56823 2011-03-22 04:30:37 -0700 REGRESSION: Allocation of memory is slow when building a 2GB tree 2011-04-07 19:59:49 -0700 1 1 1 Unclassified WebKit JavaScriptCore 528+ (Nightly build) Mac OS X 10.6 ASSIGNED InRadar P2 Normal --- 1 Tam ggaren ap edwardjsabol ggaren oldest_to_newest 371327 0 Tam 2011-03-22 04:30:37 -0700 1. Open the attached html using webkit r78249 or later. Try the same for: * Safari 5.0 * WebKit trunk r78794 or before * Any version of Chrome Notice how it is 10x slower than all others, where a pass (after the first dozen passes) takes >1000ms, rather than <100ms as other browsers do. 371435 1 ap 2011-03-22 10:03:09 -0700 There is no test case attached at the moment, please attach it. 371451 2 86471 Tam 2011-03-22 10:19:14 -0700 Created attachment 86471 Test case Really attach the test case. 371469 3 ap 2011-03-22 10:39:11 -0700 <rdar://problem/9169336> 373745 4 ggaren 2011-03-25 15:09:02 -0700 Initial times are even (between Safari 5.0.3 and TOT), but TOT starts to skyrocket upward from there. Safari 5.0.3 does have one or two very long pauses on the order of TOT, but TOT has very long pauses during each iteration. 373747 5 ggaren 2011-03-25 15:10:20 -0700 Safari 5.0.3 beginning: [A] Last pass took 43ms [A] Last pass took 48ms [A] Last pass took 65ms [A] Last pass took 30ms [A] Last pass took 31ms [A] Last pass took 101ms TOT beginning: [A] Last pass took 52ms [A] Last pass took 58ms [A] Last pass took 60ms [A] Last pass took 69ms [A] Last pass took 84ms [A] Last pass took 95ms TOT later: [A] Last pass took 1119ms [A] Last pass took 1153ms [A] Last pass took 1168ms [A] Last pass took 1184ms [A] Last pass took 1200ms [A] Last pass took 1215ms 374751 6 87197 ggaren 2011-03-28 13:57:13 -0700 Created attachment 87197 Patch 374813 7 ggaren 2011-03-28 15:00:54 -0700 Committed r82162: <http://trac.webkit.org/changeset/82162> 374817 8 ggaren 2011-03-28 15:02:12 -0700 Reopening, since my patch didn't fix the whole regression. 374889 9 ggaren 2011-03-28 16:43:29 -0700 Updating title to reflect the specific constraints of this test: it allocates a 2GB tree! Most of the run-time is dominated by OS virtual memory allocation, fault, and zero fill calls. 375893 10 ggaren 2011-03-29 16:58:33 -0700 Some regression ranges: 77699 (fast) - 80051 (slow) [actually 78312] -> The capacity() bug, which was fixed in <http://trac.webkit.org/changeset/82162>. 