locks.
* Added syscalls for a new kind of mutex. A mutex consists only of an int32 and
doesn't require any kernel resources. So it's initialization cannot fail
(it consists only of setting the mutex value to 0). An uncontended lock or
unlock operation can basically consist of an atomic_*() in userland. The
syscalls (when the mutex is contended) are a bit more expensive than semaphore
operations, though.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@36158 a95241bf-73f2-0310-859d-f6bbb57e9c96
implemented by means of an additional member in cpu_ent.
* Removed thread::keep_scheduled and the related functions. The feature
wasn't used yet and wouldn't have worked as implemented anyway.
* Resurrected an older, SMP aware version of our simple scheduler and made it
the default instead of the affine scheduler. The latter is in no state to
be used yet. It causes enormous latencies (I've seen up to 0.1s) even when
six or seven CPUs were idle at the same time, totally killing parallelism.
That's also the reason why a -j8 build was slower than a -j2. This is no
longer the case. On my machine the -j2 build takes about 10% less time now
and the -j8 build saves another 20%. The latter is not particularly
impressive (compared with Linux), but that seems to be due to lock
contention.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34615 a95241bf-73f2-0310-859d-f6bbb57e9c96
CreateAsmStructOffsetsHeader mechanism to generate a header with macros
defined to the sizes of the structures we're interested in and when compiling
in C mode define the structures as "struct { char bytes[size]; }".
It works in principle, but due to how jam works, one would have to specify the
dependency to the generated header for all sources that include it directly or
indirectly.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34441 a95241bf-73f2-0310-859d-f6bbb57e9c96
- Moved scheduler listening interface to <listeners.h> and added more
convenient to use templatized notification functions.
- Added a listener mechanism for the wait objects (semaphores, condition
variables, mutex, rw_lock).
* system profiler:
- Hopefully fixed locking issues related to notifying the profiler thread
for good. We still had an inconsistent locking order, since the scheduler
notification callbacks are invoked with the thread lock held and have to
acquire the object lock then, while the other callbacks acquired the object
lock first and as a side effect of ConditionVariable::NotifyOne() acquired
the thread lock. Now we make sure the object lock is the innermost lock.
- Track the number of dropped events due to a full buffer.
_user_system_profiler_next_buffer() returns this count now.
- When scheduling profiling events are requested also listen to wait objects
and generate the respective profiling events. We send those events lazily
and cache the infos to avoid resending an event for the same wait object.
- When starting profiling we do now generate "thread scheduled" events for
the already running threads.
- _user_system_profiler_start(): Check whether the parameters pointer is a
userland address at all.
- The system_profiler_team_added event does now also contain the team's name.
* Added a sem_get_name_unsafe() returning a semaphore's name. It is "unsafe",
since the caller has to ensure that the semaphore exists and continues to
exist as long as the returned name is used.
* Adjusted the "profile" and "scheduling_recorder" according to the system
profiling changes. The latter prints the number of dropped events, now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30345 a95241bf-73f2-0310-859d-f6bbb57e9c96
1) We now maintain a runqueue per CPU, rather than a single global shared queue. Idle threads are segregated into their own queue for simplicity.
2) Enqueueing threads is now somewhat more intelligent - if the thread is pinned, it is always enqueued onto that core. Otherwise we enqueue it on whichever CPU it previously ran, unless it either hasn't run before, or that core has been disabled via ProcessController. If so, we try to enqueue it on whichever core has been the most idle recently.
3) The above allow various simplifications to thread scheduling. Pinned threads and/or disabled cores are now no longer special cases that need to be dealt with. If a CPU has no threads ready, it looks for another one to steal a thread from, though that part still needs some tuning along with enqueueing for load balancing purposes.
The chief aim here is better load balancing and support for soft affinity. However, at the moment the overall behavior still exhibits some regressions compared to the old scheduler, so it's disabled by default. If you wish to experiment/debug with it, instructions for enabling it can be found in scheduler.cpp. Much thanks to Ingo, Axel and everyone who's helped with either code review/advice or testing so far.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29643 a95241bf-73f2-0310-859d-f6bbb57e9c96
scheduler_set_thread_priority(). Setting the thread priority was the
only situation in which it was used.
