string from a silent implicit non-global substitution to a non-silent
explicit fatal error. Archored substitutions are those containing '^'
or '$'.
The problem with changing the substitution to prevent an infinite
number of matches is that it doesn't provide the necessary feedback
to the user that there's a bug in the/a makefile. Reporting the bug
without making the condition fatal makes the feedback mostly useless
due to the way that make fails to prefix the error with program name,
makefile file name and line number information.
Note that global substitutions of the empty string anchored with '^'
(start of string) or '$' (end of string) do not cause an infinite
number of matches and are therefore not reported and hence are non-
fatal.
Suggested by: bde
Tested with: buildworld
global substitution. In general it's a makefile bug to globally
substitute the empty string, but it's a bug in make(1) if a bug
in the makefile yields an infinite running time of make(1).
Not objected to by: arch@
to Fatal errors, because the logic that we use to try to continue is far
too broken, and makes things look and act weird, because we end up pointing
past the end of a buffer boundry into freed memory in the caller, as we
don't come close to setting the lengthPtr to a sane value.
Reviewed by: make@
(This only changes failure cases which would have died horrid deaths to
explicit clean death failure cases.)
Don't gratuitously pipe thru a cat(1) if NODOCCOMPRESS.
Only create _stamp.extra when necessary.
Get rid of SOELIMPP and OBJS.
Use Groff version of soelim(1); we need its -I option
for the following to work.
Don't needlessly chdir to SRCDIR. Only a few documents
need CD_HACK, and those that need it either use refer(1)
or .PSPIC macro which internally uses the .psbb call.
to var_modify.c, for readability. constify some low hanging fruit (string
manipulation functions) and the upper layers appropriately. No longer use
the private strstr(3) implementation, while changing string code.
Tested by: lots of successful make buildworld.
data that will be modified. And do the appropriate thing now and free the
v->name buffer along with other relinquished memory.
XXX There is duplication here of destroying a Var, which is probably bogus,
and probably missed in a few places.
might help on the systems it could possibly be used as a bandaid for. In
fact, the only thing it's useful for is instrumenting free(3) calls, and in
that capacity, it's better served as a local patch, than a public wrapper.
compatability-geared util.c. These are things like message printers
and the PrintAddr function for traversing lists. Other general-purpose
utilities inside make(1) can go here, in time.
documentation already adequatedly existed in the description in most
cases. Where it did not, it was added. If no documentation existed
beforehand, then none was added. Some unused dummies for use in the
traversal functions were marked as __unused during the conversion.
Occasionally, local style fixes were applied to lines already being
modified or influenced.
Now make(1) should always build with WARNS=3.
instead of polling for them.
Unfortunately we cannot enable it yet because it panics the kernel
somewhere in kqueue.
Submitted by: Stefan Farfeleder <e0026813@stud3.tuwien.ac.at>
there from being more than 500 processes forked by make(1), to prevent a
forkbomb from happening, in a dumb and mechanical way.
PR: alane
Submitted by: bin/42772
MFC after: 2 weeks
renaming variables to not shadow libc functions or greater scope locals. Kinda
makes one wonder if the extern ones weren't meant in some of these places :)
The only thing I'd still like to do WRT this is possibly combine rstat and
status in compat.c -- that should be fine, as I do not think the codepaths
will want both around at once.
Sponsored by: Bright Path Solutions
in compat.c which doesn't even have preprocessor-conditional-hidden support
code, and add a debugging statement where we might end up with a nil list
somehow, but where I doubt it.
First confirmed userland kill for Flexelint.
Sponsored by: Bright Path Solutions
variable length arguments to a macro. Bump version as this makes DEBUG
statements *always* go to stderr rather than sometimes stdout. There are
a few stragglers, which I will take care of as soon as I can. Mostly these
relate to the need-for-death-of some of the remote job code.
