additional information, when the page is guaranteed to not belong to a
paging queue. Usually, this results in a lot of type casts which make
reasoning about the code correctness harder.
Sometimes m->object is used instead of pageq, which could cause real
and confusing bugs if non-NULL m->object is leaked. See r141955 and
r253140 for examples.
Change the pageq member into a union containing explicitly-typed
members. Use them instead of type-punning or abusing m->object in x86
pmaps, uma and vm_page_alloc_contig().
Requested and reviewed by: alc
Sponsored by: The FreeBSD Foundation
for nodes used in vm_radix.
On architectures supporting direct mapping, also avoid to pre-allocate
the KVA for such nodes.
In order to do so make the operations derived from vm_radix_insert()
to fail and handle all the deriving failure of those.
vm_radix-wise introduce a new function called vm_radix_replace(),
which can replace a leaf node, already present, with a new one,
and take into account the possibility, during vm_radix_insert()
allocation, that the operations on the radix trie can recurse.
This means that if operations in vm_radix_insert() recursed
vm_radix_insert() will start from scratch again.
Sponsored by: EMC / Isilon storage division
Reviewed by: alc (older version)
Reviewed by: jeff
Tested by: pho, scottl
Unify the 2 concept into a real, minimal, sxlock where the shared
acquisition represent the soft busy and the exclusive acquisition
represent the hard busy.
The old VPO_WANTED mechanism becames the hard-path for this new lock
and it becomes per-page rather than per-object.
The vm_object lock becames an interlock for this functionality:
it can be held in both read or write mode.
However, if the vm_object lock is held in read mode while acquiring
or releasing the busy state, the thread owner cannot make any
assumption on the busy state unless it is also busying it.
Also:
- Add a new flag to directly shared busy pages while vm_page_alloc
and vm_page_grab are being executed. This will be very helpful
once these functions happen under a read object lock.
- Move the swapping sleep into its own per-object flag
The KPI is heavilly changed this is why the version is bumped.
It is very likely that some VM ports users will need to change
their own code.
Sponsored by: EMC / Isilon storage division
Discussed with: alc
Reviewed by: jeff, kib
Tested by: gavin, bapt (older version)
Tested by: pho, scottl
into threads each processing queue in a single domain. The structure
of the pagedaemons and queues is kept intact, most of the changes come
from the need for code to find an owning page queue for given page,
calculated from the segment containing the page.
The tie between NUMA domain and pagedaemon thread/pagequeue split is
rather arbitrary, the multithreaded daemon could be allowed for the
single-domain machines, or one domain might be split into several page
domains, to further increase concurrency.
Right now, each pagedaemon thread tries to reach the global target,
precalculated at the start of the pass. This is not optimal, since it
could cause excessive page deactivation and freeing. The code should
be changed to re-check the global page deficit state in the loop after
some number of iterations.
The pagedaemons reach the quorum before starting the OOM, since one
thread inability to meet the target is normal for split queues. Only
when all pagedaemons fail to produce enough reusable pages, OOM is
started by single selected thread.
Launder is modified to take into account the segments layout with
regard to the region for which cleaning is performed.
Based on the preliminary patch by jeff, sponsored by EMC / Isilon
Storage Division.
Reviewed by: alc
Tested by: pho
Sponsored by: The FreeBSD Foundation
clearing the page's PGA_REFERENCED flag. Since we are typically
manipulating the page's act_count field when we are clearing its
PGA_REFERENCED flag, the page lock is already held everywhere that we clear
the PGA_REFERENCED flag. So, in fact, this revision only changes some
comments and an assertion. Nonetheless, it will enable later changes to
object locking in the pageout code.
Introduce vm_page_assert_locked(), which completely hides the implementation
details of the page lock from the caller, and use it in
vm_page_aflag_clear(). (The existing vm_page_lock_assert() could not be
used in vm_page_aflag_clear().) Over the coming weeks, I expect that we'll
either eliminate or replace the various uses of vm_page_lock_assert() with
vm_page_assert_locked().
Reviewed by: attilio
Sponsored by: EMC / Isilon Storage Division
reason to inline the implementation of vm_page_lock_assert() in the
!KLD_MODULES case. Use the same implementation for both KLD_MODULES and
!KLD_MODULES.
Reviewed by: kib
per-page lock rather than vm_object lock, without any further overhead.
Make the formal switch.
Sponsored by: EMC / Isilon storage division
Reviewed by: alc
Tested by: pho
Replace the per-object resident and cached pages splay tree with a
path-compressed multi-digit radix trie.
Along with this, switch also the x86-specific handling of idle page
tables to using the radix trie.
This change is supposed to do the following:
- Allowing the acquisition of read locking for lookup operations of the
resident/cached pages collections as the per-vm_page_t splay iterators
are now removed.
- Increase the scalability of the operations on the page collections.
The radix trie does rely on the consumers locking to ensure atomicity of
its operations. In order to avoid deadlocks the bisection nodes are
pre-allocated in the UMA zone. This can be done safely because the
algorithm needs at maximum one new node per insert which means the
maximum number of the desired nodes is the number of available physical
frames themselves. However, not all the times a new bisection node is
really needed.
The radix trie implements path-compression because UFS indirect blocks
can lead to several objects with a very sparse trie, increasing the number
of levels to usually scan. It also helps in the nodes pre-fetching by
introducing the single node per-insert property.
This code is not generalized (yet) because of the possible loss of
performance by having much of the sizes in play configurable.
However, efforts to make this code more general and then reusable in
further different consumers might be really done.
The only KPI change is the removal of the function vm_page_splay() which
is now reaped.
The only KBI change, instead, is the removal of the left/right iterators
from struct vm_page, which are now reaped.
