Has someone infos how to build a llvm+clang toolchain using binutils and newlib and how to use it?
host: Linux, AMD64
target: cortex-m3, stm32
c-lib: newlib
assembler: gnu as
I created a firmware framework - PolyMCU https://github.com/labapart/polymcu - that is based on CMake that support GCC and LLVM. Because it is based on CMake you can build your firmware on Linux/Windows/MacOS.
It also uses Newlib - it looks all your requirements are there!
I also wrote a blog where I compared GCC and LLVM build size on ARM Cortex-M: http://labapart.com/blogs/3-the-importance-of-the-toolchain-version-in-embedded-space
Interesting results, Clang generated code is not much bigger than GCC on Cortex-M...
Unfortunately, right now clang does not support flexible cross-compilation settings. So, most probably you will need to invoke necessary tools with all necessary arguments.
Start with building llvm + clang using --target=thumbv7-eabi configure argument (note that you will need llvm + clang as of yesterday for this). You might want to specify --enable-targets=arm as well. This will instruct clang to generate code for thumb by default. After this you can invoke clang -mcpu=cortex-m3 to generate the code for you.
You will have to provide all necessary include / library paths by hands via -I / -L, etc.
If you're happy with some C++ hacking, you can write necessary "HostInfo", so it will invoke the right tools and provide right paths automagically.
Related
When using clang v8.0.0 on Windows (from llvm prebuilt binaries) with -g or -gline-tables-only source map tables are not being picked up by gdb or lldb debuggers.
Upon including -g flag file grows in size (which is to be expected) yet neither gdb nor lldb pickes the source up
When compiled with gcc though (with -g flag) source files are detected by debugger.
I have tried running the same command (clang -g <codefile>) on macOS High Sierra (clang -v says it is Apple LLVM version 10.0.0 (clang-1000/10.44.4)) where there source files are being picked up by lldb. So I guessed it is localized to my windows instance or llvm for windows build.
P.S. output of clang -v on windows:
clang version 8.0.0 (tags/RELEASE_800/final)
Target: x86_64-pc-windows-msvc
Thread model: posix
InstalledDir: C:\Program Files\LLVM\bin
On Windows, Clang is not self-sufficient (at least not the official binaries). You need to have either GCC or MSVC installed for it to function.
As Target: x86_64-pc-windows-msvc indicates, by default your Clang is operating in some kind of MSVC-compatible mode. From what I gathered, it means using the standard library and other libraries provided by your MSVC installation, and presumably generating debug info in some MSVC-specific format.
Add --target=x86_64-w64-windows-gnu to build in GCC-compatible mode. (If you're building for 32 bits rather than 64, replace x86_64 with i686). This will make Clang use headers & libraries provided by your GCC installation, and debug info should be generated in a GCC-compatible way. I'm able to debug resulting binaries with MSYS2's GDB (and that's also where my GCC installation comes from).
If you only have GCC installed and not MSVC, you still must use this flag.
How do I know this is the right --target? This is what MSYS2's Clang uses, and I assume they know what they're doing. If you don't want to type this flag every time, you can replace the official Clang with MSYS2's one, but I'm not sure if it's the best idea.
(I think they used to provide some patches to increase compatibility with MinGW, but now the official binaries work equally well, except for the need to specify the target. Also, last time I checked their binary distribution was several GB larger, due to their inability to get dynamic linking to work. Also some of the versions they provided were prone to crashing. All those problems come from them building their Clang with MinGW, which Clang doesn't seem to support very well out of the box. In their defence, they're actively maintaining their distribution, and I think they even ship libc++ for Windows, which the official distribution doesn't do.)
I am writing some code for raspberry pi ARM target on x86 ubuntu machine. I am using the gcc-linaro-armhf toolchain. I am able to cross compile and run some independent programs on pi. Now, I want to link my code with external library such as ncurses. How can I achieve this.
Should I just link my program with the existing ncurses lib on host machine and then run on ARM? (I don't think this will work)
Do I need to get source or prebuilt version of lib for arm, put it in my lib path and then compile?
