Commit Graph

529 Commits

Author SHA1 Message Date
John Baldwin
c86b6ff551 Change the preemption code for software interrupt thread schedules and
mutex releases to not require flags for the cases when preemption is
not allowed:

The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent
switching to a higher priority thread on mutex releease and swi schedule,
respectively when that switch is not safe.  Now that the critical section
API maintains a per-thread nesting count, the kernel can easily check
whether or not it should switch without relying on flags from the
programmer.  This fixes a few bugs in that all current callers of
swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from
fast interrupt handlers and the swi_sched of softclock needed this flag.
Note that to ensure that swi_sched()'s in clock and fast interrupt
handlers do not switch, these handlers have to be explicitly wrapped
in critical_enter/exit pairs.  Presently, just wrapping the handlers is
sufficient, but in the future with the fully preemptive kernel, the
interrupt must be EOI'd before critical_exit() is called.  (critical_exit()
can switch due to a deferred preemption in a fully preemptive kernel.)

I've tested the changes to the interrupt code on i386 and alpha.  I have
not tested ia64, but the interrupt code is almost identical to the alpha
code, so I expect it will work fine.  PowerPC and ARM do not yet have
interrupt code in the tree so they shouldn't be broken.  Sparc64 is
broken, but that's been ok'd by jake and tmm who will be fixing the
interrupt code for sparc64 shortly.

Reviewed by:	peter
Tested on:	i386, alpha
2002-01-05 08:47:13 +00:00
Thomas Moestl
01f1aed259 Use the new resource_list_print_type() function.
Pass the bus device to isa_init() (this is needed for the sparc64
version).
2001-12-21 21:54:56 +00:00
John Baldwin
98f9879242 Introduce a standard name for the lock protecting an interrupt controller
and it's associated state variables: icu_lock with the name "icu".  This
renames the imen_mtx for x86 SMP, but also uses the lock to protect
access to the 8259 PIC on x86 UP.  This also adds an appropriate lock to
the various Alpha chipsets which fixes problems with Alpha SMP machines
dropping interrupts with an SMP kernel.
2001-12-20 23:48:31 +00:00
John Baldwin
7235f2b1e9 Axe stale extern for a non-existent variable. 2001-12-18 22:42:09 +00:00
John Baldwin
7e1f6dfe9d Modify the critical section API as follows:
- The MD functions critical_enter/exit are renamed to start with a cpu_
  prefix.
- MI wrapper functions critical_enter/exit maintain a per-thread nesting
  count and a per-thread critical section saved state set when entering
  a critical section while at nesting level 0 and restored when exiting
  to nesting level 0.  This moves the saved state out of spin mutexes so
  that interlocking spin mutexes works properly.
- Most low-level MD code that used critical_enter/exit now use
  cpu_critical_enter/exit.  MI code such as device drivers and spin
  mutexes use the MI wrappers.  Note that since the MI wrappers store
  the state in the current thread, they do not have any return values or
  arguments.
- mtx_intr_enable() is replaced with a constant CRITICAL_FORK which is
  assigned to curthread->td_savecrit during fork_exit().

Tested on:	i386, alpha
2001-12-18 00:27:18 +00:00
John Baldwin
0bbc882680 Overhaul the per-CPU support a bit:
- The MI portions of struct globaldata have been consolidated into a MI
  struct pcpu.  The MD per-CPU data are specified via a macro defined in
  machine/pcpu.h.  A macro was chosen over a struct mdpcpu so that the
  interface would be cleaner (PCPU_GET(my_md_field) vs.
  PCPU_GET(md.md_my_md_field)).
- All references to globaldata are changed to pcpu instead.  In a UP kernel,
  this data was stored as global variables which is where the original name
  came from.  In an SMP world this data is per-CPU and ideally private to each
  CPU outside of the context of debuggers.  This also included combining
  machine/globaldata.h and machine/globals.h into machine/pcpu.h.
- The pointer to the thread using the FPU on i386 was renamed from
  npxthread to fpcurthread to be identical with other architectures.
- Make the show pcpu ddb command MI with a MD callout to display MD
  fields.
- The globaldata_register() function was renamed to pcpu_init() and now
  init's MI fields of a struct pcpu in addition to registering it with
  the internal array and list.
- A pcpu_destroy() function was added to remove a struct pcpu from the
  internal array and list.