Further technical notes broken into mealpieces can be retrieved from the
svn branch:
http://svn.freebsd.org/base/user/attilio/vmcontention/
Sponsored by: EMC / Isilon storage division
In collaboration with: alc, jeff
Tested by: flo, pho, jhb, davide
Tested by: ian (arm)
Tested by: andreast (powerpc)
sleep, and perform the page allocations with VM_ALLOC_SYSTEM
class. Previously, the allocation was also allowed to completely drain
the reserve of the free pages, being translated to VM_ALLOC_INTERRUPT
request class for vm_page_alloc() and similar functions.
Allow the caller of malloc* to request the 'deep drain' semantic by
providing M_USE_RESERVE flag, now translated to VM_ALLOC_INTERRUPT
class. Previously, it resulted in less aggressive VM_ALLOC_SYSTEM
allocation class.
Centralize the translation of the M_* malloc(9) flags in the single
inline function malloc2vm_flags().
Discussion started by: "Sears, Steven" <Steven.Sears@netapp.com>
Reviewed by: alc, mdf (previous version)
Tested by: pho (previous version)
MFC after: 2 weeks
cache line in order to avoid manual frobbing but using
struct mtx_padalign.
The sole exception being nvme and sxfge drivers, where the author
redefined CACHE_LINE_SIZE manually, so they need to be analyzed and
dealt with separately.
Reviwed by: jimharris, alc
because the queue itself serves no purpose. When a held page is freed,
inserting the page into the hold queue has the side effect of setting the
page's "queue" field to PQ_HOLD. Later, when the page is unheld, it will
be freed because the "queue" field is PQ_HOLD. In other words, PQ_HOLD is
used as a flag, not a queue. So, this change replaces it with a flag.
To accomodate the new page flag, make the page's "flags" field wider and
"oflags" field narrower.
Reviewed by: kib
network file systems (not only NFS proper). Short reads cause pages
other then the requested one, which were not filled by read response,
to stay invalid.
Change the vm_page_readahead_finish() interface to not take the error
code, but instead to make a decision to free or to (de)activate the
page only by its validity. As result, not requested invalid pages are
freed even if the read RPC indicated success.
Noted and reviewed by: alc
MFC after: 1 week
to pull vm_param.h was removed. Other big dependency of vm_page.h on
vm_param.h are PA_LOCK* definitions, which are only needed for
in-kernel code, because modules use KBI-safe functions to lock the
pages.
Stop including vm_param.h into vm_page.h. Include vm_param.h
explicitely for the kernel code which needs it.
Suggested and reviewed by: alc
MFC after: 2 weeks
vm_page oflags by providing helper function
vm_page_readahead_finish(), which handles completed reads for pages
with indexes other then the requested one, for VOP_GETPAGES().
Reviewed by: alc
MFC after: 1 week
layer, but it is read directly by the MI VM layer. This change introduces
pmap_page_is_write_mapped() in order to completely encapsulate all direct
access to PGA_WRITEABLE in the pmap layer.
Aesthetics aside, I am making this change because amd64 will likely begin
using an alternative method to track write mappings, and having
pmap_page_is_write_mapped() in place allows me to make such a change
without further modification to the MI VM layer.
As an added bonus, tidy up some nearby comments concerning page flags.
Reviewed by: kib
MFC after: 6 weeks
for allocation of fictitious pages, for which PHYS_TO_VM_PAGE()
returns proper fictitious vm_page_t. The range should be de-registered
after consumer stopped using it.
De-inline the PHYS_TO_VM_PAGE() since it now carries code to iterate
over registered ranges.
A hash container might be developed instead of range registration
interface, and fake pages could be put automatically into the hash,
were PHYS_TO_VM_PAGE() could look them up later. This should be
considered before the MFC of the commit is done.
Sponsored by: The FreeBSD Foundation
Reviewed by: alc
MFC after: 1 month
vm_page into new interface vm_page_initfake(). Handle the case of fake
page re-initialization with changed memattr.
Sponsored by: The FreeBSD Foundation
Reviewed by: alc
MFC after: 1 month
excluding other allocations including UMA now entails the addition of
a single flag to kmem_alloc or uma zone create
Reviewed by: alc, avg
MFC after: 2 weeks
Since the address of vm_page lock mutex depends on the kernel options,
it is easy for module to get out of sync with the kernel.
No vm_page_lockptr() accessor is provided for modules. It can be added
later if needed, unless proper KPI is developed to serve the needs.
Reviewed by: attilio, alc
MFC after: 3 weeks
yielding a new public interface, vm_page_alloc_contig(). This new function
addresses some of the limitations of the current interfaces, contigmalloc()
and kmem_alloc_contig(). For example, the physically contiguous memory that
is allocated with those interfaces can only be allocated to the kernel vm
object and must be mapped into the kernel virtual address space. It also
provides functionality that vm_phys_alloc_contig() doesn't, such as wiring
the returned pages. Moreover, unlike that function, it respects the low
water marks on the paging queues and wakes up the page daemon when
necessary. That said, at present, this new function can't be applied to all
types of vm objects. However, that restriction will be eliminated in the
coming weeks.
From a design standpoint, this change also addresses an inconsistency
between vm_phys_alloc_contig() and the other vm_phys_alloc*() functions.
Specifically, vm_phys_alloc_contig() manipulated vm_page fields that other
functions in vm/vm_phys.c didn't. Moreover, vm_phys_alloc_contig() knew
about vnodes and reservations. Now, vm_page_alloc_contig() is responsible
for these things.
Reviewed by: kib
Discussed with: jhb
Use the defined types instead of int when manipulating masks.
Supposedly, it could fix support for 32KB page size in the
machine-independend VM layer.
Reviewed by: alc
MFC after: 2 weeks
word to handle the dirty mask updates in vm_page_clear_dirty_mask().
Remove the vm page queue lock around vm_page_dirty() call in vm_fault_hold()
the sole purpose of which was to protect dirty on architectures which
does not provide short or byte-wide atomics.
