(usually a couple of thousand) to 25. The measured impact on cache-hits
doesn't justify spending memory this way:
Target number of free vnodes versus namecache hit rate in % during a
make world:
10 98.5316
200 98.5479
500 98.5546
1000 98.5709
3000 98.6006
4000 98.6126
Introduce VFREE which indicates that vnode is on freelist.
Rename vholdrele() to vdrop().
Create vfree() and vbusy() to add/delete vnode from freelist.
Add vfree()/vbusy() to keep (v_holdcnt != 0 || v_usecount != 0)
vnodes off the freelist.
Generalize vhold()/v_holdcnt to mean "do not recycle".
Fix reassignbuf()s lack of use of vhold().
Use vhold() instead of checking v_cache_src list.
Remove vtouch(), the vnodes are always vget'ed soon enough
after for it to have any measuable effect.
Add sysctl debug.freevnodes to keep track of things.
Move cache_purge() up in getnewvnodes to avoid race.
Decrement v_usecount after VOP_INACTIVE(), put a vhold() on
it during VOP_INACTIVE()
Unmacroize vhold()/vdrop()
Print out VDOOMED and VFREE flags (XXX: should use %b)
Reviewed by: dyson
free list problem. Also, the vnode age flag is no longer used by the
vnode pager. (It is actually incorrect to use then.) Constructive
feedback welcome -- just be kind.
socket addresses in mbufs. (Socket buffers are the one exception.) A number
of kernel APIs needed to get fixed in order to make this happen. Also,
fix three protocol families which kept PCBs in mbufs to not malloc them
instead. Delete some old compatibility cruft while we're at it, and add
some new routines in the in_cksum family.
enough and can cause some strange performance problems. Specifically, at
or near startup time is when the problem is worst. To reproduce
the problem, run "lat_syscall stat" from the alpha lmbench code right
after bootup. A positive side effect of this mod is that the name
cache can be set to grow again by sysctl. A noticable positive
performance impact is realized due to a larger namecache being available
as needed (or tuned.)
flag wasn't being respected during vref(), et. al. Note that this
isn't the eventual fix for the locking problem. Fine grained SMP
in the VM and VFS code will require (lots) more work.
".." vnode. This is cheaper storagewise than keeping it in the
namecache, and it makes more sense since it's a 1:1 mapping.
2. Also handle the case of "." more intelligently rather than stuff
the namecache with pointless entries.
3. Add two lists to the vnode and hang namecache entries which go from
or to this vnode. When cleaning a vnode, delete all namecache
entries it invalidates.
4. Never reuse namecache enties, malloc new ones when we need it, free
old ones when they die. No longer a hard limit on how many we can
have.
5. Remove the upper limit on namelength of namecache entries.
6. Make a global list for negative namecache entries, limit their number
to a sysctl'able (debug.ncnegfactor) fraction of the total namecache.
Currently the default fraction is 1/16th. (Suggestions for better
default wanted!)
7. Assign v_id correctly in the face of 32bit rollover.
8. Remove the LRU list for namecache entries, not needed. Remove the
#ifdef NCH_STATISTICS stuff, it's not needed either.
9. Use the vnode freelist as a true LRU list, also for namecache accesses.
10. Reuse vnodes more aggresively but also more selectively, if we can't
reuse, malloc a new one. There is no longer a hard limit on their
number, they grow to the point where we don't reuse potentially
usable vnodes. A vnode will not get recycled if still has pages in
core or if it is the source of namecache entries (Yes, this does
indeed work :-) "." and ".." are not namecache entries any longer...)
11. Do not overload the v_id field in namecache entries with whiteout
information, use a char sized flags field instead, so we can get
rid of the vpid and v_id fields from the namecache struct. Since
we're linked to the vnodes and purged when they're cleaned, we don't
have to check the v_id any more.
12. NFS knew about the limitation on name length in the namecache, it
shouldn't and doesn't now.
Bugs:
The namecache statistics no longer includes the hits for ".."
and "." hits.
Performance impact:
Generally in the +/- 0.5% for "normal" workstations, but
I hope this will allow the system to be selftuning over a
bigger range of "special" applications. The case where
RAM is available but unused for cache because we don't have
any vnodes should be gone.
Future work:
Straighten out the namecache statistics.
"desiredvnodes" is still used to (bogusly ?) size hash
tables in the filesystems.
I have still to find a way to safely free unused vnodes
back so their number can shrink when not needed.
There is a few uses of the v_id field left in the filesystems,
scheduled for demolition at a later time.
Maybe a one slot cache for unused namecache entries should
be implemented to decrease the malloc/free frequency.
implementation #ifdef out. This can be used for now by NFS. As soon
as all the other filesystems' locking is fixed, this can go away.
Print the vnode address in vprint for easier debugging.
