Commit 07eea3aa4 `make --shuffle` prevented shuffling prerequisites
that use .SECONDEXPANSION, since shuffle happens before expansion.
This has two problems:
1. No shuffling happens for such prerequisites.
2. Use-after-free when outdated '->shuf' links are used.
Add a reshuffle into expansion phase right after dependency changes.
* src/file.c (expand_deps): Add reshuffle if dependencies change.
* src/shuffle.c (identity_shuffle_array): Fix comment typo.
* tests/scripts/options/shuffle: Add new SECONDEXPANSION test.
Allow build systems to choose an alternative location for make to
store its temporary files.
* NEWS: Announce the new environment variable.
* doc/make.texi (Temporary Files): Provide documentation.
* src/misc.c (get_tmpdir): Split into a new function. Compute the
temporary directory and store it in a static location.
* tests/scripts/features/jobserver: Add a test of MAKE_TMPDIR.
* src/load.c (load_file): Update .LOADED if setup returns non-0.
* tests/scripts/features/load: Change the return value based on an
environment variable. Ensure that returning -1 still adds to
.LOADED. Also add a test that verifies that make doesn't try to
rebuild the loaded file if -1 is returned.
* tests/scripts/features/load: Add function prototypes.
* tests/scripts/features/temp_stdin: Only set user execute bits.
Some configurations don't allow changing the group bits...??
If we detect a recursive variable reference when constructing the
environment for the shell function, return the original value from the
caller's environment. Other options such as failing, returning the
empty string, or returning the unexpanded make variable value have
been shown to not behave well in real-world environments. If the
variable doesn't exist in the caller's environment, return the empty
string.
Found by Sergei Trofimovich <slyich@gmail.com> when testing older
versions of autoconf.
* NEWS: Clarify this behavior.
* doc/make.texi (Shell Function): Ditto. Also add info about !=.
* src/expand.c (recursively_expand_for_file): Search the caller's
environment if we detect a recursive variable expansion.
* tests/scripts/functions/shell: Add tests for this behavior.
The fix for SV 10593 caused recursive expansion errors when exporting
a variable that contains a $(shell ...) invocation. If we see this
type of recursion, ignore it and expand to the empty string rather
than failing.
* src/variable.h (env_recursion): New global variable.
* src/variable.c (target_environment): If creating the environment
for a $(shell ...) function increment env_recursion. Remove the
check for expansion in a shell function context.
* src/expand.c (recursively_expand_for_file): Check for recursive
expansion in a $(shell ...) environment context and if present,
show the verbose message and return the empty string.
* tests/scripts/functions/shell: Add a test for this situation.
* maintMakefile: Remove the template headers as prerequisites.
* tests/scripts/features/jobserver: Only test fifo if enabled.
* tests/scripts/variables/INCLUDE_DIRS: On MacOS none of the default
directories exist so .INCLUDE_DIRS is empty by default.
* tests/scripts/features/se_explicit: Fail via exit. cp will show
different error messages on different systems.
* tests/scripts/features/se_implicit: Ditto.
* tests/scripts/features/se_statpat: Ditto.
Some POSIX systems do not allow locks to be taken on non-files, such
as pipes. This is a problem since very often make is invoked with
its stdout redirected to a pipe. Also, if stdout is redirected to a
file that already has a lock on it for some other reason (perhaps a
shared file such as /dev/null) it can cause a hang.
This means our previous method of locking stdout, although it had some
nice advantages, is not portable enough. Instead, use a temporary
file and take the lock on that. We pass the name of the file to child
make processes. On Windows we continue to use a shared mutex for
output sync.
Remove POSIX emulation functions like fcntl from Windows; instead
follow the lead of the jobserver and create an interface in os.h for
output sync, and move the OS-specific content to posixos.c and
w32os.c.
* NEWS: Add a note.
* src/makeint.h (ALL_SET): Check that all bits are set.
* src/os.h: Add bits for checking the state of stdin/stdout/stderr.
Add prototypes for OS-specific output sync methods.
* src/posixos.c (check_io_state): Determine the status of stdin,
stdout, stderr an return a suite of bits describing them.
(osync_enabled): If the global variable holding the FD of the lock
file (osync_handle) is valid return true.
(osync_setup): Create a temporary file and remember its name in a
global variable (osync_tmpfile), and set osync_handle.
(osync_get_mutex): If output sync is enabled, return the filename
of the lock file prefixed with "fnm:" to denote a filename.
(osync_parse_mutex): If the provided filename has the wrong format
disable output sync. Else open the lock file and set osync_handle.
(osync_clear): Close osync_handle. If we're the parent make, then
also unlink the temporary file.
(osync_acquire): Take a lock on the osync_handle descriptor.
(osync_release): Release the lock on the osync_handle descriptor.
(fd_set_append): Add APPEND mode to a file descriptor.
* src/w32/w32os.c: Perform the same actions as posixos.c, copying
the details from src/w32/compat/posixfcn.c. Use a mutex rather
than locking a temporary file.
* src/output.h: Remove all the OS-specific content.
* src/output.c: Remove all the OS-specific content.
(set_append_mode): Remove and replace with fd_set_append().
(sync_init): Remove and replace with check_io_state().
(acquire_semaphore): Remove and replace with osync_acquire().
(release_semaphore): Remove and replace with osync_release().
(setup_tmpfile): If the IO state is not obtained, get it. If stdout
and/or stderr are valid, set up a tempfile to capture them.
(output_init): Set io_state if not set already, and check it when
deciding whether to close stdout on exit.
* src/main.c (main): If we're syncing, set up the mutex using the
new osync_setup() / osync_parse_mutex() methods.
(prepare_mutex_handl_string): Replace with osync_parse_mutex().
(die): Call osync_clear().
* src/w32/compat/posixfcn.c: Remove implementations of fcntl(),
record_sync_mutex(), create_mutex(), and same_stream().
Reset the temp directory for every test to a local directory, then
after each test see if any new temp files were created and not
deleted: if they were then fail the test. Rather than delete the
temp files we leave them there and avoid reporting files that were
seen before, so the user can investigate them.