What is the best practice in this kind of situation?
I also want to know how it works for the c stdlib. In my program I used the stdio functions and it worked after cross compiling without doing anything special. I just provided path for my arm gcc in makefile. So, I want to know, how it got correct std headers and libs?
Regarding your general questions:
Why the C library works:
The C library is part of your cross toolchain. That's why the headers are found and the program correctly links and runs. This is also true for some other very basic system libraries like libm and libstdc++ (not in every case, depends on the toolchain configuration).
In general when dealing with cross-development you need some way to get your desired libraries cross-compiled. Using binaries in this case is very rare. That is, especially with ARM hardware, because there are so many different configurations and often everything is stripped down much in different ways. That's why binaries are not very much binary compatible between different devices and Linux configurations.
If you're running Ubuntu on the Raspberry Pi then there is a chance that you may find a suitable ncurses library on the internet or even in some Ubuntu apt repository. The typical way, however, will be to cross compile the library with the specific toolchain you have got.
In cases when a lot and complex libraries need to be cross-compiled there are solutions that make life a bit easier like buildroot or ptxdist. These programs build complete Linux kernels and root file systems for embedded devices.
In your case, however, as long as you only want ncurses you can compile the source code yourself. You just need to download the sources, run configure while specifying your toolchain using the --host option. The --prefix option will choose the installation directory. After running make and make install, considering everything went fine, you will have got a set of headers and the ARM-compiled library for your application to link against.
Regarding cross compilation you will surely find loads of information on the internet and maybe ncurses has got some pointers in its shipped documentation, too.
For the query How the C library works in cross-tools
When compiling and building cross-tool chain during configuration they will provide sysroot.
like --with-sysroot=${CLFS_CROSS_TOOLS}
--with-sysroot
--with-sysroot=dir
Tells GCC to consider dir as the root of a tree that contains (a subset of) the root filesystem of the target operating system. Target system headers, libraries and run-time object files will be searched for in there. More specifically, this acts as if --sysroot=dir was added to the default options of the built compiler. The specified directory is not copied into the install tree, unlike the options --with-headers and --with-libs that this option obsoletes. The default value, in case --with-sysroot is not given an argument, is ${gcc_tooldir}/sys-root. If the specified directory is a subdirectory of ${exec_prefix}, then it will be found relative to the GCC binaries if the installation tree is moved.
So instead of looking /lib /usr/include it will look /Toolchain/(libc) and (include files) when its compiling
you can check by
arm-linux-gnueabihf-gcc -print-sysroot
this show where to look for libc .
also
arm-linux-gnueabihf-gcc -print-search-dirs
gives you clear picture
Clearly, you will need an ncurses compiled for the ARM that you are targeting - the one on the host will do you absolutely no good at all [unless your host has an ARM processor - but you said x86, so clearly not the case].
There MAY be some prebuilt libraries available, but I suspect it's more work to find one (that works and matches your specific conditions) than to build the library yourself from sources - it shouldn't be that hard, and I expect ncurses doesn't take that many minutes to build.
As to your first question, if you intend to use ncurses library with your cross-compiler toolchain, you'll have its arm-built binaries prepared.
Your second question is how it works with std libs, well it's really NOT the system libc/libm the toolchain is using to compile/link your program is. Maybe you'll see it from --print-file-name= option of your compiler:
arm-none-linux-gnuabi-gcc --print-file-name=libm.a
...(my working folder)/arm-2011.03(arm-toolchain folder)/bin/../arm-none-linux-gnuabi/libc/usr/lib/libm.a
arm-none-linux-gnuabi-gcc --print-file-name=libpthread.so
...(my working folder)/arm-2011.03(arm-toolchain folder)/bin/../arm-none-linux-gnuabi/libc/usr/lib/libpthread.so
I think your Raspberry toolchain might be the same. You can try this out.
