d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c636c4a872
freebsd-nq/sys/amd64/isa/npx.c

1007 lines
30 KiB
C
Raw Normal View History

Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*-
* Copyright (c) 1990 William Jolitz.
* Copyright (c) 1991 The Regents of the University of California.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
ALL: Removed patch kit headers and rcsid strings, add $Id$. isa.c: Removed old #ifdef notyet isa_configure code, since it will never be used, and I have done 90% of what it attempted to. Add conflict checking code that searchs back through the devtab's looking for any device that has already been found that may conflict with what we are about to probe. Checks are mode for I/O address, memory address, IRQ, and DRQ. This should stop the screwing up of any device that has alread been found by other device probes. Print out messages when we are not going to probe a device due to a conflict so the user knows WHY something was not found. For example: aha0 not probed due to irq conflict with ahb0 at 11 Now print out a message when a device is not found so the user knows that it was probed for, but could not be found. For example: ed1 not found at 0x320 For devices that have I/O address < 0x100 say that they are on the motherboard, not on isa! The 0x100 magic number is per ISA spec. It may seem funny that pc0 and sc0 report as being on the motherboard, but this is due to the fact that the I/O address used is that of the keyboard controller which IS on the motherboard. We really need to split the keyboard probe from the display probe. It is completly legal to build a pc with out one or the other, or even with out both! npx.c: Return -1 from the probe routine if we are using the Emulator so that the i/o addresses are not printed, this is the same trick used for 486's. Do not print the ``Errors reported via Exception 16'', and ``Errors reported via IRQ 13'' messages any more, since these just lead to more user confusion that anything. It still prints the message ``Error reporting broken, using 387 emulator'' so that the person is aware that there mother board is ill.
1993-10-13 15:59:30 +00:00
* from: @(#)npx.c 7.2 (Berkeley) 5/12/91
$Id$ -> $FreeBSD$
1999-08-28 01:08:13 +00:00
* $FreeBSD$
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
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
#include "opt_cpu.h"
Make the debug options new-style. This also zaps a DPT option from lint; it wasn't referenced from anywhere.
1998-01-31 07:23:16 +00:00
#include "opt_debug_npx.h"
Choose a different name to hold the option definition.. The original one was overlapping with another file, and making some undesirable behavior a little worse - it's triggering a bug in config that appears to have been there for some time (before the options files, anyway.)
1996-01-06 23:10:57 +00:00
#include "opt_math_emulate.h"
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
#include "opt_npx.h"
convert the math emulation to use the new-style options.
1996-01-04 19:51:50 +00:00
Change all #includes to follow the current Berkeley style. Some of these ``changes'' are actually not changes at all, but CVS sometimes has trouble telling the difference. This also includes support for second-directory compiles. This is not quite complete yet, as `config' doesn't yet do the right thing. You can still make it work trivially, however, by doing the following: rm /sys/compile mkdir /usr/obj/sys/compile ln -s M-. /sys/compile cd /sys/i386/conf config MYKERNEL cd ../../compile/MYKERNEL ln -s /sys @ rm machine ln -s @/i386/include machine make depend make
1994-08-13 03:50:34 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#include <sys/bus.h>
Sorry, the last commit screwed up for me, this is the right one (I hope!) Please refer to the previous commit message about sysctl variables.
1995-10-28 13:07:28 +00:00
#include <sys/kernel.h>
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
#include <sys/lock.h>
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
#include <sys/malloc.h>
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#include <sys/module.h>
- Make npx_intr INTR_MPSAFE and move acquiring Giant into the function itself. - Remove a hack to allow acquiring Giant from the npx asm trap vector.
2001-01-20 02:30:58 +00:00
#include <sys/mutex.h>
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
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sysctl.h>
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#include <machine/bus.h>
#include <sys/rman.h>
Disabled logging of masked exceptions on exit. Keep the side effect of saving the state (see rev.1.17).
1997-01-29 13:46:28 +00:00
#ifdef NPX_DEBUG
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
#include <sys/syslog.h>
Disabled logging of masked exceptions on exit. Keep the side effect of saving the state (see rev.1.17).
1997-01-29 13:46:28 +00:00
#endif
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
#include <sys/signalvar.h>
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
#include <sys/user.h>
Finished device configuration database work for all ISA devices (except `ze') and all SCSI devices (except that it's not done quite the way I want). New information added includes: - A text description of the device - A ``state''---unknown, unconfigured, idle, or busy - A generic parent device (with support in the m.i. code) - An interrupt mask type field (which will hopefully go away) so that . ``doconfig'' can be written This requires a new version of the `lsdev' program as well (next commit).
1994-10-23 21:28:03 +00:00
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#ifndef SMP
Make the necessary changes so that an ELF kernel can be built. I have successfully built, booted, and run a number of different ELF kernel configurations, including GENERIC. LINT also builds and links cleanly, though I have not tried to boot it. The impact on developers is virtually nil, except for two things. All linker sets that might possibly be present in the kernel must be listed in "sys/i386/i386/setdefs.h". And all C symbols that are also referenced from assembly language code must be listed in "sys/i386/include/asnames.h". It so happens that failure to do these things will have no impact on the a.out kernel. But it will break the build of the ELF kernel. The ELF bootloader works, but it is not ready to commit quite yet.
1997-04-22 06:55:47 +00:00
#include <machine/asmacros.h>
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#endif
Don't include <machine/cputypes.h> or declare cputype/class interfaces in <machine/cpu.h>. Moved the declarations to <machine/cputypes.h>. Fixed style bugs in the moved code. Fixed everything that depended on the nested include. Don't include <machine/cpu.h> (in the changed files) unless something in it is used directly.
1997-10-28 11:43:57 +00:00
#include <machine/cputypes.h>
#include <machine/frame.h>
Added #include of <machine/md_var.h>. This will be needed when some declarations are moved from <machine/cpufunc.h> to better places.
1996-06-25 20:31:01 +00:00
#include <machine/md_var.h>
Include file updates.. <machine/spl.h> -> <machine/ipl.h>, add <machine/ipl.h> to those files that were depending on getting SWI_* implicitly via <machine/cpufunc.h>
1997-05-31 09:27:31 +00:00
#include <machine/pcb.h>
Don't include <machine/cputypes.h> or declare cputype/class interfaces in <machine/cpu.h>. Moved the declarations to <machine/cputypes.h>. Fixed style bugs in the moved code. Fixed everything that depended on the nested include. Don't include <machine/cpu.h> (in the changed files) unless something in it is used directly.
1997-10-28 11:43:57 +00:00
#include <machine/psl.h>
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#ifndef SMP
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
#include <machine/clock.h>
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#endif
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#include <machine/resource.h>
Change all #includes to follow the current Berkeley style. Some of these ``changes'' are actually not changes at all, but CVS sometimes has trouble telling the difference. This also includes support for second-directory compiles. This is not quite complete yet, as `config' doesn't yet do the right thing. You can still make it work trivially, however, by doing the following: rm /sys/compile mkdir /usr/obj/sys/compile ln -s M-. /sys/compile cd /sys/i386/conf config MYKERNEL cd ../../compile/MYKERNEL ln -s /sys @ rm machine ln -s @/i386/include machine make depend make
1994-08-13 03:50:34 +00:00
#include <machine/specialreg.h>
Don't include <machine/cputypes.h> or declare cputype/class interfaces in <machine/cpu.h>. Moved the declarations to <machine/cputypes.h>. Fixed style bugs in the moved code. Fixed everything that depended on the nested include. Don't include <machine/cpu.h> (in the changed files) unless something in it is used directly.
