Theoretically, not doing that could have caused still mapped pages to get
into the free queue. This could have been a cause of #3110, but the reported
circumstances don't look quite fitting.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30605 a95241bf-73f2-0310-859d-f6bbb57e9c96
again, or else the port loses capacity.
* This fixes bug #2720.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30517 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Did claim to have reserved pages when calling vm_page_allocate_page(), but
didn't have any (copy&paste bug). We cannot use it without reserved pages,
as we need to call vm_page_allocate_page() with a cache locked.
* No longer use low_resource_state() to determine whether to precache or not,
but use the new vm_page_num_used_pages() instead.
* Also don't (try to) precache when the cache already has more than 2/3 of its
pages to safe some unnecessary work.
* The size to precache was limited to the file size incorrectly.
* When precaching failed, the cache reference was not released.
* The precaching started one page too late, causing bug #3835.
* Reenabled precaching.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30515 a95241bf-73f2-0310-859d-f6bbb57e9c96
with vm_cache_acquire_locked_page_cache().
* Added new function vm_page_num_unused_pages() which returns the pages that are
actually completely free and unused.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30514 a95241bf-73f2-0310-859d-f6bbb57e9c96
* vfs_read_pages_async() must not be called with locked cache,
since it cannot be guaranteed that the operation will be
performed asynchronously. (The ISO9660 FS for example does not
implement the new IO hooks... yet.)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30476 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Implemented a way to do asynchronous pre-fetching when mapping files.
* There are slight code duplications in some places that could benefit
from cleaning up, but nothing too bad.
* Implementing smarter ways to trigger prefetching and more analysis of
the situations in the kernel would be nice. Currently up to 10 MB
of every mapped file are pre-fetched without further analysis.
* The speed improvement is nice for certain operations. On our test
system (real hardware), Firefox took 9 seconds from being launched
to display a window. Now it takes 5 seconds. Both measurements
right after booting. The same system took 35 seconds from launching
Haiku in the GRUB menu to displaying the Tracker desktop background
image. Now it takes 27 seconds.
* We didn't have the chance to check out the effects of this on the
CD boot, but potentially, they could speed it up a lot.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30465 a95241bf-73f2-0310-859d-f6bbb57e9c96
objects we're going to add, so we get less slot collisions.
* Use the "Unchecked" versions of the OpenHashTable Insert()/Remove() methods,
since we have interrupts disabled.
* Fixed wrong check that caused the wait object info events not to be filled in.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30384 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
potentially want to notify the profiler thread (i.e. wake it up), which can
cause another scheduling event notification. That is we have to make sure that
event buffer is in a consistent state at that point (i.e. wake up the thread
at the very end of the callbacks). Furthermore the ThreadEnqueuedInRunQueue()
callback can be called with spinlocks besides the thread spinlock being held. In
particular waking up threads also happens in the condition variable code with
a static spinlock being held. Trying to notify the condition variable the
profiler thread is waiting on in such a case would be a guaranteed deadlock.
Hence we avoid doing that, now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30339 a95241bf-73f2-0310-859d-f6bbb57e9c96
manipulating the queue is a particularly unsuitable place for calling the
listeners, as they wouldn't be allowed to e.g. unblock threads, since that
would screw the run queue.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30338 a95241bf-73f2-0310-859d-f6bbb57e9c96
will return consistent values. This helps with debug measurements for the time
being. Obviously we'll have to think of something different when we support
speed-stepping on models with frequency-dependent TSCs.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30287 a95241bf-73f2-0310-859d-f6bbb57e9c96
theoretically write the given page.
* page writer: Fixed the incorrect check whether a temporary page can be
written by using the new CanWritePage().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30281 a95241bf-73f2-0310-859d-f6bbb57e9c96
* The previous code would have scheduled a single page to be written out (if it
would have ever been triggered), now we schedule the complete previous write
access. This greatly speeds up a "dd if=/dev/zero of=test ..." beyond the
size of available memory.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30276 a95241bf-73f2-0310-859d-f6bbb57e9c96
passed in a structure now, so it is easier to extend it and ignore unused
parameters.
* One can now select which system profiling events one is interested in.
* Added scheduling events to the system profiling interface. Those are pretty
much the ones recorded when scheduler tracing is enabled. Still missing are
the "wait object" events that allow to interpret what a thread is waiting
for.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30243 a95241bf-73f2-0310-859d-f6bbb57e9c96
* When writing a page failed it is not a good idea to re-enqueue it at the tail
of the modified queue, since that is definitely behind the page writer's
marker and the page would be picked up again before reaching the end of the
queue. If that happened with more than 256 pages, the page writer would keep
picking up only those non-writable pages and make no more progress.
