143727 – DFG register fillSpeculate*() functions should validate incoming spill format is compatible with requested fill format

Michael Saboff

Reported 2015-04-14 14:19:31 PDT

The fillSpeculate*() functions like fillSpeculateCell() should check that current spill format for the virtual register is compatible. If it isn't, then we should force and OSR exit. There are some spot checks in these functions, but we can use the return result from the call to AbstractInterpreter<>::filter() at the top of each function to perform the checks.

Radar WebKit Bug Importer

Comment 1 2015-04-14 14:20:23 PDT

<rdar://problem/20541588>

Michael Saboff

Comment 2 2015-04-14 16:06:03 PDT

Created attachment 250751 [details] Patch Appears neutral overall on performance with a few variations. Notable is that sunspider:string-base64 is reported to be 5% faster, but that could be noise. VMs tested: "Baseline" at /Volumes/Data/src/webkit.baseline/WebKitBuild/Release/jsc (r182806) "143727" at /Volumes/Data/src/webkit/WebKitBuild/Release/jsc (r182806) Collected 4 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. Baseline 143727 SunSpider: 3d-cube 6.6951+-0.2440 ? 6.7707+-0.3619 ? might be 1.0113x slower 3d-morph 7.8648+-0.1546 ? 7.8941+-0.1884 ? 3d-raytrace 8.5908+-0.2727 ? 8.7690+-0.1159 ? might be 1.0208x slower access-binary-trees 2.9194+-0.1589 ? 3.0078+-0.0492 ? might be 1.0303x slower access-fannkuch 7.9238+-0.2817 ? 8.0009+-0.2831 ? access-nbody 4.2805+-0.1937 ? 4.3350+-0.1628 ? might be 1.0127x slower access-nsieve 4.9242+-0.1054 4.7366+-0.3863 might be 1.0396x faster bitops-3bit-bits-in-byte 1.8680+-0.1505 1.7475+-0.1081 might be 1.0690x faster bitops-bits-in-byte 5.5717+-0.0880 ? 5.5908+-0.1382 ? bitops-bitwise-and 2.7184+-0.1716 ? 2.7535+-0.2034 ? might be 1.0129x slower bitops-nsieve-bits 5.0280+-0.2265 5.0228+-0.1496 controlflow-recursive 2.7967+-0.1881 ? 2.8382+-0.1525 ? might be 1.0148x slower crypto-aes 5.7721+-0.1032 5.6432+-0.3392 might be 1.0228x faster crypto-md5 3.3722+-0.1628 ? 3.3975+-0.0320 ? crypto-sha1 3.2269+-0.1154 3.2017+-0.2253 date-format-tofte 12.3098+-0.4333 12.0205+-1.0824 might be 1.0241x faster date-format-xparb 7.6066+-0.4051 7.4769+-0.2108 might be 1.0173x faster math-cordic 4.1104+-0.1710 ? 4.1535+-0.0542 ? might be 1.0105x slower math-partial-sums 9.0890+-0.0958 ? 9.1769+-0.2290 ? math-spectral-norm 2.9006+-0.0721 2.8894+-0.0845 regexp-dna 9.5349+-0.4512 ? 9.7933+-0.3790 ? might be 1.0271x slower string-base64 6.0475+-0.1177 ^ 5.7366+-0.1585 ^ definitely 1.0542x faster string-fasta 9.2220+-0.2754 ? 9.2825+-0.2248 ? string-tagcloud 13.5172+-0.4320 13.4819+-0.3588 string-unpack-code 26.7908+-1.3774 ? 26.8724+-0.5808 ? string-validate-input 6.6559+-0.1416 ? 6.9228+-0.7314 ? might be 1.0401x slower <arithmetic> 6.9745+-0.0493 ? 6.9814+-0.0841 ? might be 1.0010x slower Baseline 143727 Octane: encrypt 0.31574+-0.00897 ? 0.32287+-0.00940 ? might be 1.0226x slower decrypt 5.65659+-0.02111 5.65017+-0.02365 deltablue x2 0.26609+-0.00050 ? 0.26661+-0.00269 ? earley 0.81587+-0.00663 0.81095+-0.00692 boyer 10.33068+-0.09405 10.31035+-0.04910 navier-stokes x2 6.41370+-0.06852 6.39819+-0.01887 raytrace x2 1.71317+-0.06833 ? 1.76390+-0.09317 ? might be 1.0296x slower richards x2 0.16647+-0.00480 ? 0.16811+-0.00521 ? splay x2 0.52492+-0.01753 0.52429+-0.00490 regexp x2 44.81353+-0.17889 44.45178+-0.34867 pdfjs x2 62.55568+-0.39999 ? 62.61614+-0.58267 ? mandreel x2 71.69652+-0.21810 71.60592+-1.53774 gbemu x2 56.09795+-0.49104 ? 56.10090+-0.39067 ? closure 0.76699+-0.00535 ? 0.77035+-0.00192 ? jquery 9.61660+-0.04316 9.61625+-0.05188 box2d x2 16.85752+-0.13343 ? 17.01420+-0.28640 ? zlib x2 526.71844+-42.07698 525.24186+-38.40360 typescript x2 1104.50177+-13.87379 1091.16248+-17.64856 might be 1.0122x faster <geometric> 9.22295+-0.04102 ? 