174993 2017-07-31 14:41:37 -0700 Fix JSCOnly ARM buildbots after r220047 and r220184 2017-08-16 03:06:19 -0700 1 1 1 Unclassified WebKit New Bugs WebKit Nightly Build Unspecified Unspecified RESOLVED FIXED InRadar P1 Blocker --- 174964 175036 1 ossy ossy ap bugs-noreply clopez commit-queue fpizlo guijemont keith_miller lforschler ossy rniwa saam webkit-bug-importer oldest_to_newest 1333545 0 ossy 2017-07-31 14:41:37 -0700 https://trac.webkit.org/changeset/220047/webkit broke the JSCOnly ARM buildbots, because they can't generate DerivedSources/JavaScriptCore/LLIntAssembly.h in the 20 minute timeout since r220047. ( The buildbots run on X86_64 inside a qemu emulated ARM environment, so ruby is emulated too, that's why it is so slow. ) We have to increase the timeout to have working buildbots again. 1333546 1 316805 ossy 2017-07-31 14:43:03 -0700 Created attachment 316805 Patch 1333660 2 316805 commit-queue 2017-07-31 22:28:08 -0700 Comment on attachment 316805 Patch Clearing flags on attachment: 316805 Committed r220087: <http://trac.webkit.org/changeset/220087> 1333661 3 commit-queue 2017-07-31 22:28:10 -0700 All reviewed patches have been landed. Closing bug. 1333662 4 webkit-bug-importer 2017-07-31 22:28:35 -0700 <rdar://problem/33646963> 1333671 5 ossy 2017-07-31 23:31:42 -0700 Could you please apply this change on build.webkit.org master? It seems it didn't restart itself automatically. 1333939 6 ossy 2017-08-01 14:08:27 -0700 (In reply to Csaba Osztrogonác from comment #5) > Could you please apply this change on build.webkit.org master? > It seems it didn't restart itself automatically. Ping? 1334056 7 lforschler 2017-08-01 17:20:19 -0700 I'm looking into this now. 1334141 8 ossy 2017-08-01 23:48:43 -0700 (In reply to Lucas Forschler from comment #7) > I'm looking into this now. Thanks. But it seems I made a mistake and increaded the timeout only for running test, but not for build step. :( I already fixed it in https://trac.webkit.org/changeset/220123/webkit Could you pick up this change too? Thanks, and sorry for the inconvenience. 1335294 9 ossy 2017-08-04 09:56:54 -0700 Reopen, because the problem is bigger now. :( Recent offlineasm changes amazingly increased the generating time. Previously a clean build took (gnerating and compiling took) ~9 minutes, https://trac.webkit.org/changeset/220047 increased it to ~30 minutes and https://trac.webkit.org/changeset/220184 increased it to ~91 minutes. I haven't measured the exact runtime of the ruby scripts separately