Tested on:	alpha, i386
Reviewed by:	peter, jake
2001-12-11 23:33:44 +00:00
Mitsuru IWASAKI
f9390180fe Some fix for the recent apm module changes.
- Now that apm loadable module can inform its existence to other kernel
   components  (e.g. i386/isa/clock.c:startrtclock()'s TCS hack).
 - Exchange priority of SI_SUB_CPU and SI_SUB_KLD for above purpose.
 - Add simple arbitration mechanism for APM vs. ACPI.  This prevents
   the kernel enables both of them.
 - Remove obsolete `#ifdef DEV_APM' related code.
 - Add abstracted interface for Powermanagement operations.  Public apm(4)
   functions, such as apm_suspend(), should be replaced new interfaces.
   Currently only power_pm_suspend (successor of apm_suspend) is implemented.

Reviewed by:	peter, arch@ and audit@
2001-11-01 16:34:07 +00:00
Luigi Rizzo
c5ca4c7e6e Backout 1.61 -- both intrcnt and intrnames are already exported
via sysctl under "hw".
2001-10-25 01:32:24 +00:00
Bruce Evans
08b00f49c3 MFi386:
- sys/pc98/pc98/npx.c 1.87 (2001/09/15; author: imp)
  I don't think pc98 has acpi at all, so ifdef the acpi attachments for
  now.

This completes merging sys/pc98/pc98/npx.c into sys/i386/isa/npx.c so
that the former can be removed.
2001-10-21 06:05:08 +00:00
Bruce Evans
abfde38316 MFpc98: fundamental differences. The magic numbers for the i/o port
and the irq are different for pc98, and are not very well handled (we
use a historical mess of hard-coded values, values from header files
and values from hints).
2001-10-21 05:56:03 +00:00
Bruce Evans
40d8c8da95 MFpc98: all changes in sys/pc98/pc98/npx.c related to FPU_ERROR_BROKEN.
- 1.58 (2000/09/01; author: kato)
  Fixed FPU_ERROR_BROKEN code.  It had old-isa code.
- 1.33 (1998/03/09; author: kato)
  Make FPU_ERROR_BROKEN a new-style option.
- 1.7 (1996/10/09; author: asami)
  Make sure FPU is recognized for non-Intel CPUs.

The log for rev.1.7 should have said something like:
Added FPU_ERROR_BROKEN option.  This forces a successful probe for
exception 16, so that hardware with a broken FPU error signal can sort
of work.
2001-10-21 05:18:30 +00:00
Bruce Evans
265e95d904 Deleted most of npxprobe(), and merged npxprobe1() back into npxprobe().
Use the normal interrupt handler (npx_intr()) instead of a special
probe-time interrupt handler, although this causes problems due to
the bus_teardown_intr() not actually even tearing down the interrupt
(these problems were avoided by doing interrupt attachment for the
special interrupt handler directly).  Fixed minor bitrot in comments.

The reason for the npxprobe()/npxprobe1() split mostly went away at
about the same time it was made (in 1992 or 1993 just before the
beginning of history).  386BSD ran all probes with interrupts completely
masked, and I didn't want to disturb this when I added an irq probe
to npxprobe().  An irq (not necessarily npx) must be acked for at least
external npx's to take the cpu out of the wait state that it enters
when an npx error occurs, so the probe must be done with a suitable
irq unmasked.  npxprobe() went to great lengths to unmask precisely
the npx irq.

Running probes with all interrupts masked was never really needed in
FreeBSD, since FreeBSD always masked interrupts well enough using
splhigh(), but it wasn't until rev.1.48 (1995/12/12) of autoconf.c
that all probes were run with CPU interrupts enabled.  This permits
npxprobe() to probe its irq using normal interrupt resources.  Note
that most drivers still can't depend on this.  It depends on the
interrupt handler being fast and the irq not being shared.
2001-10-16 14:12:35 +00:00
Bruce Evans
2504f76272 Commit my old fixes for cosmetic bugs in npxprobe() so that they aren't
lost when the buggy code goes away completely:
- don't assume that the npx irq number is >= 8.  Rev.1.73 only reversed
  part of the hard-coding of it to 13 in rev.1.66.
- backed out the part of rev.1.84 that added a highly confused comment
  about an enable_intr() being "highly bogus".  The whole reason for
  existence of npxprobe() (separate from the main probe, npxprobe1())
  is to handle the complications to make this enable_intr() safe.
- backed out the part of rev.1.94 that modified npxprobe().  It mainly
  broke the enable_intr() to restore_intr().  Restoring the interrupt
  state in a nested way is precisely what is not wanted here.  It was
  harmless in practice because npxprobe() is called with interrupts
  enabled, so restoring the interrupt state enables interrupts.  Most
  of npxprobe() is a no-op for the same reason...
2001-10-16 12:55:38 +00:00
Tor Egge
4c8f0aced5 Explicitly initialize the fpu when SSE is enabled since this no
longer happens as a side effect of calling npxsave.

