gcc/contrib/header-tools/README
Andrew MacLeod bd94906f98 count-headers: Initial file.
2015-11-11  Andrew MacLeod  <amacleod@redhat.com>

	* count-headers: Initial file.
	* gcc-order-headers: Initial file.
	* graph-header-logs: Initial file.
	* graph-include-web: Initial file.
	* headerutils.py: Initial file.
	* included-by: Initial file.
	* README: Initial file.
	* reduce-headers: Initial file.
	* replace-header: Initial file.
	* show-headers: Initial file.

From-SVN: r230171
2015-11-11 14:28:46 +00:00

284 lines
10 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

Quick start documentation for the header file utilities.
This isn't a full breakdown of the tools, just they typical use scenarios.
- Each tool accepts -h to show it's usage. Usually no parameters will also
trigger the help message. Help may specify additional functionality to what is
listed here.
- For all tools, option format for specifying filenames must have no spaces
between the option and filename.
ie.: tool -lfilename.h target.h
- Many of the tools are required to be run from the core gcc source directory
containing coretypes.h. Typically that is in gcc/gcc from a source checkout.
For these tools to work on files not in this directory, their path needs to be
specified on the command line.
ie.: tool c/c-decl.c lto/lto.c
- options can be intermixed with filenames anywhere on the command line
ie. tool ssa.h rtl.h -a is equivalent to
tool ssa.h -a rtl.h
gcc-order-headers
-----------------
This will reorder any primary backend headers files known to the tool into a
canonical order which will resolve any hidden dependencies they may have.
Any unknown headers will simply be placed after the recognized files, and
retain the same relative ordering they had.
This tool must be run in the core gcc source directory.
Simply execute the command listing any files you wish to process on the
command line.
Any files which are changed are output, and the original is saved with a
.bak extention.
ex.: gcc-order-headers tree-ssa.c c/c-decl.c
-s will list all of the known headers in their canonical order. It does not
show which of those headers include other headers, just the final canonical
ordering.
if any header files are included within a conditional code block, the tool
will issue a message and not change the file. When this happens, you can
manually inspect the file to determine if reordering it is actually OK. Then
rerun the command with the -i option. This will ignore the conditional error
condition and perform the re-ordering anyway.
If any #include line has the beginning of a multi-line comment, it will also
refuse to process the file until that is resolved by terminating the comment
on the same line, or removing it.
show-headers
------------
This will show the include structure for any given file. Each level of nesting
is indented, and when any duplicate headers are seen, they have their
duplicate number shown
-i may be used to specify additional search directories for headers to parse.
-s specifies headers to look for and emphasize in the output.
This tool must be run in the core gcc source directory.
ex.: show-headers -sansidecl.h tree-ssa.c
tree-ssa.c
config.h
auto-host.h
ansidecl.h (1) <<-------
system.h
safe-ctype.h
filenames.h
hashtab.h (1)
ansidecl.h (2) <<-------
libiberty.h
ansidecl.h (3) <<-------
hwint.h
coretypes.h
machmode.h (1)
insn-modes.h (1)
signop.h
<...>
count-headers
-------------
simply count all the headers found in the specified files. A summary is
printed showing occurrences from high to low.
ex.: count-headers tree*.c
86 : coretypes.h
86 : config.h
86 : system.h
86 : tree.h
82 : backend.h
80 : gimple.h
72 : gimple-iterator.h
70 : ssa.h
68 : fold-const.h
<...>
included-by
-----------
This tool will search all the .c,.cc and .h files and output a list of files
which include the specified header(s).
A 4 level deep 'find' of all source files is performed from the current
directory and each of those is inspected for a #include of the specified
headers. So expect a little bit of slowness.
-i limits the search to only other header files.
-c limits the search to .c and .cc files.
-a shows only source files which include all specified headers.
-f allows you to specify a file which contains a list of source files to
check rather than performing the much slower find command.
ex: included-by tree-vectorizer.h
config/aarch64/aarch64.c
config/i386/i386.c
config/rs6000/rs6000.c
tree-loop-distribution.c
tree-parloops.c
tree-ssa-loop-ivopts.c
tree-ssa-loop.c
replace-header
--------------
This tool simply replaces a single header file with one or more other headers.
-r specifies the include to replace, and one or more -f options specify the
replacement headers, in the order they occur.
This is commonly used in conjunction with 'included-by' to change all
occurrences of a header file to something else, or to insert new headers
before or after.
ex: to insert #include "before.h" before every occurence of tree.h in all
.c and .cc source files:
replace-header -rtree.h -fbefore.h -ftree.h `included-by -c tree.h`
reduce-headers
--------------