yet, but it's obvious that these changes are responsible for long running ruby scripts. (I bet the slowdown isn't problem on native builds, but I can imagine that the runtime was increased from few seconds to 1-2 minutes.) FYI, I stopped the JSCOnly ARM bots because of this blocker issue, because I won't be available for the following 2 weeks, so I won't have any chance to be able to investigate or help to fix this issue. 1335324 10 clopez 2017-08-04 10:46:15 -0700 (In reply to Csaba Osztrogonác_OOO_until_21st_Aug from comment #9) > Reopen, because the problem is bigger now. :( > > Recent offlineasm changes amazingly increased the generating time. > Previously a clean build took (gnerating and compiling took) ~9 minutes, > https://trac.webkit.org/changeset/220047 increased it to ~30 minutes and > https://trac.webkit.org/changeset/220184 increased it to ~91 minutes. > > I haven't measured the exact runtime of the ruby scripts separately yet, but > it's obvious that these changes are responsible for long running ruby > scripts. > (I bet the slowdown isn't problem on native builds, but I can imagine that > the runtime was increased from few seconds to 1-2 minutes.) > > FYI, I stopped the JSCOnly ARM bots because of this blocker issue, > because I won't be available for the following 2 weeks, so I won't > have any chance to be able to investigate or help to fix this issue. My Ruby knowledge is near zero, but I find hard to believe r220184 has increased the build time in 60 minutes (from 30 to 91)!. Even when running on an ARM CPU, the diff on the ruby script don't seem that big or complex. 1337496 11 ossy 2017-08-10 11:02:06 -0700 (In reply to Csaba Osztrogonác_OOO_until_21st_Aug from comment #9) > Reopen, because the problem is bigger now. :( > > Recent offlineasm changes amazingly increased the generating time. > Previously a clean build took (gnerating and compiling took) ~9 minutes, > https://trac.webkit.org/changeset/220047 increased it to ~30 minutes and > https://trac.webkit.org/changeset/220184 increased it to ~91 minutes. > > I haven't measured the exact runtime of the ruby scripts separately yet, but > it's obvious that these changes are responsible for long running ruby > scripts. > (I bet the slowdown isn't problem on native builds, but I can imagine that > the runtime was increased from few seconds to 1-2 minutes.) > > FYI, I stopped the JSCOnly ARM bots because of this blocker issue, > because I won't be available for the following 2 weeks, so I won't > have any chance to be able to investigate or help to fix this issue. I made a quick measurement on X86_64 Linux. The runtime of the ruby scripts (generate_offset_extractor.rb and asm.rb) was 3-4 secs/script before r220047, which was increased to 14-15 secs by r220047 and was increased to 25-26 secs by r220184. It is a huge regression, the original runtime was increased to 8x. Keith, Filip, please investigate this huge regression as the authors of r220047 and r220184 changes? Thanks. 1337499 12 ossy 2017-08-10 11:04:21 -0700 (In reply to Carlos Alberto Lopez Perez from comment #10) > My Ruby knowledge is near zero, but I find hard to believe r220184 has > increased the build time in 60 minutes (from 30 to 91)!. Even when running > on an ARM CPU, the diff on the ruby script don't seem that big or complex. But it's true, the slowdown is near ~8x on X86_64. (Back to the ARM numbers, I remember the original runtime was ~10 minutes, now it is ~90 minutes) 1338122 13 ossy 2017-08-12 00:48:12 -0700 I investigated this huge regression a little bit, it seems the offlineasm scripts (generate_offset_extractor.rb) are doing fundamentally wrong things, they generates useless things. Before r220047 LLIntDesiredOffsets.h was 12Mb, now it is 87Mb sized, because there are zillion but one useless ifdefed code in it. There are 10 valid backends now: --------------------------------- OFFLINE_ASM_X86 !OFFLINE_ASM_X86_WIN !OFFLINE_ASM_X86_64 !OFFLINE_ASM_X86_64_WIN !OFFLINE_ASM_ARM !OFFLINE_ASM_ARMv7 !OFFLINE_ASM_ARMv7_TRADITIONAL !OFFLINE_ASM_ARM64 !OFFLINE_ASM_MIPS !OFFLINE_ASM_C_LOOP There are 8 misc options now: ------------------------------ OFFLINE_ASM_ARMv7k OFFLINE_ASM_ARMv7s OFFLINE_ASM_JSVALUE64 OFFLINE_ASM_BIG_ENDIAN OFFLINE_ASM_ASSERT_ENABLED OFFLINE_ASM_COLLECT_STATS OFFLINE_ASM_EXECUTION_TRACING OFFLINE_ASM_GIGACAGE_ENABLED generate_offset_extractor.rb now generates 3 ifdef block for each misc option combinations, which means 3 * 10 * 2^8 = 7680. And only 3 of them are valid, the remaining 7677 are completely useless for anything. Adding one more misc option to offlineasm, it doubles these configurations, doubles the script runtime and doubles the size of LLIntDesiredOffsets.h. It is completely useless and unsustainable. Why do we need to generate useless codes at all? It's obvious that - OFFLINE_ASM_ARMv7k and OFFLINE_ASM_ARMv7s can be true on Apple's iOS build - JSVALUE64 can be true only on 64 bit platforms (X86_64 and AArch64) - OFFLINE_ASM_BIG_ENDIAN is clearly