Commit Graph

484 Commits

Author SHA1 Message Date
John Baldwin
35e0e5b311 Catch up to moving headers:
- machine/ipl.h -> sys/ipl.h
- machine/mutex.h -> sys/mutex.h
2000-10-20 07:58:15 +00:00
Doug Rabson
c703143638 Clear pcb_schednest in cpu_fork() for the child process. This is
is necessary since the child's call stack only includes one recursive
hold of sched_lock.
2000-10-03 08:03:03 +00:00
Jason Evans
0384fff8c5 Major update to the way synchronization is done in the kernel. Highlights
include:

* Mutual exclusion is used instead of spl*().  See mutex(9).  (Note: The
  alpha port is still in transition and currently uses both.)

* Per-CPU idle processes.

* Interrupts are run in their own separate kernel threads and can be
  preempted (i386 only).

Partially contributed by:	BSDi (BSD/OS)
Submissions by (at least):	cp, dfr, dillon, grog, jake, jhb, sheldonh
2000-09-07 01:33:02 +00:00
Andrew Gallatin
49c0f52e11 Support bounce buffers for ISA DMA on the alpha. This is required for the
irongate chipset (used in the UP1000) which does not support scatter/gather
DMA.  We'll still use scatter gather if the core logic chipset supports it.

Reviewed by: dfr
2000-06-19 18:41:27 +00:00
Alan Cox
6fba331424 cpu_fork(): Check "flags" before dereferencing "p2". Otherwise,
the call "vm_fork(p1, 0, flags);" early in fork1 can cause a kernel
panic.
2000-06-11 06:22:01 +00:00
Poul-Henning Kamp
9626b608de Separate the struct bio related stuff out of <sys/buf.h> into
<sys/bio.h>.

<sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall
not be made a nested include according to bdes teachings on the
subject of nested includes.

Diskdrivers and similar stuff below specfs::strategy() should no
longer need to include <sys/buf.> unless they need caching of data.

Still a few bogus uses of struct buf to track down.

Repocopy by:    peter
2000-05-05 09:59:14 +00:00
Poul-Henning Kamp
21144e3bf1 Remove B_READ, B_WRITE and B_FREEBUF and replace them with a new
field in struct buf: b_iocmd.  The b_iocmd is enforced to have
exactly one bit set.

B_WRITE was bogusly defined as zero giving rise to obvious coding
mistakes.

Also eliminate the redundant struct buf flag B_CALL, it can just
as efficiently be done by comparing b_iodone to NULL.

Should you get a panic or drop into the debugger, complaining about
"b_iocmd", don't continue.  It is likely to write on your disk
where it should have been reading.

This change is a step in the direction towards a stackable BIO capability.

A lot of this patch were machine generated (Thanks to style(9) compliance!)

Vinum users:  Greg has not had time to test this yet, be careful.
2000-03-20 10:44:49 +00:00
Andrew Gallatin
a6db6c48bf The kernel side of per-process unaligned access control (printing, fixing &
delivering SIGBUS).  This will allow a non-superuser to control unaligned
access behaviour on a per-process basis once a userland control program
(uac) is written.

Reviewed by: obrien
Tested by:   obrien
2000-01-16 07:07:33 +00:00
Peter Wemm
453c03ac1b Make this compile again. (missing #include for RFPROC) 1999-12-06 18:12:29 +00:00
Luoqi Chen
91c28bfde0 User ldt sharing. 1999-12-06 04:53:08 +00:00
Doug Rabson
62a50bfa38 Re-organise the code which manages the owner of the FP state (fpcurproc).
The old code was spread out through the machdep code and was sloppy about
enabling and disabling the FEN bit (which controls access to the FP
register set). This caused a DIAGNOSTIC warning "DANGER WILL ROBINSON:
FEN SET IN cpu_fork!" sometimes when operating under high loads and could
conceivably lead to processes getting incorrect FP results.

The new code is much more strict about the FEN bit and makes sure that
*only* fpcurproc ever has it enabled. This also allows us to remove a
section of code from the exception_return path which might improve
performance marginally.

