178389 2017-10-17 08:59:28 -0700 Enable tier-up in loops created by recursive tail call optimizations. 2022-03-08 18:55:17 -0800 1 1 1 Unclassified WebKit JavaScriptCore WebKit Nightly Build Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 176601 1 rmorisset rmorisset buildbot cdumez ews-watchlist jbedard keith_miller mark.lam msaboff rniwa saam tzagallo webkit-bug-importer oldest_to_newest 1361244 0 rmorisset 2017-10-17 08:59:28 -0700 See the comments on https://bugs.webkit.org/show_bug.cgi?id=176601 1363286 1 rmorisset 2017-10-23 08:33:24 -0700 I ran some tests between no optimization, optimization in both DFG and FTL, and optimization only in FTL: Disabled Normal FTLOnly FTLOnly v. Disabled n-body 679.4277+-12.1793 ^ 608.0302+-17.0313 ^ 556.8847+-9.6934 ^ definitely 1.2201x faster merge-sort 1056.9295+-13.0481 1025.3372+-19.3403 ^ 764.4767+-28.6580 ^ definitely 1.3826x faster <geometric> 847.3528+-10.6054 ^ 789.3578+-12.1466 ^ 652.3523+-16.1727 ^ definitely 1.2989x faster Two conclusions: - the win between no optimization and unconditional optimization is a lot smaller than what I had measured earlier (I had got about 20% in a less rigorous benchmark, before I got TailBench supported into the run-jsc-benchmarks script) - there is clearly a huge gain in only running it in FTL mode, which confirms our suspicion that the optimization is preventing tier-up from DFG to FTL. 1363295 2 saam 2017-10-23 09:16:07 -0700 (In reply to Robin Morisset from comment #1) > I ran some tests between no optimization, optimization in both DFG and FTL, > and optimization only in FTL: > Disabled Normal > FTLOnly FTLOnly v. Disabled > > n-body 679.4277+-12.1793 ^ 608.0302+-17.0313 ^ > 556.8847+-9.6934 ^ definitely 1.2201x faster > merge-sort 1056.9295+-13.0481 1025.3372+-19.3403 ^ > 764.4767+-28.6580 ^ definitely 1.3826x faster > > <geometric> 847.3528+-10.6054 ^ 789.3578+-12.1466 ^ > 652.3523+-16.1727 ^ definitely 1.2989x faster > > Two conclusions: > - the win between no optimization and unconditional optimization is a lot > smaller than what I had measured earlier (I had got about 20% in a less > rigorous benchmark, before I got TailBench supported into the > run-jsc-benchmarks script) > - there is clearly a huge gain in only running it in FTL mode, which > confirms our suspicion that the optimization is preventing tier-up from DFG > to FTL. I’d be okay with saying we only do this in FTL until we figure out how to OSR using this loop you introduce. 1363440 3 324603 rmorisset 2017-10-23 15:55:52 -0700 Created attachment 324603 Patch 1363441 4 rmorisset 2017-10-23 16:00:12 -0700 Some notes about that patch: - I am not sure of the necessity of the resetting of m_exitOK to false at the end of handleRecursiveTailCall - The second setting of m_exitOK to true before the placement of the LoopHint might also be unnecessary. Here are some performance numbers: "Normal" at /Users/rmorisset/Webkit2/OpenSource/WebKitBuild/Baseline/Release/jsc "FTLOnly" at /Users/rmorisset/Webkit2/OpenSource/WebKitBuild/TierUp/Release/jsc "LoopHint" at /Users/rmorisset/Webkit2/OpenSource/WebKitBuild/TierUp2/Release/jsc Collected 20 samples per benchmark/VM, with 20 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. Normal FTLOnly LoopHint LoopHint v. Normal n-body 562.0193+-4.9093 ^ 521.3055+-1.6532 ! 559.7732+-4.8271 merge-sort 963.8272+-3.6327 ^ 711.8453+-6.1834 708.6563+-6.2171 ^ definitely 1.3601x faster <geometric> 735.9597+-3.4519 ^ 609.1411+-2.7327 ! 