defined by platform ... I think each backends are clearly define all misc options, we don't need to generate any code for invalid misc option combinations. And we don't need to genarate any code for a platform we don't build right now. 1338224 14 ossy 2017-08-13 02:03:26 -0700 I found a partial fix for this issue, we can pass the backend name(s) to generate_offset_extractor.rb and it will generate only that backend(s). This feature is introduced 3 years ago in https://trac.webkit.org/changeset/172777. With only one backend generation, a complete JSCOnly ARM clean build takes only 9-10 minutes. I added this local hack to the bots and will find how to pass ARMv7, ARMv7_TRADITIONAL or ARM64 to generate_offset_extractor.rb via cmake automatically. Additionally I still think that we don't need to generate 256 combinations for each backend when only one can be valid. We should find how to do it. 1338467 15 318061 ossy 2017-08-14 14:28:20 -0700 Created attachment 318061 Patch 1338657 16 318061 clopez 2017-08-15 05:15:21 -0700 Comment on attachment 318061 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=318061&action=review > Source/JavaScriptCore/CMakeLists.txt:1173 > + elseif (ARM_TRADITIONAL_DETECTED) > + set(OFFLINE_ASM_BACKEND "ARMv7_TRADITIONAL") > + endif () > + Please add an extra "elseif" for MIPS + elseif (WTF_CPU_ARM64) + set(OFFLINE_ASM_BACKEND "MIPS") 1339113 17 318244 ossy 2017-08-16 02:24:41 -0700 Created attachment 318244 Patch for landing 1339114 18 ossy 2017-08-16 02:26:08 -0700 (In reply to Csaba Osztrogonác_OOO_until_21st_Aug from comment #17) > Created attachment 318244 [details] > Patch for landing (In reply to Carlos Alberto Lopez Perez from comment #16) > Comment on attachment 318061 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=318061&action=review > > > Source/JavaScriptCore/CMakeLists.txt:1173 > > + elseif (ARM_TRADITIONAL_DETECTED) > > + set(OFFLINE_ASM_BACKEND "ARMv7_TRADITIONAL") > > + endif () > > + > Please add an extra "elseif" for MIPS > > + elseif (WTF_CPU_ARM64) > + set(OFFLINE_ASM_BACKEND "MIPS") I bet you meant this change ;) I added it. + elseif (WTF_CPU_MIPS) + set(OFFLINE_ASM_BACKEND "MIPS") 1339119 19 clopez 2017-08-16 02:56:35 -0700 (In reply to Csaba Osztrogonác_OOO_until_21st_Aug from comment #18) > (In reply to Csaba Osztrogonác_OOO_until_21st_Aug from comment #17) > > Created attachment 318244 [details] > > Patch for landing > > (In reply to Carlos Alberto Lopez Perez from comment #16) [....] > > Please add an extra "elseif" for MIPS > > > > + elseif (WTF_CPU_ARM64) > > + set(OFFLINE_ASM_BACKEND "MIPS") > > I bet you meant this change ;) I added it. > + elseif (WTF_CPU_MIPS) > + set(OFFLINE_ASM_BACKEND "MIPS") Ups! sure :) thanks! 