Reviewed by:	peter, bde
2001-10-15 20:18:06 +00:00
John Baldwin
7106ca0d1a Add missing includes of sys/lock.h. 2001-10-11 17:52:20 +00:00
Ian Dowse
3c7bcedd06 Remove the Xresume* labels from the i386 interrupt handlers; the
code in ipl.s and icu_ipl.s that used them was removed when the
interrupt thread system was committed. Debuggers also knew about
Xresume* because these labels hide the real names of the interrupt
handlers (Xintr*), and debuggers need to special-case interrupt
handlers to get the interrupt frame.

Both gdb and ddb will now use the Xintr* and Xfastintr* symbols to
detect interrupt frames. Fast interrupt frames were never identified
correctly before, so this fixes the problem of the running stack
frame getting lost in a ddb or gdb trace generated from a fast
interrupt - e.g. when debugging a simple infinite loop in the kernel
using a serial console, the frame containing the loop would never
appear in a gdb or ddb trace.

Reviewed by:	jhb, bde
2001-10-09 19:54:52 +00:00
Robert Drehmel
1e8ff53804 Remove an unneeded variable declaration and statement.
Approved by:	jake
2001-10-09 16:06:28 +00:00
Luigi Rizzo
7d4b046991 Export interrupt statistics via sysctl.
MFC-after: 3 days
2001-10-07 17:03:56 +00:00
Yoshihiro Takahashi
f2eeb19063 Rewrite the pc98 bus_space stuff.
The type of bus_space_tag_t is now a pointer to bus_space_tag structure,
and the bus_space_tag structure saves pointers to functions for direct
access and relocate access.

Added bsh_bam member to the bus_space_handle structure, it saves access
method either direct access or relocate access which is called by
bus_space_* functions.

Added the mecia device support. If the bs_da and bs_ra in bus tag are set
NEPC_io_space_tag and NEPC_mem_space_tag respectively, new bus_space stuff
changes the register of mecia automatically for 16bit access.

Obtained from:	NetBSD/pc98
2001-10-07 10:04:18 +00:00
John Baldwin
bf2965ed32 Disable the check in icu_setup() to see if a handler was already used as
the current interrupt thread routines will guarantee the condition this is
checking for at a higher level but inthand_add() and inthand_remove() as
they currently exist don't satisfy this condition.  (Which does need to be
fixed but which will take a bit more work.)  This fixes shared interrupts.
2001-09-27 19:03:52 +00:00
Jonathan Lemon
fe16674a74 Return EINVAL if the passed intr is out of bounds.
PR: 30857
Submitted by: David Xu <davidx@viasoft.com.cn>
MFC: 1 week
2001-09-27 02:46:47 +00:00
John Baldwin
659209e636 Whitespace fixes. 2001-09-18 21:05:04 +00:00
Warner Losh
8b8a72ee71 s/thread'/thread's/ 2001-09-14 04:40:44 +00:00
Julian Elischer
b40ce4165d KSE Milestone 2
Note ALL MODULES MUST BE RECOMPILED
make the kernel aware that there are smaller units of scheduling than the
process. (but only allow one thread per process at this time).
This is functionally equivalent to teh previousl -current except
that there is a thread associated with each process.

Sorry john! (your next MFC will be a doosie!)

Reviewed by: peter@freebsd.org, dillon@freebsd.org

X-MFC after:    ha ha ha ha
2001-09-12 08:38:13 +00:00
Mitsuru IWASAKI
db2077f8e1 Reenable RTC interrupts after wakeup. Some laptops have a problem
with system statistics monitoring tools (such as systat, vmstat...)
because of stopping RTC interrupts generation.
Restore all the timers (RTC and i8254) atomically.