629.7776+-3.7344 ^ definitely 1.1686x faster We see that adding the LoopHint improves performance over the baseline, but for some reason less than running the code only in FTL for n-body. 1420101 5 rmorisset 2018-05-03 06:22:48 -0700 I've revisited this topic, and I am even more confused than before. First, here are the numbers for TailBench with the optimisation completely disabled (and with a small fix to merge-sort-cps so it takes a similar amount of time to run as merge-sort), as a kind of baseline : n-body 614.1399+-2.6044 merge-sort 476.6205+-2.4212 merge-sort-cps 462.0143+-1.4312 bf-interpreter 401.7443+-3.4122 <geometric> 482.7646+-0.9180 (all numbers are from running Tools/Scripts/run-jsc-benchmarks WebKitBuild/Release/jsc --inner 2 --outer 10 --tail-bench on my iMac, on top-of-tree as of May 2nd, 2018). Next, here are the numbers for the default configuration, where the optimisation runs, but presumably interferes with tier up: n-body 574.8102+-2.9423 merge-sort 399.7921+-4.3145 merge-sort-cps 460.7351+-5.7754 bf-interpreter 264.6061+-3.6597 <geometric> 409.0464+-2.4496 So far so good, we improve all programs except for merge-sort-cps that we don't regress at least. Now let's try several different ways to deal with the tier-up problem. With the patch that I had proposed before (inserting a LoopHint at the right place): n-body 572.7142+-2.7052 merge-sort 436.1483+-1.5764 merge-sort-cps 443.3050+-11.2810 bf-interpreter 260.9475+-4.1062 <geometric> 412.1350+-2.9962 We finally improve merge-sort-cps a tiny bit, but at the cost of a significant regression on merge-sort. Is it due to some overhead from the way the patch works (since we add a No-op at the beginning of potential target blocks to have a place to easily put the LoopHint)? No, I tested a patch with just the transformation of No-op to LoopHint disabled, and it performs as the default (current) configuration: n-body 575.9674+-2.9741 merge-sort 400.1163+-4.3708 merge-sort-cps 460.9874+-5.5892 bf-interpreter 266.0997+-3.7222 <geometric> 409.9563+-1.9411 Is it maybe due to a problem with OSR-entry at this specific point in the program? I tried making the patch insert a LoopHint but not marking it as a potential OSR entry point: n-body 573.6579+-3.0651 merge-sort 435.2522+-1.5020 merge-sort-cps 446.3558+-10.7560 bf-interpreter 263.8143+-3.2532 <geometric> 413.9548+-2.7296 No significant difference from the normal patch. So at this point I tried a completely different approach: marking the jump back (by putting 1 into its m_opInfo2), and teaching CheckTierUpInjectionPhase to treat such a jump like a return or tail call and insert a CheckTierUpAtReturn just before it: n-body 548.8219+-5.1428 merge-sort 428.9825+-1.4452 merge-sort-cps 432.5123+-4.8355 bf-interpreter 256.7629+-2.6664 <geometric> 402.0644+-2.0024 Great news: it not-only improves merge-sort-cps, it also improves bf-interpreter a tiny bit, and n-body a slightly larger amount. But it still makes merge-sort slower than the default! At last, I tried running the optimisation only in FTL mode, and it proved even better for merge-sort-cps, same as that last patch for the rest: n-body 548.3229+-5.8444 merge-sort 427.2701+-1.7264 merge-sort-cps 411.6702+-1.6944 bf-interpreter 254.7185+-2.1882 <geometric> 395.8613+-1.1237 So, here are my temporary conclusions: - the patch I first proposed is strictly dominated by only running the optimisation in FTL - the alternate patch the marks the jump is very similar to only running the optimisation in FTL, but is slower for merge-sort-cps - the current (default) situation is only faster than optimizing purely in FTL for merge-sort. - the optimisation is a win for all 4 benchmarks in all cases. I will thus submit a patch that makes the optimisation only run in FTL, even though I am still confused as to what is actually going on. 1420104 6 339404 rmorisset 2018-05-03 06:37:49 -0700 Created attachment 339404 Patch 1422364 7 339404 rmorisset 2018-05-09 11:17:32 -0700 Comment on attachment 339404 Patch I've tested the optimization with useFTLJIT=false, and it appears to have a positive impact in that case. So just running in FTL is probably not ideal, but I still recommend doing it as a simple performance improvement in the mean time. Results (on the same machine as the previous results), without FTL and without the