Rewrite the temp_stdin tests to rely on this built-in behavior
rather than implementing the checks directly.
* tests/test_driver.pl: Create a $TEMPDIR variable pointing to a
temporary directory outside the test temp directory.
(toplevel) Before starting any tests create a temp directory and set
the POSIX and Windows temp directory environment variables to use it.
(compare_output) Check the contents of the temp directory. If any
new files have appeared, fail the test.
* tests/scripts/features/temp_stdin: Remove check_tempfile() and
all users of it, plus setting of temp environment variables.
* README.git: Clarify that these methods are lightly tested.
* build_w32.bat: Don't support any config step: fail if not completed.
Move the config steps into bootstrap.bat. Don't print compile lines
by default and add a --verbose option to show them.
* bootstrap.bat: Ensure we have curl and sed before we do anything.
Pull the latest necessary files from gnulib. Create a convert.sed
script that can update the various template files, and update
Basic.mk, config.h.W32, and gmk-default.h.
* tests/run_make_tests.pl: Remove CRLF rather than using chop. If we
run perl in Git for Bash it seems to handle newlines differently.
* tests/scripts/features/temp_stdin: Remove the make copy and close
STDIN so we can delete the temp file on Windows.
* .gitignore: Ignore the convert.sed script.
Using anonymous pipes for jobserver support has some advantages:
for example there is nothing on disk that needs to be cleaned up.
However it has many obscure problems, related to the fact that in
order for it to work we need to ensure these resources are properly
passed through to child processes that want to use the jobserver.
At the same time we don't want to pass the pipe to process which
DON'T know about the jobserver.
Other processes can open file descriptors which we then think are
our jobserver, but aren't. And, we open the pipe file descriptors
in blocking mode which doesn't work for all users.
See issues such as SV 57178, SV 57242, and SV 62397
To avoid these issues, use named pipes (on systems where they are
available) instead of anonoymous pipes. This simplifies many things:
we never need to pass open file descriptors to our children; they
can open the jobserver named pipe. We don't need to worry about
recursive vs. non-recursive children. Users don't have to "pass
through" the resources if they are invoking sub-makes. Each child
can open its own file descriptor and set blocking as needed.
The downside is the named pipe exists on disk and so must be cleaned
up when the "top-level" make instance exits.
In order to allow make to continue to be used in build systems where
older versions of GNU make, or other tools that want to use the
jobserver, but don't understand named pipes, introduce a new option
--jobserver-style that allows the user to choose anonymous pipes.
* NEWS: Announce the change and the --jobserver-style option.
* doc/make.1: Add --jobserver-style documentation.
* doc/make.texi (Special Variables): Add missing items to .FEATURES.
(Options Summary): Add --jobserver-style.
(POSIX Jobserver): Named pipes, changes to --jobserver-auth, and the
--jobserver-style option.
(Windows Jobserver): Document --jobserver-style for Windows.
* configure.ac: Check for mkfifo.
* src/config.h-vms.template: Undefined HAVE_MKFIFO.
* src/config.h.W32.template: Ditto.
* src/main.c: Add jobserver-style as a new command line option.
(main): Add jobserver-fifo to .FEATURES if supported. Pass the style
option to jobserver_setup().
* src/os.h (jobserver_setup): Accept a style string option.
* src/posixos.c (enum js_type): Enumeration of the jobserver style.
(js_type): Which style we are currently using.
(fifo_name): The path to the named pipe (if in use).
(jobserver_setup): If no style is given, or "fifo" is given, set up a
named pipe: get a temporary file and use mkfifo() on it, then open it
for reading and writing. If something fails fall back to anonymous
pipes.
(jobserver_parse_auth): Parse jobserver-auth to determine the style.
If we are using a named pipe, open it. If we're using anonymous pipes
ensure they're valid as before.
(jobserver_get_invalid_auth): Don't invalidate the jobserver when
using named pipes.
(jobserver_clear): Clean up memory used for named pipes.
(jobserver_acquire_all): Unlink the named pipe when done.
* src/w32/w32os.c (jobserver_setup): Check the style argument.
* tests/scripts/features/jobserver: Use --jobserver-style to test
the anonymous pipe behavior, and also test named pipe/semaphore
behavior. Check invalid jobserver-style options.
* tests/scripts/functions/shell: Use --jobserver-style to test the
anonymous pipe behavior, and also test named pipe/semaphore
behavior.
Second-expand only the prerequisites of the targets being built.
Defer second-expanding the prerequisites of targets until we need
to decide if they should be built.
* NEWS: Mention the change in behavior.
* doc/make.texi (Secondary Expansion): Document the new behavior.
* src/filedef.h (struct file): Add flag snapped.
(expand_deps): Declare a function to second expand the
prerequisites of a target.
* src/file.c (rehash_file): Merge flag snapped.
(expand_deps): Remove qualifier static. Check flag snapped.
(snap_deps): Remove the loop which performed second expansion for all
targets.
* src/remake.c (update_file_1): Second expand the prerequisites of
the considered target.
* tests/scripts/features/se_explicit: Add tests.
* tests/scripts/features/se_implicit: Ditto.
* tests/scripts/features/se_statpat: Ditto.
Savannah issues such as SV 57242 and SV 62397 show how passing
references to closed file descriptors via the --jobserver-auth option
in MAKEFLAGS can lead to problematic outcomes.
When computing the child environment for a non-recursive shell, add
an extra option to MAKEFLAGS to disable the file descriptors for the
jobserver.
Unfortunately this doesn't modify the value of the make variable
MAKEFLAGS, it only modifies the value of the sub-shell environment
variable MAKEFLAGS. This can lead to confusion if the user is not
considering the distinction.
* src/makeint.h: Publish the jobserver-auth value. Add a global
definition of the name of the command line option.
* src/os.h (jobserver_get_invalid_auth): New function to return a
string invalidating the jobserver-auth option.
