* Added syscalls _kern_set_cpu_enabled() and _kern_cpu_enabled().
* scheduler.c::sRunQueue::tail was not maintained at all; changed sRunQueue to
be a simple thread pointer instead of a struct thread_queue.
* Turns out we're monitoring CPU activity incorrectly when we've got more
than one CPU.
* Renamed the global CPU array from "cpu" to gCPU.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16186 a95241bf-73f2-0310-859d-f6bbb57e9c96
The test application lets run a thread at the highest priority that calls
yield all the time - the system stays responsible when it runs, so it seems
to work fine :)
Changed the malloc implementation to use _kern_thread_yield() instead of
snoozing.
We should think about making this call public, too.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16166 a95241bf-73f2-0310-859d-f6bbb57e9c96
idea for real-time threads. They could completely hog the CPU in this
case. Thanks to Marcus for investigating!
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16161 a95241bf-73f2-0310-859d-f6bbb57e9c96
* BEntry::Remove() now uses _kern_remove_dir() for directories.
* Added fd parameter to _kern_remove_dir().
* Fixed LibBeAdapter's _kern_unlink() to only work on files, and
added _kern_remove_dir() for directories.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16078 a95241bf-73f2-0310-859d-f6bbb57e9c96
The time base conversion factor is the 32 bit value
2^32 * 1000000 / time base frequency,
so the system time can be computed by
system time = time base * conversion factor / 2^32.
The expression in system_time() looks more complicated now, but is
actually much faster (factor 2.5 on my Mac mini). I'm positively
surprised, how good the assembly looks, that GCC 4 generates. There's
not that much potential for optimization by hand-coding the function.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15863 a95241bf-73f2-0310-859d-f6bbb57e9c96
now run under BeOS as well (as long as they don't use any functions that are not
available under R5).
The solution is a bit messy, but we have to live with it :-)
The runtime loader now patches the __gRuntimeLoader symbol in libroot.so to point
to its exported structure instead of passing it to the init functions as an
argument.
(Hax0red by axeld and bonefish on stippi's assimilated machine -- resistence is futile)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15848 a95241bf-73f2-0310-859d-f6bbb57e9c96
device in the Open Firmware implementation of boot loader and
pass its path to the kernel, where it's opened and used for
getting/setting the real time. The expensive atomic_*64() on PPC
32-bit make things a bit more complicated. Moreover, missing
64 bit multiplication and division instructions won't really
allow system_time() to be anywhere near as fast as on x86. :-/
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15837 a95241bf-73f2-0310-859d-f6bbb57e9c96
substructure now (that's the only member actually). The system time
offset is therefore accessed via architecture specific accessor
functions.
Note, that this commit breaks the PPC build. Since I want to rename at
least one file I've already changed, I can't avoid that.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15835 a95241bf-73f2-0310-859d-f6bbb57e9c96
I wouldn't be surprised, if more wrappers are still missing.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15709 a95241bf-73f2-0310-859d-f6bbb57e9c96
support unsbrk(), and that is actually called quite often in Hoard, resulting
in out of memory quite fast.
We don't shrink the heap (yet), but at least we can reuse the memory that
was freed by Hoard (it doesn't just remove bytes from the end, it will also
free up smaller parts in the middle).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15655 a95241bf-73f2-0310-859d-f6bbb57e9c96
suboptimal (if working at all) -- review is appreciated.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15486 a95241bf-73f2-0310-859d-f6bbb57e9c96
easier to define them for now, since they are used unconditionally e.g.
in the Tracker sources.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15485 a95241bf-73f2-0310-859d-f6bbb57e9c96
but it compiles at least. :-)
* Pulled the architecture specific part out of <posix/setjmp.h> into
<posix/arch/<arch>/arch_setjmp.h>.
* Moved setjmp_save_sigs.c from the x86 specific implementation into a
"generic" sibling directory, since it is reusable (and actually used
by the PPC implementation).
* Added generic/longjmp_return.c containing a function __longjmp_return,
which is invoked at the end of siglongjmp(), resetting the signal mask
and validating the return value. It is used by the PPC implementation,
and should also be used by the x86 implementation, but I'll leave that
to someone who's motivated enough to also test it. :-)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15479 a95241bf-73f2-0310-859d-f6bbb57e9c96
functions from glibc 2.3.2. Adding those one by one is extremely
boring and time-consuming work. We should probably just remove the
math stuff again, add a verbatim copy of the glibc directories, and
shuffle things around afterwards.
