Upgrade to D v2.079.1 #2587

kinke · 2018-02-20T01:17:20Z

No description provided.

Older versions suffer from a frontend bug (for MSVC targets) causing a crash.

kinke · 2018-02-21T00:54:19Z

Status update: LDC + druntime/Phobos building fine on Win64 and Linux x64, most lit tests already green.

Actually for druntime too, where it's unused.

It's for MSVC targets only and comes with a custom win64.mak makefile, so that running (GNU) make in the test dir doesn't work. I performed the test manually on Win64, works.

* `Error: Using the result of a comma expression is not allowed` * A test for DMD issue #16022 is in dmd-testsuite's runnable/xtest46.d.

The frontend now errors out when coming across a typeid expression in betterC mode. That error is checked in dmd-testsuite's fail_compilation/betterc.d.

The expected warning now properly results in exit code != 0 with `-w`.

Compiling the std.range.package unittests requires almost 1.8 GB on Win64, that's even slightly more than the regex ones. std.algorithm.searching is at 1.7 GB.

kinke · 2018-02-24T15:48:00Z

Status update on Win64: all unittests + druntime standalone tests green; 2 failing lit-tests (PGO/hash.d and codegen/betterC_typeinfo.d).
There's a bunch of dmd-testsuite failures, primarily due to missing support for new -i and -Xi switches as well as a betterC TypeInfo regression (erroring out whenever coming across a struct in betterC mode). The new upstream minimal/pay-as-you-go druntime support will also need to be checked.

=> looking good; I expect similar results on Linux x64, but getting CI to work is a bit tedious (apparently more Phobos unittest module compilations requiring significantly more than 1 GB of memory; 32-bit std.file unittests crashing for Semaphore, possibly due to their RAM disk...).
OSX builds quickly die during CMake setup (with the intermediate compiler as host compiler) due to duplicate symbols; I haven't looked further into that.

Don't attempt to emit the TypeInfo for each struct declaration. We skip the IR emission later on for betterC anyway, but starting with 2.079, the frontend emits an error when invoking genTypeInfo() in betterC mode.

This fixes lit-test codegen/betterC_typeinfo.d.

Conflicts: .circleci/config.yml runtime/druntime runtime/phobos tests/d2/dmd-testsuite

The critical section is lowered to a `static shared` global now; previously, it was `[static] __gshared`.

Conflicts: appveyor.yml driver/main.cpp runtime/druntime runtime/phobos

kinke · 2018-04-15T17:41:53Z

Ah, feels good:

LDC - the LLVM D compiler (1.9.0git-025c69c):
  based on DMD v2.079.1 and LLVM 6.0.0

So there's one more remaining issue, runnable/testconst.d segfaulting for Travis Linux with BUILD_SHARED_LIBS=ON and enabled optimizations.

kinke · 2018-04-15T18:18:25Z

I cannot reproduce the runnable/testconst.d issue on my Linux system with LLVM 6 and BUILD_SHARED_LIBS=ON.

kinke · 2018-04-18T21:35:04Z

As we don't have SSH access for Travis (at least I don't), I quickly modified the Semaphore config to build with BUILD_SHARED_LIBS=ON and planned on interactively debugging it, but it works. It's using (LDC-)LLVM 4.0.1; Travis fails with vanilla LLVM 4.0.1 (and 3.9.1, these are the 2 Travis Linux jobs with shared libs only; the OSX shared-libs only job with LLVM 4.0.0 works fine). I'm going to experimentally set BUILD_SHARED_LIBS=ON for all Travis jobs.

kinke · 2018-04-19T00:00:04Z

So Travis Linux fails in general, not depending on LLVM version. It's based on Ubuntu 14.04, just like Semaphore. Awaiting Circle results (Ubuntu 14.04 too).

kinke · 2018-04-19T14:01:22Z

Working fine on Circle Linux too. So I guess it's somehow related to the Travis environment; I propose to publish 1.9-beta1 as-is and check if people run into any issues.

joakim-noah · 2018-04-19T15:59:27Z

Sounds good.

kinke · 2018-04-19T20:34:03Z

Using BUILD_SHARED_LIBS=ON for Circle Mac too revealed another issue; exceptions in d_do_test (then linked against the shared libs) aren't caught (_Unwind_RaiseException() returning with _URC_END_OF_STACK, as if there weren't any catch clauses in the call stack - there's at least one explicit one in that tester plus the implicit root one in druntime), leading to many dmd-testsuite failues. The symptoms are the same as for #2585 (uncaught exception + abort), so it may not be a new issue.