1339121 20 318244 commit-queue 2017-08-16 03:06:17 -0700 Comment on attachment 318244 Patch for landing Clearing flags on attachment: 318244 Committed r220791: <http://trac.webkit.org/changeset/220791> 1339122 21 commit-queue 2017-08-16 03:06:19 -0700 All reviewed patches have been landed. Closing bug. 316805 2017-07-31 14:43:03 -0700 2017-08-14 14:28:16 -0700 Patch bug-174993-20170731144302.patch text/plain 1508 ossy U3VidmVyc2lvbiBSZXZpc2lvbjogMjIwMDczCmRpZmYgLS1naXQgYS9Ub29scy9DaGFuZ2VMb2cg Yi9Ub29scy9DaGFuZ2VMb2cKaW5kZXggMDRmMDBkYmI3MTE0MjFjM2RlY2UzOGZmYzc0OTJiOWIx NzllYzRmMy4uZDNkMGU4MGI1NDczZDAzZTljY2ZkZmRhNmQ2NTYwNDk1MTgzYmIzZiAxMDA2NDQK LS0tIGEvVG9vbHMvQ2hhbmdlTG9nCisrKyBiL1Rvb2xzL0NoYW5nZUxvZwpAQCAtMSwzICsxLDEz IEBACisyMDE3LTA3LTMxICBDc2FiYSBPc3p0cm9nb27DoWMgIDxvc3N5QHdlYmtpdC5vcmc+CisK KyAgICAgICAgRml4IEpTQ09ubHkgQVJNIGJ1aWxkYm90cyBhZnRlciByMjIwMDQ3CisgICAgICAg IGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD0xNzQ5OTMKKworICAgICAg ICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAqIEJ1aWxkU2xhdmVTdXBw b3J0L2J1aWxkLndlYmtpdC5vcmctY29uZmlnL21hc3Rlci5jZmc6CisgICAgICAgIChCdWlsZEFu ZFJlbW90ZUpTQ1Rlc3RzRmFjdG9yeS5fX2luaXRfXyk6CisKIDIwMTctMDctMzAgIFl1c3VrZSBT dXp1a2kgIDx1dGF0YW5lLnRlYUBnbWFpbC5jb20+CiAKICAgICAgICAgW1dURl0gSW50cm9kdWNl IFByaXZhdGUgU3ltYm9scwpkaWZmIC0tZ2l0IGEvVG9vbHMvQnVpbGRTbGF2ZVN1cHBvcnQvYnVp bGQud2Via2l0Lm9yZy1jb25maWcvbWFzdGVyLmNmZyBiL1Rvb2xzL0J1aWxkU2xhdmVTdXBwb3J0 L2J1aWxkLndlYmtpdC5vcmctY29uZmlnL21hc3Rlci5jZmcKaW5kZXggOTdhMWZiNGI5OGI3MTdh NjQ1YjhmNjc0ZGY3ZmM5NDk2ZmUzNmQxOC4uYjE3Nzc3MzU3MDE2M2E2MjdhNjgxOThmMGFiYjQ0 MmNmN2IzMWMzMSAxMDA2NDQKLS0tIGEvVG9vbHMvQnVpbGRTbGF2ZVN1cHBvcnQvYnVpbGQud2Vi a2l0Lm9yZy1jb25maWcvbWFzdGVyLmNmZworKysgYi9Ub29scy9CdWlsZFNsYXZlU3VwcG9ydC9i dWlsZC53ZWJraXQub3JnLWNvbmZpZy9tYXN0ZXIuY2ZnCkBAIC05NTgsNyArOTU4LDcgQEAgY2xh c3MgQnVpbGRBbmRSZW1vdGVKU0NUZXN0c0ZhY3RvcnkoRmFjdG9yeSk6CiAgICAgZGVmIF9faW5p dF9fKHNlbGYsIHBsYXRmb3JtLCBjb25maWd1cmF0aW9uLCBhcmNoaXRlY3R1cmVzLCB0cmlnZ2Vy cz1Ob25lLCBhZGRpdGlvbmFsQXJndW1lbnRzPU5vbmUsIFNWTk1pcnJvcj1Ob25lKToKICAgICAg ICAgRmFjdG9yeS5fX2luaXRfXyhzZWxmLCBwbGF0Zm9ybSwgY29uZmlndXJhdGlvbiwgYXJjaGl0 ZWN0dXJlcywgRmFsc2UsIGFkZGl0aW9uYWxBcmd1bWVudHMsIFNWTk1pcnJvcikKICAgICAgICAg c2VsZi5hZGRTdGVwKENvbXBpbGVKU0NPbmx5KCkpCi0gICAgICAgIHNlbGYuYWRkU3RlcChSdW5S