174188 2017-07-05 18:07:27 -0700 NewArray in FTLLowerDFGToB3 does not handle speculating on doubles when having a bad time 2017-07-06 23:11:46 -0700 1 1 1 Unclassified WebKit JavaScriptCore Safari Technology Preview Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 1 saam saam benjamin commit-queue fpizlo ggaren gskachkov jfbastien keith_miller mark.lam msaboff ticaiolima webkit-bug-importer ysuzuki oldest_to_newest 1325853 0 saam 2017-07-05 18:07:27 -0700 ... 1325861 1 saam 2017-07-05 18:41:21 -0700 <rdar://problem/30581423> 1325862 2 314674 saam 2017-07-05 18:43:59 -0700 Created attachment 314674 patch 1325929 3 314674 mark.lam 2017-07-05 21:27:44 -0700 Comment on attachment 314674 patch View in context: https://bugs.webkit.org/attachment.cgi?id=314674&action=review r=me with question addressed. > Tools/Scripts/run-jsc-stress-tests:862 > - run("ftl-no-cjit-b3o1", *(FTL_OPTIONS + NO_CJIT_OPTIONS + B3O1_OPTIONS + optionalTestSpecificOptions)) > + run("ftl-no-cjit-b3o1", "--useArrayAllocationProfiling=false", *(FTL_OPTIONS + NO_CJIT_OPTIONS + B3O1_OPTIONS + optionalTestSpecificOptions)) Does it make sense to use --useArrayAllocationProfiling=false for all ftl-no-cjit-b3o1 tests? If not, you can just as easily add //@ runFTLNoCJITB3O1("--useArrayAllocationProfiling=false") at the top of the new test to configure that test alone to use this option. 1325930 4 314674 jfbastien 2017-07-05 21:29:31 -0700 Comment on attachment 314674 patch r=me > JSTests/stress/new-array-having-a-bad-time-double.js:12 > +Object.defineProperty(Object.prototype, "10000", {get() { return 20; }}); Ew 1325931 5 314674 saam 2017-07-05 21:33:25 -0700 Comment on attachment 314674 patch View in context: https://bugs.webkit.org/attachment.cgi?id=314674&action=review >> Tools/Scripts/run-jsc-stress-tests:862 >> + run("ftl-no-cjit-b3o1", "--useArrayAllocationProfiling=false", *(FTL_OPTIONS + NO_CJIT_OPTIONS + B3O1_OPTIONS + optionalTestSpecificOptions)) > > Does it make sense to use --useArrayAllocationProfiling=false for all ftl-no-cjit-b3o1 tests? If not, you can just as easily add //@ runFTLNoCJITB3O1("--useArrayAllocationProfiling=false") at the top of the new test to configure that test alone to use this option. The reason I added it this way is I thought we might as well run a test type this way since it’ll make our compilers hit certain code paths more often than it would otherwise 1325939 6 314674 