Uncatchable C++ exceptions (Shared libs, arm-linux-gnueabi-g++) - exception

I'm facing a strange problem related to exceptions thrown across DSO boundaries.
When the code is compiled for embedded linux board with arm-none-linux-gnueabi-g++, the exceptions can't be caught, if compiled with normal gcc compiler from ubuntu everything works fine :(
To clarify:
We have three components:
one Executeable file, which loads DSOs via dlopen(), dlsym()..
one DSO file (libMod2.so), containing a class MOD2 which throws self defined EException
(derived from std::runtime_error) when calling throwException()
one DSO file (libtest.so), containing a class MOD1 which get's a pointer to class MOD2 located and calls MOD2::throwException().
void MOD1::setMod2(IMOD2* mod2){
cout << "Calling mod2 throwException()" << endl;
try{
mod2->throwException();
}catch(EException& e){
cout << "Got you!" << endl << e.what() << endl;
}catch (...){
cout << "slippery shit..." << endl;
}
}
The problem is now that the Exception couldn't be caught by the first exception handler on arm target.
I think the problem is produced when linking:
A nm -Con the DSO show some differences when grepping for EException.
TARGET:
toterhaa#develop-TT:/var/lib/tftpboot$ /opt/freescale/usr/local/gcc-4.4.4-glibc-2.11.1-multilib-1.0/arm-fsl-linux-gnueabi/bin/arm-none-linux-gnueabi-g++ --version
arm-none-linux-gnueabi-g++ (4.4.4_09.06.2010) 4.4.4
Copyright (C) 2010 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 PURPOS
toterhaa#develop-TT:/var/lib/tftpboot$ nm -C libtest.so | grep EEx
00009ef0 V typeinfo for EException
000017f4 V typeinfo name for EException
Ubuntu:
toterhaa#develop-TT:/nfs$ g++ --version
g++ (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
Copyright (C) 2010 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.
toterhaa#develop-TT:/nfs$ nm -C libtest.so | grep EEx
0000303c d DW.ref._ZTI10EException
00002edc V typeinfo for EException
00001373 V typeinfo name for EException
The DSO created with ubuntu gcc has a additional symbol DW.ref._ZTI10EException. I think the solution is to bring this symbol also into the arm-DSO, but how?
Does anybody know this problem?

Problem solved!
The problem is not linker related, it was simpler, much simpler.
I solved it by adding RTLD_GLOBAL to the dlopen() call. It seems that the standard gcc in my ubuntu installation sets this by default and the compiler for the arm target uses RTLD_LOCAL as default.

Related

cudaErrorUnknown when calling cuda runtime or driver api

After heavy simulation that crashes my GPU, terminating the program and rebooting my computer, I cannot call any cuda api that runs correctly before rebooting.
nvidia-smi works well.
In my Ubuntu 20.04 computer, the CUDA11.6 and nvidia-driver 510.47.03 are installed.
The minimum codes for getting error in my computer are followings,
Driver api version
#include "cuda.h"
int main() {
CUresult result;
result = cuInit(0);
const char* pStr;
cuGetErrorName(result, &pStr);
if(result != CUDA_SUCCESS){
std::cerr << pStr << std::endl;
}
}
The compile command is,
nvcc a.cu -lcuda
and I got CUDA_ERROR_UNKNOWN.
Runtime api version,
#include <iostream>
int main() {
int n;
cudaError_t error = cudaGetDeviceCount(&n);
if(error){
std::cerr << cudaGetErrorName(error) << std::endl;
}
}
The compile command is,
nvcc a.cu
and I got cudaUnknownError.
Why I got this error ? Is this a problem of driver ?
looks like you trashed something (driver or cuda runtime) and you are not able to call any function related to cuda.
in my humble experience, I usually get these errors when my kernels runs for too long on a Windows machine and the Windows Display Driver Manager reset my GPU while i'm running the kernel.
Maybe you are experiencing some similiar issues on linux.
To fix this, have you tried to reset your GPU using the following bash command line ?
nvidia-smi --gpu-reset
Please also note that all CUDA errors are not recoverable and if the previous command doesn't work, you may reset your computer to fix the issue (or simply unload/reload the cuda driver).