ZW1vdGVKYXZhU2NyaXB0Q29yZVRlc3RzKCkpCisgICAgICAgIHNlbGYuYWRkU3RlcChSdW5SZW1v dGVKYXZhU2NyaXB0Q29yZVRlc3RzKHRpbWVvdXQ9NjAqNjApKQogCiAKIGNsYXNzIFRlc3RXZWJL aXQxTGVha3NGYWN0b3J5KEZhY3RvcnkpOgo= 318061 2017-08-14 14:28:20 -0700 2017-08-16 02:24:37 -0700 Patch bug-174993-20170814232819.patch text/plain 2515 ossy U3VidmVyc2lvbiBSZXZpc2lvbjogMjIwNjI4CmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2YVNjcmlw dENvcmUvQ2hhbmdlTG9nIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwppbmRleCAy MWI1YjA0ZDk3ZWEzMDYyODhlMmE3MjE3MGE4YzAzZDA1ZTEwMGM4Li44OWFmMTE3Mjc4ZGZiNTU1 Y2I0MjAzM2FhZmM2Y2RmODg4Mjk2ZmU0IDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENv cmUvQ2hhbmdlTG9nCisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKQEAgLTEs MyArMSwxMiBAQAorMjAxNy0wOC0xNCAgQ3NhYmEgT3N6dHJvZ29uw6FjICA8b3NzeUB3ZWJraXQu b3JnPgorCisgICAgICAgIEZpeCBKU0NPbmx5IEFSTSBidWlsZGJvdHMgYWZ0ZXIgcjIyMDA0NyBh bmQgcjIyMDE4NAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/ aWQ9MTc0OTkzCisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAg ICAgKiBDTWFrZUxpc3RzLnR4dDogR2VuZXJhdGUgb25seSBvbmUgYmFja2VuZCBvbiBMaW51eCB0 byBzYXZlIGJ1aWxkIHRpbWUuCisKIDIwMTctMDgtMTIgIEZpbGlwIFBpemxvICA8ZnBpemxvQGFw cGxlLmNvbT4KIAogICAgICAgICBGVEwncyBjb21waWxlR2V0VHlwZWRBcnJheUJ5dGVPZmZzZXQg bmVlZHMgdG8gZG8gY2FnaW5nCmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2YVNjcmlwdENvcmUvQ01h a2VMaXN0cy50eHQgYi9Tb3VyY2UvSmF2YVNjcmlwdENvcmUvQ01ha2VMaXN0cy50eHQKaW5kZXgg M2IzYjUxMzQwNzc5YTBkMDIyNzVhYzc1YjU4ZTgyY2IyNzZlYmM5Ny4uMGVlNTBkNWY4OGFkMTNm YjJiMzUyYTAzNjM0MjMxNmUxYjE1M2ViZCAxMDA2NDQKLS0tIGEvU291cmNlL0phdmFTY3JpcHRD b3JlL0NNYWtlTGlzdHMudHh0CisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DTWFrZUxpc3Rz LnR4dApAQCAtMTE1OCwxMSArMTE1OCwyOSBAQCBsaXN0KEFQUEVORCBKYXZhU2NyaXB0Q29yZV9I RUFERVJTCiAgICAgJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElSfS9CeXRlY29k ZXMuaAogKQogCitpZiAoKCR7Q01BS0VfU1lTVEVNX05BTUV9IFNUUkVRVUFMICJMaW51eCIpKQor ICAgIGlmIChXVEZfQ1BVX1g4NikKKyAgICAgICAgc2V0KE9GRkxJTkVfQVNNX0JBQ0tFTkQgIlg4 NiIpCisgICAgZWxzZWlmIChXVEZfQ1BVX1g4Nl82NCkKKyAgICAgICAgc2V0KE9GRkxJTkVfQVNN X0JBQ0tFTkQgIlg4Nl82NCIpCisgICAgZWxzZWlmIChXVEZfQ1BVX0FSTTY0KQorICAgICAgICBz ZXQoT0ZGTElORV9BU01fQkFDS0VORCAiQVJNNjQiKQorICAgIGVsc2VpZiAoQVJNX1RIVU1CMl9E RVRFQ1RFRCkKKyAgICAgICAgc2V0KE9GRkxJTkVfQVNNX0JBQ0tFTkQgIkFSTXY3IikKKyAgICBl bHNlaWYgKEFSTV9UUkFESVRJT05BTF9ERVRFQ1RFRCkKKyAgICAgICAgc2V0KE9GRkxJTkVfQVNN X0JBQ0tFTkQgIkFSTXY3X1RSQURJVElPTkFMIikKKyAgICBlbmRpZiAoKQorCisgICAgaWYgKE5P VCBFTkFCTEVfSklUKQorICAgICAgICBzZXQoT0ZGTElORV9BU01fQkFDS0VORCAiQ19MT09QIikK