1997-10-28 11:43:57 +00:00
#include <machine/segments.h>
Finished device configuration database work for all ISA devices (except `ze') and all SCSI devices (except that it's not done quite the way I want). New information added includes: - A text description of the device - A ``state''---unknown, unconfigured, idle, or busy - A generic parent device (with support in the m.i. code) - An interrupt mask type field (which will hopefully go away) so that . ``doconfig'' can be written This requires a new version of the `lsdev' program as well (next commit).
1994-10-23 21:28:03 +00:00
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#ifndef SMP
Change all #includes to follow the current Berkeley style. Some of these ``changes'' are actually not changes at all, but CVS sometimes has trouble telling the difference. This also includes support for second-directory compiles. This is not quite complete yet, as `config' doesn't yet do the right thing. You can still make it work trivially, however, by doing the following: rm /sys/compile mkdir /usr/obj/sys/compile ln -s M-. /sys/compile cd /sys/i386/conf config MYKERNEL cd ../../compile/MYKERNEL ln -s /sys @ rm machine ln -s @/i386/include machine make depend make
1994-08-13 03:50:34 +00:00
#include <i386/isa/icu.h>
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
#ifdef PC98
#include <pc98/pc98/pc98.h>
#else
Change all #includes to follow the current Berkeley style. Some of these ``changes'' are actually not changes at all, but CVS sometimes has trouble telling the difference. This also includes support for second-directory compiles. This is not quite complete yet, as `config' doesn't yet do the right thing. You can still make it work trivially, however, by doing the following: rm /sys/compile mkdir /usr/obj/sys/compile ln -s M-. /sys/compile cd /sys/i386/conf config MYKERNEL cd ../../compile/MYKERNEL ln -s /sys @ rm machine ln -s @/i386/include machine make depend make
1994-08-13 03:50:34 +00:00
#include <i386/isa/isa.h>
Removed an unused #included. Ifdefed #includes that are not used in the SMP case.
1997-11-18 11:32:31 +00:00
#endif
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
#endif
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
#include <i386/isa/intr_machdep.h>
Add a little do-nothing ``slopsucker'' device which gives a home to PNP0c04 (legacy ISA coprocessor support). Tourist info.
2000-05-04 23:57:32 +00:00
#include <isa/isavar.h>
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* 387 and 287 Numeric Coprocessor Extension (NPX) Driver.
*/
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
/* Configuration flags. */
#define NPX_DISABLE_I586_OPTIMIZED_BCOPY (1 << 0)
#define NPX_DISABLE_I586_OPTIMIZED_BZERO (1 << 1)
#define NPX_DISABLE_I586_OPTIMIZED_COPYIO (1 << 2)
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
#define NPX_PREFER_EMULATOR (1 << 3)
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef __GNUC__
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
#define fldcw(addr) __asm("fldcw %0" : : "m" (*(addr)))
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#define fnclex() __asm("fnclex")
#define fninit() __asm("fninit")
Finish supporting compiling with `gcc -ansi'. Fix missing `volatile's in __asm() statements while I'm here.
1998-04-15 18:58:09 +00:00
#define fnsave(addr) __asm __volatile("fnsave %0" : "=m" (*(addr)))
#define fnstcw(addr) __asm __volatile("fnstcw %0" : "=m" (*(addr)))
#define fnstsw(addr) __asm __volatile("fnstsw %0" : "=m" (*(addr)))
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
#define fp_divide_by_0() __asm("fldz; fld1; fdiv %st,%st(1); fnop")
#define frstor(addr) __asm("frstor %0" : : "m" (*(addr)))
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#ifdef CPU_ENABLE_SSE
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
#define fxrstor(addr) __asm("fxrstor %0" : : "m" (*(addr)))
#define fxsave(addr) __asm __volatile("fxsave %0" : "=m" (*(addr)))
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#define start_emulating() __asm("smsw %%ax; orb %0,%%al; lmsw %%ax" \
: : "n" (CR0_TS) : "ax")
#define stop_emulating() __asm("clts")
#else /* not __GNUC__ */
void fldcw __P((caddr_t addr));
void fnclex __P((void));
void fninit __P((void));
void fnsave __P((caddr_t addr));
void fnstcw __P((caddr_t addr));
void fnstsw __P((caddr_t addr));
void fp_divide_by_0 __P((void));
void frstor __P((caddr_t addr));
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#ifdef CPU_ENABLE_SSE
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
void fxsave __P((caddr_t addr));
void fxrstor __P((caddr_t addr));
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
void start_emulating __P((void));
void stop_emulating __P((void));
#endif /* __GNUC__ */
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
#ifdef CPU_ENABLE_SSE
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
#define GET_FPU_CW(thread) \
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
(cpu_fxsr ? \
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
(thread)->td_pcb->pcb_save.sv_xmm.sv_env.en_cw : \
(thread)->td_pcb->pcb_save.sv_87.sv_env.en_cw)
#define GET_FPU_SW(thread) \
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
(cpu_fxsr ? \
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
(thread)->td_pcb->pcb_save.sv_xmm.sv_env.en_sw : \
(thread)->td_pcb->pcb_save.sv_87.sv_env.en_sw)
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
#define GET_FPU_EXSW_PTR(pcb) \
(cpu_fxsr ? \
&(pcb)->pcb_save.sv_xmm.sv_ex_sw : \
&(pcb)->pcb_save.sv_87.sv_ex_sw)
#else /* CPU_ENABLE_SSE */
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
#define GET_FPU_CW(thread) \
(thread->td_pcb->pcb_save.sv_87.sv_env.en_cw)
#define GET_FPU_SW(thread) \
(thread->td_pcb->pcb_save.sv_87.sv_env.en_sw)
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
#define GET_FPU_EXSW_PTR(pcb) \
(&(pcb)->pcb_save.sv_87.sv_ex_sw)
#endif /* CPU_ENABLE_SSE */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
typedef u_char bool_t;
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
static int npx_attach __P((device_t dev));
Add an identify method to allow npx to arrange itself to be attached to the nexus without explicit code in the nexus to do so.
1999-08-22 19:52:51 +00:00
static void npx_identify __P((driver_t *driver, device_t parent));
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
#ifndef SMP
static void npx_intr __P((void *));
#endif
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
static int npx_probe __P((device_t dev));
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
static void fpusave __P((union savefpu *));
static void fpurstor __P((union savefpu *));
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
#ifdef I586_CPU_XXX
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
static long timezero __P((const char *funcname,
void (*func)(void *buf, size_t len)));
Ifdef declaration of a conditionally defined function "timezero".