* Renamed scheduler.cpp to scheduler_simple.cpp.
* The scheduler functions are no longer called directly. Instead there's
an operation vector now, which is initialized at kernel init time.
This allows for picking the most suitable scheduler for the machine
(e.g. a non-SMP scheduler on a non-SMP machine).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@28262 a95241bf-73f2-0310-859d-f6bbb57e9c96
scheduler tracing and scheduler analysis code into separate source
files.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@28253 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added smp_send_multicast_ici(), which sends the message to all CPUs
specified via a mask.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@27910 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added global hash table for images.
* Improved a bit of code by using the new image hash table. E.g.
_get_image_info() can return infos for images of any team, now.
* Fixed remove_images() comment: The function must not be invoked with
the team lock being held.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@27606 a95241bf-73f2-0310-859d-f6bbb57e9c96
introduces the following relevant changes:
* VMCache:
- Renamed vm_cache to VMCache, merged it with vm_store and made it a
C++ class with virtual methods (replacing the store operations).
Turned the different store implementations into subclasses.
- Introduced MergeStore() callback, changed semantics of Commit().
- Changed locking and referencing semantics. A reference can only be
acquired/released with the cache locked. An unreferenced cache is
deleted and a mergeable cache merged when it is unlocked. This
removes the "busy" state of a cache and simplifies the page fault
code.
* Added VMAnonymousCache, which will implement swap support (work by
Zhao Shuai). It is not integrated and used yet, though.
* Enabled the mutex/recursive lock holder asserts.
* Fixed DoublyLinkedList::Swap().
* Generalized the low memory handler to a low resource handler. And made
semaphores and reserved memory handled resources. Made
vm_try_resource_memory() optionally wait (with timeout), and used that
feature to reserve memory for areas.
...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@26572 a95241bf-73f2-0310-859d-f6bbb57e9c96
This is a temporary workaround to get the thing building until someone has enough cafein available to update binutils.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25895 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Current linking scheme does not always export arch specific symbols from the
kernel. Merging all compiled arch stuff into a single .o file (and linking
that to the kernel) fixes it.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25679 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Renamed main.c to main.cpp
* Use find_directory() to construct the Bootscript path.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25509 a95241bf-73f2-0310-859d-f6bbb57e9c96
address space that is fully locked and marked B_KERNEL_AREA. It can
thus be accessed by the kernel without additional checks.
* For each userland thread we do create a user_thread structure in that
area. The structure is accessible from userland via TLS, using the
private get_user_thread() function.
* Introduced private userland functions [un]defer_signals(). They can be
used to cheaply disable/re-enable signal delivery. They use the
user_thread::defer_signals/pending_signals fields which are
checked/updated by the kernel.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25451 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Renamed the old kernel_posix[_arch...].o to kernel_lib_posix...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25336 a95241bf-73f2-0310-859d-f6bbb57e9c96
Realtime option group). The implementation should be complete, but is
totally untested yet.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25326 a95241bf-73f2-0310-859d-f6bbb57e9c96
name, couldn't resist :-P). It's semantically equivalent to a mutex,
but doesn't need a semaphore (it uses thread blocking and a simple
queue instead). Initialization can't fail. In fact it is ready to use
without initialization when living in the bss segment, also in the
early boot process. It's as fast as a benaphore in cases of low lock
contention, and faster otherwise. Only disadvantage is the higher
immediate memory footprint of 16 bytes.
* Changed how the "thread" and "threads" debugger commands list the
objects they are waiting for. Cutexes are also included.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25276 a95241bf-73f2-0310-859d-f6bbb57e9c96
- If the hook of the timer we're cancelling is currently being
executed, we do now wait till it is finished. This is how the BeBook
specifies the function to behave.
- Periodic timers would not be cancelled, if their hook was being
invoked at the same time, since they weren't in the queue during
that time.
