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>
and sysv shared memory support for it. It implements a new
PG_UNMANAGED flag that has slightly different characteristics
from PG_FICTICIOUS.
A new sysctl, kern.ipc.shm_use_phys has been added to enable the
use of physically-backed sysv shared memory rather then swap-backed.
Physically backed shm segments are not tracked with PV entries,
allowing programs which use a large shm segment as a rendezvous
point to operate without eating an insane amount of KVM in the
PV entry management. Read: Oracle.
Peter's OBJT_PHYS object will also allow us to eventually implement
page-table sharing and/or 4MB physical page support for such segments.
We're half way there.
to various pmap_*() functions instead of looking up the physical address
and passing that. In many cases, the first thing the pmap code was doing
was going to a lot of trouble to get back the original vm_page_t, or
it's shadow pv_table entry.
Inspired by: John Dyson's 1998 patches.
Also:
Eliminate pv_table as a seperate thing and build it into a machine
dependent part of vm_page_t. This eliminates having a seperate set of
structions that shadow each other in a 1:1 fashion that we often went to
a lot of trouble to translate from one to the other. (see above)
This happens to save 4 bytes of physical memory for each page in the
system. (8 bytes on the Alpha).
Eliminate the use of the phys_avail[] array to determine if a page is
managed (ie: it has pv_entries etc). Store this information in a flag.
Things like device_pager set it because they create vm_page_t's on the
fly that do not have pv_entries. This makes it easier to "unmanage" a
page of physical memory (this will be taken advantage of in subsequent
commits).
Add a function to add a new page to the freelist. This could be used
for reclaiming the previously wasted pages left over from preloaded
loader(8) files.
Reviewed by: dillon
is an application space macro and the applications are supposed to be free
to use it as they please (but cannot). This is consistant with the other
BSD's who made this change quite some time ago. More commits to come.
madvise().
This feature prevents the update daemon from gratuitously flushing
dirty pages associated with a mapped file-backed region of memory. The
system pager will still page the memory as necessary and the VM system
will still be fully coherent with the filesystem. Modifications made
by other means to the same area of memory, for example by write(), are
unaffected. The feature works on a page-granularity basis.
MAP_NOSYNC allows one to use mmap() to share memory between processes
without incuring any significant filesystem overhead, putting it in
the same performance category as SysV Shared memory and anonymous memory.
Reviewed by: julian, alc, dg
eliminate an extra (useless) level of indirection in half of the page
queue accesses and (2) to use a single name for each queue throughout,
instead of, e.g., "vm_page_queue_active" in some places and
"vm_page_queues[PQ_ACTIVE]" in others.
Reviewed by: dillon
Replace various VM related page count calculations strewn over the
VM code with inlines to aid in readability and to reduce fragility
in the code where modules depend on the same test being performed
to properly sleep and wakeup.
Split out a portion of the page deactivation code into an inline
in vm_page.c to support vm_page_dontneed().
add vm_page_dontneed(), which handles the madvise MADV_DONTNEED
feature in a related commit coming up for vm_map.c/vm_object.c. This
code prevents degenerate cases where an essentially active page may
be rotated through a subset of the paging lists, resulting in premature
disposal.
This setting is also acceptable for Celerons and Pentium Pros
with less than 1MB L2 caches.
Note: PQ_L2_SIZE is a misnomer. The correct number of colors is
a function of the cache's degree of associativity as well as its size.
Submitted by: bde and alc
piecemeal, middle-of-file writes for NFS. These hacks have caused no
end of trouble, especially when combined with mmap(). I've removed
them. Instead, NFS will issue a read-before-write to fully
instantiate the struct buf containing the write. NFS does, however,
optimize piecemeal appends to files. For most common file operations,
you will not notice the difference. The sole remaining fragment in
the VFS/BIO system is b_dirtyoff/end, which NFS uses to avoid cache
coherency issues with read-merge-write style operations. NFS also
optimizes the write-covers-entire-buffer case by avoiding the
read-before-write. There is quite a bit of room for further
optimization in these areas.