* src/w32/w32os.c (jobserver_get_invaid_auth): Implement it. On
Windows we use a semaphore so there's no need to invalidate.
* src/posixos.c (jobserver_parse_auth): If we parse the invalid
auth string, don't set up the jobserver.
(jobserver_get_invalid_auth): Return an invalid option.
* src/variable.h (target_environment): Specify if the target
environment is for a recursive shell or non-recursive shell.
* src/variable.c (target_environment): Move checking for MAKELEVEL
into the loop rather than doing it at the end.
Along with this, check for MAKEFLAGS and MFLAGS, and update them
based on whether we're invoking a recursive or non-recursive child,
and also on whether it's necessary to invalidate the jobserver.
* src/function.c (func_shell_base): Shell functions can never be
recursive to pass 0 to target_environment().
* src/job.c (start_job_command): Specify whether the child is
recursive when calling target_environment().
* src/main.c: Export jobserver_auth. sync_mutex doesn't need to
be exported. Use the global definition for the option name.
* tests/scripts/variables/MAKEFLAGS: Add tests for $MAKEFLAGS.
* src/main.c (main): Don't reset the jobserver if the number of
slots has not changed.
(define_makeflags): Add all normal flags even when ALL is not set.
* tests/scripts/functions/shell: Test invoking make in $(shell ...).
* tests/scripts/variables/MAKEFLAGS: Test the value of MAKEFLAGS in
$(shell ...).
Export all variables, including exported makefile variables, when
invoking a shell for the $(shell ...) function. If we detect a
recursive variable expansion, silently ignore that variable and do
not export it. We do print a debug message.
* NEWS: Announce the potential backward-incompatibility.
* doc/make.texi (Shell Function): Document the export behavior.
* src/main.c (main): Add "shell-export" to .FEATURES.
* src/job.h: New function to free struct childbase.
* src/job.c (free_childbase): Implement it; call from free_child.
* src/function.c (func_shell_base): Use target_environment() to
obtain the proper environment for the shell function.
Use free_childbase() to free memory.
(windows32_openpipe): Don't reset the environment: the caller
already provided a proper PATH variable in envp.
* src/variable.c (target_environment): If we detect a recursive
expansion and we're called from func_shell, ignore the variable.
(sync_Path_environment): Simplify and reduce memory allocation.
* tests/scripts/functions/shell: Add tests for this.
* src/file.c (rehash_file): Fix warning message.
(rehash_file): Fix comment to match the behavior.
* tests/scripts/features/se_explicit: Fix test.
* tests/scripts/features/mult_rules: Add a new test.
Commit f2771aa614 introduced a bug where some switches were left out
of MAKEFLAGS. Instead of resetting switches, get the same results by
filtering out duplicates.
* src/makeint.h: Remove reset_switches.
* src/main.c: (reset_switches): Remove reset_switches.
* (main): Remove call to reset_switches.
* (decode_switches): Filter out duplicate flags.
* src/variable.c: (set_special_var): Remove call to reset_switches.
* tests/scripts/variables/MAKEFLAGS: Verify that duplicate flags are
properly filtered out.
If GNUMAKEFLAGS was not present in the environment when we started,
don't add it.
* src/main.c (main): Don't mess with GNUMAKEFLAGS unless it exists.
* tests/scripts/variables/GNUMAKEFLAGS: Test this behavior.
Introduce non-deterministic ordering into goal and prerequisite
traversal to help tease out inconsistent failures that may happen
when running in parallel build mode.
Introduce second order into each dependency chain:
1. Existing order is syntactic order reachable via 'dep->next'
2. New order is shuffled order stored as 'dep->shuf' in each 'dep'
When updating goals and prerequisites and '--shuffle' is provided,
use the shuffled order to walk the graph. When automatic variable
are set always use the syntactic order of parameters.
* Makefile.am: Add new src/shuffle.c and src/shuffle.h file.
* build_w32.bat: Ditto.
* builddos.bat: Ditto.
* makefile.com: Ditto.
* po/POTFILES.in: Ditto.
* doc/make.texi: Add documentation for --shuffle.
* doc/make.1: Ditto.
* src/dep.h (DEP): Add the shuf pointer.
* src/filedef.h (struct file): Add was_shuffled flag.
* src/main.c: (shuffle_mode): Global flag for the shuffle mode.
(usage): Add the --shuffle option.
(switches): Ditto.
(main): Set shuffle_mode based on the command line parameter.
Reshuffle prerequisites if requested.
* src/remake.c (update_goal_chain): Walk the shuffled list if enabled.
(update_file_1): Ditto.
* src/shuffle.h: Provide an interface for shuffling prerequisites.
* src/shuffle.c: Implement option parsing and prerequisite shuffling.
* tests/scripts/options/shuffle: Test shuffle option and modes.
During second expansion of pattern rules only the first pattern in
each "group" was being substituted. E.g. in this makefile:
.SECONDEXPANSION:
all: hello.x
%.x: $$(wordlist 1, 99, %.1 %.%.2) ; $(info $@ from $^)
hello.1 hello.\%.2 \%.1 \%.\%.2: ;
the output would build "hello.1" and "%.%.2" because each function
is considered a single "word" and only the first pattern is replaced.
Fix the expansion so each whitespace-separated string is considered a
word and the first pattern is replaced, giving "hello.1" and
"hello.%.2".
* src/rule.c (snap_implicit_rules): Keep enough space to replace %
with $(*F) if necessary.
* src/implicit.c (pattern_search): During second expansion break each
get_next_word result into individual words and replace the first % in
each with $* or $(*F) as needed.
* tests/scripts/features/patternrules: Add tests for variations.
The hash function we use can yield different results on big- and
little-endian systems which makes test output different. Choose
names to avoid this.
* tests/scripts/features/patternrules: Choose portable target names.
* tests/scripts/features/se_explicit: Ditto.
* tests/scripts/features/se_implicit: Ditto.
If -R is given on the command line it implies -r as well; make this
same assumption if -R is provided in MAKEFLAGS set in the makefile.