optimisation: n-body 964.5178+-4.1374 merge-sort 550.7901+-4.0016 merge-sort-cps 514.8657+-2.3150 bf-interpreter 600.9773+-1.5814 <geometric> 636.7378+-0.6850 Results, without FTL but with the optimization: n-body 901.5132+-1.7837 merge-sort 526.0939+-1.4784 merge-sort-cps 484.5323+-4.2516 bf-interpreter 515.0320+-5.5436 <geometric> 586.5377+-2.4870 1430579 8 339404 saam 2018-06-06 14:28:50 -0700 Comment on attachment 339404 Patch r=me Not CQ+ing since we should make sure this still applies cleanly to ToT 1848969 9 454036 rmorisset 2022-03-07 14:16:32 -0800 Created attachment 454036 Patch Just rebased for landing, thanks Mark for reminding me of this super old patch. 1849475 10 454036 rmorisset 2022-03-08 16:50:28 -0800 Comment on attachment 454036 Patch Trying to land this as all bots are green except for iOS-wk2 which appears to be down for all patches. 1849518 11 ews-feeder 2022-03-08 18:54:21 -0800 Committed r291026 (248200@main): <https://commits.webkit.org/248200@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 454036. 1849519 12 webkit-bug-importer 2022-03-08 18:55:17 -0800 <rdar://problem/90004728> 324603 2017-10-23 15:55:52 -0700 2018-05-03 06:37:46 -0700 Patch bug-178389-20171024005550.patch text/plain 4283 rmorisset U3VidmVyc2lvbiBSZXZpc2lvbjogMjIzODM0CmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2YVNjcmlw dENvcmUvQ2hhbmdlTG9nIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwppbmRleCAx Mzk1N2YwOWYyY2EwMjIzNmYzYmE2MGZmMDIwYTdiMzY2YjNhYWJjLi4zOGExMDM1MmQ0OWQ3ZTZl ZGZjNmU4NGRhMmVmZGMzYzVhMzI1ZDY4IDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENv cmUvQ2hhbmdlTG9nCisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKQEAgLTEs MyArMSwxOSBAQAorMjAxNy0xMC0yMyAgUm9iaW4gTW9yaXNzZXQgIDxybW9yaXNzZXRAYXBwbGUu Y29tPgorCisgICAgICAgIEVuYWJsZSB0aWVyLXVwIGluIGxvb3BzIGNyZWF0ZWQgYnkgcmVjdXJz aXZlIHRhaWwgY2FsbCBvcHRpbWl6YXRpb25zLgorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0 Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9MTc4Mzg5CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZ IChPT1BTISkuCisKKyAgICAgICAgVGhpcyBpcyBkb25lIGJ5IGFkZGluZyBhIExvb3BIaW50IGF0 IHRoZSB0b3Agb2YgdGhlIG5ld2x5IGNyZWF0ZWQgbG9vcCwgYW5kIG1hcmtpbmcgaXQgYXMgYW4g T1NSIGVudHJ5IHBvaW50IGlmIHdlIGFyZSBpbiB0aGUgbWFjaGluZSBjb2RlQmxvY2suCisgICAg ICAgIEluIG9yZGVyIHRvIGhhdmUgc29tZXdoZXJlIHRvIHB1dCB0aGF0IExvb3BIaW50LCB3ZSBh dXRvbWF0aWNhbGx5IGFkZCBhIENoZWNrIG5vZGUganVzdCBhZnRlciB0aGUgb3BfZW50ZXIgc3Bs aXQsIHRvIGJlIG92ZXJ3cml0dGVuIGlmIGEgcmVjdXJzaXZlIHRhaWwgY2FsbCBpcyBsYXRlciBv cHRpbWl6ZWQuCisKKyAgICAgICAgKiBkZmcvREZHQmFzaWNCbG9jay5oOgorICAgICAgICAoSlND OjpERkc6OkJhc2ljQmxvY2s6OnJlcGxhY2VGaXJzdE5vZGUpOgorICAgICAgICAqIGRmZy9ERkdC eXRlQ29kZVBhcnNlci5jcHA6CisgICAgICAgIChKU0M6OkRGRzo6Qnl0ZUNvZGVQYXJzZXI6Omhh bmRsZVJlY3Vyc2l2ZVRhaWxDYWxsKToKKyAgICAgICAgKEpTQzo6REZHOjpCeXRlQ29kZVBhcnNl cjo6cGFyc2VDb2RlQmxvY2spOgorCiAyMDE3LTEwLTIzICBZdXN1a2UgU3V6dWtpICA8dXRhdGFu ZS50ZWFAZ21haWwuY29tPgogCiAgICAgICAgIFtKU0NdIFVzZSBmYXN0Sm9pbiBpbiBBcnJheSN0 b1N0cmluZwpkaWZmIC0tZ2l0IGEvU291cmNlL0phdmFTY3JpcHRDb3JlL2RmZy9ERkdCYXNpY0Js b2NrLmggYi9Tb3VyY2UvSmF2YVNjcmlwdENvcmUvZGZnL0RGR0Jhc2ljQmxvY2suaAppbmRleCAx ZTIzZTM3NjM3YzQzMWUxOGI2ODU4ZjE2YjYzNDkyZGM3YWNmODdiLi43MzJjNjBkZThlZWY3NDJl OTAyMmFjZDM4MDE4NzMzMWE1MDQ3OWYwIDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENv cmUvZGZnL0RGR0Jhc2ljQmxvY2suaAorKysgYi9Tb3VyY2UvSmF2YVNjcmlwdENvcmUvZGZnL0RG R0Jhc2ljQmxvY2suaApAQCAtMTE1LDcgKzExNSwxMyBAQCBzdHJ1Y3QgQmFzaWNCbG9jayA6IFJl ZkNvdW50ZWQ8QmFzaWNCbG9jaz4gewogICAgICAgICBlbHNlCiAgICAgICAgICAgICBtX25vZGVz Lmluc2VydChyZXN1bHQuaW5kZXgsIG5vZGUpOwogICAgIH0KLSAgICAKKworICAgIHZvaWQgcmVw bGFjZUZpcnN0Tm9kZShOb2RlKiBub2RlKQorICAgIHsKKyAgICAgICAgUkVMRUFTRV9BU1NFUlQo KG1fbm9kZXNbMF0tPm9wKCkgPT0gQ2hlY2spIHx8IChtX25vZGVzWzBdLT5vcCgpID09IG5vZGUt Pm9wKCkpKTsKKyAgICAgICAgbV9ub2Rlc1swXSA9IG5vZGU7CisgICAgfQorCiAgICAgdm9pZCBy ZXBsYWNlVGVybWluYWwoTm9kZSopOwogICAgIAogICAgIHNpemVfdCBudW1Ob2RlcygpIGNvbnN0 