The VM system marks pages fully-valid (AKA vm_page_t->valid =
VM_PAGE_BITS_ALL) in several places, most noteably in vm_fault. This
is not correct operation. The vm_pager_get_pages() code is now
responsible for marking VM pages all-valid. A number of VM helper
routines have been added to aid in zeroing-out the invalid portions of
a VM page prior to the page being marked all-valid. This operation is
necessary to properly support mmap(). The zeroing occurs most often
when dealing with file-EOF situations. Several bugs have been fixed
in the NFS subsystem, including bits handling file and directory EOF
situations and buf->b_flags consistancy issues relating to clearing
B_ERROR & B_INVAL, and handling B_DONE.
getblk() and allocbuf() have been rewritten. B_CACHE operation is now
formally defined in comments and more straightforward in
implementation. B_CACHE for VMIO buffers is based on the validity of
the backing store. B_CACHE for non-VMIO buffers is based simply on
whether the buffer is B_INVAL or not (B_CACHE set if B_INVAL clear,
and vise-versa). biodone() is now responsible for setting B_CACHE
when a successful read completes. B_CACHE is also set when a bdwrite()
is initiated and when a bwrite() is initiated. VFS VOP_BWRITE
routines (there are only two - nfs_bwrite() and bwrite()) are now
expected to set B_CACHE. This means that bowrite() and bawrite() also
set B_CACHE indirectly.
There are a number of places in the code which were previously using
buf->b_bufsize (which is DEV_BSIZE aligned) when they should have
been using buf->b_bcount. These have been fixed. getblk() now clears
B_DONE on return because the rest of the system is so bad about
dealing with B_DONE.
Major fixes to NFS/TCP have been made. A server-side bug could cause
requests to be lost by the server due to nfs_realign() overwriting
other rpc's in the same TCP mbuf chain. The server's kernel must be
recompiled to get the benefit of the fixes.
Submitted by: Matthew Dillon <dillon@apollo.backplane.com>
unallocated parts of the last page when the file ended on a frag
but not a page boundary.
Delimitted by tags PRE_MATT_MMAP_EOF and POST_MATT_MMAP_EOF,
in files alpha/alpha/pmap.c i386/i386/pmap.c nfs/nfs_bio.c vm/pmap.h
vm/vm_page.c vm/vm_page.h vm/vnode_pager.c miscfs/specfs/spec_vnops.c
ufs/ufs/ufs_readwrite.c kern/vfs_bio.c
Submitted by: Matt Dillon <dillon@freebsd.org>
Reviewed by: Alan Cox <alc@freebsd.org>
been made but the code has been reorganized and documented to make
it more readable, reduce the size of the code, and optimize the branch
path caching capabilities that most modern processors have.
PQ_FREE. There is little operational difference other then the kernel
being a few kilobytes smaller and the code being more readable.
* vm_page_select_free() has been *greatly* simplified.
* The PQ_ZERO page queue and supporting structures have been removed
* vm_page_zero_idle() revamped (see below)
PG_ZERO setting and clearing has been migrated from vm_page_alloc()
to vm_page_free[_zero]() and will eventually be guarenteed to remain
tracked throughout a page's life ( if it isn't already ).
When a page is freed, PG_ZERO pages are appended to the appropriate
tailq in the PQ_FREE queue while non-PG_ZERO pages are prepended.
When locating a new free page, PG_ZERO selection operates from within
vm_page_list_find() ( get page from end of queue instead of beginning
of queue ) and then only occurs in the nominal critical path case. If
the nominal case misses, both normal and zero-page allocation devolves
into the same _vm_page_list_find() select code without any specific
zero-page optimizations.
Additionally, vm_page_zero_idle() has been revamped. Hysteresis has been
added and zero-page tracking adjusted to conform with the other changes.
Currently hysteresis is set at 1/3 (lo) and 1/2 (hi) the number of free
pages. We may wish to increase both parameters as time permits. The
hysteresis is designed to avoid silly zeroing in borderline allocation/free
situations.
reducing the size of vm_page_t.
SWAPBLK_NONE and SWAPBLK_MASK are defined here. These actually are
more generalized then their names imply, but their placement is somewhat
of a legacy issue from a prior test version of this code that put
the swapblk in the vm_page_t structure. That test code was eventually
thrown away. The legacy remains.
Added vm_page_flash() inline. Similar to vm_page_wakeup() except that
it does not clear PG_BUSY ( one assumes that PG_BUSY is already clear ).
Used by a number of routines to wakeup waiters.
Collapsed some of the code in inline calls to make other inline calls.
GCC will optimize this well and it reduces duplication.
vm_page_free() and vm_page_free_zero() inlines added to convert to
the proper vm_page_free_toq() call.
vm_page_sleep_busy() inline added, replacing vm_page_sleep() ( which has
been removed ). This implements a much more optimizable page-waiting
function.
changes to the VM system to support the new swapper, VM bug
fixes, several VM optimizations, and some additional revamping of the
VM code. The specific bug fixes will be documented with additional
forced commits. This commit is somewhat rough in regards to code
cleanup issues.