This tool removes any header files which are not needed from a source file.
This tool must be run for the core gcc source directory, and requires either
a native build and sometimes target builds, depending on what you are trying
to reduce.
it is good practice to run 'gcc-order-headers' on a source file before trying
to reduce it. This removes duplicates and performs some simplifications
which reduce the chances of the reduction tool missing things.
start with a completely bootstrapped native compiler.
Any desired target builds should be built in one directory using a modified
config-list.mk file which does not delete the build directory when it is done.
any target directories which do not successfully complete a 'make all-gcc'
may cause the tool to not reduce anything.
(todo - provide a config-list.mk that leaves successful target builds, but
deletes ones which do not compile)
The tool will examine all the target builds to determine which targets build
the file, and include those targets in the testing.
The tool will analyze a source file and attempt to remove each non-conditional
header from last to first in the file.:
It will first attempt to build the native all-gcc target.
If that succeeds, it will attempt to build any target build .o files
If that succeeds, it will check to see if there are any conditional
compilation dependencies between this header file and the source file or
any header which have already been determined as non-removable.
If all these tests are passed, the header file is determined to be removable
and is removed from the source file.
This continues until all headers have been checked.
At this point, a bootstrap is attempted in the native build, and if that
passes the file is considered reduced.
Any files from the config subdirectory require target builds to be present
in order to proceed.
A small subset of targets has been determined to provide excellent coverage,
at least as of Aug 31/15 . They were found by reducing all the files
contained in libbackend.a oer a full set of targets(207). All conditions
which disallowed removal of a header file were triggered by one or more of
these targets. They are also known to the tool. When building targets it
will check those targets before the rest.
This coverage can be achieved by building config-list.mk with :
LIST="aarch64-linux-gnu arm-netbsdelf avr-rtems c6x-elf epiphany-elf hppa2.0-hpux10.1 i686-mingw32crt i686-pc-msdosdjgpp mipsel-elf powerpc-eabisimaltivec rs6000-ibm-aix5.1.0 sh-superh-elf sparc64-elf spu-elf"
-b specifies the native bootstrapped build root directory
-t specifies a target build root directory that config-list.mk was run from
-f is used to limit the headers for consideration.
example:
mkdir gcc // checkout gcc in subdir gcc
mdsir build // boostrap gcc in subdir build
mkdir target // create target directory and run config-list.mk
cd gcc/gcc
reduce-headers -b../../build -t../../targets -falias.h -fexpr.h tree*.c (1)
# This will attempt to remove only alias.h and expr.h from tree*.c
reduce-headers -b../../build -t../../targets tree-ssa-live.c
# This will attempt to remove all header files from tree-ssa-live.c
the tool will generate a number of log files:
reduce-headers.log : All compilation failures from attempted reductions.
reduce-headers.sum : One line summary of what happened to each source file.
(All the remaining logs are appended to, so if the tool is run multiple times
these files are just added to. You must physically remove them yourself in
order to reset the logs.)
reduce-headers-kept.log: List of all the successful compiles that were
ignored because of conditional macro dependencies
and why it thinks that is the case
$src.c.log : for each failed header removal, the compilation
messages as to why it failed.
$header.h.log: The same log is put into the relevant header log as well.
a sample output from ira.c.log:
Compilation failed:
for shrink-wrap.h:
============================================
/gcc/2015-09-09/gcc/gcc/ira.c: In function bool split_live_ranges_for_shrink_wrap():
/gcc/2015-09-09/gcc/gcc/ira.c:4839:8: error: SHRINK_WRAPPING_ENABLED was not declared in this scope
if (!SHRINK_WRAPPING_ENABLED)
^
make: *** [ira.o] Error 1
the same message would be put into shrink-wrap.h.log.
graph-header-logs
-----------------
This tool will parse all the messages from the .C files, looking for failures
that show up in other headers... meaning there is a compilation dependency
between the 2 header files.
The tool will aggregate all these and generate a graph of the dependencies
exposed during compilation. Red lines indicate dependencies that are
present because a header file physically includes another file. Black lines
represent data dependencies causing compilation failures if the header is
not present.
ex.: graph-header-logs *.c.log
graph-include-web
-----------------
This tool can be used to visualize the include structure in files. It is
rapidly turned useless if you specify too many things, but it can be
useful for finding cycles and redundancies, or simply to see what a single
file looks like.
ex.: graph-include-web tree.c