70668 2011-10-21 20:25:51 -0700 DFG inlining sometimes fails to reset constant references 2011-10-21 20:58:10 -0700 1 1 1 Unclassified WebKit JavaScriptCore 528+ (Nightly build) All All RESOLVED FIXED P2 Normal --- 1 fpizlo webkit-unassigned barraclough fpizlo oliver oldest_to_newest 488838 0 fpizlo 2011-10-21 20:25:51 -0700 And it sometimes resets them even though it doesn't have to. The logic of resetting constant references should be moved out of parseBlock(), into a separate method, and should be called when a new block is created. 488839 1 112064 fpizlo 2011-10-21 20:48:08 -0700 Created attachment 112064 the patch This is generally neutral although it does have some interesting fluctuations that may be worth investigating in the future. Benchmark report for SunSpider, V8, and Kraken. VMs tested: "TipOfTree" at /Volumes/Data/pizlo/quinary/OpenSource/WebKitBuild/Release/jsc "ConstReset" at /Volumes/Data/pizlo/septenary/OpenSource/WebKitBuild/Release/jsc Collected 12 samples per benchmark/VM, with 4 VM invocations per benchmark. Emitted a call to gc() between sample measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific preciseTime() function to get microsecond-level timing. Reporting benchmark execution times with 95% confidence intervals in milliseconds. TipOfTree ConstReset SunSpider: 3d-cube 7.3697+-0.1885 7.3289+-0.1415 3d-morph 7.7265+-0.1192 ? 7.7378+-0.1193 ? 3d-raytrace 7.6756+-0.1582 7.5050+-0.1473 might be 1.0227x faster access-binary-trees 1.7548+-0.0548 ? 1.7575+-0.0609 ? access-fannkuch 6.5196+-0.1179 6.5059+-0.1434 access-nbody 3.5707+-0.0558 ? 3.6066+-0.0595 ? might be 1.0101x slower access-nsieve 2.5743+-0.0709 ? 2.6457+-0.0577 ? might be 1.0277x slower bitops-3bit-bits-in-byte 1.3013+-0.0298 ? 1.3021+-0.0274 ? bitops-bits-in-byte 2.4103+-0.0260 2.3469+-0.0654 might be 1.0270x faster bitops-bitwise-and 3.4038+-0.1058 ? 3.4435+-0.0980 ? might be 1.0117x slower bitops-nsieve-bits 5.3452+-0.1058 ? 5.4549+-0.1053 ? might be 1.0205x slower controlflow-recursive 2.1495+-0.0436 2.1068+-0.0419 might be 1.0203x faster crypto-aes 7.4158+-0.1735 7.4076+-0.1559 crypto-md5 2.7533+-0.0667 ? 2.8819+-0.2392 ? might be 1.0467x slower crypto-sha1 2.4421+-0.0576 ? 