9.24270+-0.07206 ? might be 1.0021x slower Baseline 143727 Kraken: ai-astar 469.689+-1.068 466.823+-12.643 audio-beat-detection 153.569+-4.686 152.523+-4.125 audio-dft 206.638+-8.557 203.708+-3.627 might be 1.0144x faster audio-fft 113.852+-8.211 112.517+-7.424 might be 1.0119x faster audio-oscillator 265.707+-4.698 263.981+-1.014 imaging-darkroom 139.822+-0.684 ? 140.755+-3.112 ? imaging-desaturate 95.119+-0.455 94.625+-0.463 imaging-gaussian-blur 156.583+-2.880 ? 156.854+-0.417 ? json-parse-financial 66.596+-1.528 65.127+-0.816 might be 1.0225x faster json-stringify-tinderbox 77.678+-1.587 76.882+-1.150 might be 1.0104x faster stanford-crypto-aes 88.093+-2.553 86.838+-5.296 might be 1.0144x faster stanford-crypto-ccm 66.703+-1.397 ? 72.410+-15.095 ? might be 1.0856x slower stanford-crypto-pbkdf2 225.513+-1.096 ? 226.186+-1.652 ? stanford-crypto-sha256-iterative 73.553+-1.066 73.502+-1.084 <arithmetic> 157.080+-1.402 156.624+-1.256 might be 1.0029x faster Baseline 143727 JSRegress: abs-boolean 3.6832+-0.0995 3.6381+-0.1115 might be 1.0124x faster adapt-to-double-divide 18.0697+-0.5780 17.6843+-0.1364 might be 1.0218x faster aliased-arguments-getbyval 1.6911+-0.0398 ? 1.7546+-0.1153 ? might be 1.0375x slower allocate-big-object 3.6746+-0.4463 ? 3.7822+-0.3241 ? might be 1.0293x slower arguments-named-and-reflective 14.6545+-0.4213 ? 14.7844+-0.2388 ? arguments-out-of-bounds 18.1047+-1.3071 17.7922+-0.4487 might be 1.0176x faster arguments-strict-mode 13.8130+-0.4097 13.3796+-0.4407 might be 1.0324x faster arguments 12.0327+-0.6359 11.8663+-0.2127 might be 1.0140x faster arity-mismatch-inlining 1.1000+-0.1074 ? 1.1740+-0.1148 ? might be 1.0673x slower array-access-polymorphic-structure 9.2925+-0.1003 9.1429+-0.4231 might be 1.0164x faster array-nonarray-polymorhpic-access 47.1545+-2.7377 47.1072+-4.0002 array-prototype-every 118.9837+-2.4395 ? 119.5703+-4.1424 ? array-prototype-forEach 117.3148+-2.9472 ? 117.8513+-1.2471 ? array-prototype-map 129.7469+-3.1162 ? 132.8765+-3.3928 ? might be 1.0241x slower array-prototype-some 119.7091+-3.7419 118.3590+-2.1894 might be 1.0114x faster array-splice-contiguous 57.3076+-1.6176 ? 58.9095+-3.2983 ? might be 1.0280x slower array-with-double-add 5.6701+-0.5087 5.6560+-0.1762 array-with-double-increment 4.0297+-0.1468 ? 4.0671+-0.1101 ? array-with-double-mul-add 7.0412+-0.0785 ? 7.2471+-0.2108 ? might be 1.0292x slower array-with-double-sum 4.4521+-0.1456 4.4016+-0.1464 might be 1.0115x faster array-with-int32-add-sub 9.5637+-0.3037 9.5309+-0.1467 array-with-int32-or-double-sum 4.4882+-0.1748 ? 4.5848+-0.6026 ? might be 1.0215x slower ArrayBuffer-DataView-alloc-large-long-lived 45.1507+-1.1367 ? 46.2133+-1.8932 ? might be 1.0235x slower ArrayBuffer-DataView-alloc-long-lived 18.5651+-0.1542 18.3342+-0.3606 might be 1.0126x faster ArrayBuffer-Int32Array-byteOffset 5.0153+-0.2657 4.9817+-0.3182 ArrayBuffer-Int8Array-alloc-large-long-lived 48.3492+-2.9078 47.4785+-1.3865 might be 1.0183x faster ArrayBuffer-Int8Array-alloc-long-lived-buffer 31.5641+-0.4340 31.3622+-0.7006 ArrayBuffer-Int8Array-alloc-long-lived 18.0937+-0.6177 17.8484+-0.3441 might be 1.0137x faster ArrayBuffer-Int8Array-alloc 15.3066+-0.1885 14.8665+-0.4276 might be 1.0296x faster asmjs_bool_bug 8.8420+-0.1137 8.7374+-0.2682 might be 1.0120x faster assign-custom-setter-polymorphic 4.4225+-0.1138 4.4219+-0.2467 assign-custom-setter 5.6971+-0.2169 ? 5.8922+-0.2404 ? might be 1.0342x slower basic-set 12.6847+-0.5399 12.1803+-0.9787 might be 1.0414x faster big-int-mul 5.8146+-0.1920 ? 5.9888+-0.4068 ? might be 1.0299x slower boolean-test 4.2387+-0.0762 ? 4.2722+-0.1509 ? branch-fold 4.5798+-0.0810 ? 4.6478+-0.1579 ? might be 1.0148x slower by-val-generic 10.8729+-0.4481 10.8701+-0.2971 call-spread-apply 40.1239+-0.4979 ? 40.8217+-1.2363 ? might be 1.0174x slower call-spread-call 33.3540+-0.9923 ! 35.4169+-0.9084 ! definitely 1.0618x slower captured-assignments 0.5679+-0.0167 ? 0.6376+-0.1101 ? might be 1.1227x slower cast-int-to-double 8.0920+-0.2307 8.0008+-0.2433 might be 1.0114x faster cell-argument 9.9703+-0.4215 9.9406+-0.2009 cfg-simplify 3.6942+-0.1573 ? 