Reviewed by: "John S. Dyson" <root@dyson.iquest.net>, "David Greenman" <dg@root.com>
Add some overflow checks to read/write (from bde).
Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags
and vm_object::paging_in_progress to use operations which are not
interruptable.
Reviewed by: Bruce Evans <bde@zeta.org.au>
FreeBSD/alpha. The most significant item is to change the command
argument to ioctl functions from int to u_long. This change brings us
inline with various other BSD versions. Driver writers may like to
use (__FreeBSD_version == 300003) to detect this change.
The prototype FreeBSD/alpha machdep will follow in a couple of days
time.
has been some bitrot and incorrect assumptions in the vfs_bio code. These
problems have manifest themselves worse on NFS type filesystems, but can
still affect local filesystems under certain circumstances. Most of
the problems have involved mmap consistancy, and as a side-effect broke
the vfs.ioopt code. This code might have been committed seperately, but
almost everything is interrelated.
1) Allow (pmap_object_init_pt) prefaulting of buffer-busy pages that
are fully valid.
2) Rather than deactivating erroneously read initial (header) pages in
kern_exec, we now free them.
3) Fix the rundown of non-VMIO buffers that are in an inconsistent
(missing vp) state.
4) Fix the disassociation of pages from buffers in brelse. The previous
code had rotted and was faulty in a couple of important circumstances.
5) Remove a gratuitious buffer wakeup in vfs_vmio_release.
6) Remove a crufty and currently unused cluster mechanism for VBLK
files in vfs_bio_awrite. When the code is functional, I'll add back
a cleaner version.
7) The page busy count wakeups assocated with the buffer cache usage were
incorrectly cleaned up in a previous commit by me. Revert to the
original, correct version, but with a cleaner implementation.
8) The cluster read code now tries to keep data associated with buffers
more aggressively (without breaking the heuristics) when it is presumed
that the read data (buffers) will be soon needed.
9) Change to filesystem lockmgr locks so that they use LK_NOPAUSE. The
delay loop waiting is not useful for filesystem locks, due to the
length of the time intervals.
10) Correct and clean-up spec_getpages.
11) Implement a fully functional nfs_getpages, nfs_putpages.
12) Fix nfs_write so that modifications are coherent with the NFS data on
the server disk (at least as well as NFS seems to allow.)
13) Properly support MS_INVALIDATE on NFS.
14) Properly pass down MS_INVALIDATE to lower levels of the VM code from
vm_map_clean.
15) Better support the notion of pages being busy but valid, so that
fewer in-transit waits occur. (use p->busy more for pageouts instead
of PG_BUSY.) Since the page is fully valid, it is still usable for
reads.
16) It is possible (in error) for cached pages to be busy. Make the
page allocation code handle that case correctly. (It should probably
be a printf or panic, but I want the system to handle coding errors
robustly. I'll probably add a printf.)
17) Correct the design and usage of vm_page_sleep. It didn't handle
consistancy problems very well, so make the design a little less
lofty. After vm_page_sleep, if it ever blocked, it is still important
to relookup the page (if the object generation count changed), and
verify it's status (always.)
18) In vm_pageout.c, vm_pageout_clean had rotted, so clean that up.
19) Push the page busy for writes and VM_PROT_READ into vm_pageout_flush.
20) Fix vm_pager_put_pages and it's descendents to support an int flag
instead of a boolean, so that we can pass down the invalidate bit.
2) Do not unnecessarily force page blocking when paging
pages out.
3) Further improve swap pager performance and correctness,
including fixing the paging in progress deadlock (except
in severe I/O error conditions.)
4) Enable vfs_ioopt=1 as a default.
5) Fix and enable the page prezeroing in SMP mode.
All in all, SMP systems especially should show a significant
improvement in "snappyness."
of the various ad-hoc schemes.
2) When bringing in UPAGES, the pmap code needs to do another vm_page_lookup.
3) When appropriate, set the PG_A or PG_M bits a-priori to both avoid some
processor errata, and to minimize redundant processor updating of page
tables.
4) Modify pmap_protect so that it can only remove permissions (as it
originally supported.) The additional capability is not needed.
5) Streamline read-only to read-write page mappings.
6) For pmap_copy_page, don't enable write mapping for source page.
7) Correct and clean-up pmap_incore.
8) Cluster initial kern_exec pagin.