2.4994+-0.0617 ? might be 1.0234x slower date-format-tofte 10.0696+-0.2276 ? 10.1237+-0.1775 ? date-format-xparb 9.4134+-0.2859 ? 9.5952+-0.2443 ? might be 1.0193x slower math-cordic 7.5517+-1.1382 6.5432+-0.1072 might be 1.1541x faster math-partial-sums 7.6746+-0.1174 ? 7.7438+-0.1316 ? math-spectral-norm 2.5707+-0.0578 ? 2.6218+-0.0599 ? might be 1.0199x slower regexp-dna 11.4736+-0.1521 ? 11.7047+-0.1798 ? might be 1.0201x slower string-base64 4.4023+-0.1100 ? 4.4344+-0.1046 ? string-fasta 6.3179+-0.1273 ? 6.4466+-0.0765 ? might be 1.0204x slower string-tagcloud 11.7143+-0.2527 11.5438+-0.2062 might be 1.0148x faster string-unpack-code 20.5107+-0.3544 20.3937+-0.2201 string-validate-input 5.2263+-0.0780 ? 5.2982+-0.1235 ? might be 1.0137x slower <arithmetic> * 6.2053+-0.0459 6.1915+-0.0270 <geometric> 5.0059+-0.0332 ? 5.0118+-0.0273 ? <harmonic> 4.0049+-0.0254 ? 4.0191+-0.0299 ? TipOfTree ConstReset V8: crypto 74.8418+-0.4605 74.7426+-0.3154 deltablue 176.9986+-1.5209 ! 182.5245+-2.7802 ! definitely 1.0312x slower earley-boyer 94.6350+-1.8084 94.4009+-1.6468 raytrace 61.7280+-0.5734 ^ 60.5946+-0.5558 ^ definitely 1.0187x faster regexp 106.1436+-0.4098 ^ 105.2642+-0.2244 ^ definitely 1.0084x faster richards 145.7465+-3.0841 142.3034+-0.5496 might be 1.0242x faster splay 95.0169+-0.5543 ^ 94.0349+-0.3270 ^ definitely 1.0104x faster <arithmetic> 107.8729+-0.4137 107.6950+-0.5193 <geometric> * 101.8415+-0.3820 101.3519+-0.4472 <harmonic> 96.4127+-0.3657 95.7241+-0.4252 TipOfTree ConstReset Kraken: ai-astar 501.7759+-2.3059 ? 505.1881+-3.9941 ? audio-beat-detection 197.0312+-2.5677 196.1807+-1.7175 audio-dft 271.7374+-3.5796 ? 272.6768+-2.4040 ? audio-fft 125.6273+-0.5593 ? 125.8057+-0.5705 ? audio-oscillator 283.1211+-31.4860 253.8578+-1.2954 might be 1.1153x faster imaging-darkroom 404.2532+-1.7140 ? 406.2973+-3.6529 ? imaging-desaturate 222.0366+-0.6368 ? 222.2356+-0.7487 ? imaging-gaussian-blur 563.3645+-1.8204 ? 564.6778+-1.8245 ? json-parse-financial 58.3606+-0.3201 ? 58.7054+-0.5319 ? json-stringify-tinderbox 68.9683+-0.4043 ! 70.1906+-0.4180 ! definitely 1.0177x slower stanford-crypto-aes 133.2210+-1.7591 ? 133.7263+-4.4229 ? stanford-crypto-ccm 103.4855+-0.8125 103.2851+-1.0559 stanford-crypto-pbkdf2 196.1213+-2.8207 ? 