Call stack from Circle SSH, just in case:

  * frame #0: 0x00007fffa34d0d42 libsystem_kernel.dylib`__pthread_kill + 10
    frame #1: 0x00007fffa35be457 libsystem_pthread.dylib`pthread_kill + 90
    frame #2: 0x00007fffa3436420 libsystem_c.dylib`abort + 129
    frame #3: 0x00000001008b4c83 libdruntime-ldc-shared.dylib`_d_throw_exception + 227
    frame #4: 0x00000001002bafb5 libphobos2-ldc-shared.dylib`_D3std9exception__T7enforceZ__TQmTbZQrFNaNfbLAxaAyamZb + 261
    frame #5: 0x0000000100005a35 d_do_test`_D9d_do_test7executeFKS3std5stdio4FileAyabQeZQh(f=0x00007fff5fbfe380, command=(length = 46, ptr = 0x0000000100d02140), expectpass=true, result_path=(length = 26, ptr = 0x0000000100d00080)) at d_do_test.d:406
    frame #6: 0x000000010000a67a d_do_test`_D9d_do_test7tryMainFAAyaZ15testCombinationMFbQymQBbZEQCaQBtFQBoZ6Result(autoCompileImports=false, argSet=(length = 0, ptr = 0x00000001000851c8), permuteIndex=0, permutedArgs=(length = 0, ptr = 0x0000000000000000)) at d_do_test.d:789
    frame #7: 0x000000010000891f d_do_test`_D9d_do_test7tryMainFAAyaZi(args=string[] @ 0x00007fff5fbfef88) at d_do_test.d:848
    frame #8: 0x00000001000071ad d_do_test`_Dmain(args=string[] @ 0x00007fff5fbfeff0) at d_do_test.d:517
    frame #9: 0x00000001008b4640 libdruntime-ldc-shared.dylib`_D2rt6dmain211_d_run_mainUiPPaPUAAaZiZ6runAllMFZv + 112
    frame #10: 0x00000001008b44a1 libdruntime-ldc-shared.dylib`_d_run_main + 433
    frame #11: 0x00000001000815a5 d_do_test`main(argc=4, argv=0x00007fff5fbff1d8) at __entrypoint.d:8
    frame #12: 0x00007fffa33a2235 libdyld.dylib`start + 1
    frame #13: 0x00007fffa33a2235 libdyld.dylib`start + 1

joakim-noah · 2018-04-20T10:27:19Z

I'm seeing a strange error when compiling for Android/ARM, which isn't reproducible when building for Linux/x64. Here's the command and error:

./bin/ldc2 -c --output-o -conf= -w -mcpu=cortex-a8 -mtriple=armv7-none-linux-android -O3 -release -unittest -d-version=StdUnittest -I../runtime/druntime/src -I../runtime/phobos -of=timezone.o ../runtime/phobos/std/datetime/timezone.d

std/datetime/timezone.d(2845): Error: undefined identifier `DateTimeException`

You should be able to reproduce with that command, as long as your llvm has the ARM backend. No idea why this frontend error would vary by arch.

kinke · 2018-04-20T11:57:25Z

std/datetime/timezone.d(2845)

Notice the relative path - it's not ../runtime/phobos/std/datetime/timezone.d. Try running the compilation command from another working directory.

joakim-noah · 2018-04-20T12:43:10Z

No, that's not the problem. I took the generated command spit out by ninja and modified it so you could run it from a build directory easily, ie without my absolute paths for druntime and Phobos. Our config builds that file from the Phobos directory, hence that error I cut and pasted, but this command should spit out the same error but with the relative path, at least it did when I just tried it.

kinke · 2018-04-20T13:16:19Z

This looks like an (upstream?) regression for Android, as that line 2845 is in a version(Android) block. In other places, e.g., here, the symbol is imported before using it.

joakim-noah · 2018-04-20T16:01:46Z

Yes, you're right, it's obviously a regression upstream from scoping the imports and not testing with Android. I'm so used to there being almost no Android differences in Phobos that I didn't even look at that file, but assumed it was a dmd frontend issue once it worked fine for linux.