80051 (fast) - 80052 (slow) -> Changing the block size to 16KB. 375914 11 ggaren 2011-03-29 17:14:15 -0700 Testing in other browsers: Firefox aggressively grows the heap to 2GB, and never shrinks it. This is the fastest possible policy for this stress test (~2s total), but a very bad policy for memory footprint! Chrome is pretty fast (~3.5s total) on its first run, but gets very slow (>30s total or complete hang) on later runs. 375924 12 ggaren 2011-03-29 17:29:53 -0700 > Firefox aggressively grows the heap to 2GB, and never shrinks it. Correction: It looks like Firefox will eventually shrink from 2GB if you give it enough cajoling. 376007 13 ggaren 2011-03-29 19:56:57 -0700 It looks like the fundamental underlying problem here is that vm_map is O(N^2). 381948 14 87197 ggaren 2011-04-07 19:59:49 -0700 Comment on attachment 87197 Patch Clearing review flag to get this out of the list. 86471 2011-03-22 10:19:14 -0700 2011-03-22 10:19:14 -0700 Test case gc3.html text/html 1486 Tam PGh0bWw+CjxoZWFkPgo8dGl0bGU+VGVzdCBHQzwvdGl0bGU+CjwvaGVhZD4KPGJvZHk+Cjx0ZXh0 YXJlYSBpZD0icHJvZ3Jlc3MiIHJvd3M9IjEiIGNvbHM9IjgwIj5QcmVzcyBhbnkga2V5IHRvIHN0 YXJ0LjwvdGV4dGFyZWE+Cjxocj4KPHRleHRhcmVhIGlkPSJvdXRwdXQiIHJvd3M9IjYwIiBjb2xz PSI4MCI+PC90ZXh0YXJlYT4KPHNjcmlwdCB0eXBlPSJ0ZXh0L2phdmFzY3JpcHQiPgp2YXIgTElT VF9TSVpFID0gMTAwMDsKdmFyIFBFUk1BTkVOVF9QT09MX1NJWkUgPSAxMDAwMDsKdmFyIFBFUk1B TkVOVF9TVEVQID0gMTAwOwoKLy8gQSBzaXplLVBFUk1BTkVOVF9QT09MX1NJWkUgYXJyYXkgb2Yg TElTVC1TSVpFLW5vZGUgc3RydWN0dXJlCnZhciBwZXJtYW5lbnRQb29sID0gW107CnZhciBwZXJt YW5lbnRDdXJyZW50Q291bnQgPSAwOwp2YXIgcGVybWFuZW50SW5kZXggPSAwOwoKdmFyIHByb2dy ZXNzID0gZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoJ3Byb2dyZXNzJyk7CnZhciBvdXRwdXQgPSBk b2N1bWVudC5nZXRFbGVtZW50QnlJZCgnb3V0cHV0Jyk7CgpmdW5jdGlvbiBhbGxvY2F0ZVBlcm1h bmVudCgpIHsKICAgIHByb2dyZXNzLnZhbHVlID0gIkFsbG9jYXRpbmcgaXRlbSAiICsgcGVybWFu ZW50SW5kZXggKyAiIC4uLiI7CiAgICAvLyBQb3B1bGF0ZSBhIGZldyBwYWdlcyBmb3IgZXZlcnkg aW52b2NhdGlvbgogICAgdmFyIGVuZCA9IHBlcm1hbmVudEluZGV4ICsgUEVSTUFORU5UX1NURVA7 CiAgICBsYXN0VGltZXN0YW1wID0gRGF0ZS5ub3coKTsKICAgIGZvciAoOyBwZXJtYW5lbnRJbmRl eCA8IGVuZDsgKytwZXJtYW5lbnRJbmRleCkgewogICAgICAgIGlmIChwZXJtYW5lbnRJbmRleCA9 PT0gUEVSTUFORU5UX1BPT0xfU0laRSkgeyAvLyBFbm91Z2gKICAgICAgICAgICAgcHJvZ3Jlc3Mu dmFsdWUgPSAiRG9uZSI7CiAgICAgICAgICAgIHJldHVybjsKICAgICAgICB9CiAgICAgICAgdmFy IGxpc3QgPSBbXTsKICAgICAgICB2YXIgcHJldiA9IHsgczogJycgfTsKICAgICAgICBmb3IgKHZh ciBqID0gMDsgaiA8IExJU1RfU0laRTsgKytqKSB7CiAgICAgICAgICAgIGxpc3Rbal0gPSB7IHBy ZXY6IHByZXYsIHM6IHByZXYucyArICdQJyArIHBlcm1hbmVudEluZGV4ICsgJyAnICsgaiB9Owog ICAgICAgICAgICBwcmV2ID0gbGlzdFtqXTsKICAgICAgICAgICAgKytwZXJtYW5lbnRDdXJyZW50 Q291bnQ7CiAgICAgICAgfQogICAgICAgIHBlcm1hbmVudFBvb2xbcGVybWFuZW50SW5kZXhdID0g bGlzdDsKICAgIH0KICAgIGRpZmYgPSBEYXRlLm5vdygpIC0gbGFzdFRpbWVzdGFtcDsKICAgIG91 dHB1dC52YWx1ZSArPSAiW0FdIExhc3QgcGFzcyB0b29rICIgKyBNYXRoLmZsb29yKGRpZmYpICsg