198.2128+-2.4398 ? might be 1.0107x slower stanford-crypto-sha256-iterative 71.9108+-0.4195 ? 72.0350+-0.4113 ? <arithmetic> * 228.6439+-2.3124 227.3625+-0.7480 <geometric> 179.9488+-1.3376 179.3734+-0.7063 <harmonic> 141.5599+-0.5082 ? 141.7610+-0.5971 ? TipOfTree ConstReset All benchmarks: <arithmetic> 87.6056+-0.6634 87.1898+-0.2779 <geometric> 22.7894+-0.0946 22.7666+-0.0863 <harmonic> 7.0532+-0.0438 ? 7.0772+-0.0515 ? TipOfTree ConstReset Geomean of preferred means: <scaled-result> 52.4719+-0.1849 52.2530+-0.1425 488847 2 fpizlo 2011-10-21 20:58:10 -0700 Landed in http://trac.webkit.org/changeset/98188 112064 2011-10-21 20:48:08 -0700 2011-10-21 20:49:58 -0700 the patch fixinlconst_patch_1.diff text/plain 6358 fpizlo SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gOTgxODYpCisrKyBTb3VyY2Uv SmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEsMTkgQEAK KzIwMTEtMTAtMjEgIEZpbGlwIFBpemxvICA8ZnBpemxvQGFwcGxlLmNvbT4KKworICAgICAgICBE RkcgaW5saW5pbmcgc29tZXRpbWVzIGZhaWxzIHRvIHJlc2V0IGNvbnN0YW50IHJlZmVyZW5jZXMK KyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTcwNjY4CisK KyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisgICAgICAgIAorICAgICAgICBS ZXNldCBjb25zdGFudCByZWZlcmVuY2VzIHdoZW4gd2UgbmVlZCB0byAobmV3IGJsb2NrIGNyZWF0 ZWQpIGFuZCBub3QKKyAgICAgICAgd2hlbiB3ZSBkb24ndCAoY2hhbmdlIG9mIGlubGluaW5nIGRl cHRoKS4KKworICAgICAgICAqIGRmZy9ERkdCeXRlQ29kZVBhcnNlci5jcHA6CisgICAgICAgIChK U0M6OkRGRzo6Qnl0ZUNvZGVQYXJzZXI6OmhhbmRsZUlubGluaW5nKToKKyAgICAgICAgKEpTQzo6 REZHOjpCeXRlQ29kZVBhcnNlcjo6cHJlcGFyZVRvUGFyc2VCbG9jayk6CisgICAgICAgIChKU0M6 OkRGRzo6Qnl0ZUNvZGVQYXJzZXI6OnBhcnNlQmxvY2spOgorICAgICAgICAoSlNDOjpERkc6OkJ5 dGVDb2RlUGFyc2VyOjpwYXJzZUNvZGVCbG9jayk6CisKIDIwMTEtMTAtMjEgIEZpbGlwIFBpemxv ICA8ZnBpemxvQGFwcGxlLmNvbT4KIAogICAgICAgICBERkcgc2hvdWxkIGhhdmUgaW5saW5pbmcK SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwCj09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIu Y3BwCShyZXZpc2lvbiA5ODE4NCkKKysrIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0 ZUNvZGVQYXJzZXIuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC04Miw2ICs4Miw4IEBAIHByaXZhdGU6 CiAgICAgYm9vbCBoYW5kbGVJbmxpbmluZyhib29sIHVzZXNSZXN1bHQsIGludCBjYWxsVGFyZ2V0 