CUDA Kernel does not launch, CMAKE Visual Studio 2015 project

I have a relatively simple CUDA kernel and I immediately call the kernel in the main method of my program in the following way:
__global__ void block() {
for (int i = 0; i < 20; i++) {
printf("a");
}
}
int main(int argc, char** argv) {
block << <1, 1 >> > ();
cudaError_t cudaerr = cudaDeviceSynchronize();
printf("Kernel executed!\n");
if (cudaerr != cudaSuccess)
printf("kernel launch failed with error \"%s\".\n",
cudaGetErrorString(cudaerr));
}
This program is compiled and launched using Visual Studio 2015, and the project being executed has been generated with CMAKE using the following CMakeLists.txt file:
project (Comparison)
cmake_minimum_required (VERSION 2.6)
find_package(CUDA REQUIRED)
set(
CUDA_NVCC_FLAGS
${CUDA_NVCC_FLAGS};
-arch=compute_30 -code=sm_30 -g -G
)
cuda_add_executable(Comparison kernel.cu)
I would expect the output of this program to print 20 A's to the console and then end with printing kernel executed. However, the A's are never printed to the console and the line Kernel executed shows up immediately. Even if I replace the for loop by a while(true) loop.
Even when running the code with the Nsight debugger attached and a breakpoint in the for loop of the kernel nothing happens. Leading me to believe that the kernel is never actually launched. Does anyone know how to make this kernel behave as expected?
The reason the kernel was not running correctly when compiled with the given CMakeLists.txt file was due to these flags:
-arch=compute_30 -code=sm_30
combined with the GPU that was being used (GTX 970, a cc 5.2 GPU).
Those flags specify the generation of cc 3.0 SASS code only, and such code is not compatible with a cc 5.2 device. The fix would be to modify the flags to something like:
-arch=sm_30
or
-arch=sm_52
or
-arch=compute_52 -code=sm_52
I would recommend the first or second approach, as it will include PTX support for future devices.
The kernel error was not evident because the error checking after the kernel was incomplete. Refer to the canonical/question answer.

Connecting MYSQL using C program in Eclipse

I am connecting the Mysql using c program in eclipse, I am using CYGWIN compiler to compile program. I have installed the mysql c connector and added the include files to the compiler and linker
but i am getting error
**** Build of configuration Debug for project DbConnectionC ****
make all
Building file: ../connection.c
Invoking: Cygwin C Compiler
gcc -I"C:\Program Files (x86)\MySQL\MySQL Connector C 6.0.2\include" -include"C:\Program Files (x86)\MySQL\MySQL Connector C 6.0.2\include\mysql.h" -O0 -g3 -Wall -c -fmessage-length=0 -MMD -MP -MF"connection.d" -MT"connection.d" -o "connection.o" "../connection.c"
cygwin warning:
MS-DOS style path detected: D:\java\workspace\DbConnectionC\Debug
Preferred POSIX equivalent is: /cygdrive/d/java/workspace/DbConnectionC/Debug
CYGWIN environment variable option "nodosfilewarning" turns off this warning.
Consult the user's guide for more details about POSIX paths:
http://cygwin.com/cygwin-ug-net/using.html#using-pathnames
../connection.c: In function `main':
../connection.c:57: warning: char format, different type arg (arg 2)
../connection.c:57: warning: char format, different type arg (arg 2)
../connection.c:61: warning: char format, different type arg (arg 2)
../connection.c:61: warning: char format, different type arg (arg 2)
Finished building: ../connection.c
Building target: DbConnectionC.exe
Invoking: Cygwin C Linker
gcc -L"C:\Program Files (x86)\MySQL\MySQL Connector C 6.0.2\lib\opt" -o "DbConnectionC.exe" ./connection.o -llibmysql.lib -lmysqlclient.lib
/usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld: cannot find -llibmysql.lib
/usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld: cannot find -lmysqlclient.lib
collect2: ld returned 1 exit status
make: *** [DbConnectionC.exe] Error 1
makefile:29: recipe for target `DbConnectionC.exe' failed
**** Build Finished ****
Please help me on this
thanks in advance
In the first part you have some warnings about possible problems:
../connection.c:39: warning: implicit declaration of function `exit'
This means function exit() is not declared, your source needs an #include <stdlib.h>
../connection.c:46: warning: control reaches end of non-void function
You don't return a value at the end of main(), add a return 0; if the program succeeds.
In the second part are linker warnings:
/usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld: cannot find -llibmysql.lib
/usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld: cannot find -lmysqlclient.lib
which means the linker needs a path to the libraries or the libraries do not exist or the library names are misspelled.
One possible solution is using -lmysql -lmysqlclient instead of -llibmysql.lib -lmysqlclient.lib, but I'm not that familiar with Windows.
Having said that, I suggest strongly, you look at https://stackoverflow.com/tags/c/info and look at some book links.