commit-queue 2017-07-05 22:01:38 -0700 Comment on attachment 314674 patch Clearing flags on attachment: 314674 Committed r219187: <http://trac.webkit.org/changeset/219187> 1325940 7 commit-queue 2017-07-05 22:01:39 -0700 All reviewed patches have been landed. Closing bug. 1326010 8 ysuzuki 2017-07-06 05:32:17 -0700 (In reply to WebKit Commit Bot from comment #7) > All reviewed patches have been landed. Closing bug. It seems that this causes regression. https://arewefastyet.com/#machine=29&view=single&suite=six-speed&subtest=object-assign-es5 https://arewefastyet.com/#machine=29&view=single&suite=six-speed&subtest=object-assign-es6 But it seems a bit strange. I'll check it when I have time. 1326125 9 saam 2017-07-06 12:04:05 -0700 (In reply to Yusuke Suzuki from comment #8) > (In reply to WebKit Commit Bot from comment #7) > > All reviewed patches have been landed. Closing bug. > > It seems that this causes regression. > https://arewefastyet.com/#machine=29&view=single&suite=six- > speed&subtest=object-assign-es5 > https://arewefastyet.com/#machine=29&view=single&suite=six- > speed&subtest=object-assign-es6 > > But it seems a bit strange. I'll check it when I have time. I'm not sure we even care about such a microbenchmark. However, there are other commits in that range. Maybe it's them? I don't see how this patch could slow that code down. 1326298 10 ysuzuki 2017-07-06 19:03:15 -0700 (In reply to Saam Barati from comment #9) > (In reply to Yusuke Suzuki from comment #8) > > (In reply to WebKit Commit Bot from comment #7) > > > All reviewed patches have been landed. Closing bug. > > > > It seems that this causes regression. > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > speed&subtest=object-assign-es5 > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > speed&subtest=object-assign-es6 > > > > But it seems a bit strange. I'll check it when I have time. > > I'm not sure we even care about such a microbenchmark. However, there are > other commits in that range. Maybe it's them? I don't see how this patch > could slow that code down. If this regression is real (I'm still suspecting b/c this change is unlikely related to this), I think we should fix it because this bench measures the time of Object.assign(). Which is super frequently used in user's code including Speedometer2.0 Redux-React. 