1999-03-28 23:28:18 +00:00
#endif /* I586_CPU */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
int hw_float; /* XXX currently just alias for npx_exists */
Sorry, the last commit screwed up for me, this is the right one (I hope!) Please refer to the previous commit message about sysctl variables.
1995-10-28 13:07:28 +00:00
SYSCTL_INT(_hw,HW_FLOATINGPT, floatingpoint,
CTLFLAG_RD, &hw_float, 0,
"Floatingpoint instructions executed in hardware");
The "easy" fixes for compiling the kernel -Wunused: remove unreferenced static and local variables, goto labels, and functions declared but not defined.
1998-12-07 21:58:50 +00:00
#ifndef SMP
static volatile u_int npx_intrs_while_probing;
static volatile u_int npx_traps_while_probing;
#endif
Preliminary support for per-cpu data pages. This eliminates a lot of #ifdef SMP type code. Things like _curproc reside in a data page that is unique on each cpu, eliminating the expensive macros like: #define curproc (SMPcurproc[cpunumber()]) There are some unresolved bootstrap and address space sharing issues at present, but Steve is waiting on this for other work. There is still some strictly temporary code present that isn't exactly pretty. This is part of a larger change that has run into some bumps, this part is standalone so it should be safe. The temporary code goes away when the full idle cpu support is finished. Reviewed by: fsmp, dyson
1997-06-22 16:04:22 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
static bool_t npx_ex16;
static bool_t npx_exists;
static bool_t npx_irq13;
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
#ifndef SMP
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
alias_for_inthand_t probetrap;
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants or hard long long constants).
1998-04-19 15:39:26 +00:00
__asm(" \n\
.text \n\
.p2align 2,0x90 \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.type " __XSTRING(CNAME(probetrap)) ",@function \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants or hard long long constants).
1998-04-19 15:39:26 +00:00
" __XSTRING(CNAME(probetrap)) ": \n\
ss \n\
incl " __XSTRING(CNAME(npx_traps_while_probing)) " \n\
fnclex \n\
iret \n\
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
");
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
#endif /* SMP */
Add an identify method to allow npx to arrange itself to be attached to the nexus without explicit code in the nexus to do so.
1999-08-22 19:52:51 +00:00
/*
* Identify routine. Create a connection point on our parent for probing.
*/
static void
npx_identify(driver, parent)
driver_t *driver;
device_t parent;
{
device_t child;
child = BUS_ADD_CHILD(parent, 0, "npx", 0);
if (child == NULL)
panic("npx_identify");
}
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
#ifndef SMP
/*
* Do minimal handling of npx interrupts to convert them to traps.
*/
static void
npx_intr(dummy)
void *dummy;
{
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
struct thread *td;
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
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
#ifndef SMP
npx_intrs_while_probing++;
#endif
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
/*
* The BUSY# latch must be cleared in all cases so that the next
* unmasked npx exception causes an interrupt.
*/
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
#ifdef PC98
outb(0xf8, 0);
#else
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
outb(0xf0, 0);
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
#endif
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
/*
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
* fpcurthread is normally non-null here. In that case, schedule an
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
* AST to finish the exception handling in the correct context
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
* (this interrupt may occur after the thread has entered the
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
* kernel via a syscall or an interrupt). Otherwise, the npx
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
* state of the thread that caused this interrupt must have been
s/thread'/thread's/
2001-09-14 04:40:44 +00:00
* pushed to the thread's pcb, and clearing of the busy latch
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
* above has finished the (essentially null) handling of this
* interrupt. Control will eventually return to the instruction
* that caused it and it will repeat. We will eventually (usually
* soon) win the race to handle the interrupt properly.
*/
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
td = PCPU_GET(fpcurthread);
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
if (td != NULL) {
td->td_pcb->pcb_flags |= PCB_NPXTRAP;
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
mtx_lock_spin(&sched_lock);
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
td->td_kse->ke_flags |= KEF_ASTPENDING;
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
mtx_unlock_spin(&sched_lock);
}
}
#endif /* !SMP */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* Probe routine. Initialize cr0 to give correct behaviour for [f]wait
* whether the device exists or not (XXX should be elsewhere). Set flags
* to tell npxattach() what to do. Modify device struct if npx doesn't
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
* need to use interrupts. Return 0 if device exists.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
static int
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
npx_probe(dev)
device_t dev;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
The "easy" fixes for compiling the kernel -Wunused: remove unreferenced static and local variables, goto labels, and functions declared but not defined.
1998-12-07 21:58:50 +00:00
#ifndef SMP
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
struct gate_descriptor save_idt_npxtrap;
struct resource *ioport_res, *irq_res;
void *irq_cookie;
int ioport_rid, irq_num, irq_rid;
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
u_short control;
u_short status;
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
save_idt_npxtrap = idt[16];
setidt(16, probetrap, SDT_SYS386TGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL));
ioport_rid = 0;
ioport_res = bus_alloc_resource(dev, SYS_RES_IOPORT, &ioport_rid,
IO_NPX, IO_NPX, IO_NPXSIZE, RF_ACTIVE);
if (ioport_res == NULL)
panic("npx: can't get ports");
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
#ifdef PC98
if (resource_int_value("npx", 0, "irq", &irq_num) != 0)
irq_num = 8;
#else
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
if (resource_int_value("npx", 0, "irq", &irq_num) != 0)
irq_num = 13;
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
#endif
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
irq_rid = 0;
irq_res = bus_alloc_resource(dev, SYS_RES_IRQ, &ioport_rid, irq_num,
irq_num, 1, RF_ACTIVE);
if (irq_res == NULL)
panic("npx: can't get IRQ");
if (bus_setup_intr(dev, irq_res, INTR_TYPE_MISC | INTR_FAST, npx_intr,
NULL, &irq_cookie) != 0)
panic("npx: can't create intr");
#endif /* !SMP */
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* Partially reset the coprocessor, if any. Some BIOS's don't reset
* it after a warm boot.
*/
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
#ifdef PC98
outb(0xf8,0);
#else
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(0xf1, 0); /* full reset on some systems, NOP on others */
outb(0xf0, 0); /* clear BUSY# latch */
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
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* Prepare to trap all ESC (i.e., NPX) instructions and all WAIT
* instructions. We must set the CR0_MP bit and use the CR0_TS
* bit to control the trap, because setting the CR0_EM bit does
* not cause WAIT instructions to trap. It's important to trap
* WAIT instructions - otherwise the "wait" variants of no-wait
* control instructions would degenerate to the "no-wait" variants
* after FP context switches but work correctly otherwise. It's
* particularly important to trap WAITs when there is no NPX -
* otherwise the "wait" variants would always degenerate.
*
* Try setting CR0_NE to get correct error reporting on 486DX's.
* Setting it should fail or do nothing on lesser processors.
*/
load_cr0(rcr0() | CR0_MP | CR0_NE);
/*
* But don't trap while we're probing.
*/
stop_emulating();
/*
* Finish resetting the coprocessor, if any. If there is an error
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
* pending, then we may get a bogus IRQ13, but npx_intr() will handle
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* it OK. Bogus halts have never been observed, but we enabled
* IRQ13 and cleared the BUSY# latch early to handle them anyway.