3.7335+-0.0345 ? might be 1.0106x slower chain-getter-access 12.8613+-0.4463 ? 12.8750+-0.1904 ? cmpeq-obj-to-obj-other 13.0319+-0.5420 12.7548+-0.4316 might be 1.0217x faster constant-test 7.6102+-0.2010 ? 7.8897+-0.5390 ? might be 1.0367x slower create-lots-of-functions 32.1183+-1.1508 31.6409+-0.3933 might be 1.0151x faster DataView-custom-properties 53.5148+-2.3179 53.0017+-2.7890 deconstructing-parameters-overridden-by-function 0.6980+-0.1974 ? 0.6983+-0.1646 ? delay-tear-off-arguments-strictmode 18.7353+-0.1462 ? 18.8301+-0.0972 ? deltablue-varargs 256.8453+-2.0385 256.1028+-6.0280 destructuring-arguments 21.9749+-0.2221 ? 22.3100+-1.0657 ? might be 1.0152x slower destructuring-swap 7.6900+-0.1418 7.6169+-0.1377 direct-arguments-getbyval 1.4437+-0.0987 ? 1.6404+-0.1972 ? might be 1.1363x slower div-boolean-double 5.4897+-0.1071 5.4175+-0.1312 might be 1.0133x faster div-boolean 9.7052+-0.1638 ? 9.8074+-0.0878 ? might be 1.0105x slower double-get-by-val-out-of-bounds 5.7213+-0.2227 ! 6.1357+-0.0678 ! definitely 1.0724x slower double-pollution-getbyval 9.6314+-0.2787 ? 9.7254+-0.1475 ? double-pollution-putbyoffset 5.4895+-0.2774 ? 5.6146+-0.2425 ? might be 1.0228x slower double-to-int32-typed-array-no-inline 2.9285+-0.1471 ? 2.9311+-0.1301 ? double-to-int32-typed-array 2.5053+-0.1455 2.4415+-0.2163 might be 1.0261x faster double-to-uint32-typed-array-no-inline 2.9061+-0.1481 ? 2.9935+-0.2730 ? might be 1.0301x slower double-to-uint32-typed-array 2.5685+-0.2304 ? 2.5690+-0.2103 ? elidable-new-object-dag 54.9875+-1.2415 ? 55.0366+-1.4285 ? elidable-new-object-roflcopter 60.2272+-0.7276 ? 61.3841+-3.0177 ? might be 1.0192x slower elidable-new-object-then-call 52.9955+-3.8982 50.5580+-1.3398 might be 1.0482x faster elidable-new-object-tree 64.0610+-1.0193 63.2074+-2.0112 might be 1.0135x faster empty-string-plus-int 7.5704+-0.2562 7.4539+-0.1685 might be 1.0156x faster emscripten-cube2hash 45.3801+-0.4851 44.7535+-0.8213 might be 1.0140x faster exit-length-on-plain-object 18.7822+-0.2135 18.4063+-0.5068 might be 1.0204x faster external-arguments-getbyval 1.6140+-0.1814 1.5614+-0.2621 might be 1.0337x faster external-arguments-putbyval 3.1501+-0.0992 ? 3.2399+-0.2782 ? might be 1.0285x slower fixed-typed-array-storage-var-index 1.5746+-0.1810 1.4970+-0.1606 might be 1.0518x faster fixed-typed-array-storage 1.2123+-0.1407 ? 1.2866+-0.1893 ? might be 1.0613x slower Float32Array-matrix-mult 5.8850+-0.3038 ? 5.9540+-0.2647 ? might be 1.0117x slower Float32Array-to-Float64Array-set 75.3255+-3.7368 ? 76.0145+-5.5955 ? Float64Array-alloc-long-lived 93.7083+-0.5569 ? 94.2015+-0.3238 ? Float64Array-to-Int16Array-set 101.1656+-0.3744 ? 105.6171+-4.6733 ? might be 1.0440x slower fold-double-to-int 20.5146+-0.4067 ! 21.6608+-0.6138 ! definitely 1.0559x slower fold-get-by-id-to-multi-get-by-offset-rare-int 10.3411+-0.5078 ? 10.4304+-0.4086 ? fold-get-by-id-to-multi-get-by-offset 9.0780+-0.4407 ? 9.1000+-0.2010 ? fold-multi-get-by-offset-to-get-by-offset 8.6281+-1.2641 7.8265+-1.1989 might be 1.1024x faster fold-multi-get-by-offset-to-poly-get-by-offset 8.1868+-0.8343 ? 8.4438+-0.8462 ? might be 1.0314x slower fold-multi-put-by-offset-to-poly-put-by-offset 7.3401+-0.7200 ? 7.4749+-0.7009 ? might be 1.0184x slower fold-multi-put-by-offset-to-put-by-offset 6.5112+-0.7173 6.2652+-0.7588 might be 1.0393x faster fold-multi-put-by-offset-to-replace-or-transition-put-by-offset 12.8248+-1.0691 12.3017+-1.2229 might be 1.0425x faster fold-put-by-id-to-multi-put-by-offset 8.6852+-1.0315 8.6730+-1.0906 fold-put-structure 6.2204+-0.7618 6.1633+-0.1120 for-of-iterate-array-entries 6.3524+-0.1138 6.2445+-0.0580 might be 1.0173x faster for-of-iterate-array-keys 5.1888+-0.1351 5.1508+-0.2326 for-of-iterate-array-values 5.0550+-0.1976 ? 5.0583+-0.5086 ? fround 21.7065+-0.5407 ? 