Sorry about the noise, adding the import easily fixes it.

kinke · 2018-04-20T19:06:33Z

Glad you noticed so we can fix it before tagging beta1.

joakim-noah · 2018-04-21T08:37:25Z

Alright, with the simple fix for that regression that I merged to this branch at ldc-developers/phobos@eb79925, all the stdlib unit tests pass for Android/ARM, except for the single long-standing ARM EH inlining issue. Only two more tests fail from the dmd testsuite after updating from ldc 1.8 to this 1.9 branch, likely nothing substantial but haven't looked into them.

I also tried building the stdlib tests for Android/AArch64, pretty much the same results as 1.8, with only two different floating-point-related failures in std.complex.

Conflicts: runtime/druntime runtime/phobos

Conflicts: runtime/druntime

kinke · 2018-04-21T16:06:21Z

Using BUILD_SHARED_LIBS=ON for Circle Mac too revealed another issue

... which is gone with #2659, see #2660.

Conflicts: driver/cl_options.h driver/configfile.h ldc2.conf.in ldc2_phobos.conf.in

kinke added 10 commits February 20, 2018 02:04

Upgrade to D v2.079.0-beta.1

df4f854

druntime: Tiny fix for ldc.eh_msvc

fabdc55

AppVeyor: Upgrade host compiler to v1.8.0-beta1

30f7fab

Older versions suffer from a frontend bug (for MSVC targets) causing a crash.

dmd.declaration: Fix invalid override (additional const)

758ca55

dmd.dsymbol: Fix overrides (missing Loc => const Loc &)

7bfb72b

Fix more front-end C++ header regressions (by-ref Loc)

b8909f4

And some more C++ header fixes

050d9e9

Another C++ header fix wrt. FuncDeclaration vtable

c7c681d

druntime: Proper fix for ldc.eh_msvc

3d59944

Adapt tests/semantic/dcompute.d to new backticks in error msg

926ae1e

kinke added 8 commits February 23, 2018 20:11

CMake: Add -d-version=StdUnittest for Phobos unittests

15a3c60

Actually for druntime too, where it's unused.

CMake: Only include new DragonFly BSD druntime modules when appropriate

ccf14b7

Disable druntime standalone test 'uuid'

a9b0184

It's for MSVC targets only and comes with a custom win64.mak makefile, so that running (GNU) make in the test dir doesn't work. I performed the test manually on Win64, works.

Remove lit-test tests/codegen/enum_vardecl.d

4d198d1

* `Error: Using the result of a comma expression is not allowed` * A test for DMD issue #16022 is in dmd-testsuite's runnable/xtest46.d.

Adapt lit-test tests/codegen/betterC_typeinfo.d

0112e20

The frontend now errors out when coming across a typeid expression in betterC mode. That error is checked in dmd-testsuite's fail_compilation/betterc.d.

Fix lit-test tests/codegen/attr_llvmFMF.d

17d9125

The expected warning now properly results in exit code != 0 with `-w`.

Phobos: Minimally relax std.complex unittest for 64-bit reals

92f2a27

Adapt CI scripts to memory-hungry Phobos unittest modules

0944965

Compiling the std.range.package unittests requires almost 1.8 GB on Win64, that's even slightly more than the regex ones. std.algorithm.searching is at 1.7 GB.

kinke force-pushed the merge-2.079 branch from 663d4fb to 0944965 Compare February 24, 2018 15:26

kinke added 5 commits February 24, 2018 18:25

dmd-testsuite: Remove superfluous -i build flag for sanitize_json tool

aa34136

Fix betterC regression for struct TypeInfos

97514dd

Don't attempt to emit the TypeInfo for each struct declaration. We skip the IR emission later on for betterC anyway, but starting with 2.079, the frontend emits an error when invoking genTypeInfo() in betterC mode.

Adapt to frontend injecting implicit 'return 0' for betterC main

1b8a3b9

This fixes lit-test codegen/betterC_typeinfo.d.

