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
enqueueing a thread to the run queue.
This mechanism is now used for the thread priority boost on semaphore
release. Also, those threads are no longer made real time threads, they
now get a temporary priority of B_FIRST_REAL_TIME_PRIORITY - 1.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16165 a95241bf-73f2-0310-859d-f6bbb57e9c96
didn't bring the app signature into its canonical lower-case form when
using it as path to the MIME DB file. Thus an application signature
MIME type didn't get any attributes, not even the path to the app's
executable, which led to all kinds of problems when finding or starting
applications by signature. Among them, Clock not finding its resources
(bug #74).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16163 a95241bf-73f2-0310-859d-f6bbb57e9c96
if it doesn't make a semantical difference. I can't believe I've written
that code.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16162 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
when the menu item under the mouse cursor has a sub-menu.
Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16160 a95241bf-73f2-0310-859d-f6bbb57e9c96
It now works fine on my P4 system as well :-)
But since the BIOS area is not identity mapped, it might work on fewer systems
than it does in BeOS R5.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16154 a95241bf-73f2-0310-859d-f6bbb57e9c96
on at least one of my systems, it only works on 3 of them...).
* Added APM safemode setting.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16152 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Menu windows now use the kMenuWindowFeel feel (but still need the B_AVOID_FOCUS
flag set, as that's currently independent from the feel).
* Minor cleanup in MenuWindow.cpp
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16150 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added --dry-run option for paranoid ones like me.
* Added support for making partition devices bootable under Linux (x86, BIOS).
When specifying respective values for HAIKU_IMAGE_{DIR,NAME} ("/dev" and
e.g. "sda7") a "jam haiku-image" will now replace the contents of your
favorite partition with a fresh Haiku installation. BE WARNED: This is nothing
for the faint of heart and I don't guarantee that my implementation is bug-free
and won't mangle the contents of your hard disk even if you use it correctly.
That being said, note that you'll probably have to run the "jam haiku-image"
as root, if you want to access the partition devices, which is a bit annoying.
Furthermore, if the build_haiku_image script fails for any reason, jam tends
to remove the image file, which in this case would be a partition device node.
The mknod command will be your friend in this case.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16148 a95241bf-73f2-0310-859d-f6bbb57e9c96
implement the B_GET_GEOMETRY command in a usable way. The bfs_shell is
able to play with partition devices under Linux, now.
* Fixed two warnings.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16146 a95241bf-73f2-0310-859d-f6bbb57e9c96
moment, but committing since it's too late to continue working, and at
least the window is resized correctly...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16142 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added a thread priority column to the thread list.
* Added "realtime" command that lists all real-time threads.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16141 a95241bf-73f2-0310-859d-f6bbb57e9c96
test it, though, as I couldn't reproduce the problem since then.
Fixed indentation in ps2_keyboard.c.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16137 a95241bf-73f2-0310-859d-f6bbb57e9c96
* We no longer try to dynamic_cast<>() a BLooper* into a BWindow* in
an unsafe context (i.e. without the looper or the looper list being
locked).
* We no longer try to quit windows that haven't been run yet (e.g.
bitmap's offscreen windows). Those windows conceptually still belong
to their creator. In the best case trying to lock such a window
simply failed (e.g. when the creator was another window that had been
told to quit earlier and deleted the window in question just not too
early). In worse cases we would dead-lock (when the owner has not
been told to quit (or refuses to do so)), cause "locker must be
locked" debugger calls or die painfully when accessing an already
deleted object.
BTW, I doubt, that the whole window quitting procedure is as it should
be. There's still a huge race condition: When a window is created after
we capture the window list at the beginning of window_quit_loop() that
completely escapes us.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16134 a95241bf-73f2-0310-859d-f6bbb57e9c96
"info threads" when one of the team's threads is waiting
uninterruptably on a semaphore).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16133 a95241bf-73f2-0310-859d-f6bbb57e9c96
or resource to be read did not exist and the method was told to
allocate a buffer, it would try to allocate the buffer with an
uninitialized size value. This basically concerned SetSupportedTypes()
and methods using that one (IsSupportedType(), Supports()).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16131 a95241bf-73f2-0310-859d-f6bbb57e9c96
different "force" levels now and updates the app file info attributes
for shared object files.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16123 a95241bf-73f2-0310-859d-f6bbb57e9c96
Axel Dörfler