1326343 11 saam 2017-07-06 20:39:10 -0700 (In reply to Yusuke Suzuki from comment #10) > (In reply to Saam Barati from comment #9) > > (In reply to Yusuke Suzuki from comment #8) > > > (In reply to WebKit Commit Bot from comment #7) > > > > All reviewed patches have been landed. Closing bug. > > > > > > It seems that this causes regression. > > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > > speed&subtest=object-assign-es5 > > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > > speed&subtest=object-assign-es6 > > > > > > But it seems a bit strange. I'll check it when I have time. > > > > I'm not sure we even care about such a microbenchmark. However, there are > > other commits in that range. Maybe it's them? I don't see how this patch > > could slow that code down. > > If this regression is real (I'm still suspecting b/c this change is unlikely > related to this), I think we should fix it because this bench measures the > time of Object.assign(). Which is super frequently used in user's code > including Speedometer2.0 Redux-React. I'll measure this locally. 1326407 12 ysuzuki 2017-07-06 23:11:46 -0700 (In reply to Saam Barati from comment #11) > (In reply to Yusuke Suzuki from comment #10) > > (In reply to Saam Barati from comment #9) > > > (In reply to Yusuke Suzuki from comment #8) > > > > (In reply to WebKit Commit Bot from comment #7) > > > > > All reviewed patches have been landed. Closing bug. > > > > > > > > It seems that this causes regression. > > > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > > > speed&subtest=object-assign-es5 > > > > https://arewefastyet.com/#machine=29&view=single&suite=six- > > > > speed&subtest=object-assign-es6 > > > > > > > > But it seems a bit strange. I'll check it when I have time. > > > > > > I'm not sure we even care about such a microbenchmark. However, there are > > > other commits in that range. Maybe it's them? I don't see how this patch > > > could slow that code down. > > > > If this regression is real (I'm still suspecting b/c this change is unlikely > > related to this), I think we should fix it because this bench measures the > > time of Object.assign(). Which is super frequently used in user's code > > including Speedometer2.0 Redux-React. > > I'll measure this locally. It seems that performance regression is fixed. I'm still not sure what happens... (My guess is C++ compiler inlining...., but idk. If such a thing can change performance so significantly, we should check the implementation in Object.assign and we should annotate NEVER_INLINE / ALWAYS_INLINE carefully since this Object.assign() is very important function.) 