(phk's) sysctl framework, and I needed special code to disambiguate
the VFS_GENERIC node from the VFS_VFSCONF leaf, so I only converted
the leaves to the FreeBSD framework. The error handling isn't quite
right. CSRGS's sysctls seem to return ENOTDIR too much and FreeBSD's
sysctls don't agree with the man page.
and getvfsbyname() interfaces. The new interfaces are now hidden from
applications unless _NEW_VFSCONF is defined. The new vfsconf interfaces
don't work yet.
- getnewvnode() and vref() were missing one simple_unlock() each.
- the Lite2 locking changes weren't merged at all in
printlockedvnodes() or sysctl_vnode(). Merging these undid
some KNF style regressions.
changes, so don't expect to be able to run the kernel as-is (very well)
without the appropriate Lite/2 userland changes.
The system boots and can mount UFS filesystems.
Untested: ext2fs, msdosfs, NFS
Known problems: Incorrect Berkeley ID strings in some files.
Mount_std mounts will not work until the getfsent
library routine is changed.
Reviewed by: various people
Submitted by: Jeffery Hsu <hsu@freebsd.org>
variable `kern.maxvnodes' which gives much better control over vnode
allocation than EXTRAVNODES (except in -current between 1995/10/28 and
1996/11/12, kern.maxvnodes was read-only and thus useless).
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.
Firstly, now our read-ahead clustering is on a file descriptor basis and not
on a per-vnode basis. This will allow multiple processes reading the
same file to take advantage of read-ahead clustering. Secondly, there
previously was a problem with large reads still using the ramp-up
algorithm. Of course, that was bogus, and now we read the entire
"chunk" off of the disk in one operation. The read-ahead clustering
algorithm should use less CPU than the previous also (I hope :-)).
NOTE: THAT LKMS MUST BE REBUILT!!!
add free vnodes back to the freelist. They must do their own vnode
management. Anyway, this change is *only* activated with their filesystem
and doesn't affect anyone else. Whoops, forgot the submitted-by lines
in my previous commits too.. :-(
Submitted-By: Tony Ardolino <tony@netcon.com>
/*
* Structure defined by POSIX.4 to be like a timeval.
*/
struct timespec {
time_t ts_sec; /* seconds */
long ts_nsec; /* and nanoseconds */
};
The correct names of the fields are tv_sec and tv_nsec.
Reminded by: James Drobina <jdrobina@infinet.com>
The interface into the "VMIO" system has changed to be more consistant
and robust. Essentially, it is now no longer necessary to call vn_open
to get merged VM/Buffer cache operation, and exceptional conditions
such as merged operation of VBLK devices is simpler and more correct.
This code corrects a potentially large set of problems including the
problems with ktrace output and loaded systems, file create/deletes,
etc.
Most of the changes to NFS are cosmetic and name changes, eliminating
a layer of subroutine calls. The direct calls to vput/vrele have
been re-instituted for better cross platform compatibility.
Reviewed by: davidg
to be allocated at boot time. This is an expensive option, as they
consume physical ram and are not pageable etc. In certain situations,
this kind of option is quite useful, especially for news servers that
access a large number of directories at random and torture the name cache.
Defining 5000 or 10000 extra vnodes should cut down the amount of vnode
recycling somewhat, which should allow better name and directory caching
etc.
This is a "your mileage may vary" option, with no real indication of
what works best for your machine except trial and error. Too many will
cost you ram that you could otherwise use for disk buffers etc.
This is based on something John Dyson mentioned to me a while ago.
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.
Unstaticize a function in scsi/scsi_base that was used, with an undocumented
option.
My last count on the LINT kernel shows:
Total symbols: 3647
unref symbols: 463
undef symbols: 4
1 ref symbols: 1751
2 ref symbols: 485
Approaching the pain threshold now.
Convert the remaining sysctl stuff to the new way of doing things.
the devconf stuff is the reason for the large number of files.
Cleaned up some compiler warnings while I were there.
it 1138 times (:-() in casts and a few more times in declarations.
This change is null for the i386.
The type has to be `typedef int vop_t(void *)' and not `typedef
int vop_t()' because `gcc -Wstrict-prototypes' warns about the
latter. Since vnode op functions are called with args of different
(struct pointer) types, neither of these function types is any use
for type checking of the arg, so it would be preferable not to use
the complete function type, especially since using the complete
type requires adding 1138 casts to avoid compiler warnings and
another 40+ casts to reverse the function pointer conversions before
calling the functions.
PR 795.
Set the size before one error return from sysctl_vnode() the same as before
the other. The caller might want to know about the amount successfully
read although the current caller doesn't.
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).
2) Removed unnecessary vm_object_lookup()/pager_cache(object, TRUE) pairs
after vnode_pager_alloc() calls - the object is already guaranteed to be
persistent.
3) Removed some gratuitous casts.
VOP_CLOSE() takes `F' (file) flags, not `IO' flags. At least that's
what close() passes. I previously fixed ttylclose() to check
FNONBLOCK instead of IO_NDELAY. This broke the call from vclean()
and cleaning of ptys sometimes deadlocked.