Nearby stylistic nits and XXX added/fixed where appropriate.
which fails the make tests (doesn't understand ${notdef:U}) and therefore
fails on __FBSDID in usr.bin/make/*. -DBOOTSTRAPPING is no help here since
this is before we are using the new share/mk/* files, and it would conflict
with the builtin -DBOOTSTRAPPING support later.. so use a different flag.
this particular GNU flag. It changes into the given directory for the
operation in question. This just goes into said directory at the time of
parsing the argument for getopt(3).
Submitted by: Rachel Hestilow <rachel@jerkcity.com>
we use in sublte ways with relative paths. Until they can be resolved,
back out these changes and put a big comment about why using realpath is
busted.
Approved by: jmallett
MFC After: 100 millifortnights
seen (somewhat) in NetBSD. This catches a few extra recursion cases that
could be hidden by expanding a NIL variable causing an existing variable to
be returned (which caused infinite looping and climbing memory usage in at
least one case).
Obtained from: NetBSD (in principle)
of ReadMakefile as __unused, it's there because this function is used by
the abstracted list interface which normally deals with item handlers which
take two arguments. Add a missing static prototype.
if a given make(1) is feature-compatible with a set of makefiles.
When merged, this will be used to replace the ugly upgrade_checks
hacks in src/Makefile.
Version has the RYYYYMMDDX format, where R is from RELENG_<R> and
X allows for 10 distinguishable changes per day.
Discussed with: bde
it, which means that relative paths will be expanded to absolute paths, and
filenames without a path will end up with their absolute path included as
well. This aids tremendously in debugging a build using our make(1) with
multiple Makefile's, such as when there is a syntax error in a file in a
sub-directory as per <bsd.subdir.mk>. Normally we'd end up with just
"Makefile" known about the Makefile in question, which means that an error
would be useless for someone trying to debug their build system, now we
end up with a complete real pathname for the Makefile.
So mostly this is useful in a debugging context, but possibly others too
(I haven't thought of them yet, but they probably are more useful if you
make Dir_FindFile use realpath(3), but that's another story).
Reviewed by: -current
MFC after: 2 weeks
With rev 1.7 one cannot build src/bin/sh -- because make fails to create
the buildtools before trying to use them. Actually it does compile the
buildtools into .o's before trying to use them, but not all the way into
binaries.
make installed) in gnu/usr.bin/groff/src/preproc/eqn (which, being a
build tool itself, is built with the original make during buildworld).
The problem seems to be that in str_concat(), the string is not
terminated when the length of the second string is 0.
This apparently can happen during null suffix rule processing.
Submitted by: tmm
test if failsafe memory allocation fails, it can't. perror -> warn. Use
failsafe memory allocation provided.
Use .Pa and .Ar. Uppercase (first letter) programname after dot.
has been determined similar to C. That is, one expects a construction
like,
.if defined(TEST) && (${TEST:L} == "test")
Never to generate an error since the second expression should never be
evaluated when TEST is undefined.
However, this was not the case. The above fails with the current
make(1) if TEST is undefined. This patch fixes the above and many
similar cases.
PR: bin/34032
Submitted by: Alan Eldridge <alane@geeksrus.net>
MFC after: 1 week
This should not affect our build process, as
find /usr/src -name Makefile | xargs grep '@[DF]'
has no matches (other than FreeBSD.org email addresses :-)
PR: bin/24377
Submitted by: Mark Valentine <mark@thuvia.demon.co.uk>
Reviewed by: Matthew Emmerton <matt@gsicomp.on.ca>
MFC after: 4 weeks
remove the concept of a 'maintainer' of our make. there really isn't a
need for any one committer to hold an exclusive lock or serve as a filter
for this code.
src/sys/modules/if_ef and possibly other things. I tested the build with
a make based on rev. 1.26, and it worked fine. Since I'm not particularly
inclined to figure out what's going on with this, it's probably prudent
just to back it out for now.
Found by: jkh
Suggested by: jhay
.endif statements but can't be placed in .elif. Basically, the problem
was that ParseSkipLine() didn't handle comments the same way that
ParseReadLine() did, and thus you had errors with comments that are on a
conditional line (i.e. "^.") rather than a non-conditional line.
