Segmentation fault during omc compilation with gcc (v1.14.0)

[florentine.rosiere@…]

Hello,

With gcc 8.3.1 and Fedora29 I get a segmentation while compiling omc v1.14.0:

make omc

...
<OM_PATH>/build/bin/omc -d=failtrace -g=MetaModelica OpenModelicaScriptingAPI.mos

Limited backtrace at point of segmentation fault
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaRuntimeC.so(+0x17c54)[0x7f179b81bc54]
/lib64/libc.so.6(+0x38600)[0x7f179b93e600]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_FGraphBuildEnv_mkProgramGraph+0x33)[0x7f179bfeed63]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Builtin_initialGraph+0x1a1)[0x7f179c2e0271]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Inst_makeEnvFromProgram+0x40)[0x7f179c1a1330]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_SymbolTable_buildEnv+0x43)[0x7f179bf4bc43]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Interactive_evaluateExpr+0x108)[0x7f179bfb0868]
<OM_PATH>/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Interactive_evaluateExprToStr+0x1b3)[0x7f179bfb0c23]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Interactive_evaluate2+0x1c5)[0x7f179bfb2e05]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Interactive_evaluateToStdOut+0x9e)[0x7f179bfb380e]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Main_translateFile+0x5d3)[0x7f179bdc20d3]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Main_main2+0x468)[0x7f179bdc3bb8]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_Main_main+0x414)[0x7f179bdc4294]
<OM_PATH>/build/bin/../lib/x86_64-linux-gnu/omc/libOpenModelicaCompiler.so(omc_main+0x1e6)[0x7f179c30ab66]
/lib64/libc.so.6(libc_start_main+0xf3)[0x7f179b92a413]
make[5]: * [Makefile.common:4: OpenModelicaScriptingAPI.mo] Erreur de segmentation (core dump créé)

with clang it compiles properly.

Thank you and Regards

[florentine.rosiere@…]

Hi,

This issue is still there in 1.14.2.

Regards

[Francesco Casella]

@florentine, we are going to release 1.16.0-beta.01 next week. Could you try with the latest 1.16.0 nightly build and report if you still have this issue?

Thanks!

[florentine.rosiere@…]

Hi,

I tried on the master yesterday and the issue was still there.
This is preventing us from using latest versions of OpenModelica.

Thank you for your response!
Regards,
Florentine

[Adrian Pop]

Can you try to configure with CFLAGS=-g?

./configure CC=gcc CXX=g++ CFLAGS='-g'

and see if you don't get a crash?

You could also try to compile with Clang:

./configure CC=clang CXX=clang++

As far as I can see it dies while using the newly build omc to generate the Qt API for OMEdit.
I'll see if I can get my hand on a VM image with Fedora29 and gcc 8.3.1 to do proper testing.

[florentine.rosiere@…]

Hi,

The compilation passes with the CFLAGS='-g' option.
It also passes with clang in all the cases, but we do need to compile with gcc as our servers does not support clang.

Thank you,
Regards,
Florentine

[Adrian Pop]

What Fedora 29 distribution do you have? Workstation? Server?

[Adrian Pop]

Also, I assume you have 64 bit OS, not 32 bit one.

[florentine.rosiere@…]

Hi,

this is Fedora 29 Workstation Edition 64 bit.
We are currently trying to compile the master with a more recent gcc version, we'll let you know if it passes asap.