*/
fninit();
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
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
device_set_desc(dev, "math processor");
I'm not sure why the #ifdef SMP became #if 1 (this overrode the npx probe and always succeeded as is required on SMP). Anyway, reverting this still compiles and appears ok.
1999-05-06 12:47:21 +00:00
#ifdef SMP
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
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
/*
* Exception 16 MUST work for SMP.
*/
npx_ex16 = hw_float = npx_exists = 1;
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
return (0);
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#else /* !SMP */
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
Submitted by: seb@erix.ericsson.se (Sebastian Strollo) Remove over-cautious early fnop() synchronization. It caused the probe to hang on systems without an FPU.
1995-02-23 17:32:38 +00:00
/*
* Don't use fwait here because it might hang.
* Don't use fnop here because it usually hangs if there is no FPU.
*/
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
DELAY(1000); /* wait for any IRQ13 */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef DIAGNOSTIC
if (npx_intrs_while_probing != 0)
printf("fninit caused %u bogus npx interrupt(s)\n",
npx_intrs_while_probing);
if (npx_traps_while_probing != 0)
printf("fninit caused %u bogus npx trap(s)\n",
npx_traps_while_probing);
#endif
/*
* Check for a status of mostly zero.
*/
status = 0x5a5a;
fnstsw(&status);
if ((status & 0xb8ff) == 0) {
/*
* Good, now check for a proper control word.
*/
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
control = 0x5a5a;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
fnstcw(&control);
if ((control & 0x1f3f) == 0x033f) {
Define new MIB variable, hw.floatingpoint, which is true if FP hardware is present, and false if an emulator is being used.
1994-09-09 23:13:03 +00:00
hw_float = npx_exists = 1;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* We have an npx, now divide by 0 to see if exception
* 16 works.
*/
control &= ~(1 << 2); /* enable divide by 0 trap */
fldcw(&control);
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
#ifdef FPU_ERROR_BROKEN
/*
* FPU error signal doesn't work on some CPU
* accelerator board.
*/
npx_ex16 = 1;
return (0);
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
npx_traps_while_probing = npx_intrs_while_probing = 0;
fp_divide_by_0();
if (npx_traps_while_probing != 0) {
/*
* Good, exception 16 works.
*/
npx_ex16 = 1;
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
goto no_irq13;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
if (npx_intrs_while_probing != 0) {
/*
* Bad, we are stuck with IRQ13.
*/
npx_irq13 = 1;
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
idt[16] = save_idt_npxtrap;
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
return (0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* Worse, even IRQ13 is broken. Use emulator.
*/
}
}
/*
* Probe failed, but we want to get to npxattach to initialize the
* emulator and say that it has been installed. XXX handle devices
* that aren't really devices better.
*/
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
/* FALLTHROUGH */
no_irq13:
idt[16] = save_idt_npxtrap;
bus_teardown_intr(dev, irq_res, irq_cookie);
/*
* XXX hack around brokenness of bus_teardown_intr(). If we left the
* irq active then we would get it instead of exception 16.
*/
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
mtx_lock_spin(&icu_lock);
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
INTRDIS(1 << irq_num);
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
mtx_unlock_spin(&icu_lock);
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
bus_release_resource(dev, SYS_RES_IRQ, irq_rid, irq_res);
bus_release_resource(dev, SYS_RES_IOPORT, ioport_rid, ioport_res);
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
return (0);
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
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
#endif /* SMP */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* Attach routine - announce which it is, and wire into system
*/
int
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
npx_attach(dev)
device_t dev;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
int flags;
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
if (resource_int_value("npx", 0, "flags", &flags) != 0)
flags = 0;
Initialize isa_devtab entries for interrupt handlers in individual device drivers, not in ioconf.c. Use a different hack in isa_device.h so that a new config(8) is not required yet. pc98 parts approved by: kato
1998-10-22 05:58:45 +00:00
Print out flags value
1999-09-21 10:51:47 +00:00
if (flags)
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "flags 0x%x ", flags);
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
if (npx_irq13) {
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "using IRQ 13 interface\n");
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
} else {
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
#if defined(MATH_EMULATE) || defined(GPL_MATH_EMULATE)
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
if (npx_ex16) {
if (!(flags & NPX_PREFER_EMULATOR))
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "INT 16 interface\n");
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
else {
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "FPU exists, but flags request "
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
"emulator\n");
hw_float = npx_exists = 0;
}
} else if (npx_exists) {
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "error reporting broken; using 387 emulator\n");
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
hw_float = npx_exists = 0;
} else
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "387 emulator\n");
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
#else
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
if (npx_ex16) {
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "INT 16 interface\n");
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
if (flags & NPX_PREFER_EMULATOR) {
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "emulator requested, but none compiled "
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
"into kernel, using FPU\n");
}
} else
Use device_printf() instead of device_print_prettyname().
2000-01-25 21:39:20 +00:00
device_printf(dev, "no 387 emulator in kernel and no FPU!\n");
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
#endif
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
npxinit(__INITIAL_NPXCW__);
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
People are still having problems with i586_* on UP machines and SMP machines, so just hack it to disable them for now until it can be fixed. Inspired by hair pulling of: asmodai
2001-04-13 17:14:53 +00:00
#ifdef I586_CPU_XXX
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
if (cpu_class == CPUCLASS_586 && npx_ex16 && npx_exists &&
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
timezero("i586_bzero()", i586_bzero) <
timezero("bzero()", bzero) * 4 / 5) {
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
if (!(flags & NPX_DISABLE_I586_OPTIMIZED_BCOPY)) {
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
bcopy_vector = i586_bcopy;
ovbcopy_vector = i586_bcopy;
}
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
if (!(flags & NPX_DISABLE_I586_OPTIMIZED_BZERO))
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
bzero = i586_bzero;
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
if (!(flags & NPX_DISABLE_I586_OPTIMIZED_COPYIO)) {
Replaced I586_OPTIMIZED_BCOPY and I586_OPTIMIZED_BZERO with boot-time negative-logic flags (flags 0x01 and 0x02 for npx0, defaulting to unset = on). This changes the default from off to on. The options have been in current for several months with no problems reported. Added a boot-time negative-logic flag for the old I5886_FAST_BCOPY option which went away too soon (flag 0x04 for npx0, defaulting to unset = on). Added a boot-time way to set the memory size (iosiz in config, iosize in userconfig for npx0). LINT: Removed old options. Documented npx0's flags and iosiz. options.i386: Removed old options. identcpu.c: Don't set the function pointers here. Setting them has to be delayed until after userconfig has had a chance to disable them and until after a good npx0 has been detected. machdep.c: Use npx0's iosize instead of MAXMEM if it is nonzero. support.s: Added vectors and glue code for copyin() and copyout(). Fixed ifdefs for i586_bzero(). Added ifdefs for i586_bcopy(). npx.c: Set the function pointers here. Clear hw_float when an npx exists but is too broken to use. Restored style from a year or three ago in npxattach().
