On my computer when starting a shell in GDB it would give a very standard shell (looks like "bash-4.4$")- ignoring most of my environment variables. The real problem for me is I cannot adjust the LD_LIBRARY_PATH variable because of this reason.
More posts on SO suggest to check .bashrc and .profile, I did that. I even created a new user and the problem persist. So some script is loaded by GDB changing my PS1 and unsetting LD_LIBRARY_PATH. However, I have no idea which process. For some reason Root does not have this problem.
This works:
delruej#delxps:~$ export LD_LIBRARY_PATH=Hello
delruej#delxps:~$ bash
delruej#delxps:~$ echo $LD_LIBRARY_PATH
Hello
However this does not:
delruej#delxps:~$ export LD_LIBRARY_PATH=Hello
delruej#delxps:~$ gdb
(gdb) shell
bash-4.4$ echo $LD_LIBRARY_PATH
It should be some script loaded by every user, but not by root.
So some script is loaded by GDB changing my PS1 and unsetting LD_LIBRARY_PATH
No. GDB just loads your $SHELL, and your $SHELL unsets these environment variables for some reason.
You should look at values of $SHELL and possibly $ENV.
Note that:
Using LD_LIBRARY_PATH is generally a mistake; you should strive to build your programs such that they don't require it.
You can set any environment variable inside GDB (you don't have to inherit them from parent shell) with (gdb) set env FOO bar.
I purged GDB, removed /usr/share/gdb/, reinstalled and the problem was fixed.
Related
I'm receiving this gdb error on any code I try to debug any program with gdb. Here's the simplest process that reproduces the error
Create a main.cpp file with this content:
int main(){
return 0;
}
Run g++ -g main.cpp
Run gdb a.out
Inside gdb set a break point at line 2 with break 2
In gdb run the program with run
Output:
Starting program: /tmp/test/a.out
During startup program exited normally.
This is all done with gdb on the command line. I've tried using g++ and gcc with the same result. I'm not really sure where to go from here.
gdb version = 9.2
g++ version = 9.3.0
EDIT: I figured out what is causing the issue, but not how to fix it. The issue seems to be something related to my SHELL variable. I'm currently using xonsh as my shell but when I set my SHELL environment variable back to /bin/bash everything works as expected. Is there anything I can do to fix this while using xonsh? Should I report this to xonsh, gdb, both or neither?
I'm currently using xonsh as my shell but when I set my SHELL environment variable back to /bin/bash everything works as expected. Is there anything I can do to fix this while using xonsh? Should I report this to xonsh, gdb, both or neither?
This might be your xonsh startup problem, or it might be xonsh problem, or it could be that xonsh doesn't do what GDB expects it to do.
Normally, GDB forks / execs $SHELL -c "/path/to/your/exe $args" and expects the $SHELL to exec your program (this is done so shell redirection still works under GDB).
Only after that exec will GDB start setting breakpoints, etc.
If you have some xonsh init-file, which e.g. causes xonsh to exec something else, things could go bad. So I suggest trying to remove any such ~/.xonshrc or whatever it's called file, and seeing whether that fixes the problem.
If it doesn't, it could be that xonsh e.g. forks and execs your binary in a child (grandchild of GDB) instead of doing it directly, or it could be that xonsh doesn't understand the -c ... syntax.
If you don't care about redirection, you could also ask GDB to not use $SHELL at all: set startup-with-shell off. Documentation.
I installed flatpak using guix, but it segfaulted on startup. I wanted to debug it, but guix installs a wrapper script for flatpak, so I get this error when trying to run it under gdb:
"/home/user/.guix-profile/bin/flatpak": not in executable format: file format not recognized
and I tried to edit the wrapper script to call gdb, but this wrapper script is not even editable by root, because it is owned by root and has read-only permissions.
Simply copy the script to your current working directory:
cp /home/user/.guix-profile/bin/flatpak .
Mark it as writable:
chmod +w flatpak
Edit the script with your favourite text editor, to replace the string exec -a with exec gdb --args.
And finally, run it with any arguments you provided before, when it misbehaved:
./flatpak remote-add flathub https://flathub.org/repo/flathub.flatpakrepo
In this particular case, this wasn't immediately super-useful, because a debug symbol output hasn't been built for this package. But at least I could get a backtrace out of gdb.
How do I give commands to run before starting gdb debugging in Eclipse ?
Actually I want to execute few scripts that set environment variables (export vars) and execute a bunch of other programs before gdb process is launched from eclipse to debug my program.
I tried doing the following in debugger tab option:
<command> && <path-to-gdb-executable>
But I got the error that eclipse cannot execute gdb as given in above statement.
Please help - I actually want to execute a script called "before-launch-commands.sh" before debugging is started by gdb. I am trying to execute a cpp program under eclipse kepler.
Thanks.
