1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-24 11:29:10 +00:00
Commit Graph

5055 Commits

Author SHA1 Message Date
Alan Cox
b416fa1041 o Lock accesses to the page queues.
o Add a comment explaining why hoisting the page queue lock outside
   of a particular loop is not possible.
2002-07-13 04:09:45 +00:00
John Baldwin
03d7a9fffb - Change chroot_refuse_vdir_fds() to require that the passed in struct
filedesc is already locked rather than having chroot() unlock the
  filedesc so chroot_refuse_vdir_fds() can immediately relock it.
- Reorder chroot() a bitso that we do the namei lookup before checking
  the process's struct filedesc.  This closes at least one potential race
  and allows us to only acquire the filedsec lock once in chroot().
- Push down Giant slightly into chroot().
2002-07-13 04:07:12 +00:00
John Baldwin
63c9e754e0 We don't need to clear oldcred here since newcred is not NULL yet. 2002-07-13 03:13:15 +00:00
Alan Cox
ae0ffa73cc Lock accesses to the page queues by sendfile() and friends. 2002-07-13 03:10:55 +00:00
Matthew Dillon
fbcf77c2ea Re-enable the idle page-zeroing code. Remove all IPIs from the idle
page-zeroing code as well as from the general page-zeroing code and use a
lazy tlb page invalidation scheme based on a callback made at the end
of mi_switch.

A number of people came up with this idea at the same time so credit
belongs to Peter, John, and Jake as well.

Two-way SMP buildworld -j 5 tests (second run, after stabilization)
    2282.76 real  2515.17 user  704.22 sys	before peter's IPI commit
    2266.69 real  2467.50 user  633.77 sys	after peter's commit
    2232.80 real  2468.99 user  615.89 sys	after this commit

Reviewed by:	peter, jhb
Approved by:	peter
2002-07-12 20:17:06 +00:00
Julian Elischer
40e550266d also set the KSE state for the idle KSE/thread case. 2002-07-12 20:16:46 +00:00
John Baldwin
33d7ad1abe Set the thread state of the newly chosen to run thread to TDS_RUNNING in
choosethread() in MI C code instead of doing it in in assembly in all the
various cpu_switch() functions.  This fixes problems on ia64 and sparc64.

Reviewed by:	julian, peter, benno
Tested on:	i386, alpha, sparc64
2002-07-12 18:34:22 +00:00
Alan Cox
a4e80b6b64 Lock accesses to the page queues. 2002-07-12 17:21:22 +00:00
Thomas Moestl
5c85966098 Fix ptrace(PT_READ_*, ...) for non-little-endian architectures where
sizeof(register_t) != sizeof(int).
2002-07-12 16:48:05 +00:00
Peter Wemm
f1b665c8fe Revive backed out pmap related changes from Feb 2002. The highlights are:
- It actually works this time, honest!
- Fine grained TLB shootdowns for SMP on i386.  IPI's are very expensive,
  so try and optimize things where possible.
- Introduce ranged shootdowns that can be done as a single IPI.
- PG_G support for i386
- Specific-cpu targeted shootdowns.  For example, there is no sense in
  globally purging the TLB cache for where we are stealing a page from
  the local unshared process on the local cpu.  Use pm_active to track
  this.
- Add some instrumentation for the tlb shootdown code.
- Rip out SMP code from <machine/cpufunc.h>
- Try and fix some very bogus PG_G and PG_PS interactions that were bad
  enough to cause vm86 bios calls to break.  vm86 depended on our existing
  bugs and this was the cause of the VESA panics last time.
- Fix the silly one-line error that caused the 'panic: bad pte' last time.
- Fix a couple of other silly one-line errors that should have caused more
  pain than they did.

Some more work is needed:
- pmap_{zero,copy}_page[_idle].  These can be done without IPI's if we
  have a hook in cpu_switch.
- The IPI handlers need some cleanup.  I have a bogus %ds load that can
  be avoided.
- APTD handling is rather bogus and appears to be a large source of
  global TLB IPI shootdowns for no really good reason.

I see speedups of between 1.5% and ~4% on buildworlds in a while 1 loop.
I expect to see a bigger difference when there is significant pageout
activity or the system otherwise has memory shortages.

I have backed out a few optimizations that I had been using over the last
few days in order to be a little more conservative.  I'll revisit these
again over the next few days as the dust settles.

New option:  DISABLE_PG_G - In case I missed something.
2002-07-12 07:56:11 +00:00
Alfred Perlstein
d11a56617d regen for freebsd4_sendfile(2) compat. 2002-07-12 06:52:44 +00:00
Alfred Perlstein
9c34129662 Create a bug-for-bug FreeBSD4 compatible version of sendfile and move the
fixed sendfile over.  This is needed to preserve binary compatibility from
4.x to 5.x.
2002-07-12 06:51:57 +00:00
Alfred Perlstein
074453c230 Introduce syscall.master option 'COMPAT4' which allows one to wrap
syscalls for FreeBSD 4 compatibility.
Add kernel option COMPAT_FREEBSD4 to enable these syscalls.
2002-07-12 06:38:34 +00:00
Kenneth D. Merry
a720097d24 Fix compilation with ENABLE_VFS_IOOPT turned on and ZERO_COPY_SOCKETS
turned off.