Reviewed by:	bde
MFC after:	1 week
2001-09-04 16:02:06 +00:00
Mike Smith
5f063c7b09 Add ACPI attachments. 2001-08-30 09:17:03 +00:00
John Baldwin
e5dfa86a70 Axe a commented, unused #define related to the old giant lock. 2001-08-24 16:52:31 +00:00
Peter Wemm
3b703181e3 Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't
active.
2001-08-23 01:03:56 +00:00
John Baldwin
688ebe120c - Close races with signals and other AST's being triggered while we are in
the process of exiting the kernel.  The ast() function now loops as long
  as the PS_ASTPENDING or PS_NEEDRESCHED flags are set.  It returns with
  preemption disabled so that any further AST's that arrive via an
  interrupt will be delayed until the low-level MD code returns to user
  mode.
- Use u_int's to store the tick counts for profiling purposes so that we
  do not need sched_lock just to read p_sticks.  This also closes a
  problem where the call to addupc_task() could screw up the arithmetic
  due to non-atomic reads of p_sticks.
- Axe need_proftick(), aston(), astoff(), astpending(), need_resched(),
  clear_resched(), and resched_wanted() in favor of direct bit operations
  on p_sflag.
- Fix up locking with sched_lock some.  In addupc_intr(), use sched_lock
  to ensure pr_addr and pr_ticks are updated atomically with setting
  PS_OWEUPC.  In ast() we clear pr_ticks atomically with clearing
  PS_OWEUPC.  We also do not grab the lock just to test a flag.
- Simplify the handling of Giant in ast() slightly.

Reviewed by:	bde (mostly)
2001-08-10 22:53:32 +00:00
Peter Wemm
3e02a8711a MASK_FPU_SW didn't do what it was expected to do. 2001-07-26 23:47:04 +00:00
Takanori Watanabe
6161544ca7 Add ACPI S2-S4BIOS Suspend/Resume code.
Some problems may remain.

Reviewed by:iwasaki
2001-07-20 06:07:34 +00:00
Tor Egge
e55bc0a096 The per-cpu temporary buffers are not needed since the pcb_save areas have
the proper alignment.  Change dummy variable in npxinit from stack to bss
to ensure proper alignment.

Reviewed by:	bde
2001-07-17 13:06:47 +00:00
Tor Egge
a5d00fe997 Use PCPU_GET(cpuid) instead of curproc->p_oncpu.
Reviewed by:	peter
2001-07-16 06:00:23 +00:00
Peter Wemm
ad1b7ffaee Fix another missed pcb_savefpu reference (inside NPX_DEBUG) 2001-07-12 12:21:53 +00:00
Peter Wemm
9d146ac5d1 Activate SSE/SIMD. This is the extra context switching support that
we are required to do if we let user processes use the extra 128 bit
registers etc.

This is the base part of the diff I got from:
  http://www.issei.org/issei/FreeBSD/sse.html
I believe this is by:  Mr. SUZUKI Issei <issei@issei.org>
SMP support apparently by: Takekazu KATO <kato@chino.it.okayama-u.ac.jp>
Test code by: NAKAMURA Kazushi <kaz@kobe1995.net>, see
  http://kobe1995.net/~kaz/FreeBSD/SSE.en.html

I have fixed a couple of style(9) deviations.  I have some followup
commits to fix a couple of non-style things.
2001-07-12 06:32:51 +00:00
Matthew Dillon
0cddd8f023 With Alfred's permission, remove vm_mtx in favor of a fine-grained approach
(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.
2001-07-04 16:20:28 +00:00
Warner Losh
1b0a8621e6 Repo copy i8237.h to dev/ic so we can get rid of some of the final vestiges
of includes of i386 files from non-i386 ports.
2001-06-30 05:29:11 +00:00
Yoshihiro Takahashi
6c551e8d24 Don't assume that resource type is ioport and rid equal 0. 2001-06-17 13:33:59 +00:00
Peter Wemm
17e904e08d Fix warnings:
908: warning: long unsigned int format, unsigned int arg (arg 3)
887: warning: `timezero' defined but not used
2001-06-15 07:53:20 +00:00
Thomas Moestl
d279178df7 Clean up the code exporting interrupt statistics via sysctl a bit:
- move the sysctl code to kern_intr.c
- do not use INTRCNT_COUNT, but rather eintrcnt - intrcnt to determine
  the length of the intrcnt array
- move the declarations of intrnames, eintrnames, intrcnt and eintrcnt
  from machine-dependent include files to sys/interrupt.h
- remove the hw.nintr sysctl, it is not needed.
- fix various style bugs