IHsgcmV0dXJuIHBoaXMuc2l6ZSgpICsgc2l6ZSgpOyB9CmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2 YVNjcmlwdENvcmUvZGZnL0RGR0J5dGVDb2RlUGFyc2VyLmNwcCBiL1NvdXJjZS9KYXZhU2NyaXB0 Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwCmluZGV4IDM1Yjc2NDE5ZTE5Yjg1NzFhMjgy MGJkNjBlZTdhZTA4ZWRhZTY1YWYuLjg0YjFiNzY1MjNlOTA0MGQ3ODllNWNkZGMzMDhjNmYzN2Mw YWUwNTAgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQ YXJzZXIuY3BwCisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJz ZXIuY3BwCkBAIC0xNDcxLDEyICsxNDcxLDI0IEBAIGJvb2wgQnl0ZUNvZGVQYXJzZXI6OmhhbmRs ZVJlY3Vyc2l2ZVRhaWxDYWxsKE5vZGUqIGNhbGxUYXJnZXROb2RlLCBjb25zdCBDYWxsTGluCiAg ICAgICAgIG1fY3VycmVudEluZGV4ID0gT1BDT0RFX0xFTkdUSChvcF9lbnRlcik7CiAgICAgICAg IG1fZXhpdE9LID0gdHJ1ZTsKICAgICAgICAgcHJvY2Vzc1NldExvY2FsUXVldWUoKTsKKworICAg ICAgICBCYXNpY0Jsb2NrKiogZW50cnlCbG9ja1B0ciA9IHRyeUJpbmFyeVNlYXJjaDxCYXNpY0Js b2NrKiwgdW5zaWduZWQ+KHN0YWNrRW50cnktPm1fYmxvY2tMaW5raW5nVGFyZ2V0cywgc3RhY2tF bnRyeS0+bV9ibG9ja0xpbmtpbmdUYXJnZXRzLnNpemUoKSwgT1BDT0RFX0xFTkdUSChvcF9lbnRl ciksIGdldEJ5dGVjb2RlQmVnaW5Gb3JCbG9jayk7CisgICAgICAgIFJFTEVBU0VfQVNTRVJUKGVu dHJ5QmxvY2tQdHIpOworCisgICAgICAgIG1fZXhpdE9LID0gdHJ1ZTsKKyAgICAgICAgTm9kZSog bG9vcEhpbnQgPSBtX2dyYXBoLmFkZE5vZGUoTG9vcEhpbnQsIGN1cnJlbnROb2RlT3JpZ2luKCks IEVkZ2UoKSwgRWRnZSgpLCBFZGdlKCkpOworICAgICAgICAoKmVudHJ5QmxvY2tQdHIpLT5yZXBs YWNlRmlyc3ROb2RlKGxvb3BIaW50KTsKKyAgICAgICAgLy8gV2UgbmV2ZXIgZG8gT1NSIGludG8g YW4gaW5saW5lZCBjb2RlIGJsb2NrLiBUaGF0IGNvdWxkIG5vdCBoYXBwZW4sIHNpbmNlIE9TUgor ICAgICAgICAvLyBsb29rcyB1cCB0aGUgY29kZSBibG9jayB0aGF0IGlzIHRoZSByZXBsYWNlbWVu dCBmb3IgdGhlIGJhc2VsaW5lIEpJVCBjb2RlCisgICAgICAgIC8vIGJsb2NrLiBIZW5jZSwgbWFj aGluZSBjb2RlIGJsb2NrID0gdHJ1ZSBjb2RlIGJsb2NrID0gbm90IGlubGluZSBjb2RlIGJsb2Nr LgorICAgICAgICBpZiAoIXN0YWNrRW50cnktPm1fY2FsbGVyKQorICAgICAgICAgICAgKCplbnRy eUJsb2NrUHRyKS0+aXNPU1JUYXJnZXQgPSB0cnVlOworCisKICAgICAgICAgbV9jdXJyZW50SW5k ZXggPSBvbGRJbmRleDsKICAgICAgICAgbV9pbmxpbmVTdGFja1RvcCA9IG9sZFN0YWNrVG9wOwog ICAgICAgICBtX2V4aXRPSyA9IGZhbHNlOwogCi0gICAgICAgIEJhc2ljQmxvY2sqKiBlbnRyeUJs b2NrUHRyID0gdHJ5QmluYXJ5U2VhcmNoPEJhc2ljQmxvY2sqLCB1bnNpZ25lZD4oc3RhY2tFbnRy eS0+bV9ibG9ja0xpbmtpbmdUYXJnZXRzLCBzdGFja0VudHJ5LT5tX2Jsb2NrTGlua2luZ1Rhcmdl dHMuc2l6ZSgpLCBPUENPREVfTEVOR1RIKG9wX2VudGVyKSwgZ2V0Qnl0ZWNvZGVCZWdpbkZvckJs b2NrKTsKLSAgICAgICAgUkVMRUFTRV9BU1NFUlQoZW50cnlCbG9ja1B0cik7CiAgICAgICAgIGFk ZEp1bXBUbygqZW50cnlCbG9ja1B0cik7CiAgICAgICAgIHJldHVybiB0cnVlOwogICAgICAgICAv LyBJdCB3b3VsZCBiZSB1bnNvdW5kIHRvIGp1bXAgb3ZlciBhIG5vbi10YWlsIGNhbGw6IHRoZSAi dGFpbCIgY2FsbCBpcyBub3QgcmVhbGx5IGEgdGFpbCBjYWxsIGluIHRoYXQgY2FzZS4KQEAgLTY0 NTIsNiArNjQ2NCwxMSBAQCB2b2lkIEJ5dGVDb2RlUGFyc2VyOjpwYXJzZUNvZGVCbG9jaygpCiAg ICAgICAgICAgICAgICAgICAgIG1fY3VycmVudEJsb2NrLT5pc09TUlRhcmdldCA9IHRydWU7CiAg ICAgICAgICAgICAgICAgICAgIG1fZ3JhcGgubV9yb290cy5hcHBlbmQobV9jdXJyZW50QmxvY2sp OwogICAgICAgICAgICAgICAgIH0KKworICAgICAgICAgICAgICAgIC8vIEp1c3QgYWZ0ZXIgYW4g b3BfZW50ZXIsIHdlIGFkZCBhIG5vLW9wIHRvIHRoZSBncmFwaCwgc28gdGhhdCBpdCBjYW4gYmUg b3ZlcndyaXR0ZW4gYnkgYSBMb29wSGludCBpZiBhIHRhaWwgY2FsbCBpcyBvcHRpbWl6ZWQgaW50 byBhIGp1bXAgdG8gaXQuCisgICAgICAgICAgICAgICAgaWYgKG1fY3VycmVudEluZGV4ID09IE9Q Q09ERV9MRU5HVEgob3BfZW50ZXIpKQorICAgICAgICAgICAgICAgICAgICBhZGRUb0dyYXBoKENo ZWNrKTsKKwogICAgICAgICAgICAgICAgIHByZXBhcmVUb1BhcnNlQmxvY2soKTsKICAgICAgICAg ICAgIH0KIAo= 339404 2018-05-03 06:37:49 -0700 2022-03-07 14:16:32 -0800 Patch bug-178389-20180503153748.patch text/plain 6020 rmorisset U3VidmVyc2lvbiBSZXZpc2lvbjogMjMxMzAxCmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2YVNjcmlw dENvcmUvQ2hhbmdlTG9nIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwppbmRleCAy NDcwNjY4OGZiZDFlMGJlOGY0YWQ4MTcwMWVhNTYxYzRmODNhN2JlLi5jYzhlNzc3MGIyNTI3ZGU5 ZmZiN2U1ZDQ3ZjQ2MjQzMWY4ZjQzODgxIDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENv cmUvQ2hhbmdlTG9nCisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKQEAgLTEs MyArMSwxNiBAQAorMjAxOC0wNS0wMyAgUm9iaW4gTW9yaXNzZXQgIDxybW9yaXNzZXRAYXBwbGUu