9) Removal of some minor lint from kern_malloc.
10) Correct some ioopt code.
11) Remove some dead code from the MI swapout routine.
12) Correct vm_object_deallocate (to remove backing_object ref.)
13) Fix dead object handling, that had problems under heavy memory load.
14) Add minor vm_page_lookup improvements.
15) Some pages are not in objects, and make sure that the vm_page.c can
properly support such pages.
16) Add some more page deficit handling.
17) Some minor code readability improvements.
This will make a number of things easier in the future, as well as (finally!)
avoiding the Id-smashing problem which has plagued developers for so long.
Boy, I'm glad we're not using sup anymore. This update would have been
insane otherwise.
64K. The change has essentially neutral effect on those machines with
little or no cache, and has a positive effect on "normal" machines
with 256K or more cache.
The default level works with minimal overhead, but one can also enable
full, efficient use of a 512K cache. (Parameters can be generated
to support arbitrary cache sizes also.)
problem with the 'shell scripts' was found, but there was a 'strange'
problem found with a 486 laptop that we could not find. This commit
backs the code back to 25-jul, and will be re-entered after the snapshot
in smaller (more easily tested) chunks.
performance issues.
1) The pmap module has had too many inlines, and so the
object file is simply bigger than it needs to be.
Some common code is also merged into subroutines.
2) Removal of some *evil* PHYS_TO_VM_PAGE macro calls.
Unfortunately, a few have needed to be added also.
The removal caused the need for more vm_page_lookups.
I added lookup hints to minimize the need for the
page table lookup operations.
3) Removal of some bogus performance improvements, that
mostly made the code more complex (tracking individual
page table page updates unnecessarily). Those improvements
actually hurt 386 processors perf (not that people who
worry about perf use 386 processors anymore :-)).
4) Changed pv queue manipulations/structures to be TAILQ's.
5) The pv queue code has had some performance problems since
day one. Some significant scalability issues are resolved
by threading the pv entries from the pmap AND the physical
address instead of just the physical address. This makes
certain pmap operations run much faster. This does
not affect most micro-benchmarks, but should help loaded system
performance *significantly*. DG helped and came up with most
of the solution for this one.
6) Most if not all pmap bit operations follow the pattern:
pmap_test_bit();
pmap_clear_bit();
That made for twice the necessary pv list traversal. The
pmap interface now supports only pmap_tc_bit type operations:
pmap_[test/clear]_modified, pmap_[test/clear]_referenced.
Additionally, the modified routine now takes a vm_page_t arg
instead of a phys address. This eliminates a PHYS_TO_VM_PAGE
operation.
7) Several rewrites of routines that contain redundant code to
use common routines, so that there is a greater likelihood of
keeping the cache footprint smaller.
Re-enables the RSS limiting, and the routine is now tail-recursive,
making it much more safe (eliminates the possiblity of kernel stack
overflow.) Also, the RSS limiting is a little more intelligent about
finding the likely objects that are pushing the process over the limit.
Added some sysctls that help with VM system tuning.
New sysctl features:
1) Enable/disable lru pageout algorithm.
vm.pageout_algorithm = 0, default algorithm that works
well, especially using X windows and heavy
memory loading. Can have adverse effects,
sometimes slowing down program loading.
vm.pageout_algorithm = 1, close to true LRU. Works much
better than clock, etc. Does not work as well as
the default algorithm in general. Certain memory
"malloc" type benchmarks work a little better with
this setting.
Please give me feedback on the performance results
associated with these.
2) Enable/disable swapping.
vm.swapping_enabled = 1, default.
vm.swapping_enabled = 0, useful for cases where swapping
degrades performance.
The config option "NO_SWAPPING" is still operative, and
takes precedence over the sysctl. If "NO_SWAPPING" is
specified, the sysctl still exists, but "vm.swapping_enabled"
is hard-wired to "0".
Each of these can be changed "on the fly."
queue in vm_fault.
Move the PG_BUSY in vm_fault to the correct place.
Remove redundant/unnecessary code in pmap.c.
Properly block on rundown of page table pages, if they are busy.
I think that the VM system is in pretty good shape now, and the following
individuals (among others, in no particular order) have helped with this
recent bunch of bugs, thanks! If I left anyone out, I apologize!
Stephen McKay, Stephen Hocking, Eric J. Chet, Dan O'Brien, James Raynard,
Marc Fournier.
some problems with the page-table page management code, since it can't
deal with the notion of page-table pages being paged out or in transit.