* src/main.c (main): Check no_builtin_variables_flag after reading
makefiles.
* tests/scripts/options/dash-r: Add a test.
Previously we always used the file->stem value as our stem in
set_file_variables(); when that wasn't correct we had to temporarily
set that value while the function was called, then reset it afterward.
This led to issues (for example when we assumed the stem was a cached
string but it wasn't).
Avoid this by passing in the stem as an argument so that different
values can be provided.
Add tests to verify this.
* src/commands.c (set_file_variables): Take second parameter stem to
relieve the callers of set_file_variables() from setting/restoring
file->stem.
* src/commands.h (set_file_variables): Ditto.
(execute_file_commands): Pass file->stem to set_file_variables().
* src/file.c (expand_deps): Pass d->stem to set_file_variables() and
remove set and restore of file->stem.
* src/implicit.c (pattern_search): Pass stem to set_file_variables()
and remove set and restore of file->stem.
* tests/scripts/features/se_explicit: Add new tests.
* tests/scripts/features/se_implicit: Ditto.
* tests/scripts/features/se_statpat: Ditto.
* tests/scripts/variables/automatic: Ditto.
During normal pattern rule expansion only the first pattern (%) is
expanded; however during secondary expansion all patterns were
expanded. Modify secondary expansion to match the behavior of normal
expansion.
Implementation tweaked by Paul Smith <psmith@gnu.org>
* src/file.c (expand_deps): Don't use subst_expand() which replaces
all % with $*: instead replace only the first one, by hand.
Fix a memory leak where the dep structure was not always freed.
* tests/scripts/features/statipattrules: Use .RECIPEPREFIX not TAB.
Add a series of tests verifying that static pattern rules with and
without secondary expansion both return the same results.
Original patch from Dmitry Goncharov <dgoncharov@users.sf.net>
Fix the ar flags to match the default values in make. If the
U option is supported, add it to the end not the beginning,
since ARFLAGS now starts with a "-".
* tests/test_driver.pl: Add defaults for global variables.
(get_osname): Preserve $osname if it's set from $^O
* tests/scripts/features/archive: Set the default arflags.
If "U" is available add it to the end of the flags.
If the re-exec fails, be sure to remove a temp makefile that was
created to read from stdin.
* src/job.c (exec_command): Return on failure.
(child_execute_job): Call exit if exec_command returns.
* src/job.h (exec_command): Don't mark as NORETURN.
* src/main.c (main): Unlink stdin temporary file if re-exec fails.
* tests/run_make_tests.pl: Get value for ERR_nonexe_file/ERR_exe_dir.
* tests/scripts/features/temp_stdin: Test that temp file unlink works.
The -f, -file, and --makefile options were not properly handled when
re-exec'ing due to makefile updates. This problem, plus a patch and
tests, was reported by Dmitry Goncharov <dgoncharov@users.sf.net>.
While examining this I found another bug: after re-exec we forgot the
batch file was temporary and never deleted it.
I decided to fix all these problems at once using a different fix
than Dmitry's: I created a new internal-only command-line option,
--temp-stdin. When reconstructing the make options for a re-exec,
replace the -f/--file/--makefile option that reads from stdin with
--temp-stdin=<filename> so that the re-exec'd version of make knows
it's a temporary batch file and will delete it.
We no longer need to add the -o options because the re-exec'd make
knows this is a temporary makefile and treats it as such.
To simplify, replace the --file and --makefile options taking a
filename, with just -f<filename> on re-exec.
Some examples of the rewrite:
User command line Re-exec command line
----------------- --------------------
-f- --temp-stdin=<batch>
--file - --temp-stdin=<batch>
-f - --makefile a.mk --temp-stdin=<batch> -fa.mk
--file=a.mk -fa.mk
-fa.mk -fa.mk
-Rf a.mk -Rf a.mk
-Rf- -R --temp-stdin=<batch>
* src/main.c (stdin_offset): Remember the offset into the makefiles
list of the batch file read from stdin. Remove stdin_nm.
(struct command_switch): Create a new --temp-stdin option, which
also updates the makefiles list.
(main): Add the temporary filename to the string cache.
Move the tempfile handling after checking makefile arguments for "-"
so that files provided via --temp-stdin are also handled specially.
When rewriting re-exec options, we may need one more than we had
originally so create a new argv list. Walk through the original
list and convert it to the new list, following the above process.
(decode_switches): Set the stdin_offset flag if we see --temp-stdin.
* tests/scripts/options/dash-f: Add many more tests, provided by
Dmitry Goncharov <dgoncharov@users.sf.net>.
The fix for SV 60595 causes make to re-execute when the makefile is
read from stdin. E.g.
$ printf 'all:; $(info hello)' | make -sf -
Reported by Sergei Trofimovich <siarheit@google.com>
Test added by Paul Smith <psmith@gnu.org>
* src/main.c (main): Set the mtime of the stdin temporary file.
* tests/scripts/features/reinvoke: Add a test.
We already use undef makefile strings to mean "re-use the previous
makefile", so if the string is empty we'll assume it means "don't
use a makefile at all" (don't add -f).
* tests/run_make_tests.pl (run_make_test): If the makefile string
is empty, don't create a makefile or add -f.
* tests/scripts/features/include: Change empty makefile to "\n".
* tests/scripts/misc/close_stdout: Ditto.
* tests/scripts/options/dash-r: Ditto.
* src/main.c (main): Show the temp filename on error. Also on
Windows prefer TMP, then TEMP, and only lastly TMPDIR.
* tests/test_driver.pl: Add TMP, TEMP, and USERPROFILE to the list
of environment variables to preserve.
If a $(shell ...) invocation failed due to a command-not-found error,
make wrote the stdout of that shell to our stderr for some reason.
That seems very wrong.
If the command's stderr was not redirected then its output would have
already been written to its stderr, and if it was redirected then we
shouldn't take it upon ourselves to force it to go to stderr!
* src/function.c (func_shell_base): Append shell stdout even if the
shell command failed.
