search

llvm / llvm-project

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
http://llvm.org
Other
29.06k stars 11.98k forks source link

Workaround "error: ‘(9.223372036854775807e+18 / 1.0e+9)’ is not a constant expression" #39044

Closed 70311826-e6d8-4d29-8604-c4ec4bfa7f0d closed 2 years ago

70311826-e6d8-4d29-8604-c4ec4bfa7f0d commented 5 years ago
Bugzilla Link 39696
Resolution FIXED
Resolved on Jun 10, 2020 10:36
Version 7.0
OS Linux
CC @ikitayama,@jdenny-ornl,@kencu,@mclow,@noloader

Extended Description

I'm trying to build LLVM and components on GCC112 from the compiler farm. GCC112 is ppc64-le, and ships with GCC 4.8.5:

$ gcc --version gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-16) Copyright (C) 2015 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

LLVM configuration and build goes well for a while. Then it encounters https://bugzilla.redhat.com/show_bug.cgi?id=1538817:

[ 35%] Building CXX object projects/libcxx/lib/CMakeFiles/cxx_objects.dir//src/condition_variable.cpp.o cd /home/noloader/llvm_build/projects/libcxx/lib && /usr/bin/c++ -DNDEBUG -D_DEBUG -D_GNU_SOURCE -D_LIBCPP_BUILDING_LIBRARY -D_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER -DSTDC_CONSTANT_MACROS -DSTDC_FORMAT_MACROS -DSTDC_LIMIT_MACROS -I/home/noloader/llvm_build/projects/libcxx/lib -I/home/noloader/llvm_source/llvm/projects/libcxx/lib -I/home/noloader/llvm_build/include -I/home/noloader/llvm_source/llvm/include -I/home/noloader/llvm_build/projects/libcxx/include/c++build -I/home/noloader/llvm_source/llvm/projects/libcxx/include -fPIC -fvisibility-inlines-hidden -std=c++11 -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wno-missing-field-initializers -Wno-long-long -Wno-maybe-uninitialized -Wdelete-non-virtual-dtor -Wno-comment -g -DLIBCXX_BUILDING_LIBCXXABI -std=c++11 -nostdinc++ -fvisibility-inlines-hidden -Wall -Wextra -W -Wwrite-strings -Wno-unused-parameter -Wno-long-long -Werror=return-type -Wno-literal-suffix -Wno-c++14-compat -Wno-noexcept-type -Wno-error -fPIC -o CMakeFiles/cxx_objects.dir//src/condition_variable.cpp.o -c /home/noloader/llvm_source/llvm/projects/libcxx/src/condition_variable.cpp In file included from /home/noloader/llvm_source/llvm/projects/libcxx/src/condition_variable.cpp:15:0: /home/noloader/llvm_source/llvm/projects/libcxx/include/thread: In function ‘void std::1::this_thread::sleep_for(const std::1::chrono::duration<_Rep, _Period>&)’: /home/noloader/llvm_source/llvm/projects/libcxx/include/thread:438:73: in constexpr expansion of ‘std::1::chrono::duration((*(const std::1::chrono::duration<long long int, std::1::ratio<1l, 1000000000l> >*)(& std::1::chrono::duration<_Rep, _Period>::max<long long int, std::__1::ratio<1l, 1000000000l> >())), 0u)’ /home/noloader/llvm_source/llvm/projects/libcxx/include/chrono:1055:68: in constexpr expansion of ‘std::1::chrono::duration_cast<std::1::chrono::duration, long long int, std::1::ratio<1l, 1000000000l> >(( & d))’ /home/noloader/llvm_source/llvm/projects/libcxx/include/chrono:907:72: in constexpr expansion of ‘std::1::chrono::duration_cast<std::1::chrono::duration<long long int, std::1::ratio<1l, 1000000000l> >, std::1::chrono::duration, std::1::ratio<1l, 1000000000l>, true, false>().std::1::chrono::duration_cast<_FromDuration, _ToDuration, _Period, true, false>::operator()<std::1::chrono::duration<long long int, std::1::ratio<1l, 1000000000l> >, std::1::chrono::duration, std::__1::ratio<1l, 1000000000l> >(( & __fd))’ /home/noloader/llvm_source/llvm/projects/libcxx/include/thread:438:73: error: ‘(9.223372036854775807e+18 / 1.0e+9)’ is not a constant expression _LIBCPP_CONSTEXPR duration _Max = nanoseconds::max(); ^ At global scope: cc1plus: warning: unrecognized command line option "-Wno-noexcept-type" [enabled by default] cc1plus: warning: unrecognized command line option "-Wno-c++14-compat" [enabledby default] make[2]: [projects/libcxx/lib/CMakeFiles/cxx_objects.dir/__/src/condition_variable.cpp.o] Error 1 make[2]: Leaving directory `/home/noloader/llvm_build' make[1]: [projects/libcxx/lib/CMakeFiles/cxx_objects.dir/all] Error 2 make[1]: Leaving directory `/home/noloader/llvm_build' make: *** [all] Error 2 Failed to make LLVM sources