Also, clean up some stylistic issues per some suggestions from
Stephen McKay.
Speed up for vfs_bio -- addition of a routine bqrelse to greatly diminish
overhead for merged cache.
Efficiency improvement for vfs_cluster. It used to do alot of redundant
calls to cluster_rbuild.
Correct the ordering for vrele of .text and release of credentials.
Use the selective tlb update for 486/586/P6.
Numerous fixes to the size of objects allocated for files. Additionally,
fixes in the various pagers.
Fixes for proper positioning of vnode_pager_setsize in msdosfs and ext2fs.
Fixes in the swap pager for exhausted resources. The pageout code
will not as readily thrash.
Change the page queue flags (PG_ACTIVE, PG_INACTIVE, PG_FREE, PG_CACHE) into
page queue indices (PQ_ACTIVE, PQ_INACTIVE, PQ_FREE, PQ_CACHE),
thereby improving efficiency of several routines.
Eliminate even more unnecessary vm_page_protect operations.
Significantly speed up process forks.
Make vm_object_page_clean more efficient, thereby eliminating the pause
that happens every 30seconds.
Make sequential clustered writes B_ASYNC instead of B_DELWRI even in the
case of filesystems mounted async.
Fix a panic with busy pages when write clustering is done for non-VMIO
buffers.
proc or any VM system structure will have to be rebuilt!!!
Much needed overhaul of the VM system. Included in this first round of
changes:
1) Improved pager interfaces: init, alloc, dealloc, getpages, putpages,
haspage, and sync operations are supported. The haspage interface now
provides information about clusterability. All pager routines now take
struct vm_object's instead of "pagers".
2) Improved data structures. In the previous paradigm, there is constant
confusion caused by pagers being both a data structure ("allocate a
pager") and a collection of routines. The idea of a pager structure has
escentially been eliminated. Objects now have types, and this type is
used to index the appropriate pager. In most cases, items in the pager
structure were duplicated in the object data structure and thus were
unnecessary. In the few cases that remained, a un_pager structure union
was created in the object to contain these items.
3) Because of the cleanup of #1 & #2, a lot of unnecessary layering can now
be removed. For instance, vm_object_enter(), vm_object_lookup(),
vm_object_remove(), and the associated object hash list were some of the
things that were removed.
4) simple_lock's removed. Discussion with several people reveals that the
SMP locking primitives used in the VM system aren't likely the mechanism
that we'll be adopting. Even if it were, the locking that was in the code
was very inadequate and would have to be mostly re-done anyway. The
locking in a uni-processor kernel was a no-op but went a long way toward
making the code difficult to read and debug.
5) Places that attempted to kludge-up the fact that we don't have kernel
thread support have been fixed to reflect the reality that we are really
dealing with processes, not threads. The VM system didn't have complete
thread support, so the comments and mis-named routines were just wrong.
We now use tsleep and wakeup directly in the lock routines, for instance.
6) Where appropriate, the pagers have been improved, especially in the
pager_alloc routines. Most of the pager_allocs have been rewritten and
are now faster and easier to maintain.
7) The pagedaemon pageout clustering algorithm has been rewritten and
now tries harder to output an even number of pages before and after
the requested page. This is sort of the reverse of the ideal pagein
algorithm and should provide better overall performance.
8) Unnecessary (incorrect) casts to caddr_t in calls to tsleep & wakeup
have been removed. Some other unnecessary casts have also been removed.
9) Some almost useless debugging code removed.
10) Terminology of shadow objects vs. backing objects straightened out.
The fact that the vm_object data structure escentially had this
backwards really confused things. The use of "shadow" and "backing
object" throughout the code is now internally consistent and correct
in the Mach terminology.
11) Several minor bug fixes, including one in the vm daemon that caused
0 RSS objects to not get purged as intended.
12) A "default pager" has now been created which cleans up the transition
of objects to the "swap" type. The previous checks throughout the code
for swp->pg_data != NULL were really ugly. This change also provides
the rudiments for future backing of "anonymous" memory by something
other than the swap pager (via the vnode pager, for example), and it
allows the decision about which of these pagers to use to be made
dynamically (although will need some additional decision code to do
this, of course).
13) (dyson) MAP_COPY has been deprecated and the corresponding "copy
object" code has been removed. MAP_COPY was undocumented and non-
standard. It was furthermore broken in several ways which caused its
behavior to degrade to MAP_PRIVATE. Binaries that use MAP_COPY will
continue to work correctly, but via the slightly different semantics
of MAP_PRIVATE.
