| Age | Commit message (Collapse) | Author |
|---|
|
change both the nointr and default pool allocators to using uvm_km_getpage.
change pools to default to a maxpages value of 8, so they hoard less memory.
change mbuf pools to use default pool allocator.
pools are now more efficient, use less of kmem_map, and a bit faster.
tested mcbride, deraadt, pedro, drahn, miod to work everywhere
|
|
we're looking for. change small page_header hash table to a splay tree.
from Chuck Silvers.
tested by brad grange henning mcbride naddy otto
|
|
- Allow a pool to be initialized with PR_DEBUG which will cause it to
allocate with malloc_debug.
- sprinkle some splassert.
|
|
- uvm_km_alloc_poolpage1 has its own spl protection, no need to add
additional layer around it.
|
|
One relevant change: round up pool element size to the alignment.
VS: ----------------------------------------------------------------------
|
|
give us pages. PR_NOWAIT most likely means "hey, we're coming from an
interrupt, don't mess with stuff that doesn't have proper protection".
- pool_allocator_free is called in too many places so I don't feel
comfortable without that added protection from splvm (and besides,
pool_allocator_free is rarely called anyway, so the extra spl will be
unnoticeable). It shouldn't matter when fiddling with those flags, but
you never know.
- Remove a wakeup without a matching tsleep. It's a left-over from
some other code path that I've been investigating when reworking the
pool a while ago and it should have been removed before that commit.
deraadt@ ok
|
|
|
|
Diff generated by Chris Kuethe.
|
|
Just because the pool allocates from intrsafe memory doesn't mean that the
pool has to be protected by splvm. We can have an intrsafe pools at splbio
or splsoftnet.
pool_page_alloc and pool_page_free must du their own splvm protection.
|
|
When trying the drain hook just in pool_allocator_alloc, don't leak memory
when the drain succeeds and don't avoid draining other pools if this
pool doesn't have a drain hook.
|
|
|
|
From thorpej@netbsd.org
|
|
the current size of the pool. ok art@
|
|
let other parts of the kernel call it.
|
|
|
|
and don't static inline big functions that are called multiple
times and are not time critical.
|
|
using printf(). Makes ddb sessions more fruitful.
|
|
PR_MALLOC wasn't used at all in the code
and PR_STATIC was missing pieces and should be solved with allocators.
|
|
1. When a pool hit the hard limit. Just before bailing out/sleeping.
2. When an allocator fails to allocate memory (with PR_NOWAIT).
3. Just before trying to reclaim some page in pool_reclaim.
The function called form the hook should try to free some items to the
pool if possible.
Convert m_reclaim hooks that were embedded in MCLGET, MGET and MGETHDR
into a pool drain hook (making the code much cleaner).
|
|
|
|
well (not at all) with shortages of the vm_map where the pages are mapped
(usually kmem_map).
Try to deal with it:
- group all information the backend allocator for a pool in a separate
struct. The pool will only have a pointer to that struct.
- change the pool_init API to reflect that.
- link all pools allocating from the same allocator on a linked list.
- Since an allocator is responsible to wait for physical memory it will
only fail (waitok) when it runs out of its backing vm_map, carefully
drain pools using the same allocator so that va space is freed.
(see comments in code for caveats and details).
- change pool_reclaim to return if it actually succeeded to free some
memory, use that information to make draining easier and more efficient.
- get rid of PR_URGENT, noone uses it.
|
|
From NetBSD.
|
|
- use ltsleep instead of simple_unlock ; tsleep
|
|
|
|
|
|
|
|
(Look ma, I might have broken the tree)
|
|
|
|
|
|
|
|
|
|
art@ ok
|
|
|
|
|
|
- pool_cache similar to the slab allocator in Solaris.
- clean up locking a bit.
- Don't pass __LINE__ and __FILE__ to pool_get and pool_put unless
POOL_DIAGNOSTIC is defined.
|
|
old vm system and I hoped that it would make people help me to switch all
archs to uvm. But that didn't help.
Fix pool to work with the old vm system (not optimal, ha!).
|
|
|
|
|
|
|
|
|