It would be nice if LLVM could work around the issue.

mclow commented 2 years ago

mentioned in issue llvm/llvm-bugzilla-archive#40671

jdenny-ornl commented 4 years ago

Fixed: https://reviews.llvm.org/D81438

jdenny-ornl commented 4 years ago

Ping.I suppose I should just take the patch to phabricator.

jdenny-ornl commented 4 years ago

Ping. What's the best way to move this forward?

jdenny-ornl commented 4 years ago

We have recently encountered this bug and have found that the patch in llvm/llvm-project#39044 #c38 fixes it. What stands in the way of this being pushed? Would it help if I created a phab review for it?

Ping.

jdenny-ornl commented 4 years ago

We have recently encountered this bug and have found that the patch in llvm/llvm-project#39044 #c38 fixes it. What stands in the way of this being pushed? Would it help if I created a phab review for it?

llvmbot commented 4 years ago

I can also confirm that this is an issue in current master / trunk (llvm 10) Here is my patch which I have incorporated in an spack package. diff --git a/libcxx/include/thread b/libcxx/include/thread index 02da703..d1677a1 100644 --- a/projects/libcxx/include/thread +++ b/projects/libcxx/include/thread @@ -368,9 +368,9 @@ sleep_for(const chrono::duration<_Rep, _Period>& __d) {

if defined(_LIBCPP_COMPILER_GCC) && (powerpc || POWERPC)

 //  GCC's long double const folding is incomplete for IBM128 long doubles.
llvmbot commented 5 years ago

I can confirm that as of now (SHA cda334ba5417d7702be755adc2f8414c877f0482) this fix is still reversed on master. Adding a NOT for the if condition (or switching the bodies for if and else) would fix it

llvmbot commented 5 years ago

201908190440 clang-8.0.1 build log on ppc64le - power9 I've attached the build log. I deleted the install messages, since I can only post attachments upto 1MB here.

ikitayama commented 5 years ago

I'd like to see the nightly build logs as I am on POWER8 and do build LLVM routinely by myself, when failture occurs I'd like to cross-check against it.

llvmbot commented 5 years ago

I'll rebuild it again and attach the logs in an hour or so. What do you hope to see in the logs?

ikitayama commented 5 years ago

Created attachment 22388 [details] Build clang-8.0.1 on linux shell script.

Test on IBM Power9 Ubuntu-18.04 for ppc64le.

