On NetBSD 8.0 it still worked. Maybe gcov doesn't support .c files as
arguments anymore. Using the .gcda files works and is more reliable
anyway since it covers the inline functions in the headers as well.
The make code is not supposed to use constructs such as "if (strcmp(s1,
s2))" or "if (p && *p)", instead all boolean expressions have a form
that would be accepted by a C#, Go or Java compiler as well. This also
ensures that pre-C99 compilers generate equivalent code as C99
compilers, at least regarding boolean expressions.
This is only used interactively, not in the official builds, therefore
the additional dependency on Perl doesn't matter. The same result could
have been achieved in any other programming language, but probably not
as concisely.
Since unmade_cohorts is a signed number (the code contains an underflow
check), the result of the '%' operator could be negative and result in
"#-999999", which would overflow the buffer, truncating the last digit.
The bug had been introduced with dir.c 1.155 on 2020-10-02 22:20:25. In
that commit, openDirectories was replaced with a combination of a list
with a hash table, for more efficient lookup by name.
Upon cleanup, OpenDirs_Done is called, which in turn called
Dir_ClearPath. Dir_ClearPath takes full ownership of the given list and
empties it. This was no problem before since afterwards the list was
empty and calling Lst_Free just frees the remaining list pointer.
With OpenDirs, this list was combined with a hash table, and the hash
table contains the list nodes, assuming that the OpenDirs functions have
full ownership of both the list and the hash table. This assumption was
generally correct, except for the one moment during cleanup where full
ownership of the list was passed to Dir_ClearPath, while the hash table
still contained pointers to the (now freed) list nodes. This by itself
was not a problem since the hash table would be freed afterwards. But
as part of Dir_ClearPath, OpenDirs_Remove was called, which looked up
the freed directory by name and now found the freed list node, trying to
free it again. Boom.
Fixed by replacing the call to Dir_ClearPath with code that only frees
the directories, without giving up control over the list.
Even though it is commented out in this Makefile, it is still run by
atf, on purpose. This test is broken on FreeBSD, but on NetBSD it
should still be run.
Without removing the archive first, it could be left over from a
previous test failure and thus be newer than all its members, leading to
${.OODATE} being empty.
When it became disabled 9 years ago, there were 15 instances of UNCONST
in the code. These have been removed in the latest cleanup sessions.
Tested with GCC 5 from NetBSD 8 and GCC 10 from pkgsrc.
Before a modifier is applied to a variable, it is not yet parsed,
therefore it is only possible to log a rough estimate of the modifier.
But after applying it, the parsing position has advanced, and the full
modifier can be logged.
In addition, to fully understand how the modifiers work, it's not enough
to just know the variable names and values, there are also some flags
that influence how the modifiers behave. The most influential is
VARE_WANTRES.
Thanks to sjg for the extensive review and valuable feedback on the
first drafts.
NetBSD make is intended to be maximally portable, therefore it uses only
C89. This was not declared in the Makefile before.
There are still a few places in parse.c and metachar.c that use
end-of-line comments. These will be fixed in a follow-up commit.
The unit tests failed on Ubuntu, and by looking closely at the code, it
was trivial to see the out-of-bounds memory read.
Other modifiers may have the same problem and will be inspected later.
This makes it immediately obvious what happens after a modifier has been
applied, instead of having to translate single-character mnemonics or
booleans to their actual intention.
This also reduces the size of the binary since there are fewer jumps.
This prevents partial builds after changing a header. The declared
dependencies are more than strictly necessary, but that's still better
than having inconsistent partial builds because too few dependencies are
declared.
The list library is only used in make(1). Having it spread out over 28
files made it look more complex than it really is. In fact, it's just a
versatile generic data type like in hash.c.
Having all the list functions in a single file reduces the code size,
both by omitting the many RCS Ids and by inlining commonly used code.
We need to call meta_job_output() from meta_compat_parent()
Also, the need for extra fds for meta mode only applies when
using filemon_ktrace.c.
Since we hope to see more interfaces like that, test for
USE_FILEMON_DEV being not defined.
Reviewed by: riastradh
By default we use the ktrace back end, but the /dev/filemon back end
is available as a compile-time option, by setting USE_FILEMON=dev in
make. sjg raised concerns about ktrace performance and would like to
continue using /dev/filemon on FreeBSD (which has seen more
maintenance kernel-side) without forking make.