* tests/run_make_tests.pl: Determine the error generated for
command-not-found situations.
* tests/scripts/functions/shell: Verify that redirecting stderr to
stdout will behave properly if the command is not found.
If a prereq of a pattern is explicitly mentioned as a prereq of an
unrelated rule, it should not be considered an intermediate file.
* tests/scripts/features/double_colon: Add tests mentioning unrelated
explicit targets.
* tests/scripts/features/grouped_targets: Ditto.
* tests/scripts/features/implicit_search: Ditto.
* tests/scripts/features/patternrules: Ditto.
* tests/scripts/features/se_implicit: Ditto.
* tests/scripts/features/statipattrules: Ditto.
The fix for SV 60595 introduced a backward-incompatibility: rules that
purported to rebuild included files, but didn't actually do so, were
treated as errors whereas before they were ignored. This breaks a
common idiom in makefiles where an empty recipe is created for an
included makefile so make doesn't complain if it doesn't exist.
Unfortunately this means make cannot diagnose some types of errors.
Extra tests supplied by Dmitry Goncharov <dgoncharov@users.sf.net>.
* doc/make.texi (Including Other Makefiles): Clarify this behavior.
* src/main.c (main): Don't run the new check-for-errors behavior.
* tests/scripts/features/reinvoke: Reset tests of the "old" behavior
and add new tests for this situation.
In SV 43677 we forced targets to be secondary if we found an
intermediate file that was listed as a prerequisite of another
target. This overrides .INTERMEDIATE settings, so doesn't work.
Now that we have an is_explicit flag in targets, use that instead.
* src/implicit.c (pattern_search): Remove setting of secondary.
Preserve the value of the is_explicit flag when creating a new
file target, and consider it when setting the intermediate flag.
* tests/scripts/features/patternrules: Add a test w/out INTERMEDIATE
* tests/scripts/targets/INTERMEDIATE: Add a test with INTERMEDIATE
Target-specific variables used to define the target as "ought to
exist" so they could never be intermediate. Now they can be, so
merge the target-specific variables from the intermediate target
so they're not lost.
* src/implicit.c (pattern_search): Use merge_variable_set_lists
to merge target-specific variables.
* tests/scripts/features/implicit_search: Add tests of target-
specific variable assignments and implicit rules.
We don't need to parse strings into C integer values to compare them.
* src/function.c (parse_textint): Find boundaries of a numeric string.
(func_intcmp): Use parse_textint() to compare integers textually.
* tests/scripts/functions/intcmp: Test with extra-large numbers.
* src/function.c (parse_numeric): Check for empty value and error.
If we find ERANGE just print our own error, not strerror.
(func_word): Use a generic "not good" error message.
(func_wordlist): Ditto
(func_intcmp): Ditto
* tests/run_make_tests.pl: Remove code to find strerror(ERANGE)
* tests/scrips/functions/intcmp: Update the error message.
* tests/scrips/functions/word: Ditto.
* tests/scripts/features/errors: Match Windows-specific errors.
* tests/scripts/features/targetvars: Match Windows-specific errors.
* tests/scripts/misc/general4: Match Windows-specific errors.
Modify make functions that parse integer values to use long long
values instead of long: on Windows long is the same as int (4 bytes)
and we don't want behavior to differ between different platforms.
* bootstrap.conf: Change strtol module to strtoll module.
* src/function.c (parse_numeric): Use strtoll() and return long long.
(func_word): Use long long.
(func_wordlist): Use long long. Verify second argument is >= 0.
(func_intcmp): Use long long.
* src/config.ami.template: Don't define HAVE_STRTOLL.
* src/config-vms.template: Define HAVE_STRTOLL.
* src/config.W32.template: Define HAVE_STRTOLL.
* tests/run_make_tests.pl: Set $ERR_out_of_range to the proper string.
* tests/scripts/functions/word: Rework to use the new style and avoid
TAB characters. Verify trailing whitespace is ignored. Add a test
for a negative second argument to wordlist. Add tests for max signed
integer values. Use $ERR_out_of_range for the error string.
* tests/scripts/functions/intcmp: Add tests for min and max signed
integer values. Use $ERR_out_of_range for the error string.
Numbers can come from $(words ...), automatic variables such as
$(MAKELEVEL), from environment variables, or from shell output such as
through $(shell expr ...). The $(intcmp ...) function allows
conditional evaluation controlled by numerical variables.
* NEWS: Announce this feature.
* doc/make.texi (Functions for Conditionals): Document 'intcmp'.
* src/function.c (func_intcmp): Create the 'intcmp' built-in function.
* tests/scripts/functions/intcmp: Test the 'intcmp' built-in function.
strtol() is part of C89 and a fallback is provided by gnulib.
* src/function.c (func_word, func_wordlist): Change atoi to strtol.
* test/scripts/functions/word: Add out-of-range verification testing.
The traditional definition of "ought to exist" in the user's manual can
lead to make choosing a poor series of chained implicit rules. Modify
that rule so that we first attempt to resolve implicit rules by
considering targets which are only (a) targets or (b) explicit
prerequisites of the current target.
For backward-compatibility, if we don't find a solution using that
algorithm use the previous definition where "ought to exist" means
"mentioned as a prerequisite of any target".
* src/implicit.c (pattern_search): An extra argument controls whether
to perform the backward-compatible rule search or not. If a prereq is
a target, or if it's a prerequisite of THIS target, then choose it.
If we get to the end and haven't found a match, but we have found a
potential compatibility rule, then retry with compat rules enabled.
If we're searching with compat rules enabled and we find a file
mentioned as a prerequisite of ANY target, then use it.
* tests/features/implicit_search: Provide a large suite of tests for
different steps of implicit rule search.
* tests/scripts/options/dash-I: Use string comparison operator.
* tests/test_driver.pl (cmd2str): New method to convert a
command line to a string with proper quoting.
* tests/run_make_tests.pl (create_command): Call cmd2str() to format
the command. Add the filename/line number to the output.
