* vm_copy_on_write_area() now does no longer overwrite the ref_count, but keeps
it's reference to the cache until it has unlocked it.
* It now also locks its reference from the start, preventing any other thread
to interfere.
* vm_cache_remove_consumer() now detects if it has to remove a foreign busy
page itself in order to preserve a mapped page.
* vm_soft_fault() now keeps a reference to the cache that owns the page to be
mapped until it has actually mapped it.
* vm_unmap_pages() removed the mappings of all pages of the area instead of only
those that are within the requested range.
* Kept (disabled) debug output for convenience.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21482 a95241bf-73f2-0310-859d-f6bbb57e9c96
block everytime it's called - that caused negative reference counts in the block
cache, causing all sorts of problems once they were flushed.
* Changed order of includes in system_dependencies.h to what I prefer: descending from
private to public (resp. from most specific to most generic) headers.
* Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21480 a95241bf-73f2-0310-859d-f6bbb57e9c96
Gerald.
* Added Gerald Zajac and Jan Kloetzke to the list of contributors.
* Added the s3savage driver and accelerant to the image.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21478 a95241bf-73f2-0310-859d-f6bbb57e9c96
Disabled by default, but all kernel devs are *highly* recommended to turn them on for your builds and see if it trips anything, and then fix it.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21477 a95241bf-73f2-0310-859d-f6bbb57e9c96
When iterating through modules, the iterator was loading the module file, inserting it into the module image hash. Then, the first time get_module() was called on a module contained in the image, it was trying to load the image again. It probably actually was. Changed the logic to call get_module_image() which checks to see if it's already loaded.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21476 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Reworked the internal layout. _ValidateLayoutData() computes and caches the
layout related data and all other methods just use those values. Now, in
layout-aware mode the class should properly work not only when using the
layout items. And when using layout items, the class does actually do
internal layout; it was basically good luck that it worked in the tests,
before. Vertical resizing is supported, too.
* We do a few mean tricks to get the probably mostly preferred layout behavior:
By default our own explicit max width and that of the menu bar layout item is
set to unlimited and the horizontal menu bar alignment to left aligned. This
allows to horizontally resize a BMenuField beyond its preferred size,
although both label and menu bar have a limited max width. The user can, of
course, override those explicit sizes/alignments to get a different behavior,
if desired.
* Fixed invalidation in SetDivider(). When having the focus, the left and top
border of the blue frame were not invalidated.
* The label is no longer drawn at vertical position font ascent + descent
+ leading + 2 (not sure how this calculation was supposed to work), but
vertically centers the label around the ascent. With big fonts the label is
shown a bit too far to the bottom. Not sure how to fix this in a generic way.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21466 a95241bf-73f2-0310-859d-f6bbb57e9c96
the maximum width. The latter supports unlimited maximum width. The
_BMCMenuBar_ draws fine when resized wider than its min/preferred width, but
not the complete "button" area will cause the menu to open when being pressed.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21465 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Addad global list where poses that need periodic updates can be registered with a callback
* Use this mechanism for poses with a volume space bar
* Create only one BVolume when the BPose is created for a volume, instead of every time the free space is calculated
* On Pulse() the global list is used to update all of the registered periodic update poses
* As the poses know their volume, it is no longer necessary to use a BVolumeRoster to loop through each volume on each Pulse()
* Removed the now superfluous SendNotices() mechanism
* Removed corresponding watching / handling of these notices in BPoseView
The BPoseView did a linear search for each volume pose on each Pulse() before. What's more it did this once for each mounted volume as it did get one individual notice for each of them. To get these volumes a BVolumeRoster was used to loop through the volumes, but then the BPose did still create a new BVolume to actually calculate the free space! I'm surprised that it did not suck away more performance with this method...
Anyway, this should bring down BVolume construction and update overhead down to a minimum and hopefully fix ticket #1247.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21462 a95241bf-73f2-0310-859d-f6bbb57e9c96
stupid errors, since I don't use exceptions usually. Feel free to beat
me on this. Moved uninitialization to _DisposeData(). Corrected some
styling issues pointed out by axel. Used fprintf instead of printf.
Turned off debugging.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21450 a95241bf-73f2-0310-859d-f6bbb57e9c96
still needs some work. The sync code was never executed, as
if (len < chunkSize) (len = signed, chunkSize = unsigned)
was compiled into an unsigned compare and thus always false
for len=-1.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21441 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Use Object{Cc,C++}Flags instead of setting {CC,C++}FLAGS on the
objects directly.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21438 a95241bf-73f2-0310-859d-f6bbb57e9c96
Ignore empty 'strn' chunks instead of treating them as an error. This
also fixes loading of the "The party at the end of the earth.divx" file.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21434 a95241bf-73f2-0310-859d-f6bbb57e9c96