Vinay's answer is pretty solid. Just a correction when compiling the ncurses library for raspberry pi the option to set your rootfs is --sysroot=<dir> and not --with-sysroot . Thats what I found when I was using the following compiler:
arm-linux-gnueabihf-gcc --version
arm-linux-gnueabihf-gcc (crosstool-NG linaro-1.13.1+bzr2650 - Linaro GCC 2014.03) 4.8.3 20140303 (prerelease)
Copyright (C) 2013 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.
There are instructions on llvm.org on how to add the Gold plugin to LLVM-gcc; however since version 3.2 they are not using LLVM-gcc anymore, using Clang instead, and it's unclear on how to use it.
Does anyone know how to add the Gold plugin to LLVM 3.2?
Do we need to install this plug-in at all, or is it embedded in Clang?
I've read that if you don't have Gold plugin as linker, if you try -O4 for LTO it will treats it as -O3, but I suppose it is when you use gcc system linker and not when using Clang, right?
1-If you don't want to change any thing in home/bin directory, then you can compile and install binutils and LLVM in same place it will automatically generate ld,ar and (not ld_new) in llvm/bin directory and since llvm first looks to its own directory to find tools (clang -print-search-dirs) so you can be sure that clang will invoke ld gold in link time first. Its not mentioned in LLVMgold.html but its needed to create bfd-plugins directory in lib and install both LLVMgold.so and linLTO.so there.
2- to support LTO in LLVM we need gold plugin and need to be rebuilt to support it. -O4 is similar to -O3 -flto
I've compiled clang to use it as a cross compiler for ARM (by configuring it with ./configure --target=armv7l-unknown-linux-gnueabihf ), but when I try to compile any C code, it tries to use /usr/bin/as. I already have binutils compiled for ARM, and they are in a separate directory. How do I direct clang (or llvm) to use the assembler that I specify?
try passing the --host option to configure which will cause all the cc ar etc utilities to prefix with armv7l-unknown-linux-gnueabihf-
eg:
./configure --host=armv7l-unknown-linux-gnueabihf --build=i686-unknown-linux-gnu
Since you are using configure with hopefully autotools take a look at:
automake Cross compiling
I've always had trouble configuring from the source directory using ./configure and gave up in the end. These days I always configure from a separate directory i.e. ..//configure although I'm told it's recommended to use an absolute path for configure.
Your ARM binutils should be installed in the same prefix you're using for clang and make sure that they're in the path when you configure & build clang - i.e. PATH=/some/prefix/bin:$PATH; /configure --target=armv7l-unknown-linux-gnueabihf. If you're keeping them in separate directories for packaging purposes then make install DESTDIR= should help.
I don't generally build clang but the buildscripts I use for devkitARM might be helpful - http://sourceforge.net/p/devkitpro/buildscripts/ci/c372699fc7b4de90eb044314ce5bed04db640291/tree/
I have some old hardware with an old version of say SuSE linux running on it. Now I have this fancy development machine running Ubuntu 9.10. Some of the tools I use to compile my C app (written in Python 2.6.x) are not available on the old SuSe box. So... is it possible to compile for that old machine on my dev box?
I have the following steps in mind, but would like to cross-check before venturing off into this quest:
1. Find out which static/shared libs my app needs and find/build target version of them
2. Also find the corresponding header files
3. Feed the correct flags to gcc to use the target headers and libraries
4. Feed the correct flags to gcc to use the correct architecture (i386/i686), or do I need a cross-compilation toolchain.
5. Compile, upload and enjoy ;-)
I regularly use avr-gcc and cc65, both are cross compiling. I know that you set up a coss compiler for developing something like a gumstix, so it should be possible to do the same for old/other Linux distros, not?
C
The way I would approach this is grab your oldmachine:/usr/lib and oldmachine:/usr/include so you have e.g. newmachine:/oldmachinecompiler/usr/{lib|include} then build a cross compiler setting --sysroot to newmachine:/oldmachinecompiler/
This is really the only way to ensure that any library requirements (including libc) in your program are compatible with oldmachine.