22.2723+-1.2071 ? might be 1.0261x slower ftl-library-inlining-dataview 90.9430+-1.9200 90.2015+-3.0495 ftl-library-inlining 112.4117+-15.6981 ? 114.6644+-4.8106 ? might be 1.0200x slower function-dot-apply 2.8652+-0.1710 ? 2.8729+-0.2057 ? function-test 4.8790+-0.0822 ? 4.9973+-0.4062 ? might be 1.0242x slower function-with-eval 133.1198+-1.1478 ? 134.3550+-3.0813 ? gcse-poly-get-less-obvious 25.0759+-0.9704 25.0520+-0.8056 gcse-poly-get 24.6019+-0.0802 ? 24.7795+-0.4499 ? gcse 6.4210+-0.2558 ? 6.5438+-0.1969 ? might be 1.0191x slower get-by-id-bimorphic-check-structure-elimination-simple 3.2976+-0.1312 ? 3.3403+-0.1473 ? might be 1.0129x slower get-by-id-bimorphic-check-structure-elimination 8.3458+-0.1079 8.3195+-0.2952 get-by-id-chain-from-try-block 11.3985+-0.0976 11.3826+-0.1749 get-by-id-check-structure-elimination 7.3347+-0.2285 ? 7.6223+-0.1275 ? might be 1.0392x slower get-by-id-proto-or-self 22.5237+-1.3097 22.1998+-0.9897 might be 1.0146x faster get-by-id-quadmorphic-check-structure-elimination-simple 3.9831+-0.0616 ? 4.0162+-0.1143 ? get-by-id-self-or-proto 22.6560+-1.5152 22.4652+-1.4889 get-by-val-out-of-bounds 5.7322+-0.3710 ? 5.9993+-0.4309 ? might be 1.0466x slower get_callee_monomorphic 5.1404+-0.4548 ? 5.1851+-0.2689 ? get_callee_polymorphic 4.3513+-0.3403 4.2195+-0.1752 might be 1.0312x faster getter-no-activation 5.7789+-0.1406 5.7550+-0.0979 getter-richards 128.2052+-2.5427 ? 128.2664+-5.1315 ? getter 7.1855+-0.0562 ? 7.2230+-0.1067 ? global-var-const-infer-fire-from-opt 1.2318+-0.1453 ? 1.3099+-0.0891 ? might be 1.0635x slower global-var-const-infer 1.2999+-0.1511 1.1618+-0.2845 might be 1.1188x faster HashMap-put-get-iterate-keys 34.7606+-0.5577 34.3834+-0.6389 might be 1.0110x faster HashMap-put-get-iterate 34.8658+-1.4332 ? 35.1343+-0.7995 ? HashMap-string-put-get-iterate 35.4775+-2.0143 ? 35.6600+-1.1483 ? hoist-make-rope 15.2250+-1.7818 13.4758+-0.7035 might be 1.1298x faster hoist-poly-check-structure-effectful-loop 6.7720+-0.2944 ? 7.0695+-0.5495 ? might be 1.0439x slower hoist-poly-check-structure 4.8701+-0.0936 ? 4.9697+-0.0656 ? might be 1.0204x slower imul-double-only 10.0265+-0.4813 ? 10.1379+-0.8136 ? might be 1.0111x slower imul-int-only 11.7772+-1.0445 ? 12.4283+-0.7395 ? might be 1.0553x slower imul-mixed 9.7637+-0.7230 ? 9.9225+-0.9500 ? might be 1.0163x slower in-four-cases 23.7491+-0.3667 ? 23.8334+-0.4580 ? in-one-case-false 12.5684+-0.2618 12.5215+-0.1952 in-one-case-true 12.4813+-0.2361 ? 12.4912+-0.1845 ? in-two-cases 12.9849+-0.1977 12.9382+-0.1264 indexed-properties-in-objects 3.7418+-0.0593 3.6436+-0.1166 might be 1.0269x faster infer-closure-const-then-mov-no-inline 4.3319+-0.1404 ? 4.3530+-0.0263 ? infer-closure-const-then-mov 24.9031+-1.0847 ? 24.9205+-1.1894 ? infer-closure-const-then-put-to-scope-no-inline 17.1152+-0.1801 16.9490+-0.1379 infer-closure-const-then-put-to-scope 30.0213+-1.0400 ? 30.0497+-0.2930 ? infer-closure-const-then-reenter-no-inline 80.7958+-0.2172 ? 81.3198+-2.1794 ? infer-closure-const-then-reenter 30.1017+-0.6374 ? 30.3847+-0.6250 ? infer-constant-global-property 37.3363+-0.7653 ? 37.4379+-1.0401 ? infer-constant-property 3.3432+-0.1572 3.3255+-0.1098 infer-one-time-closure-ten-vars 15.0338+-0.4418 14.7272+-0.3288 might be 1.0208x faster infer-one-time-closure-two-vars 14.3091+-0.4445 14.1312+-0.4135 might be 1.0126x faster infer-one-time-closure 14.2449+-0.4014 13.9420+-0.4348 might be 1.0217x faster infer-one-time-deep-closure 24.8617+-0.4569 24.5095+-0.3968 might be 1.0144x faster inline-arguments-access 5.6998+-0.2500 ? 5.8360+-0.2594 ? might be 1.0239x slower inline-arguments-aliased-access 5.7664+-0.1101 ? 5.8210+-0.2753 ? inline-arguments-local-escape 5.9590+-0.5793 5.9349+-0.2441 inline-get-scoped-var 5.7143+-0.1840 5.7037+-0.1525 inlined-put-by-id-transition 14.8660+-0.4295 14.5943+-0.4865 might be 1.0186x faster int-or-other-abs-then-get-by-val 6.8052+-0.3257 ? 