Reviewed by: alc, attilio
Tested by: flo (sparc64)
MFC after: 2 weeks
flags field. Updates to the atomic flags are performed using the atomic
ops on the containing word, do not require any vm lock to be held, and
are non-blocking. The vm_page_aflag_set(9) and vm_page_aflag_clear(9)
functions are provided to modify afalgs.
Document the changes to flags field to only require the page lock.
Introduce vm_page_reference(9) function to provide a stable KPI and
KBI for filesystems like tmpfs and zfs which need to mark a page as
referenced.
Reviewed by: alc, attilio
Tested by: marius, flo (sparc64); andreast (powerpc, powerpc64)
Approved by: re (bz)
to VPO_UNMANAGED (and also making the flag protected by the vm object
lock, instead of vm page queue lock).
- Mark the fake pages with both PG_FICTITIOUS (as it is now) and
VPO_UNMANAGED. As a consequence, pmap code now can use use just
VPO_UNMANAGED to decide whether the page is unmanaged.
Reviewed by: alc
Tested by: pho (x86, previous version), marius (sparc64),
marcel (arm, ia64, powerpc), ray (mips)
Sponsored by: The FreeBSD Foundation
Approved by: re (bz)
(Saying that the lock on the object that the page belongs to must be held
only represents one aspect of the rules.)
Eliminate the use of the page queues lock for atomically performing read-
modify-write operations on the dirty field when the underlying architecture
supports atomic operations on char and short types.
Document the fact that 32KB pages aren't really supported.
Reviewed by: attilio, kib
vm_page_undirty(). The assert is not precise due to VPO_BUSY owner
to tracked, so assertion does not catch the case when VPO_BUSY is
owned by other thread.
Reviewed by: alc
The current implementation of vm_page_alloc_freelist() does not handle
order > 0 correctly. Remove order parameter to the function and use it
only for order 0 pages.
Submitted by: alc
alc@.
The UMA zone based allocation is replaced by a scheme that creates
a new free page list for the KSEG0 region, and a new function
in sys/vm that allocates pages from a specific free page list.
This also fixes a race condition introduced by the UMA based page table
page allocation code. Dropping the page queue and pmap locks before
the call to uma_zfree, and re-acquiring them afterwards will introduce
a race condtion(noted by alc@).
The changes are :
- Revert the earlier changes in MIPS pmap.c that added UMA zone for
page table pages.
- Add a new freelist VM_FREELIST_HIGHMEM to MIPS vmparam.h for memory that
is not directly mapped (in 32bit kernel). Normal page allocations will first
try the HIGHMEM freelist and then the default(direct mapped) freelist.
- Add a new function 'vm_page_t vm_page_alloc_freelist(int flind, int
order, int req)' to vm/vm_page.c to allocate a page from a specified
freelist. The MIPS page table pages will be allocated using this function
from the freelist containing direct mapped pages.
- Move the page initialization code from vm_phys_alloc_contig() to a
new function vm_page_alloc_init(), and use this function to initialize
pages in vm_page_alloc_freelist() too.
- Split the function vm_phys_alloc_pages(int pool, int order) to create
vm_phys_alloc_freelist_pages(int flind, int pool, int order), and use
this function from both vm_page_alloc_freelist() and vm_phys_alloc_pages().
Reviewed by: alc
the maintenance of vm_pageout_deficit can be localized to just two places:
vm_page_alloc() and vm_pageout_scan().
This change also corrects an off-by-one error in the maintenance of
vm_pageout_deficit. Historically, the buffer cache functions, allocbuf()
and vm_hold_load_pages(), have not taken into account that vm_page_alloc()
already increments vm_pageout_deficit by one.
Reviewed by: kib
flag is always provided, and unconditionally retry after sleep for the
busy page or failed allocation.
The intent is to remove VM_ALLOC_RETRY eventually.
Proposed and reviewed by: alc
specify the increment of vm_pageout_deficit when sleeping due to page
shortage. Then, in allocbuf(), the code to allocate pages when extending
vmio buffer can be replaced by a call to vm_page_grab().
Suggested and reviewed by: alc
MFC after: 2 weeks
is ordered by page index. This greatly simplifies the implementation,
since we no longer need to mark the pages with VPO_CLEANCHK to denote
the progress. It is enough to remember the current position by index
before dropping the object lock.
Remove VPO_CLEANCHK and VM_PAGER_IGNORE_CLEANCHK as unused.
Garbage-collect vm.msync_flush_flags sysctl.
Suggested and reviewed by: alc
Tested by: pho
vm_pageout_clean(). When iterating over a range of pages, these functions
can be cheaper than vm_page_lookup() because their implementation takes
advantage of the vm_object's memq being ordered.
Reviewed by: kib@
MFC after: 3 weeks
PG_REFERENCED changes in vm_pageout_object_deactivate_pages().
Simplify this function's inner loop using TAILQ_FOREACH(), and shorten
some of its overly long lines. Update a stale comment.
Assert that PG_REFERENCED may be cleared only if the object containing
the page is locked. Add a comment documenting this.
Assert that a caller to vm_page_requeue() holds the page queues lock,
and assert that the page is on a page queue.
Push down the page queues lock into pmap_ts_referenced() and
pmap_page_exists_quick(). (As of now, there are no longer any pmap
functions that expect to be called with the page queues lock held.)
Neither pmap_ts_referenced() nor pmap_page_exists_quick() should ever
be passed an unmanaged page. Assert this rather than returning "0"
and "FALSE" respectively.
ARM:
Simplify pmap_page_exists_quick() by switching to TAILQ_FOREACH().
Push down the page queues lock inside of pmap_clearbit(), simplifying
pmap_clear_modify(), pmap_clear_reference(), and pmap_remove_write().