Y29tPgorCisgICAgICAgIEVuYWJsZSB0aWVyLXVwIGluIGxvb3BzIGNyZWF0ZWQgYnkgcmVjdXJz aXZlIHRhaWwgY2FsbCBvcHRpbWl6YXRpb25zLgorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0 Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9MTc4Mzg5CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZ IChPT1BTISkuCisKKyAgICAgICAgTWFkZSB0aGUgcmVjdXJzaXZlIHRhaWwgY2FsbCBvcHRpbWl6 YXRpb24gb25seSBydW4gaW4gRlRMIG1vZGUsIHNpbmNlIGl0IGlzIGEgc2lnbmlmaWNhbnQgcHJv Z3Jlc3Npb24gb24gVGFpbEJlbmNoLgorICAgICAgICBBbHNvIHNvbWUgdHJpdmlhbCByZWZhY3Rv cmluZyBvZiBoYW5kbGVSZWN1cnNpdmVUYWlsQ2FsbC4KKworICAgICAgICAqIGRmZy9ERkdCeXRl Q29kZVBhcnNlci5jcHA6CisgICAgICAgIChKU0M6OkRGRzo6Qnl0ZUNvZGVQYXJzZXI6OmhhbmRs ZVJlY3Vyc2l2ZVRhaWxDYWxsKToKKwogMjAxOC0wNS0wMyAgRG9taW5payBJbmZ1ZWhyICA8ZGlu ZnVlaHJAaWdhbGlhLmNvbT4KIAogICAgICAgICBEaXNhYmxlIHVzYWdlIG9mIGZ1c2VkIG11bHRp cGx5LWFkZCBpbnN0cnVjdGlvbnMgZm9yIEpTQyB3aXRoIGNvbXBpbGVyIGZsYWcKZGlmZiAtLWdp dCBhL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwIGIvU291 cmNlL0phdmFTY3JpcHRDb3JlL2RmZy9ERkdCeXRlQ29kZVBhcnNlci5jcHAKaW5kZXggNGYzZWRl MWRhMWZiN2YxNmY1NTFmMjM0ODg4OGRhMzg5M2U2YzVhNC4uMDc0YTNmZjM0NmVjYmY4OTMzOTRi ZTE5ZTUzY2YxNzIwZTU1NmQwYyAxMDA2NDQKLS0tIGEvU291cmNlL0phdmFTY3JpcHRDb3JlL2Rm Zy9ERkdCeXRlQ29kZVBhcnNlci5jcHAKKysrIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL2RmZy9E RkdCeXRlQ29kZVBhcnNlci5jcHAKQEAgLTEzNjEsNiArMTM2MSwxMSBAQCBib29sIEJ5dGVDb2Rl UGFyc2VyOjpoYW5kbGVSZWN1cnNpdmVUYWlsQ2FsbChOb2RlKiBjYWxsVGFyZ2V0Tm9kZSwgQ2Fs bFZhcmlhbnQgYwogICAgIGlmIChVTkxJS0VMWSghT3B0aW9uczo6b3B0aW1pemVSZWN1cnNpdmVU YWlsQ2FsbHMoKSkpCiAgICAgICAgIHJldHVybiBmYWxzZTsKIAorICAgIC8vIFRoaXMgb3B0aW1p c2F0aW9uIGJyaW5ncyBtb3JlIHBlcmZvcm1hbmNlIGlmIGl0IG9ubHkgcnVucyBpbiBGVEwsIHBy b2JhYmx5IGJlY2F1c2UgaXQgaW50ZXJmZXJlcyB3aXRoIHRpZXItdXAuCisgICAgLy8gU2VlIGh0 dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD0xNzgzODkgZm9yIGRldGFpbHMu CisgICAgaWYgKCFpc0ZUTChtX2dyYXBoLm1fcGxhbi5tb2RlKSkKKyAgICAgICAgcmV0dXJuIGZh bHNlOworCiAgICAgYXV0byB0YXJnZXRFeGVjdXRhYmxlID0gY2FsbFZhcmlhbnQuZXhlY3V0YWJs ZSgpOwogICAgIElubGluZVN0YWNrRW50cnkqIHN0YWNrRW50cnkgPSBtX2lubGluZVN0YWNrVG9w OwogICAgIGRvIHsKQEAgLTEzNzMsNiArMTM3OCwxMCBAQCBib29sIEJ5dGVDb2RlUGFyc2VyOjpo YW5kbGVSZWN1cnNpdmVUYWlsQ2FsbChOb2RlKiBjYWxsVGFyZ2V0Tm9kZSwgQ2FsbFZhcmlhbnQg YwogICAgICAgICAgICAgLy8gV2UgImNvbnRpbnVlIiBpbnN0ZWFkIG9mIHJldHVybmluZyBmYWxz ZSBpbiBjYXNlIGFub3RoZXIgc3RhY2sgZW50cnkgZnVydGhlciBvbiB0aGUgc3RhY2sgaGFzIHRo ZSByaWdodCBudW1iZXIgb2YgYXJndW1lbnRzLgogICAgICAgICAgICAgaWYgKGFyZ3VtZW50Q291 bnRJbmNsdWRpbmdUaGlzICE9IHN0YXRpY19jYXN0PGludD4oY2FsbEZyYW1lLT5hcmd1bWVudENv dW50SW5jbHVkaW5nVGhpcykpCiAgICAgICAgICAgICAgICAgY29udGludWU7CisgICAgICAgICAg ICAvLyBJZiBhbiBJbmxpbmVDYWxsRnJhbWUgaXMgbm90IGEgY2xvc3VyZSwgaXQgd2FzIG9wdGlt aXplZCB1c2luZyBhIGNvbnN0YW50IGNhbGxlZS4KKyAgICAgICAgICAgIC8vIENoZWNrIGlmIHRo aXMgaXMgdGhlIHNhbWUgY2FsbGVlIHRoYXQgd2UgYXJlIGRlYWxpbmcgd2l0aC4KKyAgICAgICAg ICAgIGlmICghY2FsbEZyYW1lLT5pc0Nsb3N1cmVDYWxsICYmIGNhbGxGcmFtZS0+Y2FsbGVlQ29u c3RhbnQoKSAhPSBjYWxsVmFyaWFudC5mdW5jdGlvbigpKQorICAgICAgICAgICAgICAgIGNvbnRp bnVlOwogICAgICAgICB9IGVsc2UgewogICAgICAgICAgICAgLy8gV2UgYXJlIGluIHRoZSBtYWNo aW5lIGNvZGUgZW50cnkgKGkuZS4gdGhlIG9yaWdpbmFsIGNhbGxlcikuCiAgICAgICAgICAgICAv LyBJZiB3ZSBoYXZlIG1vcmUgYXJndW1lbnRzIHRoYW4gdGhlIG51bWJlciBvZiBwYXJhbWV0ZXJz IHRvIHRoZSBmdW5jdGlvbiwgaXQgaXMgbm90IGNsZWFyIHdoZXJlIHdlIGNvdWxkIHB1dCB0aGVt IG9uIHRoZSBzdGFjay4KQEAgLTEzODAsMjQgKzEzODksMTYgQEAgYm9vbCBCeXRlQ29kZVBhcnNl cjo6aGFuZGxlUmVjdXJzaXZlVGFpbENhbGwoTm9kZSogY2FsbFRhcmdldE5vZGUsIENhbGxWYXJp YW50IGMKICAgICAgICAgICAgICAgICByZXR1cm4gZmFsc2U7CiAgICAgICAgIH0KIAotICAgICAg ICAvLyBJZiBhbiBJbmxpbmVDYWxsRnJhbWUgaXMgbm90IGEgY2xvc3VyZSwgaXQgd2FzIG9wdGlt aXplZCB1c2luZyBhIGNvbnN0YW50IGNhbGxlZS4KLSAgICAgICAgLy8gQ2hlY2sgaWYgdGhpcyBp cyB0aGUgc2FtZSBjYWxsZWUgdGhhdCB3ZSB0cnkgdG8gaW5saW5lIGhlcmUuCi0gICAgICAgIGlm IChzdGFja0VudHJ5LT5tX2lubGluZUNhbGxGcmFtZSAmJiAhc3RhY2tFbnRyeS0+bV9pbmxpbmVD YWxsRnJhbWUtPmlzQ2xvc3VyZUNhbGwpIHsKLSAgICAgICAgICAgIGlmIChzdGFja0VudHJ5LT5t