Problem with MySQL driver for unixODBC on Debian Lenny

On OpenSuse 11.2, I successfully compiled, linked, and ran the following code which installs a data source for a MySQL database with unixODBC:
#include <iostream>
#include <sql.h>
#include <sqlext.h>
#include <odbcinst.h>
/* Add a data source for the following MySQL db: db=testdb, username=test, password = test. */
void inst()
{
BOOL ret = SQLConfigDataSource(NULL, ODBC_ADD_DSN, "MySQL driver",
"DSN=mysource\0UID=test\0PWD=test\0DATABASE=testdb\0\0");
if (!ret) {
DWORD errCode;
char errBuf[SQL_MAX_MESSAGE_LENGTH];
WORD msgLen;
SQLInstallerError(1, &errCode, errBuf, SQL_MAX_MESSAGE_LENGTH, &msgLen);
std::cerr << errBuf << std::endl;
}
}
int main()
{
inst();
return 0;
}
With the same code on Debian Lenny, I have had problems. First, I compiled this code the following way:
c++ -o main main.cc -lodbc -lodbcinst -L/usr/lib/odbc -lmyodbc
It went ok. But when I attempted to run the resulting binary, I got a linker error which in fact was confirmed by typing ldd main:
libmyodbc3_r-3.51.15.so => not found
Although I correctly installed unixODBC and the associated MySQL driver (myodbc) on my host (Debian Lenny) the simplest way (i.e. via aptitude), I could not find this shared library.
I wrongly thought, well, I will create a symlink on /usr/lib/odbc/libmyodbc.so. Anyway now my program returns the following message:
General installer error
So I feel the file libmyodbc3_r-3.51.15.so is really missing.
Note: on Debian Lenny, the version of unixODBC is 2.2.11, and the version of MySQL is 5.0.51a
Anyone ever ran into such a situation ? Any help would be appreciated.
The option
-L/usr/lib/odbc
tells the compiler where to find the library for linking.
But the system doesn't know where to find the library when you run the executable.
You need to either statically link against libmyodbc, or tell the system where to find the library.
The first can be done by changing
-lmyodbc
to
-static -lmyodbc
The second can be done by editing /etc/ld.so.conf (or adding to /etc/ld.so.conf.d) and re-running ldconfig or by setting the LD_LIBRARY_PATH environment variable to include /usr/lib/odbc

Win7 MInGW QT MySQL program screams "cannot find -lqsqlmysqld"; where is the missing library?