Additionally, this allows for avoiding the acquisition of the page
queues lock in some cases.
PowerPC/AIM:
moea*_page_exits_quick() and moea*_page_wired_mappings() will never be
called before pmap initialization is complete. Therefore, the check
for moea_initialized can be eliminated.
Push down the page queues lock inside of moea*_clear_bit(),
simplifying moea*_clear_modify() and moea*_clear_reference().
The last parameter to moea*_clear_bit() is never used. Eliminate it.
PowerPC/BookE:
Simplify mmu_booke_page_exists_quick()'s control flow.
Reviewed by: kib@
an ordering dependence: A pmap operation that clears PG_WRITEABLE and calls
vm_page_dirty() must perform the call first. Otherwise, pmap_is_modified()
could return FALSE without acquiring the page queues lock because the page
is not (currently) writeable, and the caller to pmap_is_modified() might
believe that the page's dirty field is clear because it has not seen the
effect of the vm_page_dirty() call.
When I pushed down the page queues lock into pmap_is_modified(), I
overlooked one place where this ordering dependence is violated:
pmap_enter(). In a rare situation pmap_enter() can be called to replace a
dirty mapping to one page with a mapping to another page. (I say rare
because replacements generally occur as a result of a copy-on-write fault,
and so the old page is not dirty.) This change delays clearing PG_WRITEABLE
until after vm_page_dirty() has been called.
Fixing the ordering dependency also makes it easy to introduce a small
optimization: When pmap_enter() used to replace a mapping to one page with a
mapping to another page, it freed the pv entry for the first mapping and
later called the pv entry allocator for the new mapping. Now, pmap_enter()
attempts to recycle the old pv entry, saving two calls to the pv entry
allocator.
There is no point in setting PG_WRITEABLE on unmanaged pages, so don't.
Update a comment to reflect this.
Tidy up the variable declarations at the start of pmap_enter().
independent code. Move this code into mincore(), and eliminate the
page queues lock from pmap_mincore().
Push down the page queues lock into pmap_clear_modify(),
pmap_clear_reference(), and pmap_is_modified(). Assert that these
functions are never passed an unmanaged page.
Eliminate an inaccurate comment from powerpc/powerpc/mmu_if.m:
Contrary to what the comment says, pmap_mincore() is not simply an
optimization. Without a complete pmap_mincore() implementation,
mincore() cannot return either MINCORE_MODIFIED or MINCORE_REFERENCED
because only the pmap can provide this information.
Eliminate the page queues lock from vfs_setdirty_locked_object(),
vm_pageout_clean(), vm_object_page_collect_flush(), and
vm_object_page_clean(). Generally speaking, these are all accesses
to the page's dirty field, which are synchronized by the containing
vm object's lock.
Reduce the scope of the page queues lock in vm_object_madvise() and
vm_page_dontneed().
Reviewed by: kib (an earlier version)
here, make the style of assertion used by pmap_enter() consistent
across all architectures.
On entry to pmap_remove_write(), assert that the page is neither
unmanaged nor fictitious, since we cannot remove write access to
either kind of page.
With the push down of the page queues lock, pmap_remove_write() cannot
condition its behavior on the state of the PG_WRITEABLE flag if the
page is busy. Assert that the object containing the page is locked.
This allows us to know that the page will neither become busy nor will
PG_WRITEABLE be set on it while pmap_remove_write() is running.
Correct a long-standing bug in vm_page_cowsetup(). We cannot possibly
do copy-on-write-based zero-copy transmit on unmanaged or fictitious
pages, so don't even try. Previously, the call to pmap_remove_write()
would have failed silently.
pmap_page_is_mapped() in preparation for removing page queues locking
around calls to vm_page_free(). Setting aside the assertion that calls
pmap_page_is_mapped(), vm_page_free_toq() now acquires and holds the page
queues lock just long enough to actually add or remove the page from the
paging queues.
Update vm_page_unhold() to reflect the above change.
managed pages that didn't already have that lock held. (Freeing an
unmanaged page, such as the various pmaps use, doesn't require the page
lock.)
This allows a change in vm_page_remove()'s locking requirements. It now
expects the page lock to be held instead of the page queues lock.
Consequently, the page queues lock is no longer required at all by callers
to vm_page_rename().
Discussed with: kib
- re-assign page queue lock "Q"
- assign page lock "P"
- update several uncommented fields
- observe that hold_count is now protected by the page lock "P"
architecture from page queue lock to a hashed array of page locks
(based on a patch by Jeff Roberson), I've implemented page lock
support in the MI code and have only moved vm_page's hold_count
out from under page queue mutex to page lock. This changes
pmap_extract_and_hold on all pmaps.
Supported by: Bitgravity Inc.
Discussed with: alc, jeffr, and kib
following changes:
Rename vfs_page_set_valid() to vfs_page_set_validclean() to reflect
what this function actually does. Suggested by: tegge
Introduce a new version of vfs_page_set_valid() that does no more than
what the function's name implies. Specifically, it does not update
the page's dirty mask, and thus it does not require the page queues
lock to be held.
Update two of the three callers to the old vfs_page_set_valid() to
call vfs_page_set_validclean() instead because they actually require
the page's dirty mask to be cleared.
Introduce vm_page_set_valid().
Reviewed by: tegge
of the counter, that may happen when too many sendfile(2) calls are
being executed with this vnode [1].
To keep the size of the struct vm_page and offsets of the fields
accessed by out-of-tree modules, swap the types and locations
of the wire_count and cow fields. Add safety checks to detect cow
overflow and force fallback to the normal copy code for zero-copy
sockets. [2]
Reported by: Anton Yuzhaninov <citrin citrin ru> [1]
Suggested by: alc [2]
Reviewed by: alc
MFC after: 2 weeks
vm/vm_contig.c, vm/vm_page.c, and vm/vm_pageq.c. Today, vm/vm_pageq.c
has withered to the point that it contains only four short functions,
two of which are only used by vm/vm_page.c. Since I can't foresee any
reason for vm/vm_pageq.c to grow, it is time to fold the remaining
contents of vm/vm_pageq.c back into vm/vm_page.c.
