* device_attach() must not load the network stack. Besides being completely
unbalanced, this was also one reason why the stack could not be unloaded
anymore. Instead, it's now done in compat_open(), as before.
* This also fixes network booting from FreeBSD drivers - the stack apparently
could not be initialized that early.
* Replaced the previous network stack based callout implementation with one
that mostly copies its functionality, but has no dependencies. Furthermore,
it runs at a higher priority (the one of the network timer should also be
revisited, though).
* Fixed mtx_owned() to work without KDEBUG as well. It's not a good idea to
introduce code that behaves completely different based on debug settings.
* Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37580 a95241bf-73f2-0310-859d-f6bbb57e9c96
Add NULL guard to the color schemes struct, although the code was safe
neverthless (for now)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37575 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added debug output.
* Rounding up the size argument could cause an overflow. Use uint64 and block
indexes, now.
* The iteration index wasn't incremented in the main loop.
* The in-block index was ignored when filling the first vector. Shouldn't
really matter since we pass back complete blocks only, so the file map
should never supply an unaligned offset, but anyway.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37573 a95241bf-73f2-0310-859d-f6bbb57e9c96
B_BUFFER_OVERFLOW. It will have returned valid vectors in this case.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37572 a95241bf-73f2-0310-859d-f6bbb57e9c96
* After an unsuccessful find_port() wait a while before trying again.
find_port() is quite expensive and particularly during the boot process it
was called quite busily. Essentially a find_port() per written syslog
message was performed.
* Added TODOs regarding using a semaphore and using find_port().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37571 a95241bf-73f2-0310-859d-f6bbb57e9c96
Note: Should we create symlinks for the *.la libtool related files?
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37561 a95241bf-73f2-0310-859d-f6bbb57e9c96
Added a TODO about creating the symlinks for gcc-agnostic packages.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37560 a95241bf-73f2-0310-859d-f6bbb57e9c96
the file already exists, unless O_EXCL has been specified. At least that's
how our file systems with write support implement it.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37554 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Use them in the time preflet to display and edit the date in a local-aware way
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37551 a95241bf-73f2-0310-859d-f6bbb57e9c96
See http://clang.llvm.org/compatibility.html#c++ why it is needed. Note that HashMap.h Key and Value are typenames as well.
Afaict this is correctly done, builds and runs on gcc4. This fixes#5892.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37550 a95241bf-73f2-0310-859d-f6bbb57e9c96
features:
* Some cosmetic adjustments of the API, like using references instead of
pointers, argument order, method names, etc.
* Added convenience template methods for archiving and unarchiving to BArchiver
and BUnarchiver.
* BUnarchiver (respectively the private BUnarchiveManager) explicitly deals with
object ownership, now. This is necessary since an error while unarchiving
could leave an already unarchived object without owning object, which would
result in it being leaked. The BUnarchiver::{Get,Find}Object() methods do now
have an optional parameter to specify whether object ownership shall be
transferred.
* Fixed incorrect header guard in headers/private/binary_compatibility/Global.h.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37538 a95241bf-73f2-0310-859d-f6bbb57e9c96
"white on black" and "black on white". For some reason these aren't
applied live, like when you chose a custom combo.
I'm not really fond of the code, and will rework it in the next few days.
But at least it works!
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37535 a95241bf-73f2-0310-859d-f6bbb57e9c96
incrementing the cache color cycle. Using the fixed value (8) would
potentially misalign the object again.
* Don't use CACHE_ALIGN_ON_SIZE for object caches any longer -- we have the
alignment parameter anyway (the flag is still used for the MemoryManager,
though).
* ObjectCache::InitSlab(): Slab coloring *was* done when CACHE_ALIGN_ON_SIZE
was given, i.e. exactly the wrong way around. Also the cache_color_cycle
computation was weird -- color 0 was used twice in a row.
* The "slabs" and "slab_cache" KDL commands also print the alignment, now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37534 a95241bf-73f2-0310-859d-f6bbb57e9c96
- fixes ticket #6186: Web+ would crash if pgup/pgdn was hit in its URL input while the choice list wasn't visible.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37533 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Align the allocated objects to 16 byte. This is required by the x86
arch_thread structure. Haiku only didn't crash and burn since the until
recently used heap allocator apparently aligned the structures to 16 byte
anyway and the now used slab allocator has a bug preventing slab coloring
for that object size.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37532 a95241bf-73f2-0310-859d-f6bbb57e9c96
* updated icu-devel package to 4.4.1
* create appropriate devel lib links when icu-devel is being installed
(which only works during build, not when using installoptionalpackage)
* dropped ICU headers from repository and instead fetch them from the
icu-devel package when needed during the build
* adjusted Jamfiles that make use of ICU-headers accordingly (hopefully
I didn't miss any)
* fixed license in OptionalPackage description of icu packages
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@37531 a95241bf-73f2-0310-859d-f6bbb57e9c96