Thank you,
Regards,
Florentine

[gautier.bureau@…]

The error can be reproduced with the following steps:

docker run -it fedora:29

dnf install -y git gcc gcc-c++ gcc-gfortran autoconf automake make libtool cmake hwloc java-1.8.0-openjdk-devel blas-devel lapack-devel lpsolve-devel expat-devel glibc-devel sqlite-devel xerces-c-devel libarchive-devel zlib-devel qt-devel gettext patch wget python-devel clang llvm-devel ncurses-devel readline-devel unzip perl-Digest-MD5 vim gcovr python-pip python-psutil boost-devel lcov gtest-devel gmock-devel xz rsync python-lxml graphviz clang-devel OpenSceneGraph-devel qtwebkit qtwebkit-devel qt5-qtwebkit qt5-qtwebkit-devel qwt python-sphinx qwt-devel OpenSceneGraph-qt-devel OpenSceneGraph-devel OpenSceneGraph-libs OpenSceneGraph-qt OpenSceneGraph-qt-devel qwt5-qt4-devel glibc-static libstdc++-static omniORB-devel poppler-devel flex bison maven curl-devel uuid-devel qt5-linguist qt5-qtsvg-devel qt5-qtxmlpatterns-devel

git clone ​https://github.com/OpenModelica/OpenModelica.git \
&& cd OpenModelica \
&& git submodule init \
&& git submodule update \
&& git submodule foreach --recursive 'git submodule init' \
&& git submodule foreach --recursive 'git submodule update'

autoconf \
&& ./configure CC=gcc CXX=g++ --disable-modelica3d --disable-omnotebook --disable-omshell-terminal --with-qwt=NO \
&& make omc

A similar procedure with an ubuntu:20.04 docker has been tested and worked, gcc version is 9.3.0.

[Adrian Pop]

I could reproduce it on my Fedora 29 Workstation VirtualBox VM. I'm debugging it now.

[Adrian Pop]

Slicing down through gcc -O2 flags to see which one is to blame is really fun. I started with the last one that is different from -O1: -ftree-vrp.

.txt"