1996-11-11 20:39:03 +00:00
copyin_vector = i586_copyin;
copyout_vector = i586_copyout;
}
}
#endif
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
return (0); /* XXX unused */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* Initialize floating point unit.
*/
void
npxinit(control)
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
u_short control;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
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
static union savefpu dummy;
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
critical_t savecrit;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (!npx_exists)
return;
/*
* fninit has the same h/w bugs as fnsave. Use the detoxified
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
* fnsave to throw away any junk in the fpu. npxsave() initializes
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
* the fpu and sets fpcurthread = NULL as important side effects.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
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
savecrit = cpu_critical_enter();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
npxsave(&dummy);
stop_emulating();
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
#ifdef CPU_ENABLE_SSE
/* XXX npxsave() doesn't actually initialize the fpu in the SSE case. */
if (cpu_fxsr)
fninit();
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
fldcw(&control);
Use PCPU_GET, PCPU_PTR and PCPU_SET to access all per-cpu variables other then curproc.
2001-01-10 04:43:51 +00:00
if (PCPU_GET(curpcb) != NULL)
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
fpusave(&PCPU_GET(curpcb)->pcb_save);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
start_emulating();
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
cpu_critical_exit(savecrit);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* Free coprocessor (if we have it).
*/
void
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
npxexit(td)
struct thread *td;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
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
critical_t savecrit;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
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
savecrit = cpu_critical_enter();
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
if (td == PCPU_GET(fpcurthread))
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
npxsave(&PCPU_GET(curpcb)->pcb_save);
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
cpu_critical_exit(savecrit);
Disabled logging of masked exceptions on exit. Keep the side effect of saving the state (see rev.1.17).
1997-01-29 13:46:28 +00:00
#ifdef NPX_DEBUG
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
if (npx_exists) {
u_int masked_exceptions;
Fix another missed pcb_savefpu reference (inside NPX_DEBUG)
2001-07-12 12:21:53 +00:00
masked_exceptions = PCPU_GET(curpcb)->pcb_save.sv_87.sv_env.en_cw
& PCPU_GET(curpcb)->pcb_save.sv_87.sv_env.en_sw & 0x7f;
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
/*
Disabled logging of masked exceptions on exit. Keep the side effect of saving the state (see rev.1.17).
1997-01-29 13:46:28 +00:00
* Log exceptions that would have trapped with the old
* control word (overflow, divide by 0, and invalid operand).
Log processes that exit with an masked npx exception that would trap with the current default exception (un)mask. There should be no such processes unless you change the mask. Someday the mask should be changed to the IEEE default of everything masked. The npx state gets saved so that it can be checked and this may have the side effect of fixing a bug that was reported for 1.1.5. (npx exceptions may sometimes leak across exits and clobber another process. I can't see how this can happen.) Get some missing/wrong declarations from headers now that the headers have them.
1994-11-14 14:59:06 +00:00
*/
if (masked_exceptions & 0x0d)
log(LOG_ERR,
Whitespace fixes.
2001-09-18 21:05:04 +00:00
"pid %d (%s) exited with masked floating point exceptions 0x%02x\n",
td->td_proc->p_pid, td->td_proc->p_comm,
masked_exceptions);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
Disabled logging of masked exceptions on exit. Keep the side effect of saving the state (see rev.1.17).
1997-01-29 13:46:28 +00:00
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
/*
* The following mechanism is used to ensure that the FPE_... value
* that is passed as a trapcode to the signal handler of the user
* process does not have more than one bit set.
*
* Multiple bits may be set if the user process modifies the control
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* word while a status word bit is already set. While this is a sign
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* of bad coding, we have no choise than to narrow them down to one
* bit, since we must not send a trapcode that is not exactly one of
* the FPE_ macros.
*
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* The mechanism has a static table with 127 entries. Each combination
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* of the 7 FPU status word exception bits directly translates to a
* position in this table, where a single FPE_... value is stored.
* This FPE_... value stored there is considered the "most important"
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* of the exception bits and will be sent as the signal code. The
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* precedence of the bits is based upon Intel Document "Numerical
* Applications", Chapter "Special Computational Situations".
*
* The macro to choose one of these values does these steps: 1) Throw
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* away status word bits that cannot be masked. 2) Throw away the bits
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* currently masked in the control word, assuming the user isn't
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* interested in them anymore. 3) Reinsert status word bit 7 (stack
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* fault) if it is set, which cannot be masked but must be presered.
* 4) Use the remaining bits to point into the trapcode table.
*