314674 2017-07-05 18:43:59 -0700 2017-07-05 22:01:38 -0700 patch c-backup.diff text/plain 7208 saam SW5kZXg6IEpTVGVzdHMvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIEpTVGVzdHMvQ2hhbmdlTG9n CShyZXZpc2lvbiAyMTkxNTcpCisrKyBKU1Rlc3RzL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpA QCAtMSwzICsxLDE1IEBACisyMDE3LTA3LTA1ICBTYWFtIEJhcmF0aSAgPHNiYXJhdGlAYXBwbGUu Y29tPgorCisgICAgICAgIE5ld0FycmF5IGluIEZUTExvd2VyREZHVG9CMyBkb2VzIG5vdCBoYW5k bGUgc3BlY3VsYXRpbmcgb24gZG91YmxlcyB3aGVuIGhhdmluZyBhIGJhZCB0aW1lCisgICAgICAg IGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD0xNzQxODgKKyAgICAgICAg PHJkYXI6Ly9wcm9ibGVtLzMwNTgxNDIzPgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgICogc3RyZXNzL25ldy1hcnJheS1oYXZpbmctYS1iYWQtdGltZS1k b3VibGUuanM6IEFkZGVkLgorICAgICAgICAoYXNzZXJ0KToKKyAgICAgICAgKGZvbyk6CisKIDIw MTctMDctMDMgIFNhYW0gQmFyYXRpICA8c2JhcmF0aUBhcHBsZS5jb20+CiAKICAgICAgICAgU2tp cCB1bnNoaWZ0Q291bnRTbG93Q2FzZS1jb3JyZWN0LXBvc3RDYXBhY2l0eS5qcyBvbiBkZWJ1ZyBi dWlsZHMgc2luY2UgaXQgdGFrZXMgYSBsb25nIHRpbWUgdG8gcnVuLgpJbmRleDogSlNUZXN0cy9z dHJlc3MvbmV3LWFycmF5LWhhdmluZy1hLWJhZC10aW1lLWRvdWJsZS5qcwo9PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0t LSBKU1Rlc3RzL3N0cmVzcy9uZXctYXJyYXktaGF2aW5nLWEtYmFkLXRpbWUtZG91YmxlLmpzCShu b25leGlzdGVudCkKKysrIEpTVGVzdHMvc3RyZXNzL25ldy1hcnJheS1oYXZpbmctYS1iYWQtdGlt ZS1kb3VibGUuanMJKHdvcmtpbmcgY29weSkKQEAgLTAsMCArMSwyNSBAQAorZnVuY3Rpb24gYXNz ZXJ0KGIpIHsKKyAgICBpZiAoIWIpCisgICAgICAgIHRocm93IG5ldyBFcnJvcigiQmFkIik7Cit9 CisKK2Z1bmN0aW9uIGZvbyhhLCBiLCBjKSB7CisgICAgbGV0IHggPSBhICsgYiArIGM7CisgICAg cmV0dXJuIFthLCBiLCBjLCB4XTsKK30KK25vSW5saW5lKGZvbyk7CisKK09iamVjdC5kZWZpbmVQ cm9wZXJ0eShPYmplY3QucHJvdG90eXBlLCAiMTAwMDAiLCB7Z2V0KCkgeyByZXR1cm4gMjA7IH19 KTsKKworZm9yIChsZXQgaSA9IDA7IGkgPCAxMDAwMDsgKytpKSB7CisgICAgbGV0IGEgPSAxMC41 OworICAgIGxldCBiID0gMS4xOworICAgIGxldCBjID0gMS4yOworICAgIGxldCB4ID0gYSArIGIg KyBjOworICAgIGxldCByZXN1bHQgPSBmb28oYSwgYiwgYyk7CisgICAgYXNzZXJ0KHJlc3VsdC5s ZW5ndGggPT09IDQpOworICAgIGFzc2VydChyZXN1bHRbMF0gPT09IGEpOworICAgIGFzc2VydChy ZXN1bHRbMV0gPT09IGIpOworICAgIGFzc2VydChyZXN1bHRbMl0gPT09IGMpOworICAgIGFzc2Vy dChyZXN1bHRbM10gPT09IHgpOworfQpJbmRleDogU291cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5n ZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvSmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCShyZXZp c2lvbiAyMTkxNTcpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCSh3b3JraW5n IGNvcHkpCkBAIC0xLDMgKzEsMjcgQEAKKzIwMTctMDctMDUgIFNhYW0gQmFyYXRpICA8c2JhcmF0 aUBhcHBsZS5jb20+CisKKyAgICAgICAgTmV3QXJyYXkgaW4gRlRMTG93ZXJERkdUb0IzIGRvZXMg bm90IGhhbmRsZSBzcGVjdWxhdGluZyBvbiBkb3VibGVzIHdoZW4gaGF2aW5nIGEgYmFkIHRpbWUK KyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTE3NDE4OAor ICAgICAgICA8cmRhcjovL3Byb2JsZW0vMzA1ODE0MjM+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkg Tk9CT0RZIChPT1BTISkuCisKKyAgICAgICAgV2Ugd2VyZSBjYWxsaW5nIGxvd0pTVmFsdWUoZWRn ZSkgd2hlbiB3ZSB3ZXJlIHNwZWN1bGF0aW5nIHRoZQorICAgICAgICBlZGdlIGFzIGRvdWJsZS4g VGhpcyBpc24ndCBhbGxvd2VkLiBXZSBzaG91bGQgaGF2ZSBiZWVuIHVzaW5nCisgICAgICAgIGxv d0RvdWJsZS4KKyAgICAgICAgCisgICAgICAgIFRoaXMgcGF0Y2ggYWxzbyBhZGRzIGEgbmV3IG9w dGlvbiwgY2FsbGVkIHVzZUFycmF5QWxsb2NhdGlvblByb2ZpbGluZywKKyAgICAgICAgd2hpY2gg ZGVmYXVsdHMgdG8gdHJ1ZS4gV2hlbiBmYWxzZSwgaXQgd2lsbCBtYWtlIHRoZSBhcnJheSBhbGxv Y2F0aW9uCisgICAgICAgIHByb2ZpbGUgbm90IGFjdHVhbGx5IHNhbXBsZSBzZWVuIGFycmF5cy4g SXQnbGwgZm9yY2UgdGhlIGFsbG9jYXRpb24KKyAgICAgICAgcHJvZmlsZSdzIHByZWRpY3RlZCBp bmRleGluZyB0eXBlIHRvIGJlIEFycmF5V2l0aFVuZGVjaWRlZC4gQWRkaW5nCisgICAgICAgIHRo aXMgb3B0aW9uIG1hZGUgaXQgdHJpdmlhbCB0byB3cml0ZSBhIHRlc3QgZm9yIHRoaXMgYnVnLgor CisgICAgICAgICogYnl0ZWNvZGUvQXJyYXlBbGxvY2F0aW9uUHJvZmlsZS5jcHA6CisgICAgICAg IChKU0M6OkFycmF5QWxsb2NhdGlvblByb2ZpbGU6OnVwZGF0ZUluZGV4aW5nVHlwZSk6CisgICAg ICAgICogZnRsL0ZUTExvd2VyREZHVG9CMy5jcHA6CisgICAgICAgIChKU0M6OkZUTDo6REZHOjpM b3dlckRGR1RvQjM6OmNvbXBpbGVOZXdBcnJheSk6CisgICAgICAgICogcnVudGltZS9PcHRpb25z Lmg6CisKIDIwMTctMDctMDMgIE15bGVzIEMuIE1heGZpZWxkICA8bW1heGZpZWxkQGFwcGxlLmNv bT4KIAogICAgICAgICBSZW1vdmUgY29weSBvZiBJQ1UgaGVhZGVycyBmcm9tIFdlYktpdApJbmRl eDogU291cmNlL0phdmFTY3JpcHRDb3JlL2J5dGVjb2RlL0FycmF5QWxsb2NhdGlvblByb2ZpbGUu Y3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9ieXRlY29kZS9BcnJheUFs bG9jYXRpb25Qcm9maWxlLmNwcAkocmV2aXNpb24gMjE5MTU3KQorKysgU291cmNlL0phdmFTY3Jp cHRDb3JlL2J5dGVjb2RlL0FycmF5QWxsb2NhdGlvblByb2ZpbGUuY3BwCSh3b3JraW5nIGNvcHkp CkBAIC00OSw3ICs0OSw4IEBAIHZvaWQgQXJyYXlBbGxvY2F0aW9uUHJvZmlsZTo6dXBkYXRlSW5k ZXgKICAgICBKU0FycmF5KiBsYXN0QXJyYXkgPSBtX2xhc3RBcnJheTsKICAgICBpZiAoIWxhc3RB cnJheSkKICAgICAgICAgcmV0dXJuOwotICAgIG1fY3VycmVudEluZGV4aW5nVHlwZSA9IGxlYXN0 VXBwZXJCb3VuZE9mSW5kZXhpbmdUeXBlcyhtX2N1cnJlbnRJbmRleGluZ1R5cGUsIGxhc3RBcnJh eS0+aW5kZXhpbmdUeXBlKCkpOworICAgIGlmIChMSUtFTFkoT3B0aW9uczo6dXNlQXJyYXlBbGxv