MFC candidate for 4.3-STABLE and 3.5-STABLE.
PR: 25627
Bug found by: jhs
Fix submitted by: Seth Kingsley <sethk@osd.bsdi.com> (thanks!!)
list was cleared.
Rules with null suffixes would not be rebuilt when the suffixes were
added again.
Adding null suffix rules would fail when a rule for the same source was
declared before the suffix list was cleared.
PR: 23328, 24102
Reviewed by: will
Approved by: rwatson
through the use of a new build directive, MACHINE_CPU, which contains a
list of the CPU generations/features for which optimizations are desired.
This feature will be extended to cover the ports tree in the future.
Currently OpenSSL provides optimizations for i386, i586 and i686-class
CPUs. Currently it has not been tested on an i386 or i486.
Teach make(1) to provide sensible defaults for MACHINE_CPU if it is not
defined (namely, the lowest common denominator CPU we support for each
architecture). Currently this is i386 for the i386 architecture and ev4
for the alpha. sys.mk also sets the variable as a last resort for
consistency with MACHINE_ARCH and bootstrapping from very old versions of
make.
Benchmarks show a significant speed increase even in the i386 case, with
additional improvements for i586 and i686 systems. For maximum performance
define MACHINE_CPU=i686 i586 i386 in /etc/make.conf.
Based on a patch submitted by: Mike Silbersack <silby@silby.com>
Reviewed by: current
was in. This shall be MFC'd in about three days (probably not a good idea
to MFC the stylistic changes though - see below).
PR: 19978
Submitted by: Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
Patch by: roam (slightly modified by me to use NULL not NIL)
it to make.h so both dir.c and util.c can use it, although bde didn't
particularly like this part of the idea, IMO it's cleaner than it was.
Submitted by: bde
actually NOT '.' and '..'. Apparently this isn't the case when accessing
a directory via XFS over NFS on SGI systems. Since I don't have access to
an environment like that, this will sit out in -current for at least six
weeks. However, the patch proposed by the submitter seems acceptable, so
I've decided to commit it to the tree, in the hope that it will solve some
problems without bringing up others.
PR: 23300
Submitted by: Jim Pirzyk <Jim.Pirzyk@disney.com>
all-lower or all-upper case characters, respectively. These were added
to further reduce differences between NetBSD/OpenBSD and FreeBSD make(1)
to propagate OpenPackages.
PR: 19959
Submitted by: Gaspar Chilingarav <nm@web.am>
but allows for regex. I removed NO_REGEX since no one could give a reason
to have it, and since we don't use make in bootstrap tools, it's not needed.
This is mostly added to synch up with NetBSD/OpenBSD so as to eliminate
roadblocks in the OpenPackages project (see http://www.openpackages.org/).
It's also quite useful, and costs us only about 3 kilobytes of space.
PR: 21605
Submitted by: Hubert Feyrer <hubertf@NetBSD.org>
Reviewed by: silence on -arch
Obtained from: NetBSD
to override @-prefixed commands in Makefiles. It is especially useful for
debugging ports and/or complex Makefiles in such a manner that is basically
a last resort, but is quite effective if the output is well-handled.
I'll update the manpage after dinner. ;-)
Better patch submitted by: steve
Reviewed by: phk, steve, chuckr, obrien,
Lyndon Nerenberg <lyndon@orthanc.ab.ca>
make(1) uses sysctlbyname() to find out if it is running on a PC98
machine. This check has been added on 1998/9/9. The MIB variable was
added on 1998/8/31. At that time __FreeBSD_version was 300003. So,
only perform the check *if* __FreeBSD_version is defined and if it's
greater than 300003.
For the record: sysctlbyname was added on 1997/5/30 so this change
automaticly handles that...
