|Version 9 (modified by dterei, 5 years ago) (diff)|
Modifying the build system
This section covers making changes to the GHC build system. We'll give some general advice on how to work with the build system, and then describe a few common scenarios, such as how to add a new source file.
Note that before making any non-trivial changes to the build system you should acquaint yourself with the overall architecture. Even if you're already familiar with GNU make, the GHC build system is probably quite different from most Makefile-based build systems you've encountered before.
Incedentally, it's a good idea to have a copy of the GNU make documentation to hand when working with the build system.
When the build system doesn't do what you want, the results can be pretty cryptic. Often the problem is that something is being built in the wrong order, or some variable isn't being propagated to the places you thought it was. How do you go about debugging the build system?
Here are the techniques that we use. Note, for many of these diagnosis techniques you may want to invoke make on ghc.mk directly using make -f ghc.mk, to bypass the phase ordering machinery of the top-level Makefile.
- $(warning ... message ...)
- equivalent to "printf-debugging` in a C program: this causes make to print a message when it reads the $(warning ..) expression, and the message can include variable references. Very useful for finding out what make thinks the value of a variable is at a particular point in the Makefile, or for finding out the parameters for a particular macro call. Want to test your hypothesis about what the variable rts_C_SRCS contains? Just add a $(warning $(rts_C_SRCS)) somewhere.
- make show VALUE=VAR
- prints the value of variable VAR. Useful for quick diagnosis.
- make TRACE=1
- prints messages about certain macros that are called and their arguments. This is basically a system of $(warning) calls enabled when the value of $(TRACE) is non-empty. To see how it works, look at the file rules/trace.mk, and feel free to add trace calls to more places in the build system.
- make --debug=b --debug=m
- causes make to show the sequence of dependencies that it is following, which will often tell you why something is being built. This can help to track down missing or incorrect dependencies.
- make -p
- prints out all the generated rules and variables. The output can be huge; so pipe it to a file, and search through it for the bits of interest.
The following are a collection of common changes you might want to make to the build system, with the exact steps required in each case.
Adding a source file to GHC
GHC is in two parts:
- the ghc package, which comprises most of the GHC source code, can be found in the compiler directory.
- the ghc program itself, consists of a single Main module that imports the ghc package. It is found in the ghc directory.
Like any Cabal package, the ghc package has a ghc.cabal file, except that in this case the file is preprocessed by configure from the original: compiler/ghc.cabal.in. Be careful not to modify the preprocessed version, ghc.cabal, as it will be overwritten next time you run configure.
To add a module to the ghc package:
- Add your module to the exposed-modules section of compiler/ghc.cabal.in
- cd compiler; make stage2
Removing a source file from GHC
To retire a GHC source file that is no longer needed:
- Remove the working copy of the file (git will notice it is gone), or use git rm.
- Remove the module from the list of modules in compiler/ghc.cabal.in.
- To remove all mention of the file from derived dependency files, it is necessary to do something on the order of
cd $TOP sh config.status make all_compiler_stage1or make all_compiler_stage2 if you prefer.
If you wish something slower but more confident, you may cd $TOP; ./configure; make.
Adding a source file to the RTS
The RTS picks up source files automatically, using make's $(wildcard) function. So to add files to the RTS, just put your source file in rts, and
$ cd rts; make
To have the change propagated to the stage 2 compiler, either go and make stage 2 explicitly, or issue a top-level make.
Adding a new package
Adding a new package is quite straightforward:
- To arrange that the package is checked out by sync-all, add an entry to the packages file.
- run perl boot to generate the ghc.mk and GNUmakefile files in your package's build.
- Add an entry to the PACKAGES variable in ghc.mk. The list in PACKAGES is kept in dependency order: each package must appear after the packages it depends on.
That's it: doing a top-level make should now build your package and bring everything else up to date.
Adding a program
Utility programs live in utils, for example utils/hpc. Each program has a simple ghc.mk file, for example the one for hpc looks like this:
utils/hpc_dist_MODULES = Main HpcCombine HpcDraft HpcFlags HpcLexer HpcMap \ HpcMarkup HpcOverlay HpcParser HpcReport HpcSet \ HpcShowTix HpcUtils utils/hpc_dist_HC_OPTS = -cpp -package hpc utils/hpc_dist_INSTALL = YES utils/hpc_dist_PROG = hpc$(exeext) $(eval $(call build-prog,utils/hpc,dist,1)) $(eval $(call bindist,utils/hpc,ghc.mk))
This is all that is necessary to build and install a program. Taking each of these lines in turn:
utils/hpc_dist_MODULES = Main HpcCombine HpcDraft HpcFlags HpcLexer HpcMap \ HpcMarkup HpcOverlay HpcParser HpcReport HpcSet \ HpcShowTix HpcUtils
remember that variable names all begin with directory_distdir (see Idiom: variable names), and in this case the directory is utils/hpc, and the distdir is just dist. The variable utils/hpc_dist_MODULES specifies the list of Haskell modules that make up the hpc program.
utils/hpc_dist_HC_OPTS = -cpp -package hpc
This defines any extra options that need to be passed to GHC when compiling hpc.
utils/hpc_dist_INSTALL = YES
This line indicates that we want the hpc program to be installed by make install. The default is not to install programs.
utils/hpc_dist_PROG = hpc$(exeext)
This specifies the name of the program. The variable $(exeext) contains .exe on Windows, and is otherwise empty.
$(eval $(call build-prog,utils/hpc,dist,1))
This is a call to the build-prog macro, whose definition can be found in rules/build-prog.mk. This expands to the code for actually building the program, and it takes three arguments: the directory, the distdir, and the GHC stage to use for building (in this case we're using stage 1). The build-prog macro expects to find certain variables defined, namely directory_distdir_PROG, and directory_distdir_MODULES.
Finally, for programs that we want to install, we need to include the ghc.mk file in a binary distribution:
$(eval $(call bindist,utils/hpc,ghc.mk))