* The 6 maskable bits in order of their preference, as stated in the
* above referenced Intel manual:
* 1 Invalid operation (FP_X_INV)
* 1a Stack underflow
* 1b Stack overflow
* 1c Operand of unsupported format
* 1d SNaN operand.
* 2 QNaN operand (not an exception, irrelavant here)
* 3 Any other invalid-operation not mentioned above or zero divide
* (FP_X_INV, FP_X_DZ)
* 4 Denormal operand (FP_X_DNML)
* 5 Numeric over/underflow (FP_X_OFL, FP_X_UFL)
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
* 6 Inexact result (FP_X_IMP)
*/
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
static char fpetable[128] = {
0,
Various formatting fixes on my FPE trapcode commit. Submitted by: BDE
1999-07-26 05:47:31 +00:00
FPE_FLTINV, /* 1 - INV */
FPE_FLTUND, /* 2 - DNML */
FPE_FLTINV, /* 3 - INV | DNML */
FPE_FLTDIV, /* 4 - DZ */
FPE_FLTINV, /* 5 - INV | DZ */
FPE_FLTDIV, /* 6 - DNML | DZ */
FPE_FLTINV, /* 7 - INV | DNML | DZ */
FPE_FLTOVF, /* 8 - OFL */
FPE_FLTINV, /* 9 - INV | OFL */
FPE_FLTUND, /* A - DNML | OFL */
FPE_FLTINV, /* B - INV | DNML | OFL */
FPE_FLTDIV, /* C - DZ | OFL */
FPE_FLTINV, /* D - INV | DZ | OFL */
FPE_FLTDIV, /* E - DNML | DZ | OFL */
FPE_FLTINV, /* F - INV | DNML | DZ | OFL */
FPE_FLTUND, /* 10 - UFL */
FPE_FLTINV, /* 11 - INV | UFL */
FPE_FLTUND, /* 12 - DNML | UFL */
FPE_FLTINV, /* 13 - INV | DNML | UFL */
FPE_FLTDIV, /* 14 - DZ | UFL */
FPE_FLTINV, /* 15 - INV | DZ | UFL */
FPE_FLTDIV, /* 16 - DNML | DZ | UFL */
FPE_FLTINV, /* 17 - INV | DNML | DZ | UFL */
FPE_FLTOVF, /* 18 - OFL | UFL */
FPE_FLTINV, /* 19 - INV | OFL | UFL */
FPE_FLTUND, /* 1A - DNML | OFL | UFL */
FPE_FLTINV, /* 1B - INV | DNML | OFL | UFL */
FPE_FLTDIV, /* 1C - DZ | OFL | UFL */
FPE_FLTINV, /* 1D - INV | DZ | OFL | UFL */
FPE_FLTDIV, /* 1E - DNML | DZ | OFL | UFL */
FPE_FLTINV, /* 1F - INV | DNML | DZ | OFL | UFL */
FPE_FLTRES, /* 20 - IMP */
FPE_FLTINV, /* 21 - INV | IMP */
FPE_FLTUND, /* 22 - DNML | IMP */
FPE_FLTINV, /* 23 - INV | DNML | IMP */
FPE_FLTDIV, /* 24 - DZ | IMP */
FPE_FLTINV, /* 25 - INV | DZ | IMP */
FPE_FLTDIV, /* 26 - DNML | DZ | IMP */
FPE_FLTINV, /* 27 - INV | DNML | DZ | IMP */
FPE_FLTOVF, /* 28 - OFL | IMP */
FPE_FLTINV, /* 29 - INV | OFL | IMP */
FPE_FLTUND, /* 2A - DNML | OFL | IMP */
FPE_FLTINV, /* 2B - INV | DNML | OFL | IMP */
FPE_FLTDIV, /* 2C - DZ | OFL | IMP */
FPE_FLTINV, /* 2D - INV | DZ | OFL | IMP */
FPE_FLTDIV, /* 2E - DNML | DZ | OFL | IMP */
FPE_FLTINV, /* 2F - INV | DNML | DZ | OFL | IMP */
FPE_FLTUND, /* 30 - UFL | IMP */
FPE_FLTINV, /* 31 - INV | UFL | IMP */
FPE_FLTUND, /* 32 - DNML | UFL | IMP */
FPE_FLTINV, /* 33 - INV | DNML | UFL | IMP */
FPE_FLTDIV, /* 34 - DZ | UFL | IMP */
FPE_FLTINV, /* 35 - INV | DZ | UFL | IMP */
FPE_FLTDIV, /* 36 - DNML | DZ | UFL | IMP */
FPE_FLTINV, /* 37 - INV | DNML | DZ | UFL | IMP */
FPE_FLTOVF, /* 38 - OFL | UFL | IMP */
FPE_FLTINV, /* 39 - INV | OFL | UFL | IMP */
FPE_FLTUND, /* 3A - DNML | OFL | UFL | IMP */
FPE_FLTINV, /* 3B - INV | DNML | OFL | UFL | IMP */
FPE_FLTDIV, /* 3C - DZ | OFL | UFL | IMP */
FPE_FLTINV, /* 3D - INV | DZ | OFL | UFL | IMP */
FPE_FLTDIV, /* 3E - DNML | DZ | OFL | UFL | IMP */
FPE_FLTINV, /* 3F - INV | DNML | DZ | OFL | UFL | IMP */
FPE_FLTSUB, /* 40 - STK */
FPE_FLTSUB, /* 41 - INV | STK */
FPE_FLTUND, /* 42 - DNML | STK */
FPE_FLTSUB, /* 43 - INV | DNML | STK */
FPE_FLTDIV, /* 44 - DZ | STK */
FPE_FLTSUB, /* 45 - INV | DZ | STK */
FPE_FLTDIV, /* 46 - DNML | DZ | STK */
FPE_FLTSUB, /* 47 - INV | DNML | DZ | STK */
FPE_FLTOVF, /* 48 - OFL | STK */
FPE_FLTSUB, /* 49 - INV | OFL | STK */
FPE_FLTUND, /* 4A - DNML | OFL | STK */
FPE_FLTSUB, /* 4B - INV | DNML | OFL | STK */
FPE_FLTDIV, /* 4C - DZ | OFL | STK */
FPE_FLTSUB, /* 4D - INV | DZ | OFL | STK */
FPE_FLTDIV, /* 4E - DNML | DZ | OFL | STK */
FPE_FLTSUB, /* 4F - INV | DNML | DZ | OFL | STK */
FPE_FLTUND, /* 50 - UFL | STK */
FPE_FLTSUB, /* 51 - INV | UFL | STK */
FPE_FLTUND, /* 52 - DNML | UFL | STK */
FPE_FLTSUB, /* 53 - INV | DNML | UFL | STK */
FPE_FLTDIV, /* 54 - DZ | UFL | STK */
FPE_FLTSUB, /* 55 - INV | DZ | UFL | STK */
FPE_FLTDIV, /* 56 - DNML | DZ | UFL | STK */
FPE_FLTSUB, /* 57 - INV | DNML | DZ | UFL | STK */
FPE_FLTOVF, /* 58 - OFL | UFL | STK */
FPE_FLTSUB, /* 59 - INV | OFL | UFL | STK */
FPE_FLTUND, /* 5A - DNML | OFL | UFL | STK */
FPE_FLTSUB, /* 5B - INV | DNML | OFL | UFL | STK */
FPE_FLTDIV, /* 5C - DZ | OFL | UFL | STK */
FPE_FLTSUB, /* 5D - INV | DZ | OFL | UFL | STK */
FPE_FLTDIV, /* 5E - DNML | DZ | OFL | UFL | STK */
FPE_FLTSUB, /* 5F - INV | DNML | DZ | OFL | UFL | STK */
FPE_FLTRES, /* 60 - IMP | STK */
FPE_FLTSUB, /* 61 - INV | IMP | STK */
FPE_FLTUND, /* 62 - DNML | IMP | STK */
FPE_FLTSUB, /* 63 - INV | DNML | IMP | STK */
FPE_FLTDIV, /* 64 - DZ | IMP | STK */
FPE_FLTSUB, /* 65 - INV | DZ | IMP | STK */
FPE_FLTDIV, /* 66 - DNML | DZ | IMP | STK */
FPE_FLTSUB, /* 67 - INV | DNML | DZ | IMP | STK */
FPE_FLTOVF, /* 68 - OFL | IMP | STK */
FPE_FLTSUB, /* 69 - INV | OFL | IMP | STK */
FPE_FLTUND, /* 6A - DNML | OFL | IMP | STK */
FPE_FLTSUB, /* 6B - INV | DNML | OFL | IMP | STK */
FPE_FLTDIV, /* 6C - DZ | OFL | IMP | STK */
FPE_FLTSUB, /* 6D - INV | DZ | OFL | IMP | STK */
FPE_FLTDIV, /* 6E - DNML | DZ | OFL | IMP | STK */
FPE_FLTSUB, /* 6F - INV | DNML | DZ | OFL | IMP | STK */
FPE_FLTUND, /* 70 - UFL | IMP | STK */
FPE_FLTSUB, /* 71 - INV | UFL | IMP | STK */
FPE_FLTUND, /* 72 - DNML | UFL | IMP | STK */
FPE_FLTSUB, /* 73 - INV | DNML | UFL | IMP | STK */
FPE_FLTDIV, /* 74 - DZ | UFL | IMP | STK */
FPE_FLTSUB, /* 75 - INV | DZ | UFL | IMP | STK */
FPE_FLTDIV, /* 76 - DNML | DZ | UFL | IMP | STK */
FPE_FLTSUB, /* 77 - INV | DNML | DZ | UFL | IMP | STK */
FPE_FLTOVF, /* 78 - OFL | UFL | IMP | STK */
FPE_FLTSUB, /* 79 - INV | OFL | UFL | IMP | STK */
FPE_FLTUND, /* 7A - DNML | OFL | UFL | IMP | STK */
FPE_FLTSUB, /* 7B - INV | DNML | OFL | UFL | IMP | STK */
FPE_FLTDIV, /* 7C - DZ | OFL | UFL | IMP | STK */
FPE_FLTSUB, /* 7D - INV | DZ | OFL | UFL | IMP | STK */
FPE_FLTDIV, /* 7E - DNML | DZ | OFL | UFL | IMP | STK */
FPE_FLTSUB, /* 7F - INV | DNML | DZ | OFL | UFL | IMP | STK */
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
};
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
* Preserve the FP status word, clear FP exceptions, then generate a SIGFPE.
