We've got this legacy Fortran code running on a board with an A9 processor, and I've noticed some funny behavior when using gdb. Threads continue to run when it is in all-stop mode, and it seems gdb crashes when you switch it to non-stop mode. It also can't do watchpoints on symbols in the Fortran code, not sure if that's due to the port or the binaries not belonging to the same toolchain. You can set a watchpoint on the address cast as a C type, however.
We've got toolchain binaries from CodeSourcery installed to the stage directory during the rootfs build, then as part of our application build Linaro binaries overwrite some of the libraries in the targetroot. The libraries that are overwritten are (some of?) the only ones needed by the application and the ones it is linked against (libstdc++, libgfortran, and libpthread).
Using either gdb from CodeSourcery or Linaro seems to function equivalently, though I didn't try setting hardware assisted watchpoints with the Linaro compiler. Neither seem to be configured to work outside the documented behavior, which unless I'm mistaken what I've observed runs counter to it.
So is this fine? I mean, it runs, but if gdb is at least slightly broken I don't know what other utilities would be. I'd like to see about making use of Eclipse+Photran and TCF Agent, and I was looking into where the CodeSourcery toolchain came from, which led me to Yocto. I've been trying some builds with Yocto at home, and I'm wondering if it's worth my time to bring in a VM with all of this set up to try run our application with the toolchain built into the rootfs from the ground up.
Probably too late to switch anything over but if I can just have a smoother environment under which to work then I can just toss the source back into our current solution when it's time to turn in a fix.
Related
I have a Linux application written in C++ with proprietary source code so this question is about any mechanisms that could cause what I describe below to happen.
Anyway, the application uses some library functions that perform some data processing and spits out a result. On my machine, I get the expected result. On others, the app either segfaults or outputs nonsense.
It is distributed as a bit of source code and a static library, built on my machine, which can then be built with make and run (and possibly distributed). The OS environment is Ubuntu 18.04.4 LTS (WSL1) and is essentially the same on all of the machines.
The ldd utility indicates that the same libraries are being used and the Ubuntu and g++ versions are identical as well.
The application was run in gdb on one of the machines where it segfaults and it turned out that a specific buffer hadn't been allocated and was nullptr. The size parameter being passed to the constructor in this case is a #define macro.
I am aware that the preferred approach on Linux is to distribute source code, not precompiled binaries, but for IP protection purposes, that might not always be possible. I have also read that there could be subtle library version or kernel-level incompatibilities that can cause this kind of weirdness, but like I said, that doesn't seem to be the case here.
Q: Does anybody know why I might be observing this strange behavior?
I'm a whiz with Visual C++, but Linux development is new to me. In Visual Studio, it's easy to trace into any code implemented by the C run time libraries. I just need to make sure the sources are installed and I can step right into any function I'd like -- malloc(), cout::operator<<(), whatever.
I'm trying to develop using Eclipse's C++ package. How can I step into C run time routines there? Since Linux is open-source, how do I step into operating system routines? Seems like it should e possible -- am I missing debug information, source code, or both? Something in my configuration?
I'm using Ubuntu 12.10 at the moment. I'm using g++. I believe I'm using the Eclipse build system as I never imported a makefile project; I just started with a simple "Hello World" project from the C++ project wizard in Eclipse.
After hacking at this a bit:
I've installed the libstdc++6-4.2-dbg package thinking it would be debug symbols for the libstdc library:
sudo apt-get install libstdc++6-4.2-dbg
I've also installed dpkg-dev, since the next step said I needed it:
sudo apt-get install dpkg-dev
I tried installing libc6 sources into a directory under my home:
apt-get source libc6
At this point, trying to step into printf() tells me that printf.c is missing. I can't step into malloc or strlen, which suggests that I don't understand how the C runtime libraries are factored in Linux. How are libc, glib, and libstdc++ different? Which packages do I need?