X2lubGluZUNhbGxGcmFtZS0+Y2FsbGVlQ29uc3RhbnQoKSAhPSBjYWxsVmFyaWFudC5mdW5jdGlv bigpKQotICAgICAgICAgICAgICAgIGNvbnRpbnVlOwotICAgICAgICB9Ci0KICAgICAgICAgLy8g V2UgbXVzdCBhZGQgc29tZSBjaGVjayB0aGF0IHRoZSBwcm9maWxpbmcgaW5mb3JtYXRpb24gd2Fz IGNvcnJlY3QgYW5kIHRoZSB0YXJnZXQgb2YgdGhpcyBjYWxsIGlzIHdoYXQgd2UgdGhvdWdodC4K ICAgICAgICAgZW1pdEZ1bmN0aW9uQ2hlY2tJZk5lZWRlZCgpOwogICAgICAgICAvLyBXZSBmbHVz aCBldmVyeXRoaW5nLCBhcyBpZiB3ZSB3ZXJlIGluIHRoZSBiYWNrZWRnZSBvZiBhIGxvb3AgKHNl ZSB0cmVhdG1lbnQgb2Ygb3Bfam1wIGluIHBhcnNlQmxvY2spLgogICAgICAgICBmbHVzaEZvclRl cm1pbmFsKCk7CiAKICAgICAgICAgLy8gV2UgbXVzdCBzZXQgdGhlIGNhbGxlZSB0byB0aGUgcmln aHQgdmFsdWUKLSAgICAgICAgaWYgKHN0YWNrRW50cnktPm1faW5saW5lQ2FsbEZyYW1lKSB7Ci0g ICAgICAgICAgICBpZiAoc3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUtPmlzQ2xvc3VyZUNh bGwpCi0gICAgICAgICAgICAgICAgc2V0RGlyZWN0KHN0YWNrRW50cnktPnJlbWFwT3BlcmFuZChW aXJ0dWFsUmVnaXN0ZXIoQ2FsbEZyYW1lU2xvdDo6Y2FsbGVlKSksIGNhbGxUYXJnZXROb2RlLCBO b3JtYWxTZXQpOwotICAgICAgICB9IGVsc2UKKyAgICAgICAgaWYgKCFzdGFja0VudHJ5LT5tX2lu bGluZUNhbGxGcmFtZSkKICAgICAgICAgICAgIGFkZFRvR3JhcGgoU2V0Q2FsbGVlLCBjYWxsVGFy Z2V0Tm9kZSk7CisgICAgICAgIGVsc2UgaWYgKHN0YWNrRW50cnktPm1faW5saW5lQ2FsbEZyYW1l LT5pc0Nsb3N1cmVDYWxsKQorICAgICAgICAgICAgc2V0RGlyZWN0KHN0YWNrRW50cnktPnJlbWFw T3BlcmFuZChWaXJ0dWFsUmVnaXN0ZXIoQ2FsbEZyYW1lU2xvdDo6Y2FsbGVlKSksIGNhbGxUYXJn ZXROb2RlLCBOb3JtYWxTZXQpOwogCiAgICAgICAgIC8vIFdlIG11c3Qgc2V0IHRoZSBhcmd1bWVu dHMgdG8gdGhlIHJpZ2h0IHZhbHVlcwogICAgICAgICBpZiAoIXN0YWNrRW50cnktPm1faW5saW5l Q2FsbEZyYW1lKQpAQCAtMTQzNCw3ICsxNDM1LDYgQEAgYm9vbCBCeXRlQ29kZVBhcnNlcjo6aGFu ZGxlUmVjdXJzaXZlVGFpbENhbGwoTm9kZSogY2FsbFRhcmdldE5vZGUsIENhbGxWYXJpYW50IGMK ICAgICAgICAgLy8gSXQgd291bGQgYmUgdW5zb3VuZCB0byBqdW1wIG92ZXIgYSBub24tdGFpbCBj YWxsOiB0aGUgInRhaWwiIGNhbGwgaXMgbm90IHJlYWxseSBhIHRhaWwgY2FsbCBpbiB0aGF0IGNh c2UuCiAgICAgfSB3aGlsZSAoc3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUgJiYgc3RhY2tF bnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUtPmtpbmQgPT0gSW5saW5lQ2FsbEZyYW1lOjpUYWlsQ2Fs bCAmJiAoc3RhY2tFbnRyeSA9IHN0YWNrRW50cnktPm1fY2FsbGVyKSk7CiAKLSAgICAvLyBUaGUg dGFpbCBjYWxsIHdhcyBub3QgcmVjdXJzaXZlCiAgICAgcmV0dXJuIGZhbHNlOwogfQogCmRpZmYg LS1naXQgYS9QZXJmb3JtYW5jZVRlc3RzL0NoYW5nZUxvZyBiL1BlcmZvcm1hbmNlVGVzdHMvQ2hh bmdlTG9nCmluZGV4IDcxNDBhOWIxYmEyYTBhMTVlYWViOTNjMmM3YjI0NDc4M2Y5YTMwZWMuLmFi ZmY4NDlhYjEwNjkxYjJiMTJhNjNkZGE2MzVlNWEzOWE0YWVhOGIgMTAwNjQ0Ci0tLSBhL1BlcmZv cm1hbmNlVGVzdHMvQ2hhbmdlTG9nCisrKyBiL1BlcmZvcm1hbmNlVGVzdHMvQ2hhbmdlTG9nCkBA IC0xLDMgKzEsMTQgQEAKKzIwMTgtMDUtMDMgIFJvYmluIE1vcmlzc2V0ICA8cm1vcmlzc2V0QGFw cGxlLmNvbT4KKworICAgICAgICBFbmFibGUgdGllci11cCBpbiBsb29wcyBjcmVhdGVkIGJ5IHJl Y3Vyc2l2ZSB0YWlsIGNhbGwgb3B0aW1pemF0aW9ucy4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndl YmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTE3ODM4OQorCisgICAgICAgIFJldmlld2VkIGJ5IE5P Qk9EWSAoT09QUyEpLgorCisgICAgICAgIEp1c3QgY2hhbmdlZCBhIGNvbnN0YW50IGZvciBjb25z aXN0ZW5jeSB3aXRoIFRhaWxCZW5jaDkwMDAvbWVyZ2Utc29ydC5qcworCisgICAgICAgICogVGFp bEJlbmNoOTAwMC9tZXJnZS1zb3J0LWNwcy5qczoKKwogMjAxOC0wNC0zMCAgTXlsZXMgQy4gTWF4 ZmllbGQgIDxtbWF4ZmllbGRAYXBwbGUuY29tPgogCiAgICAgICAgIEltcHJvdmUgdGhlIHBlcmZv cm1hbmNlIG9mIEZvbnRDYXNjYWRlRGVzY3JpcHRpb24ncyBlZmZlY3RpdmVGYW1pbGllcwpkaWZm IC0tZ2l0IGEvUGVyZm9ybWFuY2VUZXN0cy9UYWlsQmVuY2g5MDAwL21lcmdlLXNvcnQtY3BzLmpz IGIvUGVyZm9ybWFuY2VUZXN0cy9UYWlsQmVuY2g5MDAwL21lcmdlLXNvcnQtY3BzLmpzCmluZGV4 IGZmZjMxYTZiOGQwMTRiYjZlYTE5MTdmZGIyYzU1NWIxOTE4NWYzMzcuLjcxMGFmZjVkMDQ1ZGIz NWY4ZjUzOWZlZGNmODc3NTkwMWFiODY5MGQgMTAwNjQ0Ci0tLSBhL1BlcmZvcm1hbmNlVGVzdHMv VGFpbEJlbmNoOTAwMC9tZXJnZS1zb3J0LWNwcy5qcworKysgYi9QZXJmb3JtYW5jZVRlc3RzL1Rh aWxCZW5jaDkwMDAvbWVyZ2Utc29ydC1jcHMuanMKQEAgLTEzMSw3ICsxMzEsNyBAQCBmdW5jdGlv biBidWlsZEFycmF5KGxlbmd0aCwgdmFsdWUpCiBsZXQgYXJyYXlzID0gWwogICAgIGJ1aWxkQXJy YXkoMTAsIHggPT4geCksCiAgICAgYnVpbGRBcnJheSgxMCwgeCA9PiAteCksCi0gICAgYnVpbGRB cnJheSgxMDAwLCB4ID0+IHJhbmRvbSgpKQorICAgIGJ1aWxkQXJyYXkoMTAwMDAsIHggPT4gcmFu ZG9tKCkpCiBdOwogCiBmdW5jdGlvbiB0ZXN0KGluZGV4KQo= 