I am attempting to make a qt program on Windows 7 that uses a MySQL plugin.
I have compiled both qt and the mysql plugin with no problems using my minGW 32bit compiler.
However, I keep on getting an error like this:
mingw32-make -f Makefile.Debug
mingw32-make[1]: Entering directory `C:/Users/dhatt/Desktop/testdb2'
g++ -c -g -frtti -fexceptions -mthreads -Wall -DUNICODE -DQT_LARGEFILE_SUPPORT -DQT_DLL -DQT_SQL_LIB -DQT_GUI_LIB -DQT_CORE_LIB -DQT_THREAD_SUPPORT -DQT_NEEDS_QMAIN - I"..\..\..\..\QT\qt\include\QtCore" -I"..\..\..\..\QT\qt\include\QtGui" -I"..\..\..\..\QT\qt\include\QtSql" -I"..\..\..\..\QT\qt\include" -I"..\..\..\..\MySQL\bin" -I"..\..\..\..\QT\qt\include\ActiveQt" -I"debug" -I"..\..\..\..\QT\qt\mkspecs\win32-g++" -o debug\database.o database.cpp
g++ -enable-stdcall-fixup -Wl,-enable-auto-import -Wl,-enable-runtime-pseudo-reloc -mthreads -Wl -Wl,-subsystem,windows -o debug\testdb2.exe debug/database.o -L"c:\QT\qt\lib" -lmingw32 -lqtmaind -L C:\MySQL\lib\opt -LC:/QT/qt/plugins/sqldrivers -lqsqlmysqld -lQtSqld4 -lQtGuid4 -lQtCored4 -LC:\MySQL\lib\opt
C:/qt/mingw/bin/../lib/gcc/mingw32/4.4.0/../../../../mingw32/bin/ld.exe: cannot find -lqsqlmysqld
collect2: ld returned 1 exit status
mingw32-make[1]: *** [debug\executable.exe] Error 1
mingw32-make[1]: Leaving directory `C:/Users/dhatt/Desktop/testdb2'
mingw32-make: *** [debug] Error 2
I apologize in advance for being very verbose of what I did, but I am doing this partly for troubleshooting, and partly so any other lost souls don't end up wasting three weeks on this particular problem. :)
Here are my specs:
Windows 7 Nokia's Open Source QT
Qt SDK for Windows (C:\Qt\2010.04\qt)
Linux MinGW Version 5.1.6 (C\Linux\MinGW)
MySQL5 with C++ files (C:\MySQL5)
If you want to know how I installed qt, just follow this site's instructions:
http://www.jiggerjuice.net/software/qt-sql-drivers.html
These other sites may hold some extra information tidbits:
http://doc.qt.nokia.com/4.6/sql-driver.html
http://www.rag.com.au/linux/qt4howto.html
http://qtnode.net/wiki?title=Qt4_on_Windows (yes, I did check with Nokia's docs!!!)
http://doc.trolltech.com/qq/qq10-windows-deployment.html
This fellow mentioned about remaking qmake, which I am not doing unless I have a good reason.
http://christopher.rasch-olsen.no/2009/04/14/qt-45-and-mysql-plugin-with-mingw-on-windows-xp/
I've already deleted the plugin cache once before, I hope I won't have to do it again...
http://doc.trolltech.com/4.2/plugins-howto.html#the-plugin-cache
http://ubuntuforums.org/showthread.php?t=1070155
If there is any confusion, between the two compilation option (creating the mysql libraries statically, or as a plugin), I chose for the plugin because it compiles quicker and I don't have to worry about licensing.
Generally, the big trouble of mysql to most people is to make a mingw compatible library. Generally, I did this with the mingw tools in ( https://olex.openlogic.com/packages/mingw-utils )...
c:\> cd MySQL\lib\opt
c:\mysql\lib\opt> reimp -d libmysql.lib
c:\mysql\lib\opt> dlltool --input-def libmysql.def --dllname libmysql.dll --output-lib libmysql.a -k
I should have done it right since in my C:\MySQL\lib\opt, it has the two files:
libmysql.a
libmysql.lib
LIBMYSQL.def (not a typo)
and in the C:\MySQL\bin directory, I have:
libmySQL.bin (not a typo)
I had compiled the mysql plugin beforehand:
cd %QTDIR%\src\plugins\sqldrivers\mysql
qmake "INCLUDEPATH+=C:\MySQL\include" "LIBS+=C:\MYSQL\lib\opt\libmysql.lib" mysql.pro
mingw32-make
As a result, I have in my C:\QT\qt\plugins\sqldrivers folder:
libqsqlmysql4.a
libqsqlmysqldq4.a
libqsqlodbc4.a
libqsqlodbcd4.a
qsqlmysql4.dll
qsqlmysqld4.dll
qsqlodbc4.dll
qsqlodbc4.dll
And in my C:\QT\bin folder
QtSql4.dll
QtSqld4.dll
So, I assume from this site ( http://www.qtforum.org/article/21352/how-to-compile-use-a-mysql-driver.html) that I got it right.
I didn't use the binaries of qt itself, I used the compiled qt files(also from Nokia), but reconfigured and recompiled them using mingw32-make. I had no errors. This was my configuration options for remaking qt.
-opensource
-nomake examples
-nomake demos
-no-sql-lite
-no-qt3support
-no-gif
-no-libpng
-no-libmng
-no-libtiff
-no-phonon
-no-phonon-backend
-no-multimedia
-no-audio-backend
-no-webkit
-no-script
-no-scripttools
-nodeclarative
-plugin-sql-mysql -l mysql -I C:\QT\qt\include -L C:\QT\qt\lib\opt
Here is my .pro file
LANGUAGE = C++
TEMPLATE = app
TARGET = executable
QT += core sql
QTPLUGIN += qsqlmysql
DEPENDPATH += .
INCLUDEPATH += C:\MySQL\bin