If I ask Eclipse to open the printf.c file I do have (at ~/eglibc-2.15/stdio-common/printf.c), it doesn't open the file (doesn't adjust the debugging window to show the source) and repaints the window that shows the error message about not being able to find the file. (Can't find a source file at "printf.c" Locate the file or edit the source lookup path to include its location.)
Whilst, as a Kernel developer on Linux, I do agree that using the individual tools separately will be a good thing to learn, and as such Basile's answer is usefuel.
However, the stepping into C runtime libraries should be equally possible with Eclipse. But just because the OS is open source doesn't mean that it supports you clambering around inside it willy nilly - in fact, you CAN NOT step into the OS itself from user-mode code. You nee KGDB (google it), and you definitely need a second computer to attach to the one being debugged, because when you step into the kernel, you will essentially lock up the machine, at the very least in the context you are stepping, but most likely also prevent other work from being done until you get back out from the kernel, so for example, if you step into open(), at some point the entire filesystem may well stop working altogether until you are back out of whatever lock you are holding. This wll certainly upset some software. Note that this is just an example of how things may work unexpectedly when debugging the kernel, not strictly "I've done this and it happened" - I have debugged kernels with debuggers several times, and you do have to be careful with what you do, and you certainly can not run the debugger on the same machine, as the machine STOPS when you are debugging.
Going back to the usermode, which you CAN debug via Eclipse, essentially all you need to do is install the source code for the runtime library you are interested in, and go... Same principle as on Windows with visual studio - except that nearly all software you ever run on a Linux system is available as source code. You may need to recompile some libraries with debugging symbols, and just like in Windows, you need to make sure the debugger knows how to find the source code. Everything else should be handled by the debugger in Eclipse. I spent about three years using Eclipse for both local and remote debugging, and in general, it works. There are quirks in places, but that's the case with almost any debugger.
Good luck.
First, you don't need Eclipse to develop software on Linux. You should better learn to do that with independent tools (command line) like emacs or gedit (as editor), git (version control), make (builder) which runs the gcc or g++ compiler (both gcc & g++ are part of GCC, the Gnu Compiler Collection).
really, you'll learn a lot more by not depending upon Eclipse; it may just hide you the real commands which are doing the job, and you should understand what they really are.
You want to pass the -g -Wall options to GCC. The -g option asks for debug information, and the -Wall options asks for almost all warnings. Improve your code till no warnings are given.
And the operating system is providing syscalls (which are operations provided by the kernel to applications; from the application's point of view, a syscall is atomic so you cannot step into it; however strace may show you all the syscalls done by some execution). If you wanted to step by step inside system libraries like libc you need the debugging variant of it (e.g. some libc6-dbg package). But there is usually no need to dive inside system libraries.
See http://advancedlinuxprogramming.com/
Then, you will use gdb to debug the binary program.
So, step by step instructions inside a terminal:
edit your source files with emacs or gedit
learn how to use GCC: for a single source C++ program compile it with g++ -Wall -g source.cc -o progbin and type ./progbin in your terminal to run it. Only when the program is debugged and satisfactory would you compile it with optimizations (by giving the -O or -O2 flag to gcc or g++)
Use gdb to debug a program (compiled with -g).
for a multi-file C++ program, consider learning how to use make
use a version control system like git
For beginners, I suggest to avoid Eclipse, because it just hides to you what is really happening underneath (Eclipse is simply running other tools like the above commands)
Software development under Linux requires a different mindset than under Windows: you really are using your own loose combination of independent tools, so better to learn a bit each of them.
NB. to step inside "system" functions like malloc (which is above syscalls like mmap) you need the debug variant of the libc package with aptitude install libc6-dbg, and you need to set LD_LIBRARY_PATH to /usr/lib/debug etc...
This question may seem blindingly obvious and I realise I am putting myself up for a large number of downvotes but I am very new to Linux dev and have only been working on it for a while.