KyAgICBlbmRpZiAoKQorZW5kaWYgKCkKKwogYWRkX2N1c3RvbV9jb21tYW5kKAogICAgIE9VVFBV VCAke0RFUklWRURfU09VUkNFU19KQVZBU0NSSVBUQ09SRV9ESVJ9L0xMSW50RGVzaXJlZE9mZnNl dHMuaAogICAgIE1BSU5fREVQRU5ERU5DWSAke0pBVkFTQ1JJUFRDT1JFX0RJUn0vb2ZmbGluZWFz bS9nZW5lcmF0ZV9vZmZzZXRfZXh0cmFjdG9yLnJiCiAgICAgREVQRU5EUyAke0xMSU5UX0FTTX0g JHtPRkZMSU5FX0FTTX0gJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElSfS9Jbml0 Qnl0ZWNvZGVzLmFzbQotICAgIENPTU1BTkQgJHtSVUJZX0VYRUNVVEFCTEV9ICR7SkFWQVNDUklQ VENPUkVfRElSfS9vZmZsaW5lYXNtL2dlbmVyYXRlX29mZnNldF9leHRyYWN0b3IucmIgLUkke0RF UklWRURfU09VUkNFU19KQVZBU0NSSVBUQ09SRV9ESVJ9LyAke0pBVkFTQ1JJUFRDT1JFX0RJUn0v bGxpbnQvTG93TGV2ZWxJbnRlcnByZXRlci5hc20gJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQ VENPUkVfRElSfS9MTEludERlc2lyZWRPZmZzZXRzLmgKKyAgICBDT01NQU5EICR7UlVCWV9FWEVD VVRBQkxFfSAke0pBVkFTQ1JJUFRDT1JFX0RJUn0vb2ZmbGluZWFzbS9nZW5lcmF0ZV9vZmZzZXRf ZXh0cmFjdG9yLnJiIC1JJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElSfS8gJHtK QVZBU0NSSVBUQ09SRV9ESVJ9L2xsaW50L0xvd0xldmVsSW50ZXJwcmV0ZXIuYXNtICR7REVSSVZF RF9TT1VSQ0VTX0pBVkFTQ1JJUFRDT1JFX0RJUn0vTExJbnREZXNpcmVkT2Zmc2V0cy5oICR7T0ZG TElORV9BU01fQkFDS0VORH0KICAgICBWRVJCQVRJTSkKIAogIyBXZSBhZGQgdGhlIGhlYWRlciBm aWxlcyBkaXJlY3RseSB0byB0aGUgQUREX0VYRUNVVEFCTEUgY2FsbCBpbnN0ZWFkIG9mIHNldHRp bmcgdGhlCg== 318244 2017-08-16 02:24:41 -0700 2017-08-16 03:06:17 -0700 Patch for landing bug-174993-20170816112440.patch text/plain 2620 ossy U3VidmVyc2lvbiBSZXZpc2lvbjogMjIwNzkwCmRpZmYgLS1naXQgYS9Tb3VyY2UvSmF2YVNjcmlw dENvcmUvQ2hhbmdlTG9nIGIvU291cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwppbmRleCA3 ODcxMTE1MTE1MGExZTU3MDU4YWRhMTMzY2Q4NmZiNWUxMzhmYWRlLi42ZjA1Nzg3YjIyZTgxZTYw MjY0OWEzYTJjMDMyM2ZjZWNiZmRhOTU0IDEwMDY0NAotLS0gYS9Tb3VyY2UvSmF2YVNjcmlwdENv cmUvQ2hhbmdlTG9nCisrKyBiL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKQEAgLTEs MyArMSwxMiBAQAorMjAxNy0wOC0xNiAgQ3NhYmEgT3N6dHJvZ29uw6FjICA8b3NzeUB3ZWJraXQu b3JnPgorCisgICAgICAgIEZpeCBKU0NPbmx5IEFSTSBidWlsZGJvdHMgYWZ0ZXIgcjIyMDA0NyBh bmQgcjIyMDE4NAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/ aWQ9MTc0OTkzCisKKyAgICAgICAgUmV2aWV3ZWQgYnkgQ2FybG9zIEFsYmVydG8gTG9wZXogUGVy ZXouCisKKyAgICAgICAgKiBDTWFrZUxpc3RzLnR4dDogR2VuZXJhdGUgb25seSBvbmUgYmFja2Vu ZCBvbiBMaW51eCB0byBzYXZlIGJ1aWxkIHRpbWUuCisKIDIwMTctMDgtMTYgIEFuZHkgRXN0ZXMg IDxhZXN0ZXNAYXBwbGUuY29tPgogCiAgICAgICAgIFtQYXltZW50IFJlcXVlc3RdIEFkZCBhbiBF TkFCTEUgZmxhZyBhbmQgYW4gZXhwZXJpbWVudGFsIGZlYXR1cmUgcHJlZmVyZW5jZQpkaWZmIC0t Z2l0IGEvU291cmNlL0phdmFTY3JpcHRDb3JlL0NNYWtlTGlzdHMudHh0IGIvU291cmNlL0phdmFT