LCBOb2RlSW5kZXggY2FsbFRhcmdldE5vZGVJbmRleCwgaW50IHJlc3VsdE9wZXJhbmQsIGJvb2wg Y2VydGFpbkFib3V0RXhwZWN0ZWRGdW5jdGlvbiwgSlNGdW5jdGlvbiosIGludCBmaXJzdEFyZywg aW50IGxhc3RBcmcsIHVuc2lnbmVkIG5leHRPZmZzZXQpOwogICAgIC8vIEhhbmRsZSBpbnRyaW5z aWMgZnVuY3Rpb25zLiBSZXR1cm4gdHJ1ZSBpZiBpdCBzdWNjZWVkZWQsIGZhbHNlIGlmIHdlIG5l ZWQgdG8gcGxhbnQgYSBjYWxsLgogICAgIGJvb2wgaGFuZGxlSW50cmluc2ljKGJvb2wgdXNlc1Jl c3VsdCwgaW50IHJlc3VsdE9wZXJhbmQsIEludHJpbnNpYywgaW50IGZpcnN0QXJnLCBpbnQgbGFz dEFyZywgUHJlZGljdGVkVHlwZSBwcmVkaWN0aW9uKTsKKyAgICAvLyBQcmVwYXJlIHRvIHBhcnNl IGEgYmxvY2suCisgICAgdm9pZCBwcmVwYXJlVG9QYXJzZUJsb2NrKCk7CiAgICAgLy8gUGFyc2Ug YSBzaW5nbGUgYmFzaWMgYmxvY2sgb2YgYnl0ZWNvZGUgaW5zdHJ1Y3Rpb25zLgogICAgIGJvb2wg cGFyc2VCbG9jayh1bnNpZ25lZCBsaW1pdCk7CiAgICAgLy8gRmluZCByZWFjaGFibGUgY29kZSBh bmQgc2V0dXAgcHJlZGVjZXNzb3IgbGlua3MgaW4gdGhlIGdyYXBoJ3MgQmFzaWNCbG9ja3MuCkBA IC0xMDI3LDYgKzEwMjksNyBAQCBib29sIEJ5dGVDb2RlUGFyc2VyOjpoYW5kbGVJbmxpbmluZyhi b29sCiAgICAgbV9pbmxpbmVTdGFja1RvcC0+bV9jYWxsZXItPm1fdW5saW5rZWRCbG9ja3MuYXBw ZW5kKFVubGlua2VkQmxvY2sobV9ncmFwaC5tX2Jsb2Nrcy5zaXplKCkpKTsKICAgICBtX2lubGlu ZVN0YWNrVG9wLT5tX2NhbGxlci0+bV9ibG9ja0xpbmtpbmdUYXJnZXRzLmFwcGVuZChtX2dyYXBo Lm1fYmxvY2tzLnNpemUoKSk7CiAgICAgbV9ncmFwaC5tX2Jsb2Nrcy5hcHBlbmQoYmxvY2sucmVs ZWFzZSgpKTsKKyAgICBwcmVwYXJlVG9QYXJzZUJsb2NrKCk7CiAgICAgCiAgICAgLy8gQXQgdGhp cyBwb2ludCB3ZSByZXR1cm4gYW5kIGNvbnRpbnVlIHRvIGdlbmVyYXRlIGNvZGUgZm9yIHRoZSBj YWxsZXIsIGJ1dAogICAgIC8vIGluIHRoZSBuZXcgYmFzaWMgYmxvY2suCkBAIC0xMTIyLDE0ICsx MTI1LDE0IEBAIGJvb2wgQnl0ZUNvZGVQYXJzZXI6OmhhbmRsZUludHJpbnNpYyhib28KICAgICB9 CiB9CiAKK3ZvaWQgQnl0ZUNvZGVQYXJzZXI6OnByZXBhcmVUb1BhcnNlQmxvY2soKQoreworICAg IGZvciAodW5zaWduZWQgaSA9IDA7IGkgPCBtX2NvbnN0YW50cy5zaXplKCk7ICsraSkKKyAgICAg ICAgbV9jb25zdGFudHNbaV0gPSBDb25zdGFudFJlY29yZCgpOworfQorCiBib29sIEJ5dGVDb2Rl UGFyc2VyOjpwYXJzZUJsb2NrKHVuc2lnbmVkIGxpbWl0KQogewotICAgIC8vIE5vIG5lZWQgdG8g cmVzZXQgc3RhdGUgaW5pdGlhbGx5LCBzaW5jZSBpdCBoYXMgYmVlbiBzZXQgYnkgdGhlIGNvbnN0 cnVjdG9yLgotICAgIGlmIChtX2N1cnJlbnRJbmRleCkgewotICAgICAgICBmb3IgKHVuc2lnbmVk IGkgPSAwOyBpIDwgbV9jb25zdGFudHMuc2l6ZSgpOyArK2kpCi0gICAgICAgICAgICBtX2NvbnN0 YW50c1tpXSA9IENvbnN0YW50UmVjb3JkKCk7Ci0gICAgfQotICAgIAogICAgIC8vIElmIHdlIGFy ZSB0aGUgZmlyc3QgYmFzaWMgYmxvY2ssIGludHJvZHVjZSBtYXJrZXJzIGZvciBhcmd1bWVudHMu