1) Files weren't properly synced on filesystems other than UFS. In some
cases, this lead to lost data. Most likely would be noticed on NFS.
The fix is to make the VM page sync/object_clean general rather than
in each filesystem.
2) Mixing regular and mmaped file I/O on NFS was very broken. It caused
chunks of files to end up as zeroes rather than the intended contents.
The fix was to fix several race conditions and to kludge up the
"b_dirtyoff" and "b_dirtyend" that NFS relies upon - paying attention
to page modifications that occurred via the mmapping.
Reviewed by: David Greenman
Submitted by: John Dyson
is more representative of worst case situations of 4 files/directory. (If
that last sentence doesn't make any sense, I'm not surprised. It's rather
compilcated how this all fits together....).
This should fix a problem that Ed Hudson has been complaining about where
directories with lots of symlinks could cause excessive disk I/O.
Fixed remaining known bugs in the buffer IO and VM system.
vfs_bio.c:
Fixed some race conditions and locking bugs. Improved performance
by removing some (now) unnecessary code and fixing some broken
logic.
Fixed process accounting of # of FS outputs.
Properly handle NFS interrupts (B_EINTR).
(various)
Replaced calls to clrbuf() with calls to an optimized routine
call vfs_bio_clrbuf().
(various FS sync)
Sync out modified vnode_pager backed pages.
ffs_vnops.c:
Do two passes: Sync out file data first, then indirect blocks.
vm_fault.c:
Fixed deadly embrace caused by acquiring locks in the wrong order.
vnode_pager.c:
Changed to use buffer I/O system for writing out modified pages. This
should fix the problem with the modification date previous not getting
updated. Also dramatically simplifies the code. Note that this is
going to change in the future and be implemented via VOP_PUTPAGES().
vm_object.c:
Fixed a pile of bugs related to cleaning (vnode) objects. The performance
of vm_object_page_clean() is terrible when dealing with huge objects,
but this will change when we implement a binary tree to keep the object
pages sorted.
vm_pageout.c:
Fixed broken clustering of pageouts. Fixed race conditions and other
lockup style bugs in the scanning of pages. Improved performance.
1. We always keep one 16th of the vnodes on the freelist, so that the
namecache doesn't get trashed. It used to be that it wasn't a problem, but
the only vnodes getting released these days are directories and things which
Clean up and improve the namecache.
1. We always keep one 16th of the vnodes on the freelist, so that the
namecache doesn't get trashed. It used to be that it wasn't a problem, but
the only vnodes getting released these days are directories and things which
gets forced out of the VM/cache. The latter is not numerous enough to keep
the pool of vnodes needed for the namecache sufficiently big.
2. Purge invalid entries in the namecache as soon as we notice them. This
avoids a stale entry pushing out a valid entry on the LRU list.
3. Speed up the lookup in the namecache by avoid a special case branch.
4. Make the cache purge routines do the thing they're supposed to, and in
a decently efficient manner.
5. Make the size of the namecache follow the number of vnodes, so that we
can always point to all the vnodes we have in core.
6. Readability has gone way up.
7. Added a "options NCH_STATISTICS" feature that will gather more
detailed statistics on the performance of the namecache.
Reviewed by: davidg
(cvs is dumping core on me :-( )
may not properly initialize this field in all cases, and this would
result in very anti-social behavior (overwriting on some other random
device/location).
Submitted by: John Dyson
Various more tweaks from John Dyson to improve read ahead calculations.
vfs_subr.c:
Only wakeup if numoutput is 0 in vwakeup().
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
cycles. While waiting there I added a lot of the extra ()'s I have, (I have
never used LISP to any extent). So I compiled the kernel with -Wall and
shut up a lot of "suggest you add ()'s", removed a bunch of unused var's
and added a couple of declarations here and there. Having a lap-top is
highly recommended. My kernel still runs, yell at me if you kernel breaks.
machdep.c:
Changed printf's a little and call vfs_unmountall() if the sync was
successful.
cd9660_vfsops.c, ffs_vfsops.c, nfs_vfsops.c, lfs_vfsops.c:
Allow dismount of root FS. It is now disallowed at a higher level.
vfs_conf.c:
Removed unused rootfs global.
vfs_subr.c:
Added new routines vfs_unmountall and vfs_unmountroot. Filesystems
are now dismounted if the machine is properly rebooted.
ffs_vfsops.c:
Toggle clean bit at the appropriate places. Print warning if an
unclean FS is mounted.
ffs_vfsops.c, lfs_vfsops.c:
Fix bug in selecting proper flags for VOP_CLOSE().
vfs_syscalls.c:
Disallow dismounting root FS via umount syscall.
- Delete redundant declarations.
- Add -Wredundant-declarations to Makefile.i386 so they don't come back.
- Delete sloppy COMMON-style declarations of uninitialized data in
header files.
- Add a few prototypes.
- Clean up warnings resulting from the above.
NB: ioconf.c will still generate a redundant-declaration warning, which
is unavoidable unless somebody volunteers to make `config' smarter.