6.8676+-0.3811 ? int-or-other-abs-zero-then-get-by-val 27.2778+-0.9229 ? 27.6646+-0.6937 ? might be 1.0142x slower int-or-other-add-then-get-by-val 6.0018+-0.2821 5.8478+-0.1617 might be 1.0263x faster int-or-other-add 7.8127+-0.2230 ? 7.9340+-0.1393 ? might be 1.0155x slower int-or-other-div-then-get-by-val 5.1697+-0.1813 ? 5.2258+-0.0554 ? might be 1.0108x slower int-or-other-max-then-get-by-val 6.5150+-0.2566 6.4810+-0.7896 int-or-other-min-then-get-by-val 5.2440+-0.2388 ? 5.4130+-0.2045 ? might be 1.0322x slower int-or-other-mod-then-get-by-val 5.0398+-0.2261 ? 5.1074+-0.1664 ? might be 1.0134x slower int-or-other-mul-then-get-by-val 4.9562+-0.2354 ? 5.0333+-0.1646 ? might be 1.0156x slower int-or-other-neg-then-get-by-val 6.2012+-0.1558 6.1985+-0.1578 int-or-other-neg-zero-then-get-by-val 26.9248+-0.3669 ? 27.2697+-0.4654 ? might be 1.0128x slower int-or-other-sub-then-get-by-val 5.9010+-0.1804 5.8435+-0.2201 int-or-other-sub 4.7698+-0.1447 ? 4.8132+-0.1799 ? int-overflow-local 5.8281+-0.1794 5.7505+-0.1700 might be 1.0135x faster Int16Array-alloc-long-lived 63.4922+-0.8037 63.3909+-0.9284 Int16Array-bubble-sort-with-byteLength 37.9282+-0.8458 37.2159+-0.2734 might be 1.0191x faster Int16Array-bubble-sort 37.3541+-0.2119 ? 37.7584+-0.8336 ? might be 1.0108x slower Int16Array-load-int-mul 2.0120+-0.1855 1.9370+-0.0846 might be 1.0387x faster Int16Array-to-Int32Array-set 73.5837+-2.1801 ? 73.6262+-0.4041 ? Int32Array-alloc-large 35.0888+-1.4214 33.9891+-1.8661 might be 1.0324x faster Int32Array-alloc-long-lived 72.5142+-0.5010 ? 72.7043+-1.0689 ? Int32Array-alloc 4.5682+-0.5480 4.4721+-0.0405 might be 1.0215x faster Int32Array-Int8Array-view-alloc 9.8177+-0.2786 ? 10.0126+-0.4405 ? might be 1.0199x slower int52-spill 8.9645+-0.2154 ? 9.2405+-0.4667 ? might be 1.0308x slower Int8Array-alloc-long-lived 57.9207+-1.6310 ? 58.1138+-1.3799 ? Int8Array-load-with-byteLength 4.7917+-0.1533 4.7891+-0.1664 Int8Array-load 4.7863+-0.0970 4.7473+-0.1287 integer-divide 14.0909+-0.1472 14.0120+-0.3021 integer-modulo 2.7492+-0.2130 ? 2.7623+-0.1777 ? large-int-captured 8.0454+-0.2286 ? 8.0660+-0.1610 ? large-int-neg 20.7856+-1.0879 20.4080+-0.1134 might be 1.0185x faster large-int 18.2613+-0.5043 18.2182+-0.4205 logical-not 6.0845+-0.3834 ? 6.3435+-0.8566 ? might be 1.0426x slower lots-of-fields 18.7712+-0.1689 18.6628+-0.3172 make-indexed-storage 4.2336+-0.4875 ? 4.3915+-0.1334 ? might be 1.0373x slower make-rope-cse 6.1550+-0.0378 6.1533+-0.0855 marsaglia-larger-ints 55.5128+-0.7541 ? 55.5353+-0.5884 ? marsaglia-osr-entry 28.1095+-0.6591 28.0898+-0.2063 max-boolean 3.3043+-0.0857 3.2988+-0.1554 method-on-number 24.7257+-0.5517 ? 25.6224+-0.9194 ? might be 1.0363x slower min-boolean 3.2396+-0.1621 3.2073+-0.0767 might be 1.0101x faster minus-boolean-double 4.1873+-0.0839 ? 4.2345+-0.1323 ? might be 1.0113x slower minus-boolean 3.2734+-0.0397 3.2038+-0.0992 might be 1.0217x faster misc-strict-eq 52.3315+-1.3247 51.2353+-1.1387 might be 1.0214x faster mod-boolean-double 11.7344+-0.0596 ? 11.9423+-0.7332 ? might be 1.0177x slower mod-boolean 8.7850+-0.2016 ? 8.8699+-0.1313 ? mul-boolean-double 4.8585+-0.0999 4.8234+-0.0935 mul-boolean 3.4742+-0.1541 3.4465+-0.1707 neg-boolean 4.3657+-0.0917 4.2991+-0.1350 might be 1.0155x faster negative-zero-divide 0.5114+-0.0992 ? 0.5437+-0.1115 ? might be 1.0631x slower negative-zero-modulo 0.5188+-0.1145 ? 0.5245+-0.0997 ? might be 1.0109x slower negative-zero-negate 0.4936+-0.0975 0.4484+-0.0883 might be 1.1007x faster nested-function-parsing 58.8552+-0.6856 ? 59.1370+-0.8599 ? new-array-buffer-dead 3.8118+-0.1075 ? 3.8547+-0.2265 ? might be 1.0113x slower new-array-buffer-push 8.9529+-0.3644 ? 9.2654+-0.2111 ? might be 1.0349x slower new-array-dead 14.9910+-1.0760 ? 15.2504+-1.8320 ? might be 1.0173x slower new-array-push 5.2435+-0.3325 ? 5.5261+-0.3861 ? might be 1.0539x slower no-inline-constructor 152.0026+-1.5529 151.6688+-0.9927 number-test 4.2701+-0.1210 4.2093+-0.1652 might be 1.0145x faster object-closure-call 7.5890+-0.1450 7.5555+-0.3864 object-test 4.4857+-0.1182 ? 