IFRoaXMgYWxsb3dzCiAgICAgLy8gdXMgdG8gdHJhY2sgaWYgYSB1c2Ugb2YgYW4gYXJndW1lbnQg bWF5IHVzZSB0aGUgYWN0dWFsIGFyZ3VtZW50IHBhc3NlZCwgYXMKICAgICAvLyBvcHBvc2VkIHRv IHVzaW5nIGEgdmFsdWUgd2Ugc2V0IGV4cGxpY2l0bHkuCkBAIC0yMzc5LDYgKzIzODIsNyBAQCBi b29sIEJ5dGVDb2RlUGFyc2VyOjpwYXJzZUNvZGVCbG9jaygpCiAgICAgICAgICAgICAgICAgICAg IG1faW5saW5lU3RhY2tUb3AtPm1fdW5saW5rZWRCbG9ja3MuYXBwZW5kKFVubGlua2VkQmxvY2so bV9ncmFwaC5tX2Jsb2Nrcy5zaXplKCkpKTsKICAgICAgICAgICAgICAgICAgICAgbV9pbmxpbmVT dGFja1RvcC0+bV9ibG9ja0xpbmtpbmdUYXJnZXRzLmFwcGVuZChtX2dyYXBoLm1fYmxvY2tzLnNp emUoKSk7CiAgICAgICAgICAgICAgICAgICAgIG1fZ3JhcGgubV9ibG9ja3MuYXBwZW5kKGJsb2Nr LnJlbGVhc2UoKSk7CisgICAgICAgICAgICAgICAgICAgIHByZXBhcmVUb1BhcnNlQmxvY2soKTsK ICAgICAgICAgICAgICAgICB9CiAgICAgICAgICAgICB9CiAKSW5kZXg6IExheW91dFRlc3RzL0No YW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9DaGFuZ2VMb2cJKHJldmlzaW9uIDk4 MTg2KQorKysgTGF5b3V0VGVzdHMvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEs MjAgQEAKKzIwMTEtMTAtMjEgIEZpbGlwIFBpemxvICA8ZnBpemxvQGFwcGxlLmNvbT4KKworICAg ICAgICBERkcgaW5saW5pbmcgc29tZXRpbWVzIGZhaWxzIHRvIHJlc2V0IGNvbnN0YW50IHJlZmVy ZW5jZXMKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTcw NjY4CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisgICAgICAgIAorICAg ICAgICBBZGRlZCBhIG5ldyB0ZXN0IHRoYXQgY292ZXJzIHRoaXMgc3BlY2lmaWMgYnVnIGFzIHdl bGwgYXMgdGhlIGdlbmVyYWwKKyAgICAgICAgYWJpbGl0eSB0byBwZXJmb3JtIGlubGluaW5nLCBh bmQgdGhlIGFiaWxpdHkgdG8gT1NSIG91dCBvZiBhbiBpbmxpbmUKKyAgICAgICAgc3RhY2suCisK KyAgICAgICAgKiBmYXN0L2pzL2RmZy1pbmxpbmUtY29uc3RhbnQtZXhwZWN0ZWQudHh0OiBBZGRl ZC4KKyAgICAgICAgKiBmYXN0L2pzL2RmZy1pbmxpbmUtY29uc3RhbnQuaHRtbDogQWRkZWQuCisg ICAgICAgICogZmFzdC9qcy9zY3JpcHQtdGVzdHMvZGZnLWlubGluZS1jb25zdGFudC5qczogQWRk ZWQuCisgICAgICAgIChmb28pOgorICAgICAgICAoYmFyKToKKwogMjAxMS0xMC0yMSAgT2phbiBW YWZhaSAgPG9qYW5AY2hyb21pdW0ub3JnPgogCiAgICAgICAgIEZpeHVwIHRlc3QgZXhwZWN0YXRp b25zIG5vdyB0aGF0IGh0dHA6Ly90cmFjLndlYmtpdC5vcmcvY2hhbmdlc2V0Lzk4MTczIGhhcyBs YW5kZWQuCkluZGV4OiBMYXlvdXRUZXN0cy9mYXN0L2pzL2RmZy1pbmxpbmUtY29uc3RhbnQtZXhw ZWN0ZWQudHh0Cj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91dFRlc3RzL2Zhc3QvanMvZGZnLWlubGluZS1j b25zdGFudC1leHBlY3RlZC50eHQJKHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9mYXN0L2pz L2RmZy1pbmxpbmUtY29uc3RhbnQtZXhwZWN0ZWQudHh0CShyZXZpc2lvbiAwKQpAQCAtMCwwICsx