@@ -5 +5 @@
   -Os                             
   -faggressive-loop-optimizations   [enabled]
   -falign-functions               [disabled]
-  -falign-functions=              1
+  -falign-functions=              16
   -falign-jumps                   [disabled]
-  -falign-jumps=                  1
-  -falign-labels                  [disabled]
-  -falign-labels=                 1
+  -falign-jumps=                  16
+  -falign-labels                  [enabled]
+  -falign-labels=                 0
   -falign-loops                   [disabled]
-  -falign-loops=                  1
+  -falign-loops=                  16
   -fassociative-math              [disabled]
   -fasynchronous-unwind-tables    [enabled]
   -fauto-inc-dec                  [enabled]
@@ -20 +20 @@
   -fbranch-target-load-optimize   [disabled]
   -fbranch-target-load-optimize2  [disabled]
   -fbtr-bb-exclusive              [disabled]
-  -fcaller-saves                  [disabled]
-  -fcode-hoisting                 [disabled]
+  -fcaller-saves                  [enabled]
+  -fcode-hoisting                 [enabled]
   -fcombine-stack-adjustments     [enabled]
   -fcompare-elim                  [enabled]
   -fconserve-stack                [disabled]
   -fcprop-registers               [enabled]
-  -fcrossjumping                  [disabled]
-  -fcse-follow-jumps              [disabled]
+  -fcrossjumping                  [enabled]
+  -fcse-follow-jumps              [enabled]
   -fcx-fortran-rules              [disabled]
   -fcx-limited-range              [disabled]
   -fdce                           [enabled]
@@ -35 +35 @@
   -fdelayed-branch                [disabled]
   -fdelete-dead-exceptions        [disabled]
   -fdelete-null-pointer-checks    [enabled]
-  -fdevirtualize                  [disabled]
-  -fdevirtualize-speculatively    [disabled]
+  -fdevirtualize                  [enabled]
+  -fdevirtualize-speculatively    [enabled]
   -fdse                           [enabled]
   -fearly-inlining                [enabled]
   -fexceptions                    [disabled]
-  -fexpensive-optimizations       [disabled]
+  -fexpensive-optimizations       [enabled]
   -ffast-math                     
   -ffinite-math-only              [disabled]
   -ffloat-store                   [disabled]
@@ -48 +48 @@
   -ffp-contract=[off|on|fast]     fast
   -ffp-int-builtin-inexact        [enabled]
   -ffunction-cse                  [enabled]
-  -fgcse                          [disabled]
+  -fgcse                          [enabled]
   -fgcse-after-reload             [disabled]
   -fgcse-las                      [disabled]
   -fgcse-lm                       [enabled]
@@ -57 +57 @@
   -fgraphite-identity             [disabled]
   -fguess-branch-probability      [enabled]
   -fhandle-exceptions             
-  -fhoist-adjacent-loads          [disabled]
+  -fhoist-adjacent-loads          [enabled]
   -fif-conversion                 [enabled]
   -fif-conversion2                [enabled]
-  -findirect-inlining             [disabled]
+  -findirect-inlining             [enabled]
   -finline                        [enabled]
   -finline-atomics                [enabled]
   -finline-functions              [disabled]
   -finline-functions-called-once  [enabled]
-  -finline-small-functions        [disabled]
-  -fipa-bit-cp                    [disabled]
-  -fipa-cp                        [disabled]
+  -finline-small-functions        [enabled]
+  -fipa-bit-cp                    [enabled]
+  -fipa-cp                        [enabled]
   -fipa-cp-clone                  [disabled]
-  -fipa-icf                       [disabled]
-  -fipa-icf-functions             [disabled]
-  -fipa-icf-variables             [disabled]
+  -fipa-icf                       [enabled]
+  -fipa-icf-functions             [enabled]
+  -fipa-icf-variables             [enabled]
   -fipa-profile                   [enabled]
   -fipa-pta                       [disabled]
   -fipa-pure-const                [enabled]
-  -fipa-ra                        [disabled]
+  -fipa-ra                        [enabled]
   -fipa-reference                 [enabled]
-  -fipa-sra                       [disabled]
-  -fipa-vrp                       [disabled]
+  -fipa-sra                       [enabled]
+  -fipa-vrp                       [enabled]
   -fira-algorithm=[CB|priority]   CB
   -fira-hoist-pressure            [enabled]
   -fira-loop-pressure             [disabled]
@@ -86 +86 @@
   -fira-share-save-slots          [enabled]
   -fira-share-spill-slots         [enabled]
   -fisolate-erroneous-paths-attribute   [disabled]
-  -fisolate-erroneous-paths-dereference   [disabled]
+  -fisolate-erroneous-paths-dereference   [enabled]
   -fivopts                        [enabled]
   -fjump-tables                   [enabled]
   -fkeep-gc-roots-live            [disabled]
@@ -98 +98 @@
   -floop-nest-optimize            [disabled]
   -floop-parallelize-all          [disabled]
   -floop-unroll-and-jam           [disabled]