4.4963+-0.0366 ? obvious-sink-pathology-taken 170.1875+-2.1206 ? 171.3854+-5.1715 ? obvious-sink-pathology 162.5636+-1.2617 ? 162.7296+-4.1510 ? obviously-elidable-new-object 45.8965+-5.0232 ? 46.9008+-4.8869 ? might be 1.0219x slower plus-boolean-arith 3.2112+-0.0738 ? 3.2177+-0.1405 ? plus-boolean-double 4.2200+-0.1809 4.2009+-0.1241 plus-boolean 3.1868+-0.1490 3.1653+-0.1357 poly-chain-access-different-prototypes-simple 3.8439+-0.1499 ? 3.8771+-0.0375 ? poly-chain-access-different-prototypes 3.2590+-0.1088 ? 3.3239+-0.0981 ? might be 1.0199x slower poly-chain-access-simpler 3.8117+-0.2083 ? 3.8688+-0.0811 ? might be 1.0150x slower poly-chain-access 3.0971+-0.1617 3.0480+-0.1245 might be 1.0161x faster poly-stricteq 68.2525+-1.2170 ? 69.3685+-1.6732 ? might be 1.0164x slower polymorphic-array-call 1.9932+-0.4922 ? 2.0095+-0.3491 ? polymorphic-get-by-id 4.0605+-0.1390 ? 4.3147+-0.7370 ? might be 1.0626x slower polymorphic-put-by-id 38.1545+-6.1494 ? 40.0676+-5.4364 ? might be 1.0501x slower polymorphic-structure 21.7722+-0.2217 21.6404+-0.3947 polyvariant-monomorphic-get-by-id 12.2614+-0.2128 ? 12.2868+-0.0974 ? proto-getter-access 13.1175+-0.4099 12.9016+-0.3751 might be 1.0167x faster put-by-id-replace-and-transition 11.8304+-0.9738 11.5258+-0.5703 might be 1.0264x faster put-by-id-slightly-polymorphic 3.5153+-0.0592 ? 3.5557+-0.1171 ? might be 1.0115x slower put-by-id 17.8141+-0.5456 ? 18.0328+-0.2828 ? might be 1.0123x slower put-by-val-direct 0.8193+-0.1323 ? 0.9025+-0.1340 ? might be 1.1015x slower put-by-val-large-index-blank-indexing-type 8.1998+-0.5175 ? 8.5166+-0.6385 ? might be 1.0386x slower put-by-val-machine-int 3.3901+-0.1955 3.3740+-0.1456 rare-osr-exit-on-local 18.4877+-0.3705 18.2715+-0.3339 might be 1.0118x faster register-pressure-from-osr 26.4394+-0.7365 ? 26.7149+-0.7900 ? might be 1.0104x slower setter 6.4805+-0.1774 ? 6.5632+-0.1305 ? might be 1.0128x slower simple-activation-demo 30.9162+-0.4224 ? 31.1877+-0.9158 ? simple-getter-access 18.0295+-0.5032 ? 18.0688+-0.3607 ? simple-poly-call-nested 10.0445+-0.1911 10.0123+-0.1814 simple-poly-call 1.6366+-0.0741 ? 1.6670+-0.0938 ? might be 1.0186x slower sin-boolean 24.2407+-4.3232 ? 25.8521+-4.1658 ? might be 1.0665x slower singleton-scope 86.7270+-0.1130 86.7228+-0.2818 sinkable-new-object-dag 93.1264+-1.3167 ? 93.3135+-1.0346 ? sinkable-new-object-taken 69.6967+-0.6111 ? 69.9041+-4.3163 ? sinkable-new-object 52.4207+-0.8008 ? 52.7608+-1.2507 ? slow-array-profile-convergence 3.8120+-0.2532 ? 3.9662+-0.2637 ? might be 1.0405x slower slow-convergence 3.7896+-0.4217 3.7825+-0.3316 sorting-benchmark 28.4336+-1.2762 ? 28.5738+-1.2388 ? sparse-conditional 1.6603+-0.0725 1.6085+-0.0565 might be 1.0322x faster splice-to-remove 21.7463+-0.4711 21.4300+-0.3675 might be 1.0148x faster string-char-code-at 20.0610+-0.3219 ? 20.1116+-0.4718 ? string-concat-object 3.1594+-0.4018 ? 3.1693+-0.2902 ? string-concat-pair-object 3.0463+-0.2800 2.9934+-0.2059 might be 1.0177x faster string-concat-pair-simple 16.0303+-0.3035 16.0302+-0.5618 string-concat-simple 16.6632+-0.1979 ^ 16.1061+-0.2265 ^ definitely 1.0346x faster string-cons-repeat 10.6597+-0.5696 10.4426+-0.1695 might be 1.0208x faster string-cons-tower 10.4323+-0.1898 ? 10.4560+-0.1287 ? string-equality 21.8538+-0.3482 ? 21.8724+-0.6109 ? string-get-by-val-big-char 9.8223+-0.2602 ? 9.9375+-0.6576 ? might be 1.0117x slower string-get-by-val-out-of-bounds-insane 6.2908+-1.8982 5.4880+-0.1652 might be 1.1463x faster string-get-by-val-out-of-bounds 6.8745+-0.0801 6.7972+-0.1949 might be 1.0114x faster string-get-by-val 4.7074+-0.0716 4.6479+-0.0384 might be 1.0128x faster string-hash 2.7343+-0.1375 ? 2.8042+-0.0717 ? might be 1.0256x slower string-long-ident-equality 17.8735+-0.2064 ? 18.2411+-1.2365 ? might be 1.0206x slower string-out-of-bounds 17.3430+-0.5082 ? 