*
* Clearing exceptions is necessary mainly to avoid IRQ13 bugs. We now
* depend on longjmp() restoring a usable state. Restoring the state
* or examining it might fail if we didn't clear exceptions.
*
On FPU exceptions, pass a useful error code (one of the FPE_... macros) to the signal handler, for old-style BSD signal handlers as the second (int) argument, for SA_SIGINFO signal handlers as siginfo_t->si_code. This is source-compatible with Solaris, except that we have no <siginfo.h> (which isn't even mentioned in POSIX 1003.1b). An rather complete example program is at http://www3.cons.org/cracauer/freebsd-signal.c This will be added to the regression tests in src/. This commit also adds code to disable the (hardware) FPU from userconfig, so that you can use a software FP emulator on a machine that has hardware floating point. See LINT.
1999-07-25 13:16:09 +00:00
* The error code chosen will be one of the FPE_... macros. It will be
* sent as the second argument to old BSD-style signal handlers and as
* "siginfo_t->si_code" (second argument) to SA_SIGINFO signal handlers.
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
*
* XXX the FP state is not preserved across signal handlers. So signal
* handlers cannot afford to do FP unless they preserve the state or
* longjmp() out. Both preserving the state and longjmp()ing may be
* destroyed by IRQ13 bugs. Clearing FP exceptions is not an acceptable
* solution for signals other than SIGFPE.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
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
int
npxtrap()
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
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
critical_t savecrit;
u_short control, status;
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
u_long *exstat;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
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
if (!npx_exists) {
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
printf("npxtrap: fpcurthread = %p, curthread = %p, npx_exists = %d\n",
PCPU_GET(fpcurthread), curthread, npx_exists);
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
panic("npxtrap from nowhere");
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
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
savecrit = cpu_critical_enter();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
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
* Interrupt handling (for another interrupt) may have pushed the
* state to memory. Fetch the relevant parts of the state from
* wherever they are.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
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
if (PCPU_GET(fpcurthread) != curthread) {
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
control = GET_FPU_CW(curthread);
status = GET_FPU_SW(curthread);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
} else {
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
fnstcw(&control);
fnstsw(&status);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
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
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
exstat = GET_FPU_EXSW_PTR(curthread->td_pcb);
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
*exstat = status;
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
if (PCPU_GET(fpcurthread) != curthread)
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
GET_FPU_SW(curthread) &= ~0x80bf;
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
else
fnclex();
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
cpu_critical_exit(savecrit);
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
return (fpetable[status & ((~control & 0x3f) | 0x40)]);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* Implement device not available (DNA) exception
*
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
* It would be better to switch FP context here (if curthread != fpcurthread)
Use sufficient parentheses in macros. Remove bogus input operands for fnsave(), fnstcw() and fnstsw(). Change all fwait's to fnop's. This might help avoid hardware bugs. Wait after fninit with an fnop. This should be safer now. Fix some spelling and formatting errors. Use natural sizes for control and status words (u_short, promotes to int). Don't clobber the SWI_CLOCK_MASK bits in npx0_imask when using IRQ13. Set the devconf state correctly (always busy, if configured). Improve code for npx_registerdev() a little (gcc can't keep id->id_unit in a register for some reason). Don't register a nonexistent npx device. Print a useful message in npxattach() again (delete references to errors and not the whole message). Don't print "387 emulator" if there is no emulator in the kernel. Use %p for pointers in error messages. Don't clobber the FPU state when there is an FPU exception. Just clear the exception flags (after saving the flags as before). This allows debuggers and SIGFPE handlers to look at the full exception state. SIGFPE handlers should normally return via longjmp(), which restores a good FPU state (as before). Returning from a SIGFPE handler may leave the FPU in the wrong state (as before). Clear the busy latch _after_ clearing the exception flags so that there is less chance of getting a bogus h/w interrupt for a control operation. Clear the saved exception status word when the next FPU instruction is excuted so that it doesn't stick around until the next exception. Clear the busy latch after fnsave() in npxsave() in case it was set when npxsave() was called.
1995-01-03 04:00:06 +00:00
* and not necessarily for every context switch, but it is too hard to
* access foreign pcb's.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
int
npxdna()
{
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
u_long *exstat;
Switch from save/disable/restore_intr() to critical_enter/exit().
2001-03-28 03:06:10 +00:00
critical_t s;
Implement preemptive scheduling of hardware interrupt threads. - If possible, context switch to the thread directly in sched_ithd(), rather than triggering a delayed ast reschedule. - Disable interrupts while restoring fpu state in the trap handler, in order to ensure that we are not preempted in the middle, which could cause migration to another cpu. Reviewed by: peter Tested by: peter (alpha)
2001-02-01 03:34:20 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (!npx_exists)
return (0);
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
if (PCPU_GET(fpcurthread) != NULL) {
printf("npxdna: fpcurthread = %p, curthread = %p\n",
PCPU_GET(fpcurthread), curthread);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
panic("npxdna");
}
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
s = cpu_critical_enter();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
stop_emulating();
/*
* Record new context early in case frstor causes an IRQ13.
*/
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
PCPU_SET(fpcurthread, curthread);
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
exstat = GET_FPU_EXSW_PTR(PCPU_GET(curpcb));
*exstat = 0;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* The following frstor may cause an IRQ13 when the state being
* restored has a pending error. The error will appear to have been
* triggered by the current (npx) user instruction even when that
* instruction is a no-wait instruction that should not trigger an
* error (e.g., fnclex). On at least one 486 system all of the
* no-wait instructions are broken the same as frstor, so our
* treatment does not amplify the breakage. On at least one
* 386/Cyrix 387 system, fnclex works correctly while frstor and
* fnsave are broken, so our treatment breaks fnclex if it is the
* first FPU instruction after a context switch.
*/
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
fpurstor(&PCPU_GET(curpcb)->pcb_save);
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
cpu_critical_exit(s);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
return (1);
}
/*
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
* Wrapper for fnsave instruction, partly to handle hardware bugs. When npx
* exceptions are reported via IRQ13, spurious IRQ13's may be triggered by
* no-wait npx instructions. See the Intel application note AP-578 for
* details. This doesn't cause any additional complications here. IRQ13's
* are inherently asynchronous unless the CPU is frozen to deliver them --
* one that started in userland may be delivered many instructions later,
* after the process has entered the kernel. It may even be delivered after
* the fnsave here completes. A spurious IRQ13 for the fnsave is handled in
* the same way as a very-late-arriving non-spurious IRQ13 from user mode:
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
* it is normally ignored at first because we set fpcurthread to NULL; it is
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
* normally retriggered in npxdna() after return to user mode.