Im1zXG4iOwogICAgc2V0VGltZW91dChhbGxvY2F0ZVBlcm1hbmVudCwgMCk7Cn07Cgpkb2N1bWVu dC5vbmtleXByZXNzID0gYWxsb2NhdGVQZXJtYW5lbnQ7Cgo8L3NjcmlwdD4KPC9ib2R5Pgo8L2h0 bWw+Cg== 87197 2011-03-28 13:57:13 -0700 2011-04-07 19:59:49 -0700 Patch bug-56823-20110328135712.patch text/plain 1862 ggaren SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gODIxNTEpCisrKyBTb3VyY2Uv SmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEsMjEgQEAK KzIwMTEtMDMtMjggIEdlb2ZmcmV5IEdhcmVuICA8Z2dhcmVuQGFwcGxlLmNvbT4KKworICAgICAg ICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBSRUdSRVNTSU9OIFtyNzg3 OTQtcjc5MjQ5XSBBbGxvY2F0aW9uIG9mIG1lbW9yeSBpcyBzbG93IHdoZW4gbnVtYmVyIG9mIGFj dGl2ZSBvYmplY3RzIGlzIGxhcmdlCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3No b3dfYnVnLmNnaT9pZD01NjgyMworICAgICAgICAKKyAgICAgICAgUGFydGlhbCBmaXggZm9yIG1v c3Qgb2YgdGhlIHByb2JsZW0uIChUT1Qgc3RpbGwgc2hvd3MgYSByZWdyZXNzaW9uLCB0aG91Z2gu KQorCisgICAgICAgICogcnVudGltZS9IZWFwLmNwcDoKKyAgICAgICAgKEpTQzo6SGVhcDo6cmVw b3J0RXh0cmFNZW1vcnlDb3N0U2xvd0Nhc2UpOiBVc2UgaGlnaFdhdGVyTWFyaygpLCBpbnN0ZWFk IG9mCisgICAgICAgIGNhcGFjaXR5KCksIHNpbmNlIGNhcGFjaXR5KCkgaXMgTyhuKSByZWxhdGl2 ZSB0byB0aGUgc2l6ZSBvZiB0aGUgaGVhcC4KKworICAgICAgICBJbiBsaW1pdGVkIGNpcmN1bXN0 YW5jZXMsIGNhcGFjaXR5KCkgaXMgYWxzbyB3b3JzZSB0aGFuIGhpZ2hXYXRlck1hcmsoKQorICAg ICAgICBmb3IgbWVhc3VyaW5nIGV4dHJhIGNvc3QgcmVsYXRpdmUgdG8gaGVhcCBzaXplLCBzaW5j ZSBjYXBhY2l0eSgpIG9ubHkKKyAgICAgICAgbWVhc3VyZXMgdGhlICpjdXJyZW50KiBjYXBhY2l0 eSBvZiB0aGUgaGVhcCwgYnV0IHRoZSBoZWFwIHdpbGwgZ3JvdyBpZgorICAgICAgICBuZWNlc3Nh cnkgdG8gYXR0YWluIGhpZ2hXYXRlck1hcmsoKS4KKwogMjAxMS0wMy0yOCAgT2xpdmVyIEh1bnQg IDxvbGl2ZXJAYXBwbGUuY29tPgogCiAgICAgICAgIFJFR1JFU1NJT04ocjgyMTMwKTogSXQgbWFk ZSBhbGwgdGVzdHMgY3Jhc2ggKFJlcXVlc3RlZCBieSBPc3N5IG9uICN3ZWJraXQpLgpJbmRleDog U291cmNlL0phdmFTY3JpcHRDb3JlL3J1bnRpbWUvSGVhcC5jcHAKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL3J1bnRpbWUvSGVhcC5jcHAJKHJldmlzaW9uIDgxOTkzKQorKysg U291cmNlL0phdmFTY3JpcHRDb3JlL3J1bnRpbWUvSGVhcC5jcHAJKHdvcmtpbmcgY29weSkKQEAg LTEwMSw3ICsxMDEsNyBAQCB2b2lkIEhlYXA6OnJlcG9ydEV4dHJhTWVtb3J5Q29zdFNsb3dDYXNl CiAgICAgLy8gaWYgYSBsYXJnZSB2YWx1ZSBzdXJ2aXZlcyBvbmUgZ2FyYmFnZSBjb2xsZWN0aW9u LCB0aGVyZSBpcyBub3QgbXVjaCBwb2ludCB0bwogICAgIC8vIGNvbGxlY3RpbmcgbW9yZSBmcmVx dWVudGx5IGFzIGxvbmcgYXMgaXQgc3RheXMgYWxpdmUuCiAKLSAgICBpZiAobV9leHRyYUNvc3Qg PiBtYXhFeHRyYUNvc3QgJiYgbV9leHRyYUNvc3QgPiBtX21hcmtlZFNwYWNlLmNhcGFjaXR5KCkg LyAyKQorICAgIGlmIChtX2V4dHJhQ29zdCA+IG1heEV4dHJhQ29zdCAmJiBtX2V4dHJhQ29zdCA+ IG1fbWFya2VkU3BhY2UuaGlnaFdhdGVyTWFyaygpIC8gMikKICAgICAgICAgY29sbGVjdEFsbEdh cmJhZ2UoKTsKICAgICBtX2V4dHJhQ29zdCArPSBjb3N0OwogfQo=