17.4003+-0.5622 ? string-repeat-arith 42.3940+-0.6312 ^ 41.5699+-0.0850 ^ definitely 1.0198x faster string-sub 81.5804+-2.5071 ? 81.7614+-0.9854 ? string-test 4.2355+-0.1774 ? 4.2424+-0.1083 ? string-var-equality 43.4485+-0.2077 43.3173+-0.2606 structure-hoist-over-transitions 3.3890+-0.0881 3.3539+-0.0909 might be 1.0104x faster substring-concat-weird 53.7742+-0.5426 53.5283+-0.4597 substring-concat 56.4106+-1.0145 56.1638+-1.5924 substring 61.8919+-0.5330 ? 62.4438+-0.7624 ? switch-char-constant 3.3887+-0.0741 3.3887+-0.1296 switch-char 7.8750+-0.1923 ? 7.9001+-0.0579 ? switch-constant 10.8351+-0.2999 ? 10.9232+-0.4263 ? switch-string-basic-big-var 25.1897+-4.4325 ? 25.9348+-1.4076 ? might be 1.0296x slower switch-string-basic-big 18.7584+-1.4635 18.6484+-1.7771 switch-string-basic-var 28.6392+-3.2773 28.3292+-3.5770 might be 1.0109x faster switch-string-basic 21.1318+-2.5262 20.3491+-2.4462 might be 1.0385x faster switch-string-big-length-tower-var 24.9944+-0.1527 24.9183+-0.1721 switch-string-length-tower-var 19.6554+-0.2339 ? 19.8168+-0.3263 ? switch-string-length-tower 14.0645+-0.3312 13.8892+-0.2916 might be 1.0126x faster switch-string-short 13.9894+-0.3528 ? 14.0222+-0.2074 ? switch 14.5886+-0.6231 ? 14.6350+-0.5577 ? tear-off-arguments-simple 4.4011+-0.1511 4.2543+-0.4407 might be 1.0345x faster tear-off-arguments 6.1454+-0.3866 5.9039+-0.4522 might be 1.0409x faster temporal-structure 16.9724+-0.1662 ? 17.0052+-0.3276 ? to-int32-boolean 21.2393+-0.3647 21.1669+-0.4563 try-catch-get-by-val-cloned-arguments 18.9767+-0.2715 18.8947+-0.1481 try-catch-get-by-val-direct-arguments 8.2734+-0.5865 ? 8.5180+-0.3788 ? might be 1.0296x slower try-catch-get-by-val-scoped-arguments 10.1182+-0.1332 ? 10.2683+-0.1584 ? might be 1.0148x slower undefined-property-access 432.1925+-10.4127 425.7377+-3.3271 might be 1.0152x faster undefined-test 4.3992+-0.0908 ? 4.5065+-0.1703 ? might be 1.0244x slower unprofiled-licm 27.0419+-0.5430 ? 27.1466+-0.8547 ? varargs-call 18.9021+-0.2435 ? 19.0057+-0.3980 ? varargs-construct-inline 29.5133+-0.4769 ? 30.1278+-0.2342 ? might be 1.0208x slower varargs-construct 42.3535+-1.6087 ? 42.4584+-1.8058 ? varargs-inline 11.9781+-0.3236 ? 12.0016+-0.3649 ? varargs-strict-mode 14.0560+-0.1409 13.8088+-0.3871 might be 1.0179x faster varargs 13.9841+-0.2656 ? 14.0099+-0.3455 ? weird-inlining-const-prop 2.8838+-0.1848 2.8153+-0.2822 might be 1.0243x faster <geometric> 11.2668+-0.0296 ? 11.2949+-0.0279 ? might be 1.0025x slower Baseline 143727 AsmBench: bigfib.cpp 656.0960+-2.4786 ? 660.0327+-4.2235 ? cray.c 622.4896+-2.7524 ? 624.2029+-5.1927 ? dry.c 640.9047+-15.1779 ? 649.0897+-12.7497 ? might be 1.0128x slower FloatMM.c 957.8546+-1.9209 956.5390+-1.3765 gcc-loops.cpp 5874.8479+-9.8641 ? 5877.9625+-20.9874 ? n-body.c 1672.8197+-4.6147 1669.5305+-1.6165 Quicksort.c 577.2261+-14.3923 ? 587.1200+-4.7258 ? might be 1.0171x slower stepanov_container.cpp 4839.1768+-19.6526 ? 4856.3395+-19.8868 ? Towers.c 376.1375+-2.3617 ? 377.3905+-1.6067 ? <geometric> 1111.4566+-5.2123 ? 1116.7242+-3.6366 ? might be 1.0047x slower Baseline 143727 CompressionBench: huffman 503.9668+-3.8761 502.3233+-5.0106 arithmetic-simple 527.0562+-2.6899 525.1830+-0.6031 arithmetic-precise 400.9268+-6.8895 400.5740+-4.6006 arithmetic-complex-precise 399.0178+-0.4564 397.0725+-3.6927 arithmetic-precise-order-0 580.8662+-9.1488 576.5144+-10.1525 arithmetic-precise-order-1 429.0465+-21.3222 419.7703+-2.6844 might be 1.0221x faster arithmetic-precise-order-2 472.9880+-4.1944 ? 473.1099+-1.3579 ? arithmetic-simple-order-1 528.6113+-3.9115 527.4030+-2.6000 arithmetic-simple-order-2 591.4443+-3.4363 590.6199+-5.4291 lz-string 418.2889+-3.5944 ? 419.0020+-8.3839 ? <geometric> 480.4041+-1.6238 478.3243+-1.1365 might be 1.0043x faster Baseline 143727 Geomean of preferred means: <scaled-result> 62.7044+-0.0601 ? 62.7367+-0.0550 ? might be 1.0005x slower