-  -flra-remat                     [disabled]
+  -flra-remat                     [enabled]
   -fmath-errno                    [enabled]
   -fmodulo-sched                  [disabled]
   -fmodulo-sched-allow-regmoves   [disabled]
@@ -107 +107 @@
   -fnothrow-opt                   [disabled]
   -fomit-frame-pointer            [enabled]
   -fopt-info                      [disabled]
-  -foptimize-sibling-calls        [disabled]
-  -foptimize-strlen               [disabled]
+  -foptimize-sibling-calls        [enabled]
+  -foptimize-strlen               [enabled]
   -fpack-struct                   [disabled]
   -fpack-struct=<number>          
-  -fpartial-inlining              [disabled]
+  -fpartial-inlining              [enabled]
   -fpatchable-function-entry=     
   -fpeel-loops                    [disabled]
   -fpeephole                      [enabled]
-  -fpeephole2                     [disabled]
+  -fpeephole2                     [enabled]
   -fplt                           [enabled]
   -fpredictive-commoning          [disabled]
   -fprefetch-loop-arrays          [enabled]
@@ -124 +124 @@
   -freg-struct-return             [enabled]
   -frename-registers              [enabled]
   -freorder-blocks                [enabled]
-  -freorder-blocks-algorithm=[simple|stc]   simple
-  -freorder-blocks-and-partition  [disabled]
-  -freorder-functions             [disabled]
-  -frerun-cse-after-loop          [disabled]
+  -freorder-blocks-algorithm=[simple|stc]   stc
+  -freorder-blocks-and-partition  [enabled]
+  -freorder-functions             [enabled]
+  -frerun-cse-after-loop          [enabled]
   -freschedule-modulo-scheduled-loops   [disabled]
   -frounding-math                 [disabled]
   -frtti                          [enabled]
@@ -149 +149 @@
   -fsched2-use-superblocks        [disabled]
   -fschedule-fusion               [enabled]
   -fschedule-insns                [disabled]
-  -fschedule-insns2               [disabled]
+  -fschedule-insns2               [enabled]
   -fsection-anchors               [disabled]
   -fsel-sched-pipelining          [disabled]
   -fsel-sched-pipelining-outer-loops  [disabled]
@@ -178 +178 @@
   -fstack-protector-strong        [disabled]
   -fstack-reuse=[all|named_vars|none]   all
   -fstdarg-opt                    [enabled]
-  -fstore-merging                 [disabled]
-  -fstrict-aliasing               [disabled]
+  -fstore-merging                 [enabled]
+  -fstrict-aliasing               [enabled]
   -fstrict-enums                  [disabled]
   -fstrict-volatile-bitfields     [enabled]
-  -fthread-jumps                  [disabled]
+  -fthread-jumps                  [enabled]
   -fno-threadsafe-statics         [enabled]
   -ftracer                        [disabled]
   -ftrapping-math                 [enabled]
@@ -210 +210 @@
   -ftree-parallelize-loops=<number>   1
   -ftree-partial-pre              [disabled]
   -ftree-phiprop                  [enabled]
-  -ftree-pre                      [disabled]
+  -ftree-pre                      [enabled]
   -ftree-pta                      [enabled]
   -ftree-reassoc                  [enabled]
   -ftree-scev-cprop               [enabled]
@@ -218 +218 @@
   -ftree-slp-vectorize            [disabled]
   -ftree-slsr                     [enabled]
   -ftree-sra                      [enabled]
-  -ftree-switch-conversion        [disabled]
-  -ftree-tail-merge               [disabled]
+  -ftree-switch-conversion        [enabled]
+  -ftree-tail-merge               [enabled]
   -ftree-ter                      [enabled]
   -ftree-vectorize                
-  -ftree-vrp                      [disabled]
+  -ftree-vrp                      [enabled]
   -funconstrained-commons         [disabled]
   -funroll-all-loops              [disabled]
   -funroll-loops                  [disabled]
@@ -234 +234 @@
   -fvar-tracking-assignments-toggle   [disabled]
   -fvar-tracking-uninit           [disabled]
   -fvariable-expansion-in-unroller  [disabled]
-  Specifies the cost model for vectorization. -fvect-cost-model=[unlimited|dynamic|cheap]   [default]
+  Specifies the cost model for vectorization. -fvect-cost-model=[unlimited|dynamic|cheap]   cheap
   -fvpt                           [disabled]
   -fweb                           [enabled]
   -fwrapv                         [disabled]

It seems to work with gcc 8.3.1 using:

./configure CFLAGS='-g -O2 -fno-tree-vrp'

I will investigate more and if this is the case I will fix configure to disable optimization -ftree-vrp after -O2. We do similar stuff for gcc 5.3 as it has some bad optimizations as well.

[Adrian Pop]

Fixed with: 62e977/OpenModelica.
Please test and reopen if needed.

[Francesco Casella]

I guess this is fixed in 1.17.0 but not in 1.16.0, unless the commit is back-ported.

@sjoelund.se, would that be easy to do?

[Adrian Pop]

Done already!
​https://github.com/OpenModelica/OpenModelica/pull/6808

[Francesco Casella]

Thank you @adrpo!