[00:21:24] <CIA-111> mmlr * r42271 /haiku/trunk/src/servers/input/InputServer.cpp: (log message trimmed)
[00:21:24] <CIA-111> A blast from the past: r15048 from way back removed the wrong boolean parameter
[00:21:24] <CIA-111> causing the message to be sent with a timeout of "true" (getting converted to 1)
[00:21:24] <CIA-111> instead of the intended 0 meaning no/infinite timeout. This caused the message
[00:21:24] <CIA-111> sending to be aborted due to the timeout if it was blocking on a full port for
[00:21:24] <CIA-111> example. Since the return value is never checked noone noticed.
[00:21:24] <CIA-111> It's possible that this was the cause of some lost input messages (mouse,
[02:53:08] <CIA-111> mmlr * r42272 /haiku/trunk/src/system/kernel/port.cpp: (log message trimmed)
[02:53:08] <CIA-111> * Growing the port heap by adding new areas was broken in various ways. For one
[02:53:08] <CIA-111> the acquired quota in sTotalSpaceInUse wasn't released in all cases leading to
[02:53:08] <CIA-111> it eventually reaching the limit (after a _very_ long time though, so this is
[02:53:08] <CIA-111> more theoretical than anything else). The sAllocatingArea flag wasn't reset in
[02:53:09] <CIA-111> the case that an area was already added in the meantime, resulting in no
[02:53:10] <CIA-111> further growing being possible. Then there were race conditions between
[03:10:31] <CIA-111> mmlr * r42273 /haiku/trunk/src/system/kernel/port.cpp:
[03:10:31] <CIA-111> * Actually include an offset to write to in writev_port_etc(). It would
[03:10:31] <CIA-111> previously just always write over the beginning of the buffer for each vector.
[03:10:31] <CIA-111> Since the writev version isn't exposed to userland by means of a syscall and
[03:10:31] <CIA-111> kernel internally nobody used it, nobody noticed so far.
[03:10:31] <CIA-111> * Merge the two loops for user and kernel copy to remove the code duplication.
[03:50:34] <CIA-111> anevilyak * r42274 /haiku/trunk/src/apps/debugger/ (5 files in 2 dirs):
[03:50:34] <CIA-111> Introduce a flag in ExpressionEvaluationContext in order to distinguish between
[03:50:34] <CIA-111> cases where no object pointer is available vs the object pointer being present
[03:50:34] <CIA-111> but NULL, which would previously not be pushed onto the stack, leading to
[03:50:34] <CIA-111> expression evaluation failures.
[14:32:36] <CIA-111> mmlr * r42275 /haiku/trunk/headers/private/device/joystick_driver.h: Tiny comment cleanup.
[15:03:12] <CIA-111> mmlr * r42276 /haiku/trunk/src/system/kernel/vm/vm.cpp:
[15:03:12] <CIA-111> Fix range check. The previous check would produce an off by one error making the
[15:03:12] <CIA-111> last byte of an unmapped-but-still-there page non-readable (i.e. from B_NO_LOCK
[15:03:12] <CIA-111> areas), causing such reads to fail in KDL.
[20:09:44] <CIA-111> anevilyak * r42277 /haiku/trunk/build/jam/HaikuImage: Remove bash completion from the image for now, since as currently implemented it breaks the build.
[22:22:34] <CIA-111> mmlr * r42278 /haiku/trunk/src/system/kernel/heap.cpp:
[22:22:35] <CIA-111> Make the heap debug functions available when USE_SLAB_ALLOCATOR_FOR_MALLOC is
[22:22:35] <CIA-111> enabled as well. As this heap implementation is still used for the port heap
[22:22:35] <CIA-111> (as it handles B_NO_LOCK areas) those are still useful.





top