To build a 32-bit SO file, you need to choose a device type with the length of "long" as 32 bits. One option in this instance would be to choose:
Device vendor: Generic
Device type: 32-bit x86 compatible
If you want to customize further, there is the target namespace, which allows the definition of custom device types. See https://mathworks.com/help/coder/ug/register-new-hardware-devices.html#mw_cea14c66-5050-43a4-a8b4-d2d175b7f64b for more details on how to do this.
If you are using a 64-bit Linux host:
You can generate a 32-bit shared library (.SO) on 64-bit Linux using 'ert_shrlib.tlc' and the ' -m32' toolchain option.
In Configuration Parameters, set 'System target file', 'Toolchain', and 'Build configuration' as follows and add '-m32' to 'C Compiler' and 'Shared Library Linker' options:
Alternatively, the following code is an example script for setting a toolchain and options programmatically:
model = 'modelName';
% set TLC for shared library generation
set_param(model, 'SystemTargetFile', 'ert_shrlib.tlc');
% set toolchain
set_param(model, 'Toolchain', 'GNU gcc/g++ | gmake (64-bit Linux)');
set_param(model, 'BuildConfiguration', 'Specify');
% get default toolchain options
opts = get_param(model, 'CustomToolchainOptions');
opts{2} = [opts{2} ' -m32']; % add -m32 to C Compiler options
opts{6} = [opts{6} ' -m32']; % add -m32 to Shared Library Linker options
% set custom toolchain options
set_param(model, 'CustomToolchainOptions', opts);
In this example script, 'opts' is a cell array with compile options that can be customized.
If you use a version after R2023b, you can also create your custom toolchain using the "target.Toolchain" objects.
To create it, you can run the following code:
toolchain = target.create('Toolchain', ...
'Name', 'GCC for 32-bit Linux', ...
'Family', 'GNU', ...
'CCompiler', 'gcc -m32', ...
'CppCompiler', 'g++ -m32', ...
'ExecutableExtension', '', ...
'SharedLibraryExtension', '.so');
ld = toolchain.Tools(strcmp({toolchain.Tools.Name}, 'GCC for 32-bit Linux - Linker'));
ld.setDirective('Shared', '-shared');
cppld = toolchain.Tools(strcmp({toolchain.Tools.Name}, 'GCC for 32-bit Linux - C++ Linker'));
cppld.setDirective('Shared', '-shared');
target.add(toolchain);
You can then set it using:
% set toolchain
set_param(model, 'Toolchain', 'GCC for 32-bit Linux');
If you are using a Windows host:
From R2024a onwards, it's possible to define a toolchain that builds on a remote Linux host from a Windows host by transferring the source code via SSH and then running CMake on the remote Linux host.
You would have to customize the toolchain appropriately to build the 32-bit binary.
See https://mathworks.com/help/releases/R2024b/rtw/ug/build-generated-code-remotely.htmlfor more details on this.