Clean up #ifdefs, and remove a bunch of unnecessary includes.

Reviewed by:	bde
Tested by:	netchild
2002-07-12 02:23:55 +00:00
Julian Elischer
5e3da64ee9 Remove debugging code that I originally only wanted to be there for a couple of days after merge.
Reminded with pointy stick by: jhb
2002-07-11 22:47:58 +00:00
John Baldwin
eb80408cff Add a missing newline during panic printf's for SMP systems that don't
have APICS.  (Like all the !i386 archs).
2002-07-11 21:56:37 +00:00
Alan Cox
97646f567d o Lock accesses to the page queues. 2002-07-11 18:48:05 +00:00
Jonathan Mini
aaa1c7715b Revert removal of cred_free_thread(): It is used to ensure that a thread's
credentials are not improperly borrowed when the thread is not current in
the kernel.

Requested by:	jhb, alfred
2002-07-11 02:18:33 +00:00
Johan Karlsson
92da2e7671 Open accounting file for appending, not general writing.
This allows accton(1) to be used with an append-only file.

PR:		7169
Reported by:	Joao Carlos Mendes Luis <jonny@jonny.eng.br>
Reviewed by:	bde
Approved by:	sheldonh (mentor)
MFC after:	2 weeks
2002-07-10 17:31:58 +00:00
Matthew Dillon
d331c5d43f Replace the global buffer hash table with per-vnode splay trees using a
methodology similar to the vm_map_entry splay and the VM splay that Alan
Cox is working on.  Extensive testing has appeared to have shown no
increase in overhead.

Disadvantages
    Dirties more cache lines during lookups.

    Not as fast as a hash table lookup (but still N log N and optimal
    when there is locality of reference).

Advantages
    vnode->v_dirtyblkhd is now perfectly sorted, making fsync/sync/filesystem
    syncer operate more efficiently.

    I get to rip out all the old hacks (some of which were mine) that tried
    to keep the v_dirtyblkhd tailq sorted.

    The per-vnode splay tree should be easier to lock / SMPng pushdown on
    vnodes will be easier.

    This commit along with another that Alan is working on for the VM page
    global hash table will allow me to implement ranged fsync(), optimize
    server-side nfs commit rpcs, and implement partial syncs by the
    filesystem syncer (aka filesystem syncer would detect that someone is
    trying to get the vnode lock, remembers its place, and skip to the
    next vnode).

Note that the buffer cache splay is somewhat more complex then other splays
due to special handling of background bitmap writes (multiple buffers with
the same lblkno in the same vnode), and B_INVAL discontinuities between the
old hash table and the existence of the buffer on the v_cleanblkhd list.

Suggested by: alc
2002-07-10 17:02:32 +00:00
Julian Elischer
ad22735e3f Don't slow every syscall and trap by doing locks and stuff if the
'stop' bits are not set. This is a temporary thing.. I think this code probably
needs to be rewritten anyhow.
2002-07-10 06:40:22 +00:00
Don Lewis
832dafad3d Rearrange the code so that it checks whether the file is something
valid to write a core dump to before doing the preparations to actually
write to the file.

Call VOP_GETATTR() before dropping the initial vnode lock.
2002-07-10 06:31:35 +00:00
Maxime Henrion
fbedc80bd1 Remove vfs_stdmount() and vfs_stdunmount(). They are not
really useful and are incompatible with nmount.
2002-07-09 22:50:29 +00:00
Jeff Roberson
50bfcee1cb - Use the new vop_lookup_{pre,post} instead of simpler locking specification. 2002-07-09 19:55:06 +00:00
Jeff Roberson
25b286d6db - Use standard locking functions in syncer's opv
- vput instead of vrele syncer vnodes in vfs_mount
 - Add vop_lookup_{pre,post} to verify locking in VOP_LOOKUP
2002-07-09 19:54:20 +00:00
John Baldwin
54a033896d Fix a minor whitespace style nit that broke 'grep ^uuidgen'. 2002-07-09 19:36:50 +00:00
Maxime Henrion
43088e9868 Add a VFS_START() call in vfs_mountroot_try() for the sake
of being correct.  None of the root mountable filesystems
do something at VFS_START().

Shorten a comment to fix a style bug while I'm here.

PR:	kern/18505
2002-07-08 19:10:15 +00:00
Bruce Evans
f20c8dde15 Fixed some printf format errors (one new one reported by gcc and 3 nearby
old ones not reported by gcc).  This helps unbreak LINT.
2002-07-08 12:21:11 +00:00
Peter Wemm
a136efe9b6 Collect all the (now equivalent) pmap_new_proc/pmap_dispose_proc/
pmap_swapin_proc/pmap_swapout_proc functions from the MD pmap code
and use a single equivalent MI version.  There are other cleanups
needed still.