Add some comments. Rename one of the functions, vm_pageq_enqueue(),
that is now static within vm/vm_page.c to vm_page_enqueue().
Eliminate PQ_MAXCOUNT as it no longer serves any purpose.
cache: vm_object_page_remove() should convert any cached pages that
fall with the specified range to free pages. Otherwise, there could
be a problem if a file is first truncated and then regrown.
Specifically, some old data from prior to the truncation might reappear.
Generalize vm_page_cache_free() to support the conversion of either a
subset or the entirety of an object's cached pages.
Reported by: tegge
Reviewed by: tegge
Approved by: re (kensmith)
ways:
(1) Cached pages are no longer kept in the object's resident page
splay tree and memq. Instead, they are kept in a separate per-object
splay tree of cached pages. However, access to this new per-object
splay tree is synchronized by the _free_ page queues lock, not to be
confused with the heavily contended page queues lock. Consequently, a
cached page can be reclaimed by vm_page_alloc(9) without acquiring the
object's lock or the page queues lock.
This solves a problem independently reported by tegge@ and Isilon.
Specifically, they observed the page daemon consuming a great deal of
CPU time because of pages bouncing back and forth between the cache
queue (PQ_CACHE) and the inactive queue (PQ_INACTIVE). The source of
this problem turned out to be a deadlock avoidance strategy employed
when selecting a cached page to reclaim in vm_page_select_cache().
However, the root cause was really that reclaiming a cached page
required the acquisition of an object lock while the page queues lock
was already held. Thus, this change addresses the problem at its
root, by eliminating the need to acquire the object's lock.
Moreover, keeping cached pages in the object's primary splay tree and
memq was, in effect, optimizing for the uncommon case. Cached pages
are reclaimed far, far more often than they are reactivated. Instead,
this change makes reclamation cheaper, especially in terms of
synchronization overhead, and reactivation more expensive, because
reactivated pages will have to be reentered into the object's primary
splay tree and memq.
(2) Cached pages are now stored alongside free pages in the physical
memory allocator's buddy queues, increasing the likelihood that large
allocations of contiguous physical memory (i.e., superpages) will
succeed.
Finally, as a result of this change long-standing restrictions on when
and where a cached page can be reclaimed and returned by
vm_page_alloc(9) are eliminated. Specifically, calls to
vm_page_alloc(9) specifying VM_ALLOC_INTERRUPT can now reclaim and
return a formerly cached page. Consequently, a call to malloc(9)
specifying M_NOWAIT is less likely to fail.
Discussed with: many over the course of the summer, including jeff@,
Justin Husted @ Isilon, peter@, tegge@
Tested by: an earlier version by kris@
Approved by: re (kensmith)
This allocator uses a binary buddy system with a twist. First and
foremost, this allocator is required to support the implementation of
superpages. As a side effect, it enables a more robust implementation
of contigmalloc(9). Moreover, this reimplementation of
contigmalloc(9) eliminates the acquisition of Giant by
contigmalloc(..., M_NOWAIT, ...).
The twist is that this allocator tries to reduce the number of TLB
misses incurred by accesses through a direct map to small, UMA-managed
objects and page table pages. Roughly speaking, the physical pages
that are allocated for such purposes are clustered together in the
physical address space. The performance benefits vary. In the most
extreme case, a uniprocessor kernel running on an Opteron, I measured
an 18% reduction in system time during a buildworld.
This allocator does not implement page coloring. The reason is that
superpages have much the same effect. The contiguous physical memory
allocation necessary for a superpage is inherently colored.
Finally, the one caveat is that this allocator does not effectively
support prezeroed pages. I hope this is temporary. On i386, this is
a slight pessimization. However, on amd64, the beneficial effects of
the direct-map optimization outweigh the ill effects. I speculate
that this is true in general of machines with a direct map.
Approved by: re
VM_PHYSSEG_SPARSE depending on whether the physical address space is
densely or sparsely populated with memory. The effect of this
definition is to determine which of two implementations of
vm_page_array and PHYS_TO_VM_PAGE() is used. The legacy
implementation is obtained by defining VM_PHYSSEG_DENSE, and a new
implementation that trades off time for space is obtained by defining
VM_PHYSSEG_SPARSE. For now, all architectures except for ia64 and
sparc64 define VM_PHYSSEG_DENSE. Defining VM_PHYSSEG_SPARSE on ia64
allows the entirety of my Itanium 2's memory to be used. Previously,
only the first 1 GB could be used. Defining VM_PHYSSEG_SPARSE on
sparc64 allows USIIIi-based systems to boot without crashing.
This change is a combination of Nathan Whitehorn's patch and my own
work in perforce.
Discussed with: kmacy, marius, Nathan Whitehorn
PR: 112194
immediately flag any page that is allocated to a OBJT_PHYS object as
unmanaged in vm_page_alloc() rather than waiting for a later call to
vm_page_unmanage(). This allows for the elimination of some uses of
the page queues lock.
Change the type of the kernel and kmem objects from OBJT_DEFAULT to
OBJT_PHYS. This allows us to take advantage of the above change to
simplify the allocation of unmanaged pages in kmem_alloc() and
kmem_malloc().
Remove vm_page_unmanage(). It is no longer used.
inlined and a procedure call is made in the rare case, i.e., when it is
necessary to sleep. In this case, inlining the test actually makes the
kernel smaller.
page queues-synchronized flag. Reduce the scope of the page queues lock in
vm_fault() accordingly.
