This adds a new header <sys/internals.h>, which provides access to LibC internals.
This is in the interest of type-checking LibC itself, as well as enabling less-hacky
access for uses like LinkDemo.
And, of course, this progresses LibC towards building cleanly with -Wmissing-declarations.
I'm not sure how else to handle this. Curiously, I can't find the string '_start'
anywhere else in the project. Could it be that we haven't NIH'd this yet?
And that we actually rely on magic from the compiler to call _start for us?
This enables a nice warning in case a function becomes dead code.
For example with the unused function malloc_good_size() :^)
I found these places by using -Wmissing-declarations.
The Kernel still shows these issues, which I think are false-positives,
but don't want to touch:
- Libraries/LibC/crt0.cpp:41:5: int _start(int, char**, char**)
Not sure how to handle this.
- Libraries/LibC/cxxabi.cpp:48:5: int __cxa_atexit(AtExitFunction, void*, void*)
- Libraries/LibC/cxxabi.cpp:58:6: void __cxa_finalize(void*)
Not sure how to tell the compiler that the compiler is already using them.
- Libraries/LibC/libcinit.cpp:36:6: void __libc_init()
- Libraries/LibC/libcinit.cpp:55:19: void __stack_chk_fail()
- Libraries/LibC/malloc.cpp:430:6: void __malloc_init()
- Libraries/LibC/stdio.cpp:562:6: void __stdio_init()
These are ninja-imported by other LibC functions.
Maybe we should have some kind of "internals.h" header.
Sometimes people write strange things like "assert(x), something();"
and this will not work if "assert(x)" expands to "".
So make it expand to ((void)0) instead.
- Remove goofy _r suffix from syscall names.
- Don't take a signed buffer size.
- Use Userspace<T>.
- Make TTY::tty_name() return a String instead of a StringView.
This syscall allows a parent process to disown a child process, setting
its parent PID to 0.
Unparented processes are automatically reaped by the kernel upon exit,
and no sys$waitid() is required. This will make it much nicer to do
spawn-and-forget which is common in the GUI environment.
This commit adds an implementation of memmem, using the Bitap text
search algorithm for needles smaller than 32 bytes, and a naive loop
search for longer needles.
"0" was interpreted as a base-8 prefix, and the parse pointer was then
unconditionally advanced, causing us to consume zero characters.
This unbreaks the git port. :^)
(We should really have tests for LibC..)
This being inline somehow broke the binutils autoconf scripts. It used
to work, so I suspect that some other change to LibC has caused those
autoconf scripts to go down a new path.
Regardless, this seems perfectly sensible.
There's no great advantage to using MMX instructions here on modern
processors, since REP MOVSB/STOSB are optimized in microcode anyway
and tend to run much faster than MMX/SSE/AVX variants.
This also makes it much easier to implement high-level emulation of
memcpy/memset in UserspaceEmulator once we get there. :^)
While profiling I noticed that gettid() was hitting gs register
twice, once for the initial fetch of s_cache_tid out of TLS for
the initialization check, and then again when we return the actual
value.
Optimize the implementation to cache the value so we avoid the
double fetch during the 99% case where it's already set. With
this change gettid() goes from being the 3rd most sampled function
in test-js, to pretty much disappearing into ~20th place.
Additionally add the same optimization to getpid().
This works the same as gettid(). No sense in making a syscall to the
kernel every time you ask for the PID since it won't change.
Just like gettid(), the cache is invalidated on fork().
For now, only the non-standard _SC_NPROCESSORS_CONF and
_SC_NPROCESSORS_ONLN are implemented.
Use them to make ninja pick a better default -j value.
While here, make the ninja package script not fail if
no other port has been built yet.
These new syscalls allow you to send and receive file descriptors over
a local domain socket. This will enable various privilege separation
techniques and other good stuff. :^)
ppoll() is similar() to poll(), but it takes its timeout
as timespec instead of as int, and it takes an additional
sigmask parameter.
Change the sys$poll parameters to match ppoll() and implement
poll() in terms of ppoll().
pselect() is similar() to select(), but it takes its timeout
as timespec instead of as timeval, and it takes an additional
sigmask parameter.
Change the sys$select parameters to match pselect() and implement
select() in terms of pselect().
And rewrite the timeval functions as inline functions.
Also add the non-standard but fairly common and useful
TIMEVAL_TO_TIMESPEC / TIMESPEC_TO_TIMEVAL functions.
posix_spawn() tries to present semantics as if no fork() is happening
behind the scenes, so running arbitrary atexit handlers of the parent
in the child seems like the wrong thing to do.