Reviewed by: gallatin
1999-11-10 21:14:25 +00:00
Alan Cox
be72f78813 The core of this patch is to vm/vm_page.h. The effects are two-fold: (1) to
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
1999-10-30 07:37:14 +00:00
Poul-Henning Kamp
923502ff91 useracc() the prequel:
Merge the contents (less some trivial bordering the silly comments)
of <vm/vm_prot.h> and <vm/vm_inherit.h> into <vm/vm.h>.  This puts
the #defines for the vm_inherit_t and vm_prot_t types next to their
typedefs.

This paves the road for the commit to follow shortly: change
useracc() to use VM_PROT_{READ|WRITE} rather than B_{READ|WRITE}
as argument.
1999-10-29 18:09:36 +00:00
Matthew Dillon
4cc712004c Fix bug in pipe code relating to writes of mmap'd but illegal address
spaces which cross a segment boundry in the page table.  pmap_kextract()
    is not designed for access to the user space portion of the page
    table and cannot handle the null-page-directory-entry case.

    The fix is to have vm_fault_quick() return a success or failure which
    is then used to avoid calling pmap_kextract().
1999-09-20 19:08:48 +00:00
Peter Wemm
c3aac50f28 $Id$ -> $FreeBSD$ 1999-08-28 01:08:13 +00:00
Andrew Gallatin
34067f9475 Fix the child's return path from fork so that fork will return 0
in the child.  This corrects a problem where linux/alpha binaries see
the child's return value of fork as the parent's pid.  This happens because
linux/alpha binaries apparently check the return value directly, rather
than looking for a non-zero value in a4, as *BSD & OSF/1 do.

Reviewed by:dfr@nlsystems.com
1999-08-27 14:47:23 +00:00
John Polstra
0226f1b80d Sync with alc's revision 1.125 of i386/i386/vm_machdep.c. This
fixes the kernel build breakage.
1999-08-05 23:38:13 +00:00
Alan Cox
3b21348301 Reduce the number of "magic constants" used for page coloring
by one: PQ_PRIME2 and PQ_PRIME3 are used to accomplish the same
thing at different places in the kernel.  Drop PQ_PRIME3.
1999-07-22 06:04:17 +00:00
Peter Wemm
9c8b8baa38 Slight reorganization of kernel thread/process creation. Instead of using
SYSINIT_KT() etc (which is a static, compile-time procedure), use a
NetBSD-style kthread_create() interface.  kproc_start is still available
as a SYSINIT() hook.  This allowed simplification of chunks of the
sysinit code in the process.  This kthread_create() is our old kproc_start
internals, with the SYSINIT_KT fork hooks grafted in and tweaked to work
the same as the NetBSD one.

One thing I'd like to do shortly is get rid of nfsiod as a user initiated
process.  It makes sense for the nfs client code to create them on the
fly as needed up to a user settable limit.  This means that nfsiod
doesn't need to be in /sbin and is always "available".  This is a fair bit
easier to do outside of the SYSINIT_KT() framework.
1999-07-01 13:21:46 +00:00
Dmitrij Tejblum
23405ee478 Replace my previous fix of saving the FP state with a much simpler one: when
we swap out fpcurproc, save its FP state.

Suggested by:	bde
1999-06-10 20:40:59 +00:00
Dmitrij Tejblum
be960acd20 Keep fpcurproc locked in memory, so that we always can save the FP state
correctly.

This should fix the "pmap_changebit didn't" panic that some people see.

Reviewed by:	dfr
1999-06-08 16:42:19 +00:00
Dmitrij Tejblum
e415107779 Fixed several (not all) warnings. 1999-04-23 19:53:38 +00:00
Dmitrij Tejblum
f998a53420 Added consts to cpu_set_fork_handler prototype. (Follow i386 version.) 1999-04-20 22:53:54 +00:00
Peter Wemm
db42d90829 unifdef -DVM_STACK - it's been on for a while for x86 and was checked
and appeared to be working for the Alpha some time ago.
1999-04-19 14:14:14 +00:00
Matthew Dillon
dd2f9956e2 Adjust idle zero-page fill hysteresis based on tests. Use 2/3 and 4/5
zero-fill levels.