Move vm_fault()'s call to vm_object_set_writeable_dirty() outside of the
scope of the page queues lock. Reviewed by: tegge
Additionally, eliminate an unnecessary dereference in computing the
argument that is passed to vm_object_set_writeable_dirty().
synchronized by the lock on the object containing the page.
Transition PG_WANTED and PG_SWAPINPROG to use the new field,
eliminating the need for holding the page queues lock when setting
or clearing these flags. Rename PG_WANTED and PG_SWAPINPROG to
VPO_WANTED and VPO_SWAPINPROG, respectively.
Eliminate the assertion that the page queues lock is held in
vm_page_io_finish().
Eliminate the acquisition and release of the page queues lock
around calls to vm_page_io_finish() in kern_sendfile() and
vfs_unbusy_pages().
- provide an interface (macros) to the page coloring part of the VM system,
this allows to try different coloring algorithms without the need to
touch every file [1]
- make the page queue tuning values readable: sysctl vm.stats.pagequeue
- autotuning of the page coloring values based upon the cache size instead
of options in the kernel config (disabling of the page coloring as a
kernel option is still possible)
MD changes:
- detection of the cache size: only IA32 and AMD64 (untested) contains
cache size detection code, every other arch just comes with a dummy
function (this results in the use of default values like it was the
case without the autotuning of the page coloring)
- print some more info on Intel CPU's (like we do on AMD and Transmeta
CPU's)
Note to AMD owners (IA32 and AMD64): please run "sysctl vm.stats.pagequeue"
and report if the cache* values are zero (= bug in the cache detection code)
or not.
Based upon work by: Chad David <davidc@acns.ab.ca> [1]
Reviewed by: alc, arch (in 2004)
Discussed with: alc, Chad David, arch (in 2004)
as opt_vmpage.h will not be available to user space library builds. A
similar existing check is present for KLD_MODULE for similar reasons.
MFC after: 3 days
that indicates that the caller does not want a page with its busy flag set.
In many places, the global page queues lock is acquired and released just
to clear the busy flag on a just allocated page. Both the allocation of
the page and the clearing of the busy flag occur while the containing vm
object is locked. So, the busy flag might as well never be set.
position that is 64-bit aligned and makes sure that the valid and
dirty fields are also 64-bit aligned. This means that if PAGE_SIZE
is 32K, the size of the vm_page structure is only increased by 8
bytes instead of 16 bytes. More importantly, the vm_page structure
is either 120 or 128 bytes on ia64. These are "interesting" sizes.
improved chance of working despite pressure from running programs.
Instead of trying to throw a bunch of pages out to swap and hope for
the best, only a range that can potentially fulfill contigmalloc(9)'s
request will have its contents paged out (potentially, not forcibly)
at a time.
The new contigmalloc operation still operates in three passes, but it
could potentially be tuned to more or less. The first pass only looks
at pages in the cache and free pages, so they would be thrown out
without having to block. If this is not enough, the subsequent passes
page out any unwired memory. To combat memory pressure refragmenting
the section of memory being laundered, each page is removed from the
systems' free memory queue once it has been freed so that blocking
later doesn't cause the memory laundered so far to get reallocated.
The page-out operations are now blocking, as it would make little sense
to try to push out a page, then get its status immediately afterward
to remove it from the available free pages queue, if it's unlikely to
have been freed. Another change is that if KVA allocation fails, the
allocated memory segment will be freed and not leaked.
There is a sysctl/tunable, defaulting to on, which causes the old
contigmalloc() algorithm to be used. Nonetheless, I have been using
vm.old_contigmalloc=0 for over a month. It is safe to switch at
run-time to see the difference it makes.
A new interface has been used which does not require mapping the
allocated pages into KVA: vm_page.h functions vm_page_alloc_contig()
and vm_page_release_contig(). These are what vm.old_contigmalloc=0
uses internally, so the sysctl/tunable does not affect their operation.
When using the contigmalloc(9) and contigfree(9) interfaces, memory
is now tracked with malloc(9) stats. Several functions have been
exported from kern_malloc.c to allow other subsystems to use these
statistics, as well. This invalidates the BUGS section of the
contigmalloc(9) manpage.
vm_pageout_scan(). Rationale: I don't like leaving a busy page in the
cache queue with neither the vm object nor the vm page queues lock held.
- Assert that the page is active in vm_pageout_page_stats().
pmap_copy_page() et al. to accept a vm_page_t rather than a physical
address. Also, this change will facilitate locking access to the vm page's
valid field.
in struct vm_page are defined as u_int for 16K pages and u_long
for 32K pages, with the implied assumption that long will at least
be 64 bits wide on platforms where we support 32K pages.
where physical addresses larger than virtual addresses, such as i386s
with PAE.
- Use this to represent physical addresses in the MI vm system and in the
i386 pmap code. This also changes the paddr parameter to d_mmap_t.
- Fix printf formats to handle physical addresses >4G in the i386 memory
detection code, and due to kvtop returning vm_paddr_t instead of u_long.
Note that this is a name change only; vm_paddr_t is still the same as
vm_offset_t on all currently supported platforms.
Sponsored by: DARPA, Network Associates Laboratories
Discussed with: re, phk (cdevsw change)
vm_page_alloc not to insert this page into an object. The pindex is
still used for colorization.
- Rework vm_page_select_* to accept a color instead of an object and
pindex to work with VM_PAGE_NOOBJ.
- Document other VM_ALLOC_ flags.
Reviewed by: peter, jake
be no major change in performance from this change at this time but this
will allow other work to progress: Giant lock removal around VM system
in favor of per-object mutexes, ranged fsyncs, more optimal COMMIT rpc's for
NFS, partial filesystem syncs by the syncer, more optimal object flushing,
etc. Note that the buffer cache is already using a similar splay tree
mechanism.
Note that a good chunk of the old hash table code is still in the tree.