*
* npxsave() must be called with interrupts disabled, so that it clears
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
* fpcurthread atomically with saving the state. We require callers to do the
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
* disabling, since most callers need to disable interrupts anyway to call
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
* npxsave() atomically with checking fpcurthread.
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
*
* A previous version of 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.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
void
npxsave(addr)
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
union savefpu *addr;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
stop_emulating();
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
fpusave(addr);
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
Made the SMP case ignore the possibility of an INT13 interface. This eliminates all the APIC code, and thus several routines that would otherwise need to be made MP-safe. Reviewed by: Bruce Evans <bde@zeta.org.au>
1997-07-21 07:57:50 +00:00
start_emulating();
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
PCPU_SET(fpcurthread, NULL);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
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
static void
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
fpusave(addr)
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
union savefpu *addr;
{
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#ifdef CPU_ENABLE_SSE
if (cpu_fxsr)
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
fxsave(addr);
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
else
#endif
fnsave(addr);
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
}
static void
Use PCPU_GET(cpuid) instead of curproc->p_oncpu. Reviewed by: peter
2001-07-16 06:00:23 +00:00
fpurstor(addr)
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
union savefpu *addr;
{
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
#ifdef CPU_ENABLE_SSE
if (cpu_fxsr)
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
fxrstor(addr);
Dont compile in SSE fxsave/fxrstor instructions if CPU_ENABLE_SSE isn't active.
2001-08-23 01:03:56 +00:00
else
#endif
frstor(addr);
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
}
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
#ifdef I586_CPU_XXX
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
static long
timezero(funcname, func)
const char *funcname;
void (*func) __P((void *buf, size_t len));
{
void *buf;
Show the bzero() bandwidth in kBps instead of Bps; use u_int32_t instead of long and int64_t; and print the result as an unsigned long. This should make the output from the bzero() test more readable, and avoid printing a negative bandwidth. Note that this doesn't change the decision process, since that is based on time elapsed, not on computed bandwidth.
2001-03-19 00:28:04 +00:00
#define BUFSIZE 1048576
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
long usec;
struct timeval finish, start;
buf = malloc(BUFSIZE, M_TEMP, M_NOWAIT);
if (buf == NULL)
return (BUFSIZE);
microtime(&start);
(*func)(buf, BUFSIZE);
microtime(&finish);
usec = 1000000 * (finish.tv_sec - start.tv_sec) +
finish.tv_usec - start.tv_usec;
if (usec <= 0)
usec = 1;
if (bootverbose)
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
printf("%s bandwidth = %u kBps\n", funcname,
Show the bzero() bandwidth in kBps instead of Bps; use u_int32_t instead of long and int64_t; and print the result as an unsigned long. This should make the output from the bzero() test more readable, and avoid printing a negative bandwidth. Note that this doesn't change the decision process, since that is based on time elapsed, not on computed bandwidth.
2001-03-19 00:28:04 +00:00
(u_int32_t)(((BUFSIZE >> 10) * 1000000) / usec));
Only use the i586-optimized copying and zeroing functions if they are actually faster (more than 20% faster for zeroing 1 MB at boot time). This fixes pessimized copying and zeroing on K6's and perhaps on other CPUs that are misclassified as i586's.
1998-02-12 21:41:10 +00:00
free(buf, M_TEMP);
return (usec);
}
#endif /* I586_CPU */
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
static device_method_t npx_methods[] = {
/* Device interface */
Add an identify method to allow npx to arrange itself to be attached to the nexus without explicit code in the nexus to do so.
1999-08-22 19:52:51 +00:00
DEVMETHOD(device_identify, npx_identify),
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
DEVMETHOD(device_probe, npx_probe),
DEVMETHOD(device_attach, npx_attach),
DEVMETHOD(device_detach, bus_generic_detach),
DEVMETHOD(device_shutdown, bus_generic_shutdown),
DEVMETHOD(device_suspend, bus_generic_suspend),
DEVMETHOD(device_resume, bus_generic_resume),
{ 0, 0 }
};
static driver_t npx_driver = {
"npx",
npx_methods,
1, /* no softc */
};
static devclass_t npx_devclass;
/*
* We prefer to attach to the root nexus so that the usual case (exception 16)
* doesn't describe the processor as being `on isa'.
*/
DRIVER_MODULE(npx, nexus, npx_driver, npx_devclass, 0, 0);
Add a little do-nothing ``slopsucker'' device which gives a home to PNP0c04 (legacy ISA coprocessor support). Tourist info.
2000-05-04 23:57:32 +00:00
/*
Add ACPI attachments.
2001-08-30 09:17:03 +00:00
* This sucks up the legacy ISA support assignments from PNPBIOS/ACPI.
Add a little do-nothing ``slopsucker'' device which gives a home to PNP0c04 (legacy ISA coprocessor support). Tourist info.
2000-05-04 23:57:32 +00:00
*/
static struct isa_pnp_id npxisa_ids[] = {
{ 0x040cd041, "Legacy ISA coprocessor support" }, /* PNP0C04 */
{ 0 }
};
static int
npxisa_probe(device_t dev)
{
Make the PnP 'slopsucker' quiet in the !bootverbose case - the real NPX probe happens much earlier, and may come to very different conclusions about the system's NPX setup.
2000-06-23 08:19:50 +00:00
int result;
if ((result = ISA_PNP_PROBE(device_get_parent(dev), dev, npxisa_ids)) <= 0) {
device_quiet(dev);
}
return(result);
Add a little do-nothing ``slopsucker'' device which gives a home to PNP0c04 (legacy ISA coprocessor support). Tourist info.
2000-05-04 23:57:32 +00:00
}
static int
npxisa_attach(device_t dev)
{
return (0);
}
static device_method_t npxisa_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, npxisa_probe),
DEVMETHOD(device_attach, npxisa_attach),
DEVMETHOD(device_detach, bus_generic_detach),
DEVMETHOD(device_shutdown, bus_generic_shutdown),
DEVMETHOD(device_suspend, bus_generic_suspend),
DEVMETHOD(device_resume, bus_generic_resume),
{ 0, 0 }
};
static driver_t npxisa_driver = {
"npxisa",
npxisa_methods,
1, /* no softc */
};
static devclass_t npxisa_devclass;
DRIVER_MODULE(npxisa, isa, npxisa_driver, npxisa_devclass, 0, 0);
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
#ifndef PC98
Add ACPI attachments.
2001-08-30 09:17:03 +00:00
DRIVER_MODULE(npxisa, acpi, npxisa_driver, npxisa_devclass, 0, 0);
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
#endif
Add a little do-nothing ``slopsucker'' device which gives a home to PNP0c04 (legacy ISA coprocessor support). Tourist info.
2000-05-04 23:57:32 +00:00
Reference in New Issue Copy Permalink