14) (dyson) Sharing maps have been removed. It's marginal usefulness in a
threads design can be worked around in other ways. Both #12 and #13
were done to simplify the code and improve readability and maintain-
ability. (As were most all of these changes)
TODO:
1) Rewrite most of the vnode pager to use VOP_GETPAGES/PUTPAGES. Doing
this will reduce the vnode pager to a mere fraction of its current size.
2) Rewrite vm_fault and the swap/vnode pagers to use the clustering
information provided by the new haspage pager interface. This will
substantially reduce the overhead by eliminating a large number of
VOP_BMAP() calls. The VOP_BMAP() filesystem interface should be
improved to provide both a "behind" and "ahead" indication of
contiguousness.
3) Implement the extended features of pager_haspage in swap_pager_haspage().
It currently just says 0 pages ahead/behind.
4) Re-implement the swap device (swstrategy) in a more elegant way, perhaps
via a much more general mechanism that could also be used for disk
striping of regular filesystems.
5) Do something to improve the architecture of vm_object_collapse(). The
fact that it makes calls into the swap pager and knows too much about
how the swap pager operates really bothers me. It also doesn't allow
for collapsing of non-swap pager objects ("unnamed" objects backed by
other pagers).
Headers should always use `__inline' for inline functions to avoid
syntax errors when modules that don't even use the offending functions
are compiled with `gcc -ansi'.
New functions create - vm_object_pip_wakeup and pagedaemon_wakeup that
are used to reduce the actual number of wakeups.
New function vm_page_protect which is used in conjuction with some new
page flags to reduce the number of calls to pmap_page_protect.
Minor changes to reduce unnecessary spl nesting.
Rewrote vm_page_alloc() to improve readability.
Various other mostly cosmetic changes.
Added hook for pmap_prefault() and use symbolic constant for new third
argument to vm_page_alloc() (vm_fault.c, various)
Changed the way that upages and page tables are held. (vm_glue.c)
Fixed architectural flaw in allocating pages at interrupt time that was
introduced with the merged cache changes. (vm_page.c, various)
Adjusted some algorithms to acheive better paging performance and to
accomodate the fix for the architectural flaw mentioned above. (vm_pageout.c)
Fixed pbuf handling problem, changed policy on handling read-behind page.
(vnode_pager.c)
Submitted by: John Dyson
much higher filesystem I/O performance, and much better paging performance. It
represents the culmination of over 6 months of R&D.
The majority of the merged VM/cache work is by John Dyson.
The following highlights the most significant changes. Additionally, there are
(mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to
support the new VM/buffer scheme.
vfs_bio.c:
Significant rewrite of most of vfs_bio to support the merged VM buffer cache
scheme. The scheme is almost fully compatible with the old filesystem
interface. Significant improvement in the number of opportunities for write
clustering.
vfs_cluster.c, vfs_subr.c
Upgrade and performance enhancements in vfs layer code to support merged
VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff.
vm_object.c:
Yet more improvements in the collapse code. Elimination of some windows that
can cause list corruption.
vm_pageout.c:
Fixed it, it really works better now. Somehow in 2.0, some "enhancements"
broke the code. This code has been reworked from the ground-up.
vm_fault.c, vm_page.c, pmap.c, vm_object.c
Support for small-block filesystems with merged VM/buffer cache scheme.
pmap.c vm_map.c
Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of
kernel PTs.
vm_glue.c
Much simpler and more effective swapping code. No more gratuitous swapping.
proc.h
Fixed the problem that the p_lock flag was not being cleared on a fork.
swap_pager.c, vnode_pager.c
Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the
code doesn't need it anymore.
machdep.c
Changes to better support the parameter values for the merged VM/buffer cache
scheme.
machdep.c, kern_exec.c, vm_glue.c
Implemented a seperate submap for temporary exec string space and another one
to contain process upages. This eliminates all map fragmentation problems
that previously existed.
ffs_inode.c, ufs_inode.c, ufs_readwrite.c
Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on
busy buffers.
Submitted by: John Dyson and David Greenman
Disable the bogus declaration of pmap_bootstrap(). Since its arg list
is machine-dependent, it must be declared in a machine-dependent header.
vm_page.h:
Change `inline' to `__inline' and old-style function parameter lists for
inlined functions to new-style.
`inline' and old-style function parameter lists should never be used in
system headers, even in very machine-dependent ones, because they cause
warnings from gcc -Wreally-all.