The Eclipse Debug Configurations can already setup environment variables for you. I'm going to assume that that isn't sufficient, or you'd have already done it.
The first thing to do is create a new script, wrapped-gdb.sh:
#!/bin/sh
# Export any variables we need.
# Note that '.' (dot) is like an "include" statement.
. /path/to/before-launch-commands.sh
# Run GDB using the parameters passed in
exec /path/to/gdb "$#"
Next, set that script executable:
chmod +x /path/to/wrapped-gdb.sh
Finally, go to the Debugger tab in the debug configuration dialog, and in the box marked "GDB Debugger" enter /path/to/wrapped-gdb.sh.
When you launch your debug session it should now Do The Right Thing.
I have built a bash script to start up some processes in my system. It simply calls the process and associated config file. Same as I would call from the command line.
#!/bin/bash
# Start specified process in a new session
setsid $1 &>/dev/null &
So to start up someprocess, I would call from the command line:
root#supercomputer:~# start someprocess
This works like a charm. Every process, every time. But when I make a system call from a different running C++ process, someprocess never starts up.
system( "start someprocess" )
This approach for 90% of my processes, except for one. The only difference in the working and not working processes is that the non-working one uses proprietary libraries underneath. I recently added the setsid option to the bash script in hopes that starting a new session would help, but it made no difference. I've also tried popen, and execv. No change.
So my question is what is the difference between calling something with system() and just making that same call from the command line?
All processes are written in C++ on Linux.
.bashrc is only invoked if bash is run as interactive, non-login shell. If it's invoked as non-interactive shell, as when using system() on a script with a bash shebang, it only reads the configuration file pointed to by $BASH_ENV.
That means you have the following options:
add -l to the shebang - causes the shell to read ~/.profile at startup
set $BASH_ENV to the script you want sourced before calling system()
add -i to the shebang - invokes bash as interactive shell and causes it to read ~/.bashrc, but will also effect how bash handles input/output.
I'd recommend the first option.
You can find a detailed explanation of how bash reads it's startup files here. I'm not sure this will solve your problem completely, but it may at leas shed some light on that part of the issue.
Check the environment variables that are used in the system() call. For example, call system to print out some of the variables, and see if they match what you see from the command line.
Likely they are not being sourced correctly.
I'm trying to debug a CUDA program, but when I'm launching gdb like so:
$ gdb -i=mi <program name>
$ r <program arguments>
I'm getting:
/home/wvxvw/Projects/cuda/exercise-1-udacity/cs344/HW2/hw:
error while loading shared libraries: libcudart.so.5.0:
cannot open shared object file: No such file or directory
Process gdb-inferior killed
(formatted for readability)
(I'm running gdb using M-xgdb) If that matters, then CUDA libraries are in the .bashrc
export PATH="/usr/local/cuda/bin:$PATH"
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda/lib64"
error while loading shared libraries: libcudart.so.5.0
This error has nothing to do with GDB: your executable, when run from inside GDB, can't find the library it needs.
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda/lib64"
GDB runs your program in a new $SHELL, so that should have worked. I wonder if there is some interaction with emacs.
In any case, this:
(gdb) set env LD_LIBRARY_PATH /usr/local/cuda/lib64
(gdb) run
should fix this problem.
Update:
as I've mentioned it before, ld path is set properly
No, it isn't. If it was, you wouldn't have the problem.
Now, I don't know why it isn't set properly. If you really want to find out, start by running GDB outside emacs (to exclude possible emacs interactions).
If the problem is still present, gdb show env, shell env, adding echo "Here" to your ~/.basrc, etc. should help you find where things are not working as you expect them.
I've had this problem as well. One way to look at it is that even if the LD_LIBRARY_PATH variable is correct when you enter show env into gdb, it may not be correct when you actually execute the program because gdb executes $SHELL -c <program> to run the program. Try this as a test, run $SHELL from the command line and then echo $LD_LIBRARY_PATH. Is it correct? If not, then you probably need to add it to your rc (.tcshrc in my case).
I had a similar problem when trying to run gdb on windows 7. I use MobaXterm to access a Linux toolbox. I installed gdb separately from http://www.gnu.org/software/gdb/ . I got it to work by making sure gdb could find the correct .dll files as mentioned by Employed Russian. If you have MobaXterm installed the .dll files should appear in your home directory in MobaXterm/slash/bin.
gdb however did not recognize the LD_LIBRARY_PATH variable. For me, it worked when I used the PATH variable instead:
(gdb) set env PATH C:\Users\Joshua\Documents\MobaXterm\slash\bin
(gdb) run
I would think using PATH instead of LD_LIBRARY_PATH might work for you provided you put the correct path to your library.
gdb is looking for a library, so why are you concerned with the include path? You may want to try to set the gdb option "solib-search-path" to point to the location of the libcudart.so.5.0 library.