Alan or I will remove it prior to the release if the new code does not
introduce unsolvable bugs, else we can revert more easily.
Submitted by: alc (this is Alan's code)
Approved by: re
- Remove all instances of the mallochash.
- Stash the slab pointer in the vm page's object pointer when allocating from
the kmem_obj.
- Use the overloaded object pointer to find slabs for malloced memory.
pmap_zero_page() and pmap_zero_page_area() were modified to accept
a struct vm_page * instead of a physical address, vm_page_zero_fill()
and vm_page_zero_fill_area() have served no purpose.
to return a wired page.
o Use VM_ALLOC_WIRED within Alpha's pmap_growkernel(). Also, because
Alpha's pmap_growkernel() calls vm_page_alloc() from within a critical
section, specify VM_ALLOC_INTERRUPT instead of VM_ALLOC_SYSTEM. (Only
VM_ALLOC_INTERRUPT is implemented entirely with a spin mutex.)
o Assert that the page queues mutex is held in vm_page_wire()
on Alpha, just like the other platforms.
o Assert that the page queues lock is held in vm_page_unwire().
o Make vm_page_lock_queues() and vm_page_unlock_queues() visible
to kernel loadable modules.
queue lock (revision 1.33 of vm/vm_page.c removed them).
o Make the free queue lock a spin lock because it's sometimes acquired
inside of a critical section.
MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes.
ti.4: Update the ti(4) man page to include information on the
TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options,
and also include information about the new character
device interface and the associated ioctls.
man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated
links.
jumbo.9: New man page describing the jumbo buffer allocator
interface and operation.
zero_copy.9: New man page describing the general characteristics of
the zero copy send and receive code, and what an
application author should do to take advantage of the
zero copy functionality.
NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS,
TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT.
conf/files: Add uipc_jumbo.c and uipc_cow.c.
conf/options: Add the 5 options mentioned above.
kern_subr.c: Receive side zero copy implementation. This takes
"disposable" pages attached to an mbuf, gives them to
a user process, and then recycles the user's page.
This is only active when ZERO_COPY_SOCKETS is turned on
and the kern.ipc.zero_copy.receive sysctl variable is
set to 1.
uipc_cow.c: Send side zero copy functions. Takes a page written
by the user and maps it copy on write and assigns it
kernel virtual address space. Removes copy on write
mapping once the buffer has been freed by the network
stack.
uipc_jumbo.c: Jumbo disposable page allocator code. This allocates
(optionally) disposable pages for network drivers that
want to give the user the option of doing zero copy
receive.
uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are
enabled if ZERO_COPY_SOCKETS is turned on.
Add zero copy send support to sosend() -- pages get
mapped into the kernel instead of getting copied if
they meet size and alignment restrictions.
uipc_syscalls.c:Un-staticize some of the sf* functions so that they
can be used elsewhere. (uipc_cow.c)
if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid
calling malloc() with M_WAITOK. Return an error if
the M_NOWAIT malloc fails.
The ti(4) driver and the wi(4) driver, at least, call
this with a mutex held. This causes witness warnings
for 'ifconfig -a' with a wi(4) or ti(4) board in the
system. (I've only verified for ti(4)).
ip_output.c: Fragment large datagrams so that each segment contains
a multiple of PAGE_SIZE amount of data plus headers.
This allows the receiver to potentially do page
flipping on receives.
if_ti.c: Add zero copy receive support to the ti(4) driver. If
TI_PRIVATE_JUMBOS is not defined, it now uses the
jumbo(9) buffer allocator for jumbo receive buffers.
Add a new character device interface for the ti(4)
driver for the new debugging interface. This allows
(a patched version of) gdb to talk to the Tigon board
and debug the firmware. There are also a few additional
debugging ioctls available through this interface.
Add header splitting support to the ti(4) driver.
Tweak some of the default interrupt coalescing
parameters to more useful defaults.
Add hooks for supporting transmit flow control, but
leave it turned off with a comment describing why it
is turned off.
if_tireg.h: Change the firmware rev to 12.4.11, since we're really
at 12.4.11 plus fixes from 12.4.13.
Add defines needed for debugging.
Remove the ti_stats structure, it is now defined in
sys/tiio.h.
ti_fw.h: 12.4.11 firmware.
ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13,
and my header splitting patches. Revision 12.4.13
doesn't handle 10/100 negotiation properly. (This
firmware is the same as what was in the tree previously,
with the addition of header splitting support.)
sys/jumbo.h: Jumbo buffer allocator interface.
sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to
indicate that the payload buffer can be thrown away /
flipped to a userland process.
socketvar.h: Add prototype for socow_setup.
tiio.h: ioctl interface to the character portion of the ti(4)
driver, plus associated structure/type definitions.
uio.h: Change prototype for uiomoveco() so that we'll know
whether the source page is disposable.
ufs_readwrite.c:Update for new prototype of uiomoveco().
vm_fault.c: In vm_fault(), check to see whether we need to do a page
based copy on write fault.
vm_object.c: Add a new function, vm_object_allocate_wait(). This
does the same thing that vm_object allocate does, except
that it gives the caller the opportunity to specify whether
it should wait on the uma_zalloc() of the object structre.
This allows vm objects to be allocated while holding a
mutex. (Without generating WITNESS warnings.)
vm_object_allocate() is implemented as a call to
vm_object_allocate_wait() with the malloc flag set to
M_WAITOK.
vm_object.h: Add prototype for vm_object_allocate_wait().
vm_page.c: Add page-based copy on write setup, clear and fault
routines.
vm_page.h: Add page based COW function prototypes and variable in
the vm_page structure.
Many thanks to Drew Gallatin, who wrote the zero copy send and receive
code, and to all the other folks who have tested and reviewed this code
over the years.
style(9)
- Minor space adjustment in cases where we have "( ", " )", if(), return(),
while(), for(), etc.