POSIX Issue 8 will require a new assignment operator, :::=.
This operator behaves similarly to the BSD make := operator: the
right-hand side is expanded immediately, but then the value is
re-escaped (all '$' are converted to '$$') and the resulting variable
is considered a recursive variable: the value is re-expanded on use.
* src/variable.h (enum variable_flavor): Add f_expand flavor.
* src/variable.c (do_variable_definition): When defining f_expand,
post-process the result to re-escape '$' characters.
Remove default: to the compiler warns about un-handled enum values.
Set recursive values for both f_recursive and f_expand.
(parse_variable_definition): Rewrite this method.
The previous version was annoying to extend to ':::='.
(print_variable): Remove default: so the compiler warns us about
un-handled enum values.
* src/function.c (func_origin): Remove default: so the compiler warns
us about un-handled enum values.
* doc/make.texi: Add documentation for :::=.
* tests/scripts/variables/define: Add a test for define :::=.
* tests/scripts/variables/flavors: Add tests for :::=.
* tests/scripts/variables/negative: Add tests for :::=.
If multiple intermediate files are built together be sure all of them
are removed after make is finished. Continue to ensure that targets
that exist before make starts are not removed even if they appear to be
intermediate.
Add a number of tests to pattern rules to verify this behavior.
* src/filedef.h (struct file): Add a new is_explicit bitfield.
* src/file.c (rehash_file): Merge the is_explicit bit.
(enter_prereqs): Set is_explicit if the file is explicitly mentioned.
* src/implicit.c (pattern_search): Set intermediate on the file if it's
not explicit.
(record_files): Set is_explicit if a file is mentioned as a target.
* src/remake.c (update_file_1): Set secondary on files that already
exist so they won't be removed.
* tests/scripts/features/double_colon: Add a test for double-colon
pattern rules.
* tests/scripts/features/patternrules: Update KGO for tests where more
files are removed.
Add new tests to verify handling removal of intermediate files in the
context of grouped pattern targets.
If we find an included makefile but it's not readable, stop
immediately with an error rather than continuing to look in other
directories.
* src/read.c (eval_makefile): Only keep searching if the fopen error
is ENOENT, else stop and fail.
* tests/scripts/features/include: Add tests to verify this behavior.
Change the handling of included makefiles which are phony targets to
be similar to double-colon rules with no prerequisites: simply don't
build them at all during the remake a makefile phase.
Ensure that any included makefile which is needed but not built
results in an error.
Update the documentation to make this clear.
Add tests to verify this behavior.
* doc/make.texi (Remaking Makefiles): Clarify double-colon exception.
Document that phony targets are handled the same way.
(Phony Targets): Ditto.
* src/main.c (main): Check for phony targets when skipping goals.
Rather than throwing out skipped but failed goals keep them
separately then report them as errors.
* src/read.c (eval): Set the file location on included makefiles even
when there's no error.
* tests/scripts/features/include: Add tests for handling included
makefiles with both phony and double-colon rules to rebuild them.
When MAKEFLAGS is set in a makefile, reparse it immediately rather
than waiting until after all makefiles have been read and parsed.
This change doesn't actually fix the SV bug referenced because, even
though we do reparse MAKEFLAGS, we don't handle the -r or -R options
immediately. Doing this will require more effort.
* NEWS: Announce the change.
* src/makeint.h: Publish reset_switches() and decode_env_switches()
from main.c
* src/main.c (main): Don't call construct_include_path(); it will be
invoked decode_switches().
Preserve the old values of builtin_rules, builtin_variables, and
job_slots before we read makefiles since they can be changed now.
(reset_switches): Publish (remove static). Set the initial value of
the stringlist list to NULL.
(decode_switches): Call construct_include_path() after decoding.
(decode_env_switches): Publish (remove static).
(define_makeflags): Set the MAKEFLAGS variable for special handling.
* src/read.c (eval_makefile): Check for empty include_directories.
(construct_include_path): Clear any old value of .INCLUDE_DIRS before
appending new values. Free the previous include_directories.
* src/variable.c (lookup_special_var): When MAKEFLAGS is set, first
reset the switches then re-parse the variable.
* tests/run_make_tests.pl: Memo-ize some default variable values.
* tests/scripts/options/dash-r: Create tests for setting -r and -R.
* tests/scripts/variables/MAKEFLAGS: Test that resetting -I from
within the makefile takes effect immediately.
Previously if an included makefile was rebuilt as a prerequisite of
another included makefile which didn't need to be rebuilt, make would
not realize that it needed to re-exec itself.
Ensure that if any included makefile target is rebuilt we re-exec.
Also ensure that if an included makefile is not readable, and our rule
for rebuilding it doesn't actually change it, we will still fail.
* src/remake.c (update_goal_chain): If a goal's update was successful
then check its status, even if no actual commands were run because it
was already up to date.
(show_goal_error): Remove superfluous cast.
* src/main.c (main): If the makefile remake did nothing, check that we
were able to successfully include all the makefiles we care about; if
not fail. When generating error messages about included makefiles be
sure to show the filename/linenumber information.
* test/scripts/features/reinvoke: Add tests for this behavior.
* test/scripts/options/dash-k: Update error messages.
Accept a "-" directory value to the -I option to clear the set of
directories to be searched up to that point, including the default
directories.
* NEWS: Announce the change.
* doc/make.texi (Summary of Options): Add documentation.
* src/read.c (construct_include_path): Check for '-' and if found,
clear the list of directories to be searched.
* tests/scripts/options/dash-I: Add tests for -I-.
* tests/scripts/variables/INCLUDE_DIRS: Add tests for -I-.
Support a new special target, .NOTINTERMEDIATE. Any file or pattern
prerequisite of this target will never be considered intermediate.
This differs from .SECONDARY in that .SECONDARY files won't be deleted
but they will still not be built if they are missing.
.NOTINTERMEDIATE files are treated the same way as a target which is
explicitly mentioned in the makefile. This is mostly useful with
patterns; obviously mentioning a target explicitly here is enough in
and of itself to make something not intermediate.