    Adjust comment for ozfod in vmmeter.h - this counter represents
    non-optimal ( on the fly ) zero fills, not prefills.
1999-02-08 02:42:13 +00:00
Matthew Dillon
d718be7be7 Add hysteresis to alpha version of vm_page_zero_idle(). 1999-02-08 00:47:32 +00:00
Matthew Dillon
faa273d5c2 Rip out PQ_ZERO queue. PQ_ZERO functionality is now combined in with
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.
1999-02-08 00:37:36 +00:00
Julian Elischer
2907af2a96 Mostly remove the VM_STACK OPTION.
This changes the definitions of a few items so that structures are the
same whether or not the option itself is enabled. This allows
people to enable and disable the option without recompilng the world.

As the author says:

|I ran into a problem pulling out the VM_STACK option.  I was aware of this
|when I first did the work, but then forgot about it.  The VM_STACK stuff
|has some code changes in the i386 branch.  There need to be corresponding
|changes in the alpha branch before it can come out completely.

what is done:
|
|1) Pull the VM_STACK option out of the header files it appears in.  This
|really shouldn't affect anything that executes with or without the rest
|of the VM_STACK patches.  The vm_map_entry will then always have one
|extra element (avail_ssize).  It just won't be used if the VM_STACK
|option is not turned on.
|
|I've also pulled the option out of vm_map.c.  This shouldn't harm anything,
|since the routines that are enabled as a result are not called unless
|the VM_STACK option is enabled elsewhere.
|
|2) Add what appears to be appropriate code the the alpha branch, still
|protected behind the VM_STACK switch.  I don't have an alpha machine,
|so we would need to get some testers with alpha machines to try it out.
|
|Once there is some testing, we can consider making the change permanent
|for both i386 and alpha.
|
[..]
|
|Once the alpha code is adequately tested, we can pull VM_STACK out
|everywhere.
|

Submitted by:	"Richard Seaman, Jr." <dick@tar.com>
1999-01-26 02:49:52 +00:00
Doug Rabson
9c0fed3dcf Various changes to support OSF1 emulation:
* Move the user stack from VM_MAXUSER_ADDRESS to a place below the 32bit
  boundary (needed to support 32bit OSF programs).  This should also save
  one pagetable per process.
* Add cvtqlsv to the set of instructions handled by the floating point
  software completion code.
* Disable all floating point exceptions by default.
* A minor change to execve to allow the OSF1 image activator to support
  dynamic loading.
1998-12-30 10:38:59 +00:00
Bruce Evans
4f2129fa86 Removed bogus casts of USRSTACK and/or the other operand in binary
expressions involving USRSTACK.
1998-12-16 15:21:51 +00:00
Doug Rabson
5b38fe900d Implement 'software completion' for floating point arithmetic. On the
alpha, operations involving non-finite numbers or denormalised numbers
or operations which should generate such numbers will cause an arithmetic
exception.  For programs which follow some strict code generation rules,
the kernel trap handler can then 'complete' the operation by emulating
the faulting instruction.

To use software completion, a program must be compiled with the arguments
'-mtrap-precision=i' and '-mfp-trap-mode=su' or '-mfp-trap-mode=sui'.
Programs compiled in this way can use non-finite and denormalised numbers
at the expense of slightly less efficient code generation of floating
point instructions.  Programs not compiled with these options will receive
a SIGFPE signal when non-finite or denormalised numbers are used or
generated.

Reviewed by: John Polstra <jdp@polstra.com>
1998-12-04 10:52:48 +00:00
Doug Rabson
e4b0567b9b Change a bogus cast to the correct one. 1998-10-15 09:53:27 +00:00
Doug Rabson
8e8c234f4d Don't bother calling pmap_emulate_reference() from cpu_fork(). It isn't
needed and it panics a DIAGNOSTIC kernel.
1998-07-12 16:30:58 +00:00
Doug Rabson
5c09fbfe64 Add missing copyrights. Thanks to Jason Thorpe for politely noting the
mistake...
1998-06-10 19:59:41 +00:00