LIBS += -L C:\MySQL\lib\opt -lmysql
# Input
SOURCES += database.cpp
I installed the plugin described in here:
C:\QT\qt
My path variables are:
%SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem;%SYSTEMROOT%\System32\WindowsPowerShell\v1.0\;C:\Program Files (x86)\QuickTime\QTSystem\;C:\Python26;C:\Linux\Cygwin\bin;C:\mingw-utils-0.3\bin;C:\QT\qt\bin;C:\MySQL\bin;C:\MySQL\include;C:\QT\mingw32\bin;C:\QT\mingw\bin;C:\QT\qt\plugins\sqldrivers
The qt command prompt added a few extra though, so I did all of this in the command prompt.
Setting up a MinGW/Qt only environment...
-- QTDIR set to C:\QT\qt
-- PATH set to C:\QT\qt\bin
-- Adding C:\QT\bin to PATH
-- Adding C:\Windows\System32 to PATH
-- QMAKESPEC set to win32-g++ (mingw is my only compiler so, this is unnecessary)
Although I either did all that already, or it is redundant. I only add this for the sake of completeness.
Here is my code (database.cpp):
#include <QtSql>
#include <iostream>
using namespace std;
int main( int argc, char ** argv )
{
QSqlDatabase db = QSqlDatabase::addDatabase("QMYSQL");
db.setHostName("---.---.---.---");
db.setDatabaseName("--------");
db.setUserName("------------");
db.setPassword("------------");
if (!db.open()) cout << "Failed to connect to mysql" << endl;
else cout << "Works finally." << endl;
QSqlDatabase::removeDatabase("QMYSQL");
exit ( 0 );
}
Very simple, yes?
I went to my directory with the example code, run
mingw32-make distclean
qmake
mingw32-make
and get the error message above. I've tried building a version with the release version only (no debug), and it still shows the same message, but with "cannot find -lqsqlmysq", so it is not that.
I've tried many things, but where should I look next to solve it; maybe someone can narrow it down for me, set me on the right path, or even better, solve his annoying problem.
Also, I plan to use python bindings with my code (I need PyQT + MySQL). If the proposed solution would prevent me from doing so, let me know.
Well, I'm going to solve my own problem, again, so let's make this fun!
This is your last chance.
After this, there is no turning back.
You download the PyQT.exe, the story ends. You wake up in your bed and you believe whatever you want to believe. You modify the .pro file, you stay in wonderland. And, I show you how deep the rabbit hole goes.
I eventually gave up and downloaded the .exe, which does have MySQL support out of the box. If mysql does not work, your application is the problem, and I recommend you read this post here ( http://lists.trolltech.com/qt-interest/2006-06/thread00292-0.html ) or follow the quote below:
The issue is that you either have to
use the addLibraryPath method or
create a QCoreApplication instance
before your first call to loading a
database
Believe me, manually installing PyQT+MySQL on Windows is a pain. But if you need some out of the way plugin to get at that the executable doesn't know, you have to go down the rabbit hole further.
Here is the new and improved .pro file:
LANGUAGE = C++
TEMPLATE = app
TARGET = executable
QT += core sql
QTPLUGIN += qsqlmysql
DEPENDPATH += .
INCLUDEPATH += C:\MySQL\bin
LIBS += -L C:\MySQL\lib\opt
# Input
SOURCES += database.cpp
Turns out I did have the right path to mysql, I was just confusing it with the .pro file that I had. After redownloading qt and following the steps above again, modifying my .pro file made all the difference.
But now I had to download SIP and PyQT. I followed the docs on there. There are a few more problems. Follow the links or the directions which are left there in case the information is removed.
If your SIP make install has an error where it is looking at Unix paths (/usr/bin) instead of DOS paths (C:\QT), look at this link http://old.nabble.com/Building-SIP-on-MinGW-:-problem-at-%22make-install%22-td28909249.html#
(short version: the problem is the sh.exe in one of your other linux compilers like cygwin or msys, change the name temporarily to force the make install to use DOS path naming):
If you configure PyQT and it spits out a file error that has to do with QTCore
Google pexports and download. Go to %QTdir%/bin. Then follow instructions or link ( http://jeethurao.com/blog/?p=18 )
pexports QtCore4.dll > QtCore4.def
dlltool –dllname QtCore4.dll –def QtCore4.def –output-lib libQtCore4.a
move libQtCore4.a ..\lib
And now you know kung-fu.
P.S: I never tried this method myself. This is a different, but untested (by me) method of compiling PyQT, done up by the trolls at Trolltech:
http://www.diotavelli.net/PyQtWiki/InstallingPyQTCommercialWin