Do you have build logs carried out on POWER9?

llvmbot commented 5 years ago

I've attached a shell script for building clang-8.0.1 on linux. It's been test on a Power9 ppc64le system.

llvmbot commented 5 years ago

Patch for building llvm-8.0.1 on ppc64el.

llvmbot commented 5 years ago

Build clang-8.0.1 on linux shell script. Test on IBM Power9 Ubuntu-18.04 for ppc64le.

ikitayama commented 5 years ago

I confirm that I am able to build one of the LLVM projects, libcxx and libcxxabi on PowerPC natively for the first time ever. Time for changing the doc to Supported on Linux and PowerPC combination?

ikitayama commented 5 years ago

And note that C++14 now that is a requirement.

ikitayama commented 5 years ago

Has this been landed on Trunk yet?

llvmbot commented 5 years ago

This is the correct patch for ppc64el. I've tested this on a Raptor System Talos II Power9 workstation.

diff --git a/libcxx/include/thread b/libcxx/include/thread index 8c0115f87..e439f60b9 100644 --- a/libcxx/include/thread +++ b/libcxx/include/thread @@ -435,7 +435,12 @@ sleep_for(const chrono::duration<_Rep, _Period>& d) using namespace chrono; if (d > duration<_Rep, _Period>::zero()) { +#if defined(_LIBCPP_COMPILER_GCC) && (powerpc || POWERPC)

kencu commented 5 years ago

I must be cross-eyed.

To me the original commit looks correct, and the corrected fix now looks backwards.

When building with gcc, on powerpc, don't we want to use

_LIBCPP_CONSTEXPR duration _Max = duration(ULLONG_MAX/1000000000ULL) ;

Because if so, we have that backwards now, don't we?

ikitayama commented 5 years ago

BTW, it builds fine on x86.

ikitayama commented 5 years ago

And fixed correctly in https://llvm.org/r357540

On ppc64le, the code base which has the Fix doen's get build:

$ cmake -G Ninja -DLLVM_ENABLE_PROJECTS="libcxx;libcxxabi" -DLLVM_TEMPORARILY_ALLOW_OLD_TOOLCHAIN=TRUE -DCMAKE_C_COMPILER=/gpfs/software/opt/gcc/5.4.0/bin/gcc -DCMAKE_CXX_COMPILER=/gpfs/software/opt/gcc/5.4.0/bin/g++ ../llvm

mclow commented 5 years ago

And fixed correctly in https://llvm.org/r357540

ikitayama commented 5 years ago

Fixed in r357478.

I don't see the Fix is in the llvm-project repo on GitHub, should I try the svn in the meantime?

mclow commented 5 years ago

Fixed in r357478.

kencu commented 5 years ago

A similar issue recently came up in gcc’s bug tracker https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88204 on their test suite, and the resolution was to disable the test.

After 14 years, it would not seem very likely that gcc is going to fix this any time soon.

Although it is undesirable to have one system working differently than all the others, for bootstrapping libc++ with gcc on powerpc there would not appear to be a lot of other available options than a patch something like Iain’s.

Is there something about that patch (other than the systems test, which needs to be broadened out a bit) that I am missing that makes it unusable?

mclow commented 5 years ago

By fully bootstrapped LLVM, it means that the linker also needs to be lld, not ld nor gold?

This is a compile-time error; it hasn't gotten to the linker yet.

ikitayama commented 5 years ago

By fully bootstrapped LLVM, it means that the linker also needs to be lld, not ld nor gold?

Am I correct this doesn’t happen when building LLVM with self bootstrapped LLVM?

That is my understanding. Building with LLVM is fine. Also note that this is not "building LLVM", but building libc++.

mclow commented 5 years ago

Am I correct this doesn’t happen when building LLVM with self bootstrapped LLVM?

That is my understanding. Building with LLVM is fine. Also note that this is not "building LLVM", but building libc++.

ikitayama commented 5 years ago

Am I correct this doesn’t happen when building LLVM with self bootstrapped LLVM?

kencu commented 5 years ago

As far as I can tell from https://github.com/llvm-mirror/libcxx/blob/master/src/condition_variable.cpp#L56 whatever superlong number we pass in here gets shortened down to "0x59682F000000E941" anyway.