Some adjustments made by psmith@gnu.org
* NEWS: Announce the new feature.
* doc/make.texi (Special Targets): Document .NOTINTERMEDIATE.
(Chained Rules): Describe how to use .NOTINTERMEDIATE.
* src/main.c (main): Add "notintermediate" to the .FEATURES variable.
* src/filedef.h (struct file): Add "notintermediate" flag.
* src/file.c (no_intermediates): Mark global .NOTINTERMEDIATE.
(snap_file): Support .NOTINTERMEDIATE special target. Throw an error
if the same target is marked both .NOTINTERMEDIATE and .SECONDARY or
.INTERMEDIATE.
(rehash_file): Merge intermediate, notintermediate, secondary flags.
(remove_intermediates): Check notintermediate flag before removing.
(print_file):
* src/implicit.c (pattern_search): Set notintermediate based on the
pattern.
* tests/scripts/targets/NOTINTERMEDIATE: Add a new test suite.
Set the $$< automatic variable as best we can during secondary
expansion of prerequisites.
* src/commands.c (set_file_variables): Don't break without setting
'less' if secondary expansion is enabled.
* tests/scripts/features/se_explicit: Test secondary expansion results.
* tests/scripts/features/se_implicit: Test secondary expansion results.
* src/dep.h (DEP): Remove extraneous semicolon.
* src/dir.c (find_directory) [W32]: Replace removed variable.
* tests/scripts/features/include: Allow the extra default makefile
searched for on Windows.
* tests/scripts/functions/file: Use $(info ...) to avoid "command
too long" errors on Windows.
Keep a count of bytes read rather than comparing pointers since the
variable_buffer might get reallocated.
Bug and patch by Ken Tossell <ken@tossell.net>
Regression tests by Dmitry Goncharov <dgoncharov@users.sf.net>
Tweaked by Paul Smith <psmith@gnu.org>
* src/function.c (func_file): Use bytes read rather than a pointer.
* tests/scripts/functions/file: Provide various tests for reading
empty files, files with/without newlines, and large files.
If a prereq of a pattern is an explicit target, it should not be
considered an intermediate file.
(Minor tweaks by Paul Smith <psmith@gnu.org>)
* src/dep.h (struct nameseq): Add is_explicit flag.
* src/implicit.c (struct patdeps): Ditto.
(pattern_search): Set the is_explicit flag appropriately for each
prerequisite, based on whether it contained a pattern or not.
Update the help output to note implicit vs. explicit prereqs.
* tests/scripts/features/double_colon: Add tests.
* tests/scripts/features/grouped_targets: Ditto.
* tests/scripts/features/patternrules: Ditto.
* tests/scripts/features/se_implicit: Ditto.
* tests/scripts/features/statipattrules: Ditto.
* src/read.c (parse_var_assignment): If called in a target-specific
variable context don't allow define/undefine as variable assignments.
* test/scripts/variables/define: Add a test.
* test/scripts/variables/undefine: Add a test.
Add a new function $(let ...) which allows lexically scoped variables.
* NEWS: Add information on this feature.
* doc/make.texi (Let Function): Document the 'let' function.
* src/function.c (func_let): Create the 'let' built-in function.
* tests/scripts/functions/let: Test the 'let' built-in function.
Add debug options to print recipes even if they would otherwise be
silent, and to print the reason that a target was considered out of
date.
Modify --trace to simply be a shorthand for --debug=print,why.
* NEWS: Announce changes.
* doc/make.texi (Summary of Options): Document the new options.
* doc/make.1: Ditto.
* src/debug.h: Add new flags DB_PRINT and DB_WHY.
* src/makeint.h: Remove the trace_flag variable.
* src/job.c (start_job_command): Check debug flags not trace_flag.
(new_job): Ditto.
* src/main.c (trace_flag): Make a static variable for switches.
(decode_debug_flags): Set DB_PRINT and DB_WHY if trace_flag is set.
* tests/scripts/variables/GNUMAKEFLAGS: Update known-good messages.
* tests/scripts/variables/MAKEFLAGS: Ditto.
If a -include file does not exist, then some subsequent operation
creates it, then allow it to be successfully included.
* src/read.c (eval_makefile): If our last mtime was NONEXISTENT then
reset it to 0 so we'll check it again.
* tests/scripts/features/include: Add a test for this behavior.
Each time we invoke a command it's possible that it will change the
filesystem in ways that were not described by the target. If that
happens but we have cached previous directory contents then we may
make decisions or report results based on obsolete information.
Keep a count of how many commands we've invoked, and remember the
current command count every time we load the contents of a directory.
If we request the directory and the current command count has changed
we know the cache is outdated so reload from scratch.
* NEWS: Announce the change.
* src/makeint.h (command_count): Create a global counter.
* src/main.c (command_count): Ditto.
* src/job.c (reap_children): Increment the counter on job completion.
* src/function.c (func_file): Increment if we write a file.
* src/dir.c (clear_directory_contents): Clear the current contents of
a cached directory.
(struct directory_contents): Remember the counter value.
(struct directory): Remember the counter value for non-existing dirs.
(find_directory): If we have a cached directory and the count hasn't
changed then return it. Else, clear the previous contents and re-read
from scratch.
* tests/scripts/features/dircache: Add tests of the directory cache.
Rewrite the environment variable algorithm to correctly inherit
export settings from parent variable sets. The new algorithm
for computing the table of environment variables is:
- Start with the most local variable set and proceed to global.
- If the variable already exists in the table and we don't know
its export status, update it with the current variable's status.
- If the variable is not in the table and it's not global, add it
regardless of its status so if it's unexported we remember that.
- If the variable is not in the table and is global, check its
export status and don't add it if we won't export it.
Then when generating the environment variables, check the export
status of each variable in case it was a target-specific variable
and we have determined it should not be exported.
Rework SHELL handling to check at the end whether we added it or
not and if we didn't, add the value from the environment.