Y2F0aW9uUHJvZmlsaW5nKCkpKQorICAgICAgICBtX2N1cnJlbnRJbmRleGluZ1R5cGUgPSBsZWFz dFVwcGVyQm91bmRPZkluZGV4aW5nVHlwZXMobV9jdXJyZW50SW5kZXhpbmdUeXBlLCBsYXN0QXJy YXktPmluZGV4aW5nVHlwZSgpKTsKICAgICBtX2xhc3RBcnJheSA9IDA7CiB9CiAKSW5kZXg6IFNv dXJjZS9KYXZhU2NyaXB0Q29yZS9mdGwvRlRMTG93ZXJERkdUb0IzLmNwcAo9PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0t LSBTb3VyY2UvSmF2YVNjcmlwdENvcmUvZnRsL0ZUTExvd2VyREZHVG9CMy5jcHAJKHJldmlzaW9u IDIxOTE1NykKKysrIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9mdGwvRlRMTG93ZXJERkdUb0IzLmNw cAkod29ya2luZyBjb3B5KQpAQCAtNDY0OCw5ICs0NjQ4LDE2IEBAIHByaXZhdGU6CiAgICAgICAg IAogICAgICAgICBmb3IgKHVuc2lnbmVkIG9wZXJhbmRJbmRleCA9IDA7IG9wZXJhbmRJbmRleCA8 IG1fbm9kZS0+bnVtQ2hpbGRyZW4oKTsgKytvcGVyYW5kSW5kZXgpIHsKICAgICAgICAgICAgIEVk Z2UgZWRnZSA9IG1fZ3JhcGgudmFyQXJnQ2hpbGQobV9ub2RlLCBvcGVyYW5kSW5kZXgpOwotICAg ICAgICAgICAgbV9vdXQuc3RvcmU2NCgKLSAgICAgICAgICAgICAgICBsb3dKU1ZhbHVlKGVkZ2Us IE1hbnVhbE9wZXJhbmRTcGVjdWxhdGlvbiksCi0gICAgICAgICAgICAgICAgbV9vdXQuYWJzb2x1 dGUoYnVmZmVyICsgb3BlcmFuZEluZGV4KSk7CisgICAgICAgICAgICBMVmFsdWUgdmFsdWVUb1N0 b3JlOworICAgICAgICAgICAgc3dpdGNoIChtX25vZGUtPmluZGV4aW5nVHlwZSgpKSB7CisgICAg ICAgICAgICBjYXNlIEFMTF9ET1VCTEVfSU5ERVhJTkdfVFlQRVM6CisgICAgICAgICAgICAgICAg dmFsdWVUb1N0b3JlID0gYm94RG91YmxlKGxvd0RvdWJsZShlZGdlKSk7CisgICAgICAgICAgICAg ICAgYnJlYWs7CisgICAgICAgICAgICBkZWZhdWx0OgorICAgICAgICAgICAgICAgIHZhbHVlVG9T dG9yZSA9IGxvd0pTVmFsdWUoZWRnZSwgTWFudWFsT3BlcmFuZFNwZWN1bGF0aW9uKTsKKyAgICAg ICAgICAgICAgICBicmVhazsKKyAgICAgICAgICAgIH0KKyAgICAgICAgICAgIG1fb3V0LnN0b3Jl NjQodmFsdWVUb1N0b3JlLCBtX291dC5hYnNvbHV0ZShidWZmZXIgKyBvcGVyYW5kSW5kZXgpKTsK ICAgICAgICAgfQogICAgICAgICAKICAgICAgICAgbV9vdXQuc3RvcmVQdHIoCkluZGV4OiBTb3Vy Y2UvSmF2YVNjcmlwdENvcmUvcnVudGltZS9PcHRpb25zLmgKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291cmNl L0phdmFTY3JpcHRDb3JlL3J1bnRpbWUvT3B0aW9ucy5oCShyZXZpc2lvbiAyMTkxNTcpCisrKyBT b3VyY2UvSmF2YVNjcmlwdENvcmUvcnVudGltZS9PcHRpb25zLmgJKHdvcmtpbmcgY29weSkKQEAg LTQ2NCw3ICs0NjQsOCBAQCB0eXBlZGVmIGNvbnN0IGNoYXIqIG9wdGlvblN0cmluZzsKICAgICB2 KGJvb2wsIHVzZVdlYkFzc2VtYmx5RmFzdFRMUywgdHJ1ZSwgTm9ybWFsLCAiSWYgdHJ1ZSwgd2Ug d2lsbCB0cnkgdG8gdXNlIGZhc3QgdGhyZWFkLWxvY2FsIHN0b3JhZ2UgaWYgYXZhaWxhYmxlIG9u IHRoZSBjdXJyZW50IHBsYXRmb3JtLiIpIFwKICAgICB2KGJvb2wsIHVzZUZhc3RUTFNGb3JXYXNt Q29udGV4dCwgdHJ1ZSwgTm9ybWFsLCAiSWYgdHJ1ZSAoYW5kIGZhc3QgVExTIGlzIGVuYWJsZWQp LCB3ZSB3aWxsIHN0b3JlIGNvbnRleHQgaW4gZmFzdCBUTFMuIElmIGZhbHNlLCB3ZSB3aWxsIHBp