It appears that make fails to read the global symbol table of the archive file,
making it think that the library needs to be rebuilt.
fix supplied in PR:
PR: bin/14167
Submitted by: Sebastian Lederer <lederer@bonn-online.com>
Reviewed by: gallatin@FreeBSD.ORG
of VPATH to be a little sexier. I'd say it looks a bit like Claudia
Schiffer now, which I consider a definate improvement. We're also a little
more subtle about saying that VPATH is for backward(s) compatibility.
Submitted-by: chris, bde, and myself
[although none of their descriptions were as sexy
M(|is|r)s Schiffer, of course].
for BSD less than 4.4 where sys/cdefs.h DNE but can be simulated with
param.h. Unfortunately, the additional include did little to actually
improve portability since we've lost portability in several other chunks
of make code (err(), sysctlbyname(), probable more).
Partly submitted by: bde
"Make no longer properly parses archive files, so using archive members
as make targets, as described in PSD:12 section 4.2, does not work."
Really should be merged into 3.3 before we ship if we can..
it's a trivial fix.
PR: bin/13039, bin/10274
Submitted by: Mikko Tyolajarvi mikko@dynas.se
that -E only operates for a specified variable. Useful since the -e option
will often pull-in many unwanted variable overrides (esp. in a make world
situation). Uses include overriding BINOWN (which cannot be done by normal
methods or through abuses of MAKEFLAGS) or likely for ports to honour CFLAGS
(provided they're running on a system whose make(1) has this option).
support. I've been building world with these changes for months w/o
ill effect. I've also managed to build the cross tool chain for MIPS
with these patches.
Please note that the extent to which these patches work is largely
dictated by how well our tool chains support the cross compilation.
Building alpha binaries on i386 doesn't work. Supposedly building
i386 binaries on alpha does work, but I've not verified it with these
patches, however.
compiled in default in case it isn't defined. This is needed to make
cross compilation work in some edge cases. It also makes cross
compiling on FreeBSD other BSD's easier as well.
Obtained from: NetBSD, OpenBSD (predates the split)
This case did not need to be tested when RANLIBMAG was defined
(as when in an `aout' environment) because Arch_StatMember() treated
the two cases of the library not being present and a member of
the library not being present the same way, forcing a rebuild
of the library. Since in the ELF environment we don't look inside
archive libraries we now need to check if the archive library is
present in order to determine its `out-of-date'-ness.
(I hope I've been able to meet the Oct 15th freeze).
Reported-by: Steve Price (and a few others whom I've forgotten, sorry)
option lets make look inside archive libraries when determining if a
library is out-of-date. I don't think that make should look inside
files, so this fix effectively tells it not to. If the decision is
made (by someone else) that make _should_ be doing this, it will need
to be taught how to recognise elf archive libraries.
Problem noticed by: Joseph Koshy <jkoshy@freebsd.org>
other, less advanced architecutres. This should minorly help porting
efforts of FreeBSD. I've done several make worlds since this came up
with this change, as well as debugging several interesting nits with
-V (which is the only thing this change will affect really).
says it pretends to make.
This bug was apparently harmless except for normal cases involving
.ORDER statements when it made debugging of -jN using -n very
confusing. E.g., for:
.ORDER: beforedepend .depend
depend: beforedepend .depend
where beforedepend depends on something so that it is not initially
up to date, `make [-n] -j2 depend' causes `make' to wait for itself
to make beforedepend. This works fine without -n. The job to make
beforedepend has normally been started, and beforedepend is marked
as made when the job completes. However, with -n, the pseudo-job
for making beforedepend has normally completed, and in any case
there was no chance of beforedepend being marked as made. `make'
actually exited almost immediately with status 0 instead of waiting
forever.
to free the suffix. I think, it is a very strange idea. (Or, maybe, it is a
POSIX requirement?) And it is done incorrectly. Apparently, it even don't
update the list of known suffixes (but it is an other bug).
PR: 4254, 4692, 4783
Reviewed by: phk
Submitted by: Dmitrij Tejblum <dima@tejblum.dnttm.rssi.ru>