I have been writing an application on ubuntu 12.04 (kernel 3.2.0) in C++ then copying this via scp to an ubuntu 8.04 (kernel 2.6.30) installation on another device. I have been noticing some very strange behaviour that I simply cannot explain. I have naively assumed that I can run this executable on a previous version, but it is beginning to dawn on me that this in fact may not be the case. In future must I ensure that the Linux version I build my application on is identical to that which it will be running on in the field?? Or must I actually build the application from source code directly on the device it will be running on??? I am very new to Linux dev but not new to C++ so I realise that this question may seem facile, but this is the kind of issue that I have simply not seen in books/tutorials etc.
Most of the time, it's not the kernel that stops you, it's glibc.
glibc is backwards compatible, meaning programs compiled and linked to an older version will work exactly the same with a newer version at runtime. The other way around is not that compatible.
Best is of course to build on the distro you want to run it. If you can't do that, build on the one with the oldest glibc install.
It's also very hard to build and link to an older glibc than the system glibc, installing/building glibc tends to mess up your system more than it's worth. Set up a VM with an old Linux, and use that instead.
Question in one sentence: How can I compile code for AIX using G++ on Ubuntu? (Assuming it is possible)
I hope that it is as simple as adding an option to the make file to specify target processor. I am a novice when it comes to most things compiler related.
Thank you in advance.
What you are looking for is a cross-compiling toolchain.
A toolchain includes a cross-compiler (a compiler that runs on the current platform but builds the binary code to run on another, on your case, AIX), the C or C++ library, and some other interesting tools.
I have successfully used buildroot in the past, which is a tool that automates the process of creating a cross-compiling toolchain. I know they support several target platforms, maybe AIX is among them.
If you want to compile your toolchain by hand, take a look at the Roll-your-own section on this page.
Another approach, probably easier on your case, would be to install a AIX system inside a virtual machine on Ubuntu. This way you would have access to a complete AIX system running inside your machine, giving the opportunity to develop and test your application under real conditions (or whatever reasons you may find interesting for doing such a thing).
You'll want to download the right version of g++ (i.e., one that generates code for POWER, or whatever you're running AIX on), and compile that to run under Ubuntu.
Try to compile with -maix32 or -maux64 (like g++ -m32 file.cxx), it it doesn't work it means that it's not supported by your compiller.
I am trying to build gdb for armv6 architecture. I will be compiling this package on a Fedora Linux-Intel x86 box. I read the process of installing the gdb, like
Download the source pachage
run configure -host
make
But I got lost in the process because I was not able to make out what will be the host, target, needed for the configure script.
I need to basically be able to debug programs running on armv6 architecture board which runs linux kernel 2.6.21.5-cfs-v19. The gdb executable which I intend to obtain after compilation of the source also needs to be able to run on above mentioned configuration.
Now to get a working gdb executable for this configuration what steps should I follow?
We (www.rockbox.org) use the arm target for a whole batch of our currently working DAPS. The target we specify is usually arm-elf, rather than arm-linux.
Be careful with arm-linux vs. arm-elf, eg.
http://sources.redhat.com/ml/crossgcc/2005-11/msg00028.html
arm-elf is a standalone toolchain which does not require an underlying OS. So you can use
it to generate programs using newlib
arm-linux is a toolchain targetted to generate code for linux OS running on an ARM machine
We sometimes say arm-elf is for "bare metal".
Unfortunately there's another "bare metal" target arm-eabi and no one knows what the difference between these two exactly is.
BTW,
The gdb executable which i intend to obtain after compilation of the source,also needs to be able to run on above mentioned configuration.
Really? Running GDB on an ARM board may be quite slow.
I recommend you either of
Remote debugging of the ARM board from an x86 PC
Saving a memory core on the ARM board, transferring it to an x86 PC and then inspecting it there
Cf.
http://elinux.org/GDB
Cross-platform, multithreaded debugging (x86 to ARM) with gdb and gdbserver not recognizing threads
http://www.chromium.org/chromium-os/how-tos-and-troubleshooting/remote-debugging
target/host is usually the target tool chain you would be using (mostly arm-linux)