While here, use the UMA zone hooks to keep a cache of preinitialized
proc structures handy, just like the thread system does.  This eliminates
one dependency on 'struct proc' being persistent even after being freed.
There are some comments about things that can be factored out into
ctor/dtor functions if it is worth it.  For now they are mostly just
doing statistics to get a feel of how it is working.
2002-07-07 23:05:27 +00:00
Jeff Roberson
41a5470d03 - Require locks for getattr. At some point this could only require shared
locks.
2002-07-07 22:37:45 +00:00
Jeff Roberson
31965a72c9 - Delay unlocking a vnode in linker_hints_lookup until we're actually done
with it.
 - Remove a now stale comment about improper vnode locking.
2002-07-07 22:35:47 +00:00
Peter Wemm
b799f5a475 Make this compile on 64 bit platforms 2002-07-07 22:27:40 +00:00
Jeff Roberson
18c48f437f - Don't hold the vn lock while calling VOP_CLOSE in vclean(). 2002-07-07 06:38:22 +00:00
Jeff Roberson
bed75d4627 - BUF_REFCNT() seems to be the preferred method for verifying a locked buf.
Tell vop_strategy_pre() to use this instead.
 - Ignore B_CLUSTER bufs.  Their components are locked but they don't really
   exist so they don't have to be.  This isn't ideal but it is safe.
2002-07-07 05:29:45 +00:00
Jeff Roberson
49244e35ff Add two asserts that prove & document getblk and geteblk's behavior of
returning locked bufs.
2002-07-07 05:27:08 +00:00
Jeff Roberson
c031d11bb4 Fix a mistake in my last commit. Don't grab an extra reference to the object
in bp->b_object.
2002-07-06 21:27:20 +00:00
Jeff Roberson
9a236af3ad Fixup uses of GETVOBJECT.
- Cache a pointer to the vnode's object in the buf.
 - Hold a reference to that object in addition to the vnode's reference just
   to be consistent.
 - Cleanup code that got the object indirectly through the vp and VOP calls.

This fixes at least one case where we were calling GETVOBJECT without a lock.
It also avoids an expensive layered call at the cost of another pointer in
struct buf.
2002-07-06 08:59:52 +00:00
Julian Elischer
fe0f1bf4b1 make this repect ps_sigintr if there is a pre-existing signal
or suspension request.

Submitted by:	David Xu
2002-07-06 08:47:24 +00:00
Jeff Roberson
0b2ed1aef7 Clean up execve locking:
- Grab the vnode object early in exec when we still have the vnode lock.
 - Cache the object in the image_params.
 - Make use of the cached object in imgact_*.c
2002-07-06 07:00:01 +00:00
Jeff Roberson
e818064e98 - Disable original vop_strategy lock specification.
- Switch to the new vop_strategy_pre for lock validation.

VOP_STRATEGY requires only that the buf is locked UNLESS the block numbers need
to be translated.  There may be other reasons, but as long as the underlying
layer uses a VOP to perform the operations they will be caught later.
2002-07-06 05:23:17 +00:00
Jeff Roberson
302c7aaab9 - Add vop_strategy_pre to validate VOP_STRATEGY locking.
- Disable original vop_strategy lock specification.
 - Switch to the new vop_strategy_pre for lock validation.

VOP_STRATEGY requires only that the buf is locked UNLESS the block numbers need
to be translated.  There may be other reasons, but as long as the underlying
layer uses a VOP to perform the operations they will be caught later.
2002-07-06 05:21:12 +00:00
Jeff Roberson
13e407efee Use the new #! directive for vop_rename. Leave the old lock specification
intact but disabled.
2002-07-06 04:41:27 +00:00
Jeff Roberson
cc8662b0f9 Add "vop_rename_pre" to do pre rename lock verification. This is enabled only
with DEBUG_VFS_LOCKS.
2002-07-06 04:39:48 +00:00
Julian Elischer
55fb7ca894 Fix at least one of the things wrong with signals
^Z should work a lot better now.

Submitted by:	peter@freebsd.org
2002-07-06 02:45:11 +00:00
Andrew Gallatin
a83560d677 Remove the advertising clause from the Duke BSD copyright on the
zero-copy files

Requested by: rwatson
Approved by: Jeff Chase (my old boss at Duke)
2002-07-06 02:44:15 +00:00
Warner Losh
e0b7446484 dd %i as an alias for %d for greater compatibility with our *BSD bretheren
Obtained from: NetBSD
Reviewed by: jake, rwatson, bosko
2002-07-05 18:36:49 +00:00
Jeff Roberson
2efc89d4dc Include systm.h before vnode.h so Debugger() and printf() are available when
full vnode lock debugging is enabled.
2002-07-05 05:15:30 +00:00
Alan Cox
70c1763634 o Resurrect vm_page_lock_queues(), vm_page_unlock_queues(), and the free
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.
2002-07-04 22:07:37 +00:00
Maxime Henrion
d7f9ecc86b Move vfs_rootmountalloc() in vfs_mount.c and remove lite2_vfs_mountroot()
which was #if 0'd and is not likely to be used now.
2002-07-03 09:27:24 +00:00
Julian Elischer
aa0fa33464 Try clean up some of the mess that resulted from layers and layers
of p4 merges from -current as things started getting different.

Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00