* When selecting pages also skip temporary pages, if there's no more swap space
available, since trying to write those pages would most likely fail anyway
(triggering the first problem).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30230 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Added "step" debugger command to single-step to the next instruction (of the
topmost iframe).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30212 a95241bf-73f2-0310-859d-f6bbb57e9c96
continue kernel breakpoints -- we would just retrigger the breakpoint when
continuing before.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30210 a95241bf-73f2-0310-859d-f6bbb57e9c96
re-entered it fast enough, we wouldn't leave the loop and thus cause a
deadlock, since we wouldn't process the ICI message telling us to halt. We do
now call smp_intercpu_int_handler() in the loop and guard the function from
being re-entered. This also has the advantage that we can execute code on all
CPUs in the kernel debugger, if we have to.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30208 a95241bf-73f2-0310-859d-f6bbb57e9c96
enter/exit code. There's no real reason not to keep kernel breakpoints
enabled when in userland (unless there are breakpoints installed for the
team, of course).
* Enabled kernel breakpoints by default (check your kernel_debug_config.h,
if you have overridden it!), since they don't really add any overhead
anymore.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30206 a95241bf-73f2-0310-859d-f6bbb57e9c96
pointed out, there is also an unsolvable race condition with BFS that other
file systems should share.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30204 a95241bf-73f2-0310-859d-f6bbb57e9c96
unload the network stack. This now works again, although I didn't manage to
do it without another lock.
* Cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30196 a95241bf-73f2-0310-859d-f6bbb57e9c96
profiling events.
* profile: Avoid using get_{team,thread}_info() in common code paths. The
system profiling mode is asynchronous, so the team or thread in question
could already be gone.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30193 a95241bf-73f2-0310-859d-f6bbb57e9c96
methods that used an "event mask" field. There was no need to introduce
a "flags" field for the same purpose.
* Renamed protected DefaultNotificationService methods (removed "_" prefix).
* Adjusted the code providing a notification service accordingly.
* Changed the event message several notification services generated by renaming
the "opcode" field to "event".
* Implemented the TEAM_ADDED event and also added a TEAM_EXEC event.
* Added notifications for threads and images.
* Added visitor-like iteration functions for teams, threads, and images.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30126 a95241bf-73f2-0310-859d-f6bbb57e9c96
data. This is available when BLOCK_CACHE_BLOCK_TRACING is 2 or greater.
* Completely untested as of now, though. Will do so soon.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30104 a95241bf-73f2-0310-859d-f6bbb57e9c96
documentation the segment offsets it shall return are relative to the
address specified in the program header.
* Added "qSupported".
* Added the mandatory commands and some additional ones. Partially implemented
only.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30038 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Fixed the read with bounce buffer case. When skipping a partial bounce
buffer before the part we're interested in, we forgot to update "offset".
* Added some more comments for readability.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29999 a95241bf-73f2-0310-859d-f6bbb57e9c96
better, though :-)
* Also fixed a mixup of the unintuitive argument order of our
atomic_test_and_set(); I guess I will change that sooner or later.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29994 a95241bf-73f2-0310-859d-f6bbb57e9c96
page failed since the cache has been shrunk, we need not only remove the
page from the cache, we also need to remove all of its area mappings and
free it. Not removing the area mappings might have been the cause of #3110,
not freeing it would cause it to be leaked for good.
* vm_page_write_modified_page_range(): When writing failed for another reason
and the page wasn't in the modified queue before, we would lose the info
in which queue it was before and setting the page state to modified would
assume the active queue. This could potentially screw up our page queue
structures.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29992 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Renamed DoublyLinkedList::Size() to Count(), since it actually counts the
items (ie. O(n)).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29979 a95241bf-73f2-0310-859d-f6bbb57e9c96
* optimized unmapping of a single page that was rather inefficient
for large areas
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29978 a95241bf-73f2-0310-859d-f6bbb57e9c96
most importantly input_server, which then closes the drivers for input devices.
This in turn will do some cleanup. The touch pad driver will then be reset to
act as normal PS/2 mouse, so that OS like BeOS can use it again.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@29948 a95241bf-73f2-0310-859d-f6bbb57e9c96