* Note that for x86 holds sizeof(long double) == 96. I removed the
generic 128 bit long double functions that were in use before and
added respective 96 bit versions to arch/x86 (or left them out, if
they weren't used). For PPC the System V ABI spec supplement specifies
sizeof(long double) == 128, but I had trouble configuring gcc to
actually use that setting (looks like there simply is no support for
TI mode for PPC 32). So we have to live with sizeof(long double) ==
sizeof(double) == 64 for the time being.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15477 a95241bf-73f2-0310-859d-f6bbb57e9c96
x86 the Haiku image still builds, so it is apparently not used. I
guess it would be best to abandon our posix_math.o completely and
use the glibc math implementations instead.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15446 a95241bf-73f2-0310-859d-f6bbb57e9c96
added a generic implementation using atomic_test_and_set().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15444 a95241bf-73f2-0310-859d-f6bbb57e9c96
glibc 2.3.2, which seems to be the version used in other places,
though I also found older files.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15402 a95241bf-73f2-0310-859d-f6bbb57e9c96
system happy for the time being. We need real setjmp()/longjmp()
implementations for PPC eventually, though.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15401 a95241bf-73f2-0310-859d-f6bbb57e9c96
Each component is built in the respective subdirectory now
and no longer in src/system/Jamfile.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15184 a95241bf-73f2-0310-859d-f6bbb57e9c96
we might miss some wrappers for some ieee754 functions (to check)
hope nothing is broken
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14968 a95241bf-73f2-0310-859d-f6bbb57e9c96
they could not be parsed again by the kernel, and thus, safemode was not working
anymore (as it was queried by the SMP stuff).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14960 a95241bf-73f2-0310-859d-f6bbb57e9c96
wonder why things worked nevertheless with the old gcc). We now also use
realloc() instead of malloc()+memcpy(), and check add_variable()'s
return code.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14957 a95241bf-73f2-0310-859d-f6bbb57e9c96
jump to sigsetjmp(). I'm a bit irritated that the previous version didn't
run as expected, but it now does.
This needs some more investigation.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14901 a95241bf-73f2-0310-859d-f6bbb57e9c96
Fixed implementation of ualarm() to support larger values.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14890 a95241bf-73f2-0310-859d-f6bbb57e9c96
categories:
* Missing includes (like <stdlib.h> and <string.h>).
* Linking against $(TARGET_LIBSTDC++) instead of libstdc++.r4.so.
* Local variables shadowing parameters.
* Default parameters in function definitions (as opposed to function
declarations).
* All C++ stuff (nothrow, map, set, vector, min, max,...) must be imported
explicitly from the std:: namespace now.
* "new (sometype)[...]" must read "new sometype[...]", even if sometype is
something like "const char *".
* __FUNCTION__ is no longer a string literal (but a string expression), i.e.
'printf(__FUNCTION__ ": ...\n")' is invalid code.
* A type cast results in a non-lvalue. E.g. "(char *)buffer += bytes"
is an invalid expression.
* "friend class SomeClass" only works when SomeClass is known before.
Otherwise the an inner class with that name is considered as friend.
gcc 4 is much pickier about scopes.
* gcc 4 is generally stricter with respect to type conversions in C.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14878 a95241bf-73f2-0310-859d-f6bbb57e9c96
Library names are now mapped for all targets but "host" (not only for
"haiku") -- added one more level of indirection to achieve that.
(TARGET_LIBRARY_NAME_MAP -> *_LIBRARY_NAME_MAP_*).
* Renamed build/HaikuBuildCompatibility.h to BeOSBuildCompatibility.h
(auto-included when compiling something that uses the Be API for platform
"host" on anon-BeOS platform), and introduced build/HaikuBuildCompatibility.h,
which can be included when compiling something that can be built for both,
Haiku and BeOS compatible platforms.
* Introduced libhaikucompat.a, a library that adds a few functions existing
under Haiku, but not under BeOS.
* New rule AddSubDirSupportedPlatforms.
* Renamed libopenbeos.so to libbe_haiku.so.
* Introduced new target platform "libbe_test", which is basically equivalent
to a BeOS compatible host platform target, with the exception, that instead
of the host platform's libbe.so a special build of Haiku's libbe.so
(libbe_haiku.so (formerly known as libopenbeos.so)) is used. Furthermore
Haiku's public app, interface, storage, and support kit headers are used
when compiling. This replaces the less nice way in which the test app server
and applications for this test environment were built.
When building for platform "libbe_test", the library name "be" is
autotranslated to "libbe_haiku.so". Thus most applications don't need
special fiddling when them building them for the app server test environment;
usually an "AddSubDirSupportedPlatforms libbe_test ;" will suffice.
* Reduced the dependencies of <syscalls.h> and fixed problems caused by this
(e.g. source files not including the needed headers directly).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@14749 a95241bf-73f2-0310-859d-f6bbb57e9c96