biBpdCB0byBhIHJlZ2lzdGVyLiIpIFwKICAgICB2KGJvb2wsIHVzZUNhbGxJQ3NGb3JXZWJBc3Nl bWJseVRvSlNDYWxscywgdHJ1ZSwgTm9ybWFsLCAiSWYgdHJ1ZSwgd2Ugd2lsbCB1c2UgQ2FsbExp bmtJbmZvIHRvIGlubGluZSBjYWNoZSBXYXNtIHRvIEpTIGNhbGxzLiIpIFwKLSAgICB2KGJvb2ws IHVzZU9iamVjdFJlc3RTcHJlYWQsIHRydWUsIE5vcm1hbCwgIklmIHRydWUsIHdlIHdpbGwgZW5h YmxlIE9iamVjdCBSZXN0L1NwcmVhZCBmZWF0dXJlLiIpCisgICAgdihib29sLCB1c2VPYmplY3RS ZXN0U3ByZWFkLCB0cnVlLCBOb3JtYWwsICJJZiB0cnVlLCB3ZSB3aWxsIGVuYWJsZSBPYmplY3Qg UmVzdC9TcHJlYWQgZmVhdHVyZS4iKSBcCisgICAgdihib29sLCB1c2VBcnJheUFsbG9jYXRpb25Q cm9maWxpbmcsIHRydWUsIE5vcm1hbCwgIklmIHRydWUsIHdlIHdpbGwgdXNlIG91ciBub3JtYWwg YXJyYXkgYWxsb2NhdGlvbiBwcm9maWxpbmcuIElmIGZhbHNlLCB0aGUgYWxsb2NhdGlvbiBwcm9m aWxlIHdpbGwgYWx3YXlzIGNsYWltIHRvIGJlIHVuZGVjaWRlZC4iKQogCiAKIGVudW0gT3B0aW9u RXF1aXZhbGVuY2UgewpJbmRleDogVG9vbHMvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFRvb2xz L0NoYW5nZUxvZwkocmV2aXNpb24gMjE5MTc0KQorKysgVG9vbHMvQ2hhbmdlTG9nCSh3b3JraW5n IGNvcHkpCkBAIC0xLDMgKzEsMTMgQEAKKzIwMTctMDctMDUgIFNhYW0gQmFyYXRpICA8c2JhcmF0 aUBhcHBsZS5jb20+CisKKyAgICAgICAgTmV3QXJyYXkgaW4gRlRMTG93ZXJERkdUb0IzIGRvZXMg bm90IGhhbmRsZSBzcGVjdWxhdGluZyBvbiBkb3VibGVzIHdoZW4gaGF2aW5nIGEgYmFkIHRpbWUK KyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTE3NDE4OAor ICAgICAgICA8cmRhcjovL3Byb2JsZW0vMzA1ODE0MjM+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkg Tk9CT0RZIChPT1BTISkuCisKKyAgICAgICAgKiBTY3JpcHRzL3J1bi1qc2Mtc3RyZXNzLXRlc3Rz OgorCiAyMDE3LTA3LTAzICBNeWxlcyBDLiBNYXhmaWVsZCAgPG1tYXhmaWVsZEBhcHBsZS5jb20+ CiAKICAgICAgICAgUmVtb3ZlIGNvcHkgb2YgSUNVIGhlYWRlcnMgZnJvbSBXZWJLaXQKSW5kZXg6 IFRvb2xzL1NjcmlwdHMvcnVuLWpzYy1zdHJlc3MtdGVzdHMKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gVG9vbHMv U2NyaXB0cy9ydW4tanNjLXN0cmVzcy10ZXN0cwkocmV2aXNpb24gMjE5MTU3KQorKysgVG9vbHMv U2NyaXB0cy9ydW4tanNjLXN0cmVzcy10ZXN0cwkod29ya2luZyBjb3B5KQpAQCAtODU5LDcgKzg1 OSw3IEBAIGRlZiBydW5GVExOb0NKSVQoKm9wdGlvbmFsVGVzdFNwZWNpZmljT3AKIGVuZAogCiBk ZWYgcnVuRlRMTm9DSklUQjNPMSgqb3B0aW9uYWxUZXN0U3BlY2lmaWNPcHRpb25zKQotICAgIHJ1 bigiZnRsLW5vLWNqaXQtYjNvMSIsICooRlRMX09QVElPTlMgKyBOT19DSklUX09QVElPTlMgKyBC M08xX09QVElPTlMgKyBvcHRpb25hbFRlc3RTcGVjaWZpY09wdGlvbnMpKQorICAgIHJ1bigiZnRs LW5vLWNqaXQtYjNvMSIsICItLXVzZUFycmF5QWxsb2NhdGlvblByb2ZpbGluZz1mYWxzZSIsICoo RlRMX09QVElPTlMgKyBOT19DSklUX09QVElPTlMgKyBCM08xX09QVElPTlMgKyBvcHRpb25hbFRl c3RTcGVjaWZpY09wdGlvbnMpKQogZW5kCiAKIGRlZiBydW5GVExOb0NKSVRWYWxpZGF0ZSgqb3B0 aW9uYWxUZXN0U3BlY2lmaWNPcHRpb25zKQo=