* NEWS: Announce support for target-specific "unexport"."
* doc/make.texi (Target-specific): Document the support.
* src/variable.h (enum variable_export): Make into a global type.
* src/read.c (struct vmodifiers): Use enum variable_export rather
than individual booleans.
(parse_var_assignment): Parse the "unexport" keyword.
(eval): Remember the vmodifier value in the variable.
(record_target_var): Ditto.
* src/variable.c (should_export): Check if the variable should be
exported.
(target_environment): Implement the above algorithm.
* tests/scripts/features/export: Test export/unexport with variable
assignments on the same line.
* tests/scripts/features/targetvars: Add a comprehensive suite of
tests for different types of target-specific export / unexport.
* tests/scripts/variables/SHELL: Update the comment.
When checking for invalid environment variable names we searched the
entire name string instead of just the first LENGTH chars; this could
cause us to incorrectly decide the variable was not exportable.
Dmitry Goncharov <dgoncharov@users.sf.net> found this bug and
provided a test case and sample fix: I used the test but chose a
slightly different fix.
* src/variable.c (define_variable_in_set): check the variable name
not the input string.
* tests/scripts/features/targetvars: Ensure environment variable
values are exported.
GNU make must recognize some special targets as they are defined.
Because of the way targets are defined, we were not recognizing these
special targets until we were handling the NEXT statement. However
that's too late for some special targets such as .POSIX etc. which can
change the behavior of make during parsing.
Check for special targets earlier, as soon as we've finished parsing
the target introduction line (before we've even parsed the recipe).
* NEWS: Mention the change.
* src/read.c (check_specials): New function to look for special
targets. Move checks from eval() and record_files() to this new
function.
(eval): Call check_specials() after we've completed parsing the target
introduction line. Move default goal detection to check_specials().
(record_files): Move handling of .POSIX, .SECONDEXPANSION, and
.ONESHELL to check_specials().
* tests/scripts/misc/bs-nl: Remove workaround for late .POSIX issue.
* tests/scripts/targets/POSIX: Add a comment.
* src/function.c (func_filter_filterout): Allocate arrays to hold
pattern and word information rather than creating linked lists on
the stack.
* tests/scripts/functions/filter-out: Test large filters.
Ensure that makefiles are rebuilt in the order in which make first
considered them, and document this behavior in the manual.
* NEWS: Add a note about the new behavior
* doc/make.text (How make Processes a Makefile): Document it.
* main.c (main): Inverse the list of makefile goals.
* read.c (read_all_makefiles): Add default makefiles to the list at
the front in reverse order, the same way other makefiles are added.
* tests/scripts/features/include: Add tests to verify rebuild order.
The "[" link may be missing during OS boostrapping.
* build.sh: Convert "[ ... ]" to "test ..."
* maintMakefile: Ditto.
* scripts/copyright-update: Ditto
* tests/scripts/features/reinvoke: Ditto
* tests/scripts/features/targetvars: Ditto
When using execvp() if $PATH is not present in the environment
it will automatically search the system default PATH string. Emulate
this by passing the system default PATH to find_in_given_path() if
we don't find PATH in the environment.
* src/job.c (child_execute_job): Use confstr(_CS_PATH) if PATH is not
found.
Rather than having an %extraENV that is added to the default %ENV
and resetting %ENV _before_ each test, allow the test setup to
modify %ENV directly as needed then reset %ENV _after_ each test.
* tests/test_driver.pl: Remove unused %extraENV.
(resetENV): Don't add in %extraENV.
(_run_command): Reset after we run the command rather than before.
* tests/scripts/features/export: Convert %extraENV to %ENV
* tests/scripts/features/jobserver: Ditto
* tests/scripts/features/parallelism: Ditto
* tests/scripts/features/targetvars: Ditto
* tests/scripts/functions/eval: Ditto
* tests/scripts/functions/foreach: Ditto
* tests/scripts/functions/origin: Ditto
* tests/scripts/misc/general4: Ditto
* tests/scripts/options/dash-e: Ditto
* tests/scripts/targets/POSIX: Ditto
* tests/scripts/variables/GNUMAKEFLAGS: Ditto
* tests/scripts/variables/SHELL: Ditto
We want to process -C options as early as possible, before we might
write informational messages, so that Entering/Leaving messages have
the correct directory.
* src/main.c (main): Move code dealing with changing directories
before parsing of the jobserver auth flag.
* tests/scripts/features/jobserver: Test the order of enter/leave.
Previously if --no-print-directory was seen anywhere even once
(environment, command line, etc.) it would always take precedence
over any --print-directory option. Change this so that the last
seen option (which will be the command line, if present there) takes
precedence.
* NEWS: Mark this change in behavior.
* src/makeint.h (print_directory): A new variable to control printing.
* src/output.c (output_dump): Use the new variable.
(output_start): Ditto.
* src/main.c: Add a new variable print_directory. Use -1 for
print_directory_flag so we know of the option was seen or not. Add a
new default_print_directory_flag set to -1 to keep options from being
added.
(switches): Use flag_off for --no-print-directory, rather than a
separate inhibit_print_directory_flag.
(main): If print_directory_flag was set by the user, use that for
print_directory. If not, compute the print_directory value based on
-s, -C, and sub-makes as before.
* tests/scripts/variables/GNUMAKEFLAGS: -w is not added automatically
* tests/scripts/options/print-directory: Add tests for overriding
print-directory options.
POSIX says that suffix rules cannot have prerequisites, but after
making this change we observed a number of makefiles "in the wild"
that were relying on this behavior and failed.
For .POSIX: makefiles, obey POSIX. Otherwise preserve the old
behavior. However, generate a warning so users know this is a
problem. In a future version we will change all behavior to be
POSIX-conforming.
* NEWS: describe the change
* src/rule.c (convert_to_pattern): If posix_pedantic don't make a
pattern rule if prereqs exist. Otherwise show a warning.
* tests/scripts/features/suffixrules: Add tests for the new behavior
including .POSIX vs. non-.POSIX.