- Add /* SYMBOL */ after a few #endifs.
Reviewed by: alc
and again in vm_page.c and vm_pageq.c.
o Delete unusused prototypes. (Mainly a result of the earlier renaming
of various functions from vm_page_*() to vm_pageq_*().)
count that would otherwise be on one of the free queues. This eliminates a
panic when broken programs unmap memory that still has pending IO from raw
devices.
Reviewed by: dillon, alc
on and off since John Dyson left his work-in-progress.
It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on.
There are some hacks here to deal with the present lack of preemption - we
yield after doing a small number of pages since we wont preempt otherwise.
This is basically Matt's algorithm [with hysteresis] with an idle process
to call it in a similar way it used to be called from the idle loop.
I cleaned up the includes a fair bit here too.
Also removed some spl's and added some VM mutexes, but they are not actually
used yet, so this commit does not really make any operational changes
to the system.
vm_page.c relates to vm_page_t manipulation, including high level deactivation,
activation, etc... vm_pageq.c relates to finding free pages and aquiring
exclusive access to a page queue (exclusivity part not yet implemented).
And the world still builds... :-)
most of these inlines had been bloated in -current far beyond their
original intent. Normalize prototypes and function declarations to be ANSI
only (half already were). And do some general cleanup.
(kernel size also reduced by 50-100K, but that isn't the prime intent)
(this commit is just the first stage). Also add various GIANT_ macros to
formalize the removal of Giant, making it easy to test in a more piecemeal
fashion. These macros will allow us to test fine-grained locks to a degree
before removing Giant, and also after, and to remove Giant in a piecemeal
fashion via sysctl's on those subsystems which the authors believe can
operate without Giant.
Tor created a while ago, removes the raw I/O piece (that has cache coherency
problems), and adds a buffer cache / VM freeing piece.
Essentially this patch causes O_DIRECT I/O to not be left in the cache, but
does not prevent it from going through the cache, hence the 80%. For
the last 20% we need a method by which the I/O can be issued directly to
buffer supplied by the user process and bypass the buffer cache entirely,
but still maintain cache coherency.
I also have the code working under -stable but the changes made to sys/file.h
may not be MFCable, so an MFC is not on the table yet.
Submitted by: tegge, dillon
vm_mtx does not recurse and is required for most low level
vm operations.
faults can not be taken without holding Giant.
Memory subsystems can now call the base page allocators safely.
Almost all atomic ops were removed as they are covered under the
vm mutex.
Alpha and ia64 now need to catch up to i386's trap handlers.
FFS and NFS have been tested, other filesystems will need minor
changes (grabbing the vm lock when twiddling page properties).
Reviewed (partially) by: jake, jhb
in 4.2-REL which I ripped out in -stable and -current when implementing the
low-memory handling solution. However, maxlaunder turns out to be the saving
grace in certain very heavily loaded systems (e.g. newsreader box). The new
algorithm limits the number of pages laundered in the first pageout daemon
pass. If that is not sufficient then suceessive will be run without any
limit.
Write I/O is now pipelined using two sysctls, vfs.lorunningspace and
vfs.hirunningspace. This prevents excessive buffered writes in the
disk queues which cause long (multi-second) delays for reads. It leads
to more stable (less jerky) and generally faster I/O streaming to disk
by allowing required read ops (e.g. for indirect blocks and such) to occur
without interrupting the write stream, amoung other things.
NOTE: eventually, filesystem write I/O pipelining needs to be done on a
per-device basis. At the moment it is globalized.
Removed most of the hacks that were trying to deal with low-memory
situations prior to now.
The new code is based on the concept that I/O must be able to function in
a low memory situation. All major modules related to I/O (except
networking) have been adjusted to allow allocation out of the system
reserve memory pool. These modules now detect a low memory situation but
rather then block they instead continue to operate, then return resources
to the memory pool instead of cache them or leave them wired.
Code has been added to stall in a low-memory situation prior to a vnode
being locked.
Thus situations where a process blocks in a low-memory condition while
holding a locked vnode have been reduced to near nothing. Not only will
I/O continue to operate, but many prior deadlock conditions simply no
longer exist.
Implement a number of VFS/BIO fixes
(found by Ian): in biodone(), bogus-page replacement code, the loop
was not properly incrementing loop variables prior to a continue
statement. We do not believe this code can be hit anyway but we
aren't taking any chances. We'll turn the whole section into a
panic (as it already is in brelse()) after the release is rolled.
In biodone(), the foff calculation was incorrectly
clamped to the iosize, causing the wrong foff to be calculated
for pages in the case of an I/O error or biodone() called without
initiating I/O. The problem always caused a panic before. Now it
doesn't. The problem is mainly an issue with NFS.
Fixed casts for ~PAGE_MASK. This code worked properly before only
because the calculations use signed arithmatic. Better to properly
extend PAGE_MASK first before inverting it for the 64 bit masking
op.
In brelse(), the bogus_page fixup code was improperly throwing
away the original contents of 'm' when it did the j-loop to
fix the bogus pages. The result was that it would potentially
invalidate parts of the *WRONG* page(!), leading to corruption.
There may still be cases where a background bitmap write is
being duplicated, causing potential corruption. We have identified
a potentially serious bug related to this but the fix is still TBD.
So instead this patch contains a KASSERT to detect the problem
and panic the machine rather then continue to corrupt the filesystem.
The problem does not occur very often.. it is very hard to
reproduce, and it may or may not be the cause of the corruption
people have reported.
Review by: (VFS/BIO: mckusick, Ian Dowse <iedowse@maths.tcd.ie>)
Testing by: (VM/Deadlock) Paul Saab <ps@yahoo-inc.com>
set equal to the number of kilobytes in your cache. The old options are
still supported for backwards compatibility.
Submitted by: Kelly Yancey <kbyanc@posi.net>