Requested by:	bde
Reviewed by:	bde (some time ago)
2001-06-01 13:23:28 +00:00
Alfred Perlstein
b99c886a7f lock vm while playing with pmap 2001-05-23 22:13:58 +00:00
Bruce Evans
1c1771cb5b Convert npx interrupts into traps instead of vice versa. This is much
simpler for npx exceptions that start as traps (no assembly required...)
and works better for npx exceptions that start as interrupts (there is
no longer a problem for nested interrupts).

Submitted by:	original (pre-SMPng) version by luoqi
2001-05-22 21:20:49 +00:00
Bruce Evans
17008f5343 Throw away the complications in npxsave() and their infrastructure.
npxsave() went to great lengths to excecute fnsave with interrupts
enabled in case executing it froze the CPU.  This case can't happen,
at least for Intel CPU/NPX's.  Spurious IRQ13's don't imply spurious
freezes.  Anyway, the complications were usually no-ops because IRQ13
is not used on i486's and newer CPUs, and because SMPng broke them in
rev.1.84.  Forcible enabling of interrupts was changed to
write_eflags(old_eflags), but since SMPng usually calls npxsave() from
cpu_switch() with interrupts disabled, write_eflags() usually just
kept interrupts disabled.
2001-05-20 20:04:40 +00:00
Bruce Evans
7010278935 Use a critical region to protect almost everything in npxinit().
npxinit() didn't have the usual race because it doesn't save to curpcb,
but it may have had a worse form of it since it uses the npx when it
doesn't "own" it.  I'm not sure if locking prevented this.  npxinit()
is normally caled with the proc lock but not sched_lock.

Use a critical region to protect pushing of curproc's npx state to
curpcb in npxexit().  Not doing so was harmless since it at worst
saved a wrong state to a dieing pcb.
2001-05-20 18:05:44 +00:00
John Baldwin
ddfbf9d259 - Axe the IMEN_BITS and APIC_IMEN_BITS constants.
- Add back in a definition of NHWI which is preferred over ICU_LEN.

Submitted by:	bde
2001-05-17 22:24:17 +00:00
John Baldwin
8bd57f8fc2 Remove unneeded includes of sys/ipl.h and machine/ipl.h. 2001-05-15 23:22:29 +00:00
John Baldwin
4966b0e91c Move the definition of HWI_MASK to the i386/isa/icu.h header right next to
the definition of ICU_LEN.
2001-05-15 23:11:48 +00:00
John Baldwin
d008c720b6 - Use ICU_LEN rather than NHWI for the size of the array of ithreads.
- Remove unneeded include of sys/ipl.h.
2001-05-15 22:31:08 +00:00
Bruce Evans
79d4e25bea Fixed panics in npx exception handling. When using IRQ13 exception
handling, SMPng always switches the npx context away from curproc
before calling the handler, so the handler always paniced.  When using
exception 16 exception handling, SMPng sometimes switches the npx
context away from curproc before calling the handler, so the handler
sometimes paniced.  Also, we didn't lock the context while using it,
so we sometimes didn't detect the switch and then paniced in a less
controlled way.

Just lock the context while using it, and return without doing anything
except clearing the busy latch if the context is not for curproc.  This
fixes the exception 16 case and makes the IRQ13 case harmless.  In both
cases, the instruction that caused the exception is restarted and the
exception repeats.  In the exception 16 case, we soon get an exception
that can be handled without doing anything special.  In the IRQ13 case,
we get an easy to kill hung process.
2001-05-02 13:06:58 +00:00
Mark Murray
fb919e4d5a Undo part of the tangle of having sys/lock.h and sys/mutex.h included in
other "system" header files.

Also help the deprecation of lockmgr.h by making it a sub-include of
sys/lock.h and removing sys/lockmgr.h form kernel .c files.

Sort sys/*.h includes where possible in affected files.

OK'ed by:	bde (with reservations)
2001-05-01 08:13:21 +00:00