Geoffrey Garen

Comment 3 2015-04-14 16:52:32 PDT

Comment on attachment 250751 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=250751&action=review > Source/JavaScriptCore/dfg/DFGSpeculativeJIT32_64.cpp:849 > + if (m_interpreter.filter(value, SpecInt32) == Contradiction) { Can you merge this block with the constant check below? If the edge is constant, and not int32, the abstract interpreter should know it, and so we shouldn't need a separate constant check. > Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:821 > + if (m_interpreter.filter(value, SpecInt32) == Contradiction) { Same question about constants. > Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:972 > + if (m_interpreter.filter(value, SpecMachineInt) == Contradiction) { Same question about constants. > Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:1106 > + if (m_interpreter.filter(value, SpecCell) == Contradiction) { Same question about constants.

Michael Saboff

Comment 4 2015-04-14 17:37:52 PDT

Created attachment 250766 [details] Updated Patch As suggested, I eliminated all other spill/fill consistency checks as the AbstractInterpreter<>::filter() covers them.

Geoffrey Garen

Comment 5 2015-04-14 17:44:08 PDT

Comment on attachment 250766 [details] Updated Patch r=me

Michael Saboff

Comment 6 2015-04-14 17:50:02 PDT

Committed r182827: <http://trac.webkit.org/changeset/182827>

Mark Lam

Comment 7 2015-05-14 15:54:46 PDT

*** Bug 145023 has been marked as a duplicate of this bug. ***

Attachments
Patch (8.76 KB, patch) 2015-04-14 16:06 PDT, Michael Saboff	no flags	Details Formatted Diff Diff
Updated Patch (13.61 KB, patch) 2015-04-14 17:37 PDT, Michael Saboff	ggaren: review+	Details Formatted Diff Diff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.