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
serve, bin sizes and page size. This minimizes the amount of "large"
allocations made in heaps that don't have a bin for the allocation size
(combining multiple pages). This is desirable as such large allocations
are generally pretty inefficient, and also because it separates larger from
smaller allocations better, making the chance of a heap becoming empty higher.
For now there are three heap classes "small", "large" and "huge", with a
predefined set of bin sizes for each. This might need some finetuning later on.
Reduce the grow size to 4MB though as the allocations should now be spread
across heap classes which each grow on their own.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@26210 a95241bf-73f2-0310-859d-f6bbb57e9c96
waiting for a heap grow.
* Use that nogrow version in the VM code to avoid a deadlock with the address
space lock when a grow operation would try to create an area while a malloc
happened from such a function in the VM.
* When waiting for a grow to happen, notify the waiting thread from the grower
also if it failed to allocate a new heap. Otherwise a thread would just sit
there and wait until another thread requested growing too and that one
succeeded (or just forever in the worst case).
* Make the dedicated grow heap growable too. If the current grow heaps run low
on memory it will instruct the grower to allocate a new grow heap. This
reduces the likelyhood of running out of memory with no way to grow to a
minimum. As the growing is done asynchronously it is still possible to
happen, but it is highly unlikely as the grow heap is solely used to
allocate memory in the process of creating new heap areas and it will even
try using normal public memory if the dedicated memory has run out.
* Reduced the dedicated grow heap from 2 to 1MB. As it can now grow itself, it
doesn't need to last so long.
* Extract heap creation into it's own function that does area creation and heap
attach and use this function for growing normal and grow heaps.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@26009 a95241bf-73f2-0310-859d-f6bbb57e9c96
reference counted memory allocations. Can be used for sharing immutable
structures.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24666 a95241bf-73f2-0310-859d-f6bbb57e9c96
in code that has interrupts disabled. The chunks of memories are queued
and free()d periodically by a kernel daemon.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@24332 a95241bf-73f2-0310-859d-f6bbb57e9c96
This eliminates the edge case where the grow thread would not be able to create
a new area because no memory could be allocated for the allocation of the area.
As this case cannot happen anymore, it is also not possible to deadlock in
memalign. Therefore the timeout (which would only have prevented the deadlock
but wouldn't have solved the edge case anyway) has been removed too.
Add options to dump the dedicated grow heap and to only print the current heap
count to the "heap" debugger command.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23994 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
