6CiAgICAgTm9kZVJhcmVEYXRhQmFzZSgpIHsgfQorCiBwcml2YXRlOgogICAgIFJlbmRl ck9iamVjdCogbV9yZW5kZXJlcjsKIH07CkluZGV4OiBTb3VyY2UvV2ViQ29yZS9kb20vTm9kZVJh cmVEYXRhLmNwcAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2ViQ29yZS9kb20vTm9kZVJhcmVEYXRh LmNwcAkocmV2aXNpb24gMTM5Nzg3KQorKysgU291cmNlL1dlYkNvcmUvZG9tL05vZGVSYXJlRGF0 YS5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTM5LDcgKzM5LDcgQEAKIG5hbWVzcGFjZSBXZWJDb3Jl IHsKIAogc3RydWN0IFNhbWVTaXplQXNOb2RlUmFyZURhdGEgewotICAgIHZvaWQqIG1fcG9pbnRl cls0XTsKKyAgICB2b2lkKiBtX3BvaW50ZXJbM107CiAjaWYgRU5BQkxFKE1JQ1JPREFUQSkKICAg ICB2b2lkKiBtX21pY3JvRGF0YTsKICNlbmRpZgpJbmRleDogU291cmNlL1dlYkNvcmUvZG9tL05v ZGVSYXJlRGF0YS5oCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9XZWJDb3JlL2RvbS9Ob2RlUmFyZURh dGEuaAkocmV2aXNpb24gMTM5Nzg3KQorKysgU291cmNlL1dlYkNvcmUvZG9tL05vZGVSYXJlRGF0 YS5oCSh3b3JraW5nIGNvcHkpCkBAIC0yODAsOSArMjgwLDYgQEAgcHVibGljOiAgICAKICAgICBO b2RlUmFyZURhdGEoKQogICAgIHsgfQogCi0gICAgdmlydHVhbCB+Tm9kZVJhcmVEYXRhKCkKLSAg ICB7IH0KLQogICAgIHZvaWQgY2xlYXJOb2RlTGlzdHMoKSB7IG1fbm9kZUxpc3RzLmNsZWFyKCk7 IH0KICAgICBOb2RlTGlzdHNOb2RlRGF0YSogbm9kZUxpc3RzKCkgY29uc3QgeyByZXR1cm4gbV9u b2RlTGlzdHMuZ2V0KCk7IH0KICAgICBOb2RlTGlzdHNOb2RlRGF0YSogZW5zdXJlTm9kZUxpc3Rz KCkKQEAgLTMwOSw3ICszMDYsOCBAQCBwdWJsaWM6ICAgIAogICAgIH0KICNlbmRpZgogCi0gICAg dmlydHVhbCB2b2lkIHJlcG9ydE1lbW9yeVVzYWdlKE1lbW9yeU9iamVjdEluZm8qKSBjb25zdDsK KyAgICAvLyBUaGlzIG1lbWJlciBmdW5jdGlvbiBpcyBpbnRlbnRpb25pYWxseSBub3QgdmlydHVh bCB0byBhdm9pZCBhZGRpbmcgYSB2dGFibGUgcG9pbnRlci4KKyAgICB2b2lkIHJlcG9ydE1lbW9y eVVzYWdlKE1lbW9yeU9iamVjdEluZm8qKSBjb25zdDsKIAogcHJpdmF0ZToKICAgICBPd25QdHI8 Tm9kZUxpc3RzTm9kZURhdGE+IG1fbm9kZUxpc3RzOwo=