Y3JpcHRDb3JlL0NNYWtlTGlzdHMudHh0CmluZGV4IDg5ODA5ZmFmZTRlODk5NDlmZjlmYjNhOTk3 NTdmYjQ0ODRlNzQ3OTEuLjA1ZjU2NzNiNjZlOTRlYjQ4YmQ2MWE0ZmE5YzU4NjAyY2QxNDBlYjMg MTAwNjQ0Ci0tLSBhL1NvdXJjZS9KYXZhU2NyaXB0Q29yZS9DTWFrZUxpc3RzLnR4dAorKysgYi9T b3VyY2UvSmF2YVNjcmlwdENvcmUvQ01ha2VMaXN0cy50eHQKQEAgLTExNjAsMTEgKzExNjAsMzEg QEAgbGlzdChBUFBFTkQgSmF2YVNjcmlwdENvcmVfSEVBREVSUwogICAgICR7REVSSVZFRF9TT1VS Q0VTX0pBVkFTQ1JJUFRDT1JFX0RJUn0vQnl0ZWNvZGVTdHJ1Y3RzLmgKICkKIAoraWYgKCgke0NN QUtFX1NZU1RFTV9OQU1FfSBTVFJFUVVBTCAiTGludXgiKSkKKyAgICBpZiAoV1RGX0NQVV9YODYp CisgICAgICAgIHNldChPRkZMSU5FX0FTTV9CQUNLRU5EICJYODYiKQorICAgIGVsc2VpZiAoV1RG X0NQVV9YODZfNjQpCisgICAgICAgIHNldChPRkZMSU5FX0FTTV9CQUNLRU5EICJYODZfNjQiKQor ICAgIGVsc2VpZiAoV1RGX0NQVV9BUk02NCkKKyAgICAgICAgc2V0KE9GRkxJTkVfQVNNX0JBQ0tF TkQgIkFSTTY0IikKKyAgICBlbHNlaWYgKEFSTV9USFVNQjJfREVURUNURUQpCisgICAgICAgIHNl dChPRkZMSU5FX0FTTV9CQUNLRU5EICJBUk12NyIpCisgICAgZWxzZWlmIChBUk1fVFJBRElUSU9O QUxfREVURUNURUQpCisgICAgICAgIHNldChPRkZMSU5FX0FTTV9CQUNLRU5EICJBUk12N19UUkFE SVRJT05BTCIpCisgICAgZWxzZWlmIChXVEZfQ1BVX01JUFMpCisgICAgICAgIHNldChPRkZMSU5F X0FTTV9CQUNLRU5EICJNSVBTIikKKyAgICBlbmRpZiAoKQorCisgICAgaWYgKE5PVCBFTkFCTEVf SklUKQorICAgICAgICBzZXQoT0ZGTElORV9BU01fQkFDS0VORCAiQ19MT09QIikKKyAgICBlbmRp ZiAoKQorZW5kaWYgKCkKKwogYWRkX2N1c3RvbV9jb21tYW5kKAogICAgIE9VVFBVVCAke0RFUklW RURfU09VUkNFU19KQVZBU0NSSVBUQ09SRV9ESVJ9L0xMSW50RGVzaXJlZE9mZnNldHMuaAogICAg IE1BSU5fREVQRU5ERU5DWSAke0pBVkFTQ1JJUFRDT1JFX0RJUn0vb2ZmbGluZWFzbS9nZW5lcmF0 ZV9vZmZzZXRfZXh0cmFjdG9yLnJiCiAgICAgREVQRU5EUyAke0xMSU5UX0FTTX0gJHtPRkZMSU5F X0FTTX0gJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElSfS9Jbml0Qnl0ZWNvZGVz LmFzbQotICAgIENPTU1BTkQgJHtSVUJZX0VYRUNVVEFCTEV9ICR7SkFWQVNDUklQVENPUkVfRElS fS9vZmZsaW5lYXNtL2dlbmVyYXRlX29mZnNldF9leHRyYWN0b3IucmIgLUkke0RFUklWRURfU09V UkNFU19KQVZBU0NSSVBUQ09SRV9ESVJ9LyAke0pBVkFTQ1JJUFRDT1JFX0RJUn0vbGxpbnQvTG93 TGV2ZWxJbnRlcnByZXRlci5hc20gJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElS fS9MTEludERlc2lyZWRPZmZzZXRzLmgKKyAgICBDT01NQU5EICR7UlVCWV9FWEVDVVRBQkxFfSAk e0pBVkFTQ1JJUFRDT1JFX0RJUn0vb2ZmbGluZWFzbS9nZW5lcmF0ZV9vZmZzZXRfZXh0cmFjdG9y LnJiIC1JJHtERVJJVkVEX1NPVVJDRVNfSkFWQVNDUklQVENPUkVfRElSfS8gJHtKQVZBU0NSSVBU Q09SRV9ESVJ9L2xsaW50L0xvd0xldmVsSW50ZXJwcmV0ZXIuYXNtICR7REVSSVZFRF9TT1VSQ0VT X0pBVkFTQ1JJUFRDT1JFX0RJUn0vTExJbnREZXNpcmVkT2Zmc2V0cy5oICR7T0ZGTElORV9BU01f QkFDS0VORH0KICAgICBWRVJCQVRJTSkKIAogIyBXZSBhZGQgdGhlIGhlYWRlciBmaWxlcyBkaXJl Y3RseSB0byB0aGUgQUREX0VYRUNVVEFCTEUgY2FsbCBpbnN0ZWFkIG9mIHNldHRpbmcgdGhlCg==