- Since we know the CPU on which the timer is scheduled (timer::cpu),
we don't have to look through any other CPU queue to find it.
- Fixed the return value. It should report whether the timer had
already fired, and was not always doing that.
* Added private add_timer() flag B_TIMER_ACQUIRE_THREAD_LOCK. It causes
the thread spinlock to be acquired before the event hook is called.
cancel_timer() doesn't wait for timers with the flag set. Instead we
check in the timer interrupt function after acquiring the thread
spinlock whether the timer was cancelled in the meantime. Calling
cancel_timer() with the thread spinlock being held does thus avoid any
race conditions and won't deadlock, if the event hook needs to acquire
the thread spinlock, too. This feature proves handy for some kernel
internal needs.
* The scheduler uses a B_TIMER_ACQUIRE_THREAD_LOCK timer now and
cancel_timer() instead of the no longer needed
_local_timer_cancel_event().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@25098 a95241bf-73f2-0310-859d-f6bbb57e9c96
* The boot splash code now checks wether debug screen output is enabled or not
using the above function.
* The boot splash code no longer maps it's own copy of the frame buffer, instead,
it will use the boot item feature as the VESA driver does. Also removed the
lock, as that's not needed at all.
* Renamed splash.cpp to boot_splash.cpp, and boot/splash.h to boot_splash.h
(it's not part of the boot loader, but the kernel).
* Removed dead code from boot_splash.cpp, added license. Replaced license
header in boot_splash.h to a style guide conforming one.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24489 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added the feature of an animated boot screen (icons lighting up at
different boot stages).
* Added first version of new boot splash images, generated by the new
hsbg tool. (Also finally contains the "new" Haiku logo.)
changes by myself:
* Added Artur to the contributors list in About System.
* Fixed some left overs in the patch, kept tracing turned off.
TODO:
* Remove the need for hard coding the icon positions. (Maybe generate
those from hsbg and put them into images.h? Have user provide icon
spacing/offsets at the command line for hsbg?)
* Rename the stages to something meaningful.
* Use hsbg as a build system tool and generate images.h during build
from PNGs provided in the artwork folder.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24434 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Fixed various warnings with GCC4 due to the double sHaikuRevision line.
* Turned system_info.c to a C++ file.
* Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24404 a95241bf-73f2-0310-859d-f6bbb57e9c96
structure. They are properly inherited and updated on
fork(), load_image(), and exec().
* Implemented the get[e]{u,g}id(), set[[r]e]{u,g}id() family for real.
* getgroups() also calls the kernel now, but only returns the effective
group ID. Supplementary groups support is still missing.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24359 a95241bf-73f2-0310-859d-f6bbb57e9c96
and pages are now kept in lists as well. This allows to return free pages once
a bin does not need them anymore. Partially filled pages are kept in a sorted
linked list so that allocation will always happen on the fullest page - this
favours having full pages and makes it more likely lightly used pages will get
completely empty so they can be returned. Generally this now goes more in the
direction of a slab allocator.
The allocation logic has been extracted, so a heap is now simply attachable to
a region of memory. This allows for multiple heaps and for dynamic growing. In
case the allocator runs out of free pages, an asynchronous growing thread is
notified to create a new area and attach a new heap to it.
By default the kernel heap is now set to 16MB and grows by 8MB each time all
heaps run full.
This should solve quite a few issues, like certain bins just claiming all pages
so that even if there is free space nothing can be allocated. Also it obviously
does aways with filling the heap page by page until it overgrows.
I think this is now a well performing and scalable allocator we can live with
for quite some time. It is well tested under emulation and real hardware and
performs as expected. If problems come up there is an extensive sanity checker
that can be enabled by PARANOID_VALIDATION that covers most aspects of the
allocator. For normal operation this is not necessary though and is therefore
disabled by default.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23939 a95241bf-73f2-0310-859d-f6bbb57e9c96
used it for an hour or so, I really wonder how we could live without it.
