[meta] 9.0.0 Release Blockers

[tstellar]

Bugzilla Link	42474
Resolution	FIXED
Resolved on	Sep 25, 2020 16:06
Version	9.0
OS	All
Depends On	#42206 #32583 #37839 #39829 #40020 #40038 #40107 #40117 #40355 #40557 #40733 #41072 #41820 #41858 #41920 #41931 #41948 #41991 #42010 #42012 #42014 #42015 #42020 #42022 #42023 #42024 #42030 #42056 #42057 #42060 #42064 #42069 #42070 #42084 #42085 #42097 #42105 #42108 #42114 #42120 #42126 #42135 #42136 #42157 #42164 #42168 #42169 #42171 #42174 #42178 #42181 #42188 #42200 #42202 #42204 #42229 #42232 #42235 #42236 #42238 #42252 #42266 #42271 #42272 #42280 #42307 #42311 #42323 #42330 #42333 #42336 #42337 #42356 #42363 #42366 #42375 #42401 #42404 #42407 #42419 #42440 #42448 #42463 #42466 #42477 #42482 #42493 #42504 #42512 #42523 #42541 #42549 #42567 #42570 #42575 #42578 #42581 #42588 #42589 #42613 #42654
CC	@alexey-bataev,@beccadax,@davidbolvansky,@efriedma-quic,@hahnjo,@JosephTremoulet,@jsonn,@JVApen,@RKSimon,@modocache,@pawelsopensource,@tlively,@wang-bin,@zoecarver

[tstellar]

assigned to @zmodem

[llvmbot]

Hans could we consider #40038 to be a blocker for 9.0?

[zmodem]

Hans could we consider #40038 to be a
blocker for 9.0?

Yes, added it.

[modocache]

I added llvm/llvm-bugzilla-archive#42827 as a blocker, please triage as appropriate. Thank you!

[jdoerfert]

Hans could we consider #40038 to be a
blocker for 9.0?

Yes, added it.

Fixed in r367387. Do I need to create a new bug as a merge request?

[zmodem]

Fixed in r367387. Do I need to create a new bug as a merge request?

(Forgot to reply here: no, I've merged it already as commented on the bug.)

[llvmbot]

I'd like to nominate #42266 as blocker. I'm hoping it's easy and low risk for someone who knows the right part of the build system.

[jsonn]

Please merge r368104 (clang) and r367750 (LLVM) part. This fixes a regression since the 8.0 release. One part for addressing the issues for -O0 is still under review, but improving the situation for optimized builds is already a huge step.

[zmodem]

Please merge r368104 (clang) and r367750 (LLVM) part. This fixes a
regression since the 8.0 release. One part for addressing the issues for -O0
is still under review, but improving the situation for optimized builds is
already a huge step.

I started with the LLVM patch, but it doesn't apply cleanly on the branch, for example, RISCVTargetLowering::getConstraintType() didn't exist when we branched.

Jörg or Bill, could you prepare a patch that applies cleanly against the branch?

[zmodem]

Please merge r368104 (clang) and r367750 (LLVM) part. This fixes a
regression since the 8.0 release. One part for addressing the issues for -O0
is still under review, but improving the situation for optimized builds is
already a huge step.

I started with the LLVM patch, but it doesn't apply cleanly on the branch,
for example, RISCVTargetLowering::getConstraintType() didn't exist when we
branched.

Jörg or Bill, could you prepare a patch that applies cleanly against the
branch?

Actually, I just ended up merging r367403 to unblock it.

I've merged r367750 in r368421 and r368104(+368202) in r368422.

[beccadax]

Should #42366 be a blocker? A reviewer of the fix suggested it should be.

[zoecarver]

is_base_of_union.pass.cpp fails in C++03. Maybe it should be a release blocker?

[zoecarver]

Nevermind, that is just on apple-clang-11.0.

[zmodem]

Should #42366 be a blocker? A reviewer
of the fix suggested it should be.

Yes, that seems reasonable. I see it's marked as a blocker now.

[llvmbot]

Can llvm/llvm-bugzilla-archive#42924 be considered a blocker? It is causing timeouts for us when building with AddressSanitizer.

[zmodem]

Can llvm/llvm-bugzilla-archive#42924 be considered a blocker? It
is causing timeouts for us when building with AddressSanitizer.

Sorry, I think it's too late to do anything about it for 9.0.0. I've put in on my list of things for 9.0.1.

[llvmbot]

Please #42188 Clang-cl -std=c++17 doesn't handle the constexpr symbol correctly. I looked at the recent clang commit. No one seems to be dealing with this issue. If this release is not blocked, clang 9.0 may fail on Windows.

[llvmbot]

Please #42188 Clang-cl -std=c++17
doesn't handle the constexpr symbol correctly. I looked at the recent clang
commit. No one seems to be dealing with this issue. If this release is not
blocked, clang 9.0 may fail on Windows.

Sorry, here is a correction, clang-cl -std=c++14 (9.0) did not handle constexpr, not -std=c++17.

[zmodem]

Please #42188 Clang-cl -std=c++17
doesn't handle the constexpr symbol correctly. I looked at the recent clang
commit. No one seems to be dealing with this issue. If this release is not
blocked, clang 9.0 may fail on Windows.

Sorry, here is a correction, clang-cl -std=c++14 (9.0) did not handle
constexpr, not -std=c++17.

I had not seen this before. I'll take a look and reply on the bug.

[JosephTremoulet]

I'd like to request that 43229 get triaged, and considered for 9.0.1 if not 9.0.0 . Thanks.

[zmodem]

I'd like to request that 43229 get triaged, and considered for 9.0.1 if not
9.0.0 . Thanks.

I've put it on my 9.0.1 list.

[zmodem]

The -final tag is in.

9.0.1 blockers are tracked by llvm/llvm-bugzilla-archive#43360

[tstellar]

mentioned in issue #41820

[llvmbot]

mentioned in issue #41858

[aheejin]

mentioned in issue #41920

[ldalessa]

mentioned in issue #41931

[bsdkurt]

mentioned in issue #41948

[llvmbot]

mentioned in issue #41991

[llvmbot]

mentioned in issue #42010

[sam-mccall]

mentioned in issue #42014