Upgrade to D v2.079.0-beta.2

1251272

Merge branch 'master' into merge-2.079

6eceb91

Conflicts: .circleci/config.yml runtime/druntime runtime/phobos tests/d2/dmd-testsuite

kinke changed the title ~~WIP: Upgrade to D v2.079.0-beta.1~~ WIP: Upgrade to D v2.079.0-beta.2 Feb 25, 2018

kinke added 3 commits February 25, 2018 16:18

Adapt lit-test codegen/gh1955.d to changed critical section mangling

13044c6

The critical section is lowered to a `static shared` global now; previously, it was `[static] __gshared`.

Merge branch 'master' into merge-2.079

5608825

Adapt lit-test PGO/hash.d to now fatal warnings with -w

112b916

kinke changed the title ~~WIP: Upgrade to D v2.079.0~~ Upgrade to D v2.079.1 Apr 15, 2018

kinke changed the title ~~Upgrade to D v2.079.1~~ WIP: Upgrade to D v2.079.1 Apr 15, 2018

Merge branch 'master' into merge-2.079

025c69c

Conflicts: appveyor.yml driver/main.cpp runtime/druntime runtime/phobos

CircleCI: Disable another test on macOS due to unexpected linker warning

ddf0031

kinke added 2 commits April 21, 2018 16:00

Phobos: Fix Android regression in std.datetime.timezone

4356f70

Merge branch 'master' into merge-2.079

fb7fd86

Conflicts: runtime/druntime runtime/phobos

kinke force-pushed the merge-2.079 branch from 49ca346 to fb7fd86 Compare April 21, 2018 14:03

kinke added 2 commits April 21, 2018 16:32

druntime: Fix another assembly warning for MIPS32

2b17fab

Merge branch 'master' into merge-2.079

4509340

Conflicts: runtime/druntime

kinke added 3 commits April 21, 2018 21:26

Upgrade dub to v1.8.1

571dc9a

Travis: Allow Linux BUILD_SHARED_LIBS=ON failures for now

8c55682

Merge branch 'master' into merge-2.079

6812e0b

Conflicts: driver/cl_options.h driver/configfile.h ldc2.conf.in ldc2_phobos.conf.in

kinke merged commit a9c76cf into master Apr 22, 2018

kinke deleted the merge-2.079 branch April 22, 2018 12:32

kinke changed the title ~~WIP: Upgrade to D v2.079.1~~ Upgrade to D v2.079.1 Apr 22, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Upgrade to D v2.079.1 #2587

Upgrade to D v2.079.1 #2587

kinke commented Feb 20, 2018

kinke commented Feb 21, 2018

kinke commented Feb 24, 2018 •

edited

Loading

kinke commented Apr 15, 2018 •

edited

Loading

kinke commented Apr 15, 2018 •

edited

Loading

kinke commented Apr 18, 2018 •

edited

Loading

kinke commented Apr 19, 2018

kinke commented Apr 19, 2018

joakim-noah commented Apr 19, 2018

kinke commented Apr 19, 2018 •

edited

Loading

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 21, 2018

kinke commented Apr 21, 2018

Upgrade to D v2.079.1 #2587

Upgrade to D v2.079.1 #2587

Conversation

kinke commented Feb 20, 2018

kinke commented Feb 21, 2018

kinke commented Feb 24, 2018 • edited Loading

kinke commented Apr 15, 2018 • edited Loading

kinke commented Apr 15, 2018 • edited Loading

kinke commented Apr 18, 2018 • edited Loading

kinke commented Apr 19, 2018

kinke commented Apr 19, 2018

joakim-noah commented Apr 19, 2018

kinke commented Apr 19, 2018 • edited Loading

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 20, 2018

kinke commented Apr 20, 2018

joakim-noah commented Apr 21, 2018

kinke commented Apr 21, 2018

kinke commented Feb 24, 2018 •

edited

Loading

kinke commented Apr 15, 2018 •

edited

Loading

kinke commented Apr 15, 2018 •

edited

Loading

kinke commented Apr 18, 2018 •

edited

Loading

kinke commented Apr 19, 2018 •

edited

Loading