:-)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23640 a95241bf-73f2-0310-859d-f6bbb57e9c96
* With post syscall debugging enabled, the x86 syscall handling didn't
remove all parameters from the stack after calling the respective user
debugger hook. Should have been harmless though, since the following
code didn't rely on the stack being in order.
* Added syscall pre/post (kernel) tracing functions
trace_{pre,post}_syscall(). They are generic, but need to be invoked
by the architecture specific syscall code. Currently only done for
x86.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23600 a95241bf-73f2-0310-859d-f6bbb57e9c96
23139 into trunk, with roughly the following changes (for details svn
log the branch):
* The int 99 syscall handler is now fully in assembly.
* Added a sysenter/sysexit handler and use it on Pentiums that support
it (via commpage).
* Got rid of i386_handle_trap(). A bit of functionality was moved into
the assembly handler which now uses a jump table to call C functions
handling the respective interrupt.
* Some optimizations to get user debugger support code out of the
interrupt handling path.
* Introduced a thread::flags fields which allows to skip handling of
rare events (signals, user debug enabling/disabling) on the
common interrupt handling path.
* Got rid of the explicit iframe stack. The iframes can still be
retrieved by iterating through the stack frames.
* Made the commpage an architecture independent feature. It's used for
the real time data stuff (instead of creating a separate area).
* The x86 CPU modules can now provide processor optimized versions for
common functions (currently memcpy() only). They are used in the
kernel and are provided to the userland via commpage entries.
* Introduced build system feature allowing easy use of C structure
member offsets in assembly code.
Changes after merging:
* Fixed merge conflict in src/system/kernel/arch/x86/arch_debug.cpp
(caused by refactoring and introduction of "call" debugger command).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23370 a95241bf-73f2-0310-859d-f6bbb57e9c96
vfs_select.h, respectively moved most of it into the new kernel
private header wait_for_objects.h.
* Added new experimental API functions wait_for_objects[_etc](). They
work pretty much like poll(), but also for semaphores, ports, and
threads.
* Removed the "ref" parameter from notify_select_events() and the
select_sync_pool functions as well as from fd_ops::fd_[de]select(). It
is no longer needed. The FS interface select() hook still has it,
though -- the VFS will always pass 0.
* de]select_fd() take a select_info* instead of a select_sync* + ref
pair, now. Added respective functions for semaphores, ports, and
threads.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22416 a95241bf-73f2-0310-859d-f6bbb57e9c96
cheap means to block threads until notified explicitely.
threads
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21874 a95241bf-73f2-0310-859d-f6bbb57e9c96
relevant for the image creation.
* The CopySetHaikuRevision propagates the value of the
HAIKU_INCLUDE_IN_IMAGE variable from the source to the target.
* Propagate the value of HAIKU_INCLUDE_IN_IMAGE from "kernel" to
"kernel_$(TARGET_ARCH)".
Now "jam update-install kernel" should work as expected.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21527 a95241bf-73f2-0310-859d-f6bbb57e9c96
libroot). It's available in the kernel through the private
get_haiku_revision() and added to the kernel syslog output.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21173 a95241bf-73f2-0310-859d-f6bbb57e9c96
The solution is less than optimal, but should work for now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20722 a95241bf-73f2-0310-859d-f6bbb57e9c96
files, too.
* Added a temporary icon for the kernel until Stephan comes up with a better
one (hint hint!) :-))
* This even fixes bug #648.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17825 a95241bf-73f2-0310-859d-f6bbb57e9c96
loader): it now supports holes between segments, and accepts any segment order.
* Renamed elf.c to elf.cpp and fixed warnings.
* Renamed elf_image_info::dynamic_ptr and eheader to dynamic_section and elf_header.
* Some cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16878 a95241bf-73f2-0310-859d-f6bbb57e9c96
longer pushing a whole path on the stack.
* open_module_list() now adds the prefix to the paths it puts on the
stack to reduce the amount of searching to do (this probably wasn't
done before because of the earlier boot method via bootfs which
didn't allow for deep path names).
* module.c is now module.cpp (for KPath usage).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16584 a95241bf-73f2-0310-859d-f6bbb57e9c96