454036 2022-03-07 14:16:32 -0800 2022-03-08 18:54:24 -0800 Patch patch178389 text/plain 5565 rmorisset ZGlmZiAtLWdpdCBhL1BlcmZvcm1hbmNlVGVzdHMvQ2hhbmdlTG9nIGIvUGVyZm9ybWFuY2VUZXN0 cy9DaGFuZ2VMb2cKaW5kZXggMzI4NTMzN2Y5N2IxLi5iZTNmMGI3MjcxZjYgMTAwNjQ0Ci0tLSBh L1BlcmZvcm1hbmNlVGVzdHMvQ2hhbmdlTG9nCisrKyBiL1BlcmZvcm1hbmNlVGVzdHMvQ2hhbmdl TG9nCkBAIC0xLDMgKzEsMTQgQEAKKzIwMjItMDMtMDcgIFJvYmluIE1vcmlzc2V0ICA8cm1vcmlz c2V0QGFwcGxlLmNvbT4KKworICAgICAgICBFbmFibGUgdGllci11cCBpbiBsb29wcyBjcmVhdGVk IGJ5IHJlY3Vyc2l2ZSB0YWlsIGNhbGwgb3B0aW1pemF0aW9ucy4KKyAgICAgICAgaHR0cHM6Ly9i dWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTE3ODM4OQorCisgICAgICAgIFJldmlld2Vk IGJ5IFNhYW0gQmFyYXRpLgorCisgICAgICAgIEp1c3QgY2hhbmdlZCBhIGNvbnN0YW50IGZvciBj b25zaXN0ZW5jeSB3aXRoIFRhaWxCZW5jaDkwMDAvbWVyZ2Utc29ydC5qcworCisgICAgICAgICog VGFpbEJlbmNoOTAwMC9tZXJnZS1zb3J0LWNwcy5qczoKKwogMjAyMi0wMy0wMSAgWXVzdWtlIFN1 enVraSAgPHlzdXp1a2lAYXBwbGUuY29tPgogCiAgICAgICAgIEZpeCBTcGVlZG9tZXRlcidzIHNl dFRpbWVvdXQgdGhyb3R0bGluZyBpc3N1ZQpkaWZmIC0tZ2l0IGEvUGVyZm9ybWFuY2VUZXN0cy9U YWlsQmVuY2g5MDAwL21lcmdlLXNvcnQtY3BzLmpzIGIvUGVyZm9ybWFuY2VUZXN0cy9UYWlsQmVu Y2g5MDAwL21lcmdlLXNvcnQtY3BzLmpzCmluZGV4IGZmZjMxYTZiOGQwMS4uNzEwYWZmNWQwNDVk IDEwMDY0NAotLS0gYS9QZXJmb3JtYW5jZVRlc3RzL1RhaWxCZW5jaDkwMDAvbWVyZ2Utc29ydC1j cHMuanMKKysrIGIvUGVyZm9ybWFuY2VUZXN0cy9UYWlsQmVuY2g5MDAwL21lcmdlLXNvcnQtY3Bz LmpzCkBAIC0xMzEsNyArMTMxLDcgQEAgZnVuY3Rpb24gYnVpbGRBcnJheShsZW5ndGgsIHZhbHVl KQogbGV0IGFycmF5cyA9IFsKICAgICBidWlsZEFycmF5KDEwLCB4ID0+IHgpLAogICAgIGJ1aWxk QXJyYXkoMTAsIHggPT4gLXgpLAotICAgIGJ1aWxkQXJyYXkoMTAwMCwgeCA9PiByYW5kb20oKSkK KyAgICBidWlsZEFycmF5KDEwMDAwLCB4ID0+IHJhbmRvbSgpKQogXTsKIAogZnVuY3Rpb24gdGVz dChpbmRleCkKZGlmZiAtLWdpdCBhL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cgYi9T b3VyY2UvSmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCmluZGV4IGQ5N2MxMjMyNjU3Yi4uZTgxZjlm M2VlNDcxIDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENvcmUvQ2hhbmdlTG9nCisrKyBi L1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKQEAgLTEsMyArMSwxNiBAQAorMjAyMi0w My0wNyAgIFJvYmluIE1vcmlzc2V0ICA8cm1vcmlzc2V0QGFwcGxlLmNvbT4KKworICAgICAgICBF bmFibGUgdGllci11cCBpbiBsb29wcyBjcmVhdGVkIGJ5IHJlY3Vyc2l2ZSB0YWlsIGNhbGwgb3B0 aW1pemF0aW9ucy4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dp P2lkPTE3ODM4OQorCisgICAgICAgIFJldmlld2VkIGJ5IFNhYW0gQmFyYXRpLgorCisgICAgICAg IE1hZGUgdGhlIHJlY3Vyc2l2ZSB0YWlsIGNhbGwgb3B0aW1pemF0aW9uIG9ubHkgcnVuIGluIEZU TCBtb2RlLCBzaW5jZSBpdCBpcyBhIHNpZ25pZmljYW50IHByb2dyZXNzaW9uIG9uIFRhaWxCZW5j aC4KKyAgICAgICAgQWxzbyBzb21lIHRyaXZpYWwgcmVmYWN0b3Jpbmcgb2YgaGFuZGxlUmVjdXJz aXZlVGFpbENhbGwuCisKKyAgICAgICAgKiBkZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwOgorICAg ICAgICAoSlNDOjpERkc6OkJ5dGVDb2RlUGFyc2VyOjpoYW5kbGVSZWN1cnNpdmVUYWlsQ2FsbCk6 CisKIDIwMjItMDMtMDcgIEFuZ2Vsb3MgT2lrb25vbW9wb3Vsb3MgIDxhbmdlbG9zQGlnYWxpYS5j b20+CiAKICAgICAgICAgQXNzZW1ibHlIZWxwZXJzU3Bvb2xlcnM6IHVzZSBsb2FkL3N0b3JlIHBh aXIgb24gQVJNdjcKZGlmZiAtLWdpdCBhL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0 ZUNvZGVQYXJzZXIuY3BwIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL2RmZy9ERkdCeXRlQ29kZVBh cnNlci5jcHAKaW5kZXggN2E0ZmUwMzk2MWU4Li5kOThmYjc2ZWNiNjEgMTAwNjQ0Ci0tLSBhL1Nv dXJjZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwCisrKyBiL1NvdXJj ZS9KYXZhU2NyaXB0Q29yZS9kZmcvREZHQnl0ZUNvZGVQYXJzZXIuY3BwCkBAIC0xNDY3LDYgKzE0 NjcsMTEgQEAgYm9vbCBCeXRlQ29kZVBhcnNlcjo6aGFuZGxlUmVjdXJzaXZlVGFpbENhbGwoTm9k ZSogY2FsbFRhcmdldE5vZGUsIENhbGxWYXJpYW50IGMKICAgICBpZiAoVU5MSUtFTFkoIU9wdGlv bnM6Om9wdGltaXplUmVjdXJzaXZlVGFpbENhbGxzKCkpKQogICAgICAgICByZXR1cm4gZmFsc2U7 CiAKKyAgICAvLyBUaGlzIG9wdGltaXNhdGlvbiBicmluZ3MgbW9yZSBwZXJmb3JtYW5jZSBpZiBp dCBvbmx5IHJ1bnMgaW4gRlRMLCBwcm9iYWJseSBiZWNhdXNlIGl0IGludGVyZmVyZXMgd2l0aCB0 aWVyLXVwLgorICAgIC8vIFNlZSBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/ aWQ9MTc4Mzg5IGZvciBkZXRhaWxzLgorICAgIGlmICghaXNGVEwobV9ncmFwaC5tX3BsYW4ubW9k ZSgpKSkKKyAgICAgICAgcmV0dXJuIGZhbHNlOworCiAgICAgYXV0byB0YXJnZXRFeGVjdXRhYmxl