So Iain's 592 years is long enough to saturate the system.

llvmbot commented 5 years ago

I'm wondering if replacing long double with double isn't the better fix here. std::chrono::nanoseconds::max() should be representable as a double.

My understanding is that the GCC bug is only triggered when using the IBM long double ABI, and that other long double ABI's are available. Is these a GCC macro we can use to detect IBM's long double specifically?

llvmbot commented 5 years ago

I'm wondering if replacing long double with double isn't the better fix here. std::chrono::nanoseconds::max() should be representable as a double.

My understanding is that the GCC bug is only triggered when using the IBM long double ABI, and that other long double ABI's are available. Is these a GCC macro we can use to detect IBM's long double specifically?

kencu commented 5 years ago

The way I understand Iain's patch, it changes the thread to sleep from thousands of years to just hundreds of years.

:>

70311826-e6d8-4d29-8604-c4ec4bfa7f0d commented 5 years ago

For what it is worth I patch-out the constexpr part (https://github.com/noloader/build-llvm/blob/master/build-llvm.sh#L566):

THIS_FILE=include/thread sed -i "s/_LIBCPP_CONSTEXPR duration _Max/const duration _Max/g" "$THIS_FILE" > "$THIS_FILE.patched" mv "$THIS_FILE.patched" "$THIS_FILE"

I patch it out because my two choices are (1) broken compile an no compiler, or (2) lose constexpr on a function I probably won't use. The value still works, its just not constexpr.

Patching-out the constexpr is a no brainer for me because I need a working compiler.

kencu commented 5 years ago

Here's my understanding:

It's a very old deficiency in gcc's handling of a unique IBM format for long long floating point values.

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26374

A patch years ago was put it to indicate this type of folding could not be done:

https://gcc.gnu.org/ml/gcc-patches/2005-01/msg01966.html

And that code is more or less still in gcc8 in fold-const.c

  /* Don't constant fold this floating point operation if the
 result may dependent upon the run-time rounding mode and
 flag_rounding_math is set, or if GCC's software emulation
 is unable to accurately represent the result.  */
  if ((flag_rounding_math
   || (MODE_COMPOSITE_P (mode) && !flag_unsafe_math_optimizations))
  && (inexact || !real_identical (&result, &value)))
return NULL_TREE;

There is some work being done to change things for PPCLE but not for PPCBE:

https://fedoraproject.org/wiki/Changes/PPC64LE_Float128_Transition

There is nobody I know of who knows more about gcc and PowerPC than Iain Sandoe, and his workaround was in the patch I referenced above.

llvmbot commented 5 years ago

Also see: https://bugzilla.redhat.com/show_bug.cgi?id=1538817

llvmbot commented 5 years ago

Here's a minimal reproducer on godbolt: https://godbolt.org/z/xFkZTR

What confuses me is that the same expression that breaks with long double works with double and `float.

ikitayama commented 5 years ago

So we either use the latest GCC version which has the fix or we apply the patch on the GCC of preference? In my environment using GCC 7.2.0 seems to be required by the CUDA toolkit.

mclow commented 5 years ago

The "proper fix" is for GCC to fix their PPC floating point codegen, because '(9.223372036854775807e+18 / 1.0e+9)' certainly is a constant expression.

ikitayama commented 5 years ago

Is the proper fix being worked on? I’d like a heads-up so I got pull from llvm-project repo.

mclow commented 5 years ago

Bug llvm/llvm-bugzilla-archive#40671 has been marked as a duplicate of this bug.

kencu commented 5 years ago

This worked for me to fix this. It is quite a minor change.

https://gist.githubusercontent.com/iains/3b2e074d65263cd4be456c36839823d1/raw/5fdd16b2284d0bfd5d628333b80eeb49b8cbbe43/gistfile1.txt

mclow commented 5 years ago

The code is like it is because of llvm/llvm-project#14093 ; so read that before attempting a fix.