LDEzIEBACitUaGlzIHRlc3RzIHRoYXQgZnVuY3Rpb24gaW5saW5pbmcgaW4gdGhlIERGRyBKSVQg ZG9lc24ndCBnZXQgY29uZnVzZWQgYnkgY29uc3RhbnRzIGJlaW5nIHJldXNlZCBiZXR3ZWVuIGlu bGluZXIgYW5kIGlubGluZWUuCisKK09uIHN1Y2Nlc3MsIHlvdSB3aWxsIHNlZSBhIHNlcmllcyBv ZiAiUEFTUyIgbWVzc2FnZXMsIGZvbGxvd2VkIGJ5ICJURVNUIENPTVBMRVRFIi4KKworCitQQVNT IGJhcig2LCAwKSBpcyAxMAorUEFTUyBiYXIoNiwgMSkgaXMgMTUKK1BBU1MgYmFyKDYsIGZhbHNl KSBpcyAxMAorUEFTUyBiYXIoNiwgdHJ1ZSkgaXMgMTUKK1BBU1Mgc3VjY2Vzc2Z1bGx5UGFyc2Vk IGlzIHRydWUKKworVEVTVCBDT01QTEVURQorCkluZGV4OiBMYXlvdXRUZXN0cy9mYXN0L2pzL2Rm Zy1pbmxpbmUtY29uc3RhbnQuaHRtbAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9mYXN0L2pz L2RmZy1pbmxpbmUtY29uc3RhbnQuaHRtbAkocmV2aXNpb24gMCkKKysrIExheW91dFRlc3RzL2Zh c3QvanMvZGZnLWlubGluZS1jb25zdGFudC5odG1sCShyZXZpc2lvbiAwKQpAQCAtMCwwICsxLDEw IEBACis8IURPQ1RZUEUgSFRNTCBQVUJMSUMgIi0vL0lFVEYvL0RURCBIVE1MLy9FTiI+Cis8aHRt bD4KKzxoZWFkPgorPHNjcmlwdCBzcmM9InJlc291cmNlcy9qcy10ZXN0LXByZS5qcyI+PC9zY3Jp cHQ+Cis8L2hlYWQ+Cis8Ym9keT4KKzxzY3JpcHQgc3JjPSJzY3JpcHQtdGVzdHMvZGZnLWlubGlu ZS1jb25zdGFudC5qcyI+PC9zY3JpcHQ+Cis8c2NyaXB0IHNyYz0icmVzb3VyY2VzL2pzLXRlc3Qt cG9zdC5qcyI+PC9zY3JpcHQ+Cis8L2JvZHk+Cis8L2h0bWw+CkluZGV4OiBMYXlvdXRUZXN0cy9m YXN0L2pzL3NjcmlwdC10ZXN0cy9kZmctaW5saW5lLWNvbnN0YW50LmpzCj09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0t IExheW91dFRlc3RzL2Zhc3QvanMvc2NyaXB0LXRlc3RzL2RmZy1pbmxpbmUtY29uc3RhbnQuanMJ KHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9mYXN0L2pzL3NjcmlwdC10ZXN0cy9kZmctaW5s aW5lLWNvbnN0YW50LmpzCShyZXZpc2lvbiAwKQpAQCAtMCwwICsxLDIzIEBACitkZXNjcmlwdGlv bigKKyJUaGlzIHRlc3RzIHRoYXQgZnVuY3Rpb24gaW5saW5pbmcgaW4gdGhlIERGRyBKSVQgZG9l c24ndCBnZXQgY29uZnVzZWQgYnkgY29uc3RhbnRzIGJlaW5nIHJldXNlZCBiZXR3ZWVuIGlubGlu ZXIgYW5kIGlubGluZWUuIgorKTsKKworZnVuY3Rpb24gZm9vKGEsIGIpIHsKKyAgICBpZiAoYikK KyAgICAgICAgcmV0dXJuIGEgKyA0OworICAgIHJldHVybiBiICsgNTsKK30KKworZnVuY3Rpb24g YmFyKGEsIGIpIHsKKyAgICByZXR1cm4gZm9vKGEsIGIpICsgNTsKK30KKworZm9yICh2YXIgaSA9 IDA7IGkgPCAxMDAwOyArK2kpCisgICAgYmFyKGksIGkgKyAxKTsKKworc2hvdWxkQmUoImJhcig2 LCAwKSIsICIxMCIpOworc2hvdWxkQmUoImJhcig2LCAxKSIsICIxNSIpOworc2hvdWxkQmUoImJh cig2LCBmYWxzZSkiLCAiMTAiKTsKK3Nob3VsZEJlKCJiYXIoNiwgdHJ1ZSkiLCAiMTUiKTsKKwor dmFyIHN1Y2Nlc3NmdWxseVBhcnNlZCA9IHRydWU7Cg==