ID0gY2FsbFZhcmlhbnQuZXhlY3V0YWJsZSgpOwogICAgIElubGluZVN0YWNrRW50cnkqIHN0YWNr RW50cnkgPSBtX2lubGluZVN0YWNrVG9wOwogICAgIGRvIHsKQEAgLTE0ODgsNiArMTQ5MywxMCBA QCBib29sIEJ5dGVDb2RlUGFyc2VyOjpoYW5kbGVSZWN1cnNpdmVUYWlsQ2FsbChOb2RlKiBjYWxs VGFyZ2V0Tm9kZSwgQ2FsbFZhcmlhbnQgYwogICAgICAgICAgICAgLy8gYSBnb29kIHRhcmdldCB0 byBqdW1wIGludG8uCiAgICAgICAgICAgICBpZiAoY2FsbEZyYW1lLT5pc1ZhcmFyZ3MoKSkKICAg ICAgICAgICAgICAgICBjb250aW51ZTsKKyAgICAgICAgICAgIC8vIElmIGFuIElubGluZUNhbGxG cmFtZSBpcyBub3QgYSBjbG9zdXJlLCBpdCB3YXMgb3B0aW1pemVkIHVzaW5nIGEgY29uc3RhbnQg Y2FsbGVlLgorICAgICAgICAgICAgLy8gQ2hlY2sgaWYgdGhpcyBpcyB0aGUgc2FtZSBjYWxsZWUg dGhhdCB3ZSBhcmUgZGVhbGluZyB3aXRoLgorICAgICAgICAgICAgaWYgKCFjYWxsRnJhbWUtPmlz Q2xvc3VyZUNhbGwgJiYgY2FsbEZyYW1lLT5jYWxsZWVDb25zdGFudCgpICE9IGNhbGxWYXJpYW50 LmZ1bmN0aW9uKCkpCisgICAgICAgICAgICAgICAgY29udGludWU7CiAgICAgICAgIH0gZWxzZSB7 CiAgICAgICAgICAgICAvLyBXZSBhcmUgaW4gdGhlIG1hY2hpbmUgY29kZSBlbnRyeSAoaS5lLiB0 aGUgb3JpZ2luYWwgY2FsbGVyKS4KICAgICAgICAgICAgIC8vIElmIHdlIGhhdmUgbW9yZSBhcmd1 bWVudHMgdGhhbiB0aGUgbnVtYmVyIG9mIHBhcmFtZXRlcnMgdG8gdGhlIGZ1bmN0aW9uLCBpdCBp cyBub3QgY2xlYXIgd2hlcmUgd2UgY291bGQgcHV0IHRoZW0gb24gdGhlIHN0YWNrLgpAQCAtMTQ5 NSwyNCArMTUwNCwxNyBAQCBib29sIEJ5dGVDb2RlUGFyc2VyOjpoYW5kbGVSZWN1cnNpdmVUYWls Q2FsbChOb2RlKiBjYWxsVGFyZ2V0Tm9kZSwgQ2FsbFZhcmlhbnQgYwogICAgICAgICAgICAgICAg IHJldHVybiBmYWxzZTsKICAgICAgICAgfQogCi0gICAgICAgIC8vIElmIGFuIElubGluZUNhbGxG cmFtZSBpcyBub3QgYSBjbG9zdXJlLCBpdCB3YXMgb3B0aW1pemVkIHVzaW5nIGEgY29uc3RhbnQg Y2FsbGVlLgotICAgICAgICAvLyBDaGVjayBpZiB0aGlzIGlzIHRoZSBzYW1lIGNhbGxlZSB0aGF0 IHdlIHRyeSB0byBpbmxpbmUgaGVyZS4KLSAgICAgICAgaWYgKHN0YWNrRW50cnktPm1faW5saW5l Q2FsbEZyYW1lICYmICFzdGFja0VudHJ5LT5tX2lubGluZUNhbGxGcmFtZS0+aXNDbG9zdXJlQ2Fs bCkgewotICAgICAgICAgICAgaWYgKHN0YWNrRW50cnktPm1faW5saW5lQ2FsbEZyYW1lLT5jYWxs ZWVDb25zdGFudCgpICE9IGNhbGxWYXJpYW50LmZ1bmN0aW9uKCkpCi0gICAgICAgICAgICAgICAg Y29udGludWU7Ci0gICAgICAgIH0KLQogICAgICAgICAvLyBXZSBtdXN0IGFkZCBzb21lIGNoZWNr IHRoYXQgdGhlIHByb2ZpbGluZyBpbmZvcm1hdGlvbiB3YXMgY29ycmVjdCBhbmQgdGhlIHRhcmdl dCBvZiB0aGlzIGNhbGwgaXMgd2hhdCB3ZSB0aG91Z2h0LgogICAgICAgICBlbWl0RnVuY3Rpb25D aGVja0lmTmVlZGVkKCk7CiAgICAgICAgIC8vIFdlIGZsdXNoIGV2ZXJ5dGhpbmcsIGFzIGlmIHdl IHdlcmUgaW4gdGhlIGJhY2tlZGdlIG9mIGEgbG9vcCAoc2VlIHRyZWF0bWVudCBvZiBvcF9qbXAg aW4gcGFyc2VCbG9jaykuCiAgICAgICAgIGZsdXNoRm9yVGVybWluYWwoKTsKIAogICAgICAgICAv LyBXZSBtdXN0IHNldCB0aGUgY2FsbGVlIHRvIHRoZSByaWdodCB2YWx1ZQotICAgICAgICBpZiAo c3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUpIHsKLSAgICAgICAgICAgIGlmIChzdGFja0Vu dHJ5LT5tX2lubGluZUNhbGxGcmFtZS0+aXNDbG9zdXJlQ2FsbCkKLSAgICAgICAgICAgICAgICBz ZXREaXJlY3QocmVtYXBPcGVyYW5kKHN0YWNrRW50cnktPm1faW5saW5lQ2FsbEZyYW1lLCBDYWxs RnJhbWVTbG90OjpjYWxsZWUpLCBjYWxsVGFyZ2V0Tm9kZSwgTm9ybWFsU2V0KTsKLSAgICAgICAg fSBlbHNlCisgICAgICAgIGlmICghc3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUpCiAgICAg ICAgICAgICBhZGRUb0dyYXBoKFNldENhbGxlZSwgY2FsbFRhcmdldE5vZGUpOworICAgICAgICBl bHNlIGlmIChzdGFja0VudHJ5LT5tX2lubGluZUNhbGxGcmFtZS0+aXNDbG9zdXJlQ2FsbCkKKyAg ICAgICAgICAgIHNldERpcmVjdChyZW1hcE9wZXJhbmQoc3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxs RnJhbWUsIENhbGxGcmFtZVNsb3Q6OmNhbGxlZSksIGNhbGxUYXJnZXROb2RlLCBOb3JtYWxTZXQp OworCiAKICAgICAgICAgLy8gV2UgbXVzdCBzZXQgdGhlIGFyZ3VtZW50cyB0byB0aGUgcmlnaHQg dmFsdWVzCiAgICAgICAgIGlmICghc3RhY2tFbnRyeS0+bV9pbmxpbmVDYWxsRnJhbWUpCkBAIC0x NTUwLDcgKzE1NTIsNiBAQCBib29sIEJ5dGVDb2RlUGFyc2VyOjpoYW5kbGVSZWN1cnNpdmVUYWls Q2FsbChOb2RlKiBjYWxsVGFyZ2V0Tm9kZSwgQ2FsbFZhcmlhbnQgYwogICAgICAgICAvLyBJdCB3 b3VsZCBiZSB1bnNvdW5kIHRvIGp1bXAgb3ZlciBhIG5vbi10YWlsIGNhbGw6IHRoZSAidGFpbCIg Y2FsbCBpcyBub3QgcmVhbGx5IGEgdGFpbCBjYWxsIGluIHRoYXQgY2FzZS4KICAgICB9IHdoaWxl IChzdGFja0VudHJ5LT5tX2lubGluZUNhbGxGcmFtZSAmJiBzdGFja0VudHJ5LT5tX2lubGluZUNh bGxGcmFtZS0+a2luZCA9PSBJbmxpbmVDYWxsRnJhbWU6OlRhaWxDYWxsICYmIChzdGFja0VudHJ5 ID0gc3RhY2tFbnRyeS0+bV9jYWxsZXIpKTsKIAotICAgIC8vIFRoZSB0YWlsIGNhbGwgd2FzIG5v dCByZWN1cnNpdmUKICAgICByZXR1cm4gZmFsc2U7CiB9CiAK