2 * Copyright (c) 1990 The Regents of the University of California.
4 * LWKT threads Copyright (c) 2003 Matthew Dillon
6 * This code is derived from software contributed to Berkeley by
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by the University of
20 * California, Berkeley and its contributors.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
37 * $FreeBSD: src/sys/i386/i386/swtch.s,v 1.89.2.10 2003/01/23 03:36:24 ps Exp $
38 * $DragonFly: src/sys/i386/i386/Attic/swtch.s,v 1.34 2004/05/05 19:26:38 dillon Exp $
43 #include <sys/rtprio.h>
45 #include <machine/asmacros.h>
46 #include <machine/ipl.h>
48 #include <machine/pmap.h>
49 #include <machine/smptests.h> /** GRAB_LOPRIO */
50 #include <machine/apicreg.h>
51 #include <machine/lock.h>
56 #define MPLOCKED lock ;
65 #if defined(SWTCH_OPTIM_STATS)
66 .globl swtch_optim_stats, tlb_flush_count
67 swtch_optim_stats: .long 0 /* number of _swtch_optims */
68 tlb_flush_count: .long 0
75 * cpu_heavy_switch(next_thread)
77 * Switch from the current thread to a new thread. This entry
78 * is normally called via the thread->td_switch function, and will
79 * only be called when the current thread is a heavy weight process.
81 * Some instructions have been reordered to reduce pipeline stalls.
83 * YYY disable interrupts once giant is removed.
85 ENTRY(cpu_heavy_switch)
89 movl PCPU(curthread),%ecx
90 movl (%esp),%eax /* (reorder optimization) */
91 movl TD_PCB(%ecx),%edx /* EDX = PCB */
92 movl %eax,PCB_EIP(%edx) /* return PC may be modified */
93 movl %ebx,PCB_EBX(%edx)
94 movl %esp,PCB_ESP(%edx)
95 movl %ebp,PCB_EBP(%edx)
96 movl %esi,PCB_ESI(%edx)
97 movl %edi,PCB_EDI(%edx)
100 movl %ecx,%ebx /* EBX = curthread */
101 movl TD_PROC(%ecx),%ecx
102 movl PCPU(cpuid), %eax
103 movl P_VMSPACE(%ecx), %ecx /* ECX = vmspace */
104 MPLOCKED btrl %eax, VM_PMAP+PM_ACTIVE(%ecx)
107 * Push the LWKT switch restore function, which resumes a heavy
108 * weight process. Note that the LWKT switcher is based on
109 * TD_SP, while the heavy weight process switcher is based on
110 * PCB_ESP. TD_SP is usually two ints pushed relative to
111 * PCB_ESP. We push the flags for later restore by cpu_heavy_restore.
114 pushl $cpu_heavy_restore
115 movl %esp,TD_SP(%ebx)
118 * Save debug regs if necessary
120 movb PCB_FLAGS(%edx),%al
122 jz 1f /* no, skip over */
123 movl %dr7,%eax /* yes, do the save */
124 movl %eax,PCB_DR7(%edx)
125 andl $0x0000fc00, %eax /* disable all watchpoints */
128 movl %eax,PCB_DR6(%edx)
130 movl %eax,PCB_DR3(%edx)
132 movl %eax,PCB_DR2(%edx)
134 movl %eax,PCB_DR1(%edx)
136 movl %eax,PCB_DR0(%edx)
141 * Save the FP state if we have used the FP. Note that calling
142 * npxsave will NULL out PCPU(npxthread).
144 cmpl %ebx,PCPU(npxthread)
146 pushl TD_SAVEFPU(%ebx)
147 call npxsave /* do it in a big C function */
148 addl $4,%esp /* EAX, ECX, EDX trashed */
150 #endif /* NNPX > 0 */
153 * Switch to the next thread, which was passed as an argument
154 * to cpu_heavy_switch(). Due to the eflags and switch-restore
155 * function we pushed, the argument is at 12(%esp). Set the current
156 * thread, load the stack pointer, and 'ret' into the switch-restore
159 * The switch restore function expects the new thread to be in %eax
160 * and the old one to be in %ebx.
162 * There is a one-instruction window where curthread is the new
163 * thread but %esp still points to the old thread's stack, but
164 * we are protected by a critical section so it is ok.
166 movl 12(%esp),%eax /* EAX = newtd, EBX = oldtd */
167 movl %eax,PCPU(curthread)
168 movl TD_SP(%eax),%esp
174 * The switch function is changed to this when a thread is going away
175 * for good. We have to ensure that the MMU state is not cached, and
176 * we don't bother saving the existing thread state before switching.
178 * At this point we are in a critical section and this cpu owns the
179 * thread's token, which serves as an interlock until the switchout is
182 ENTRY(cpu_exit_switch)
184 * Get us out of the vmspace
192 movl PCPU(curthread),%ebx
194 * Switch to the next thread. RET into the restore function, which
195 * expects the new thread in EAX and the old in EBX.
197 * There is a one-instruction window where curthread is the new
198 * thread but %esp still points to the old thread's stack, but
199 * we are protected by a critical section so it is ok.
202 movl %eax,PCPU(curthread)
203 movl TD_SP(%eax),%esp
207 * cpu_heavy_restore() (current thread in %eax on entry)
209 * Restore the thread after an LWKT switch. This entry is normally
210 * called via the LWKT switch restore function, which was pulled
211 * off the thread stack and jumped to.
213 * This entry is only called if the thread was previously saved
214 * using cpu_heavy_switch() (the heavy weight process thread switcher),
215 * or when a new process is initially scheduled. The first thing we
216 * do is clear the TDF_RUNNING bit in the old thread and set it in the
219 * YYY theoretically we do not have to restore everything here, a lot
220 * of this junk can wait until we return to usermode. But for now
221 * we restore everything.
223 * YYY the PCB crap is really crap, it makes startup a bitch because
224 * we can't switch away.
226 * YYY note: spl check is done in mi_switch when it splx()'s.
229 ENTRY(cpu_heavy_restore)
231 movl TD_PCB(%eax),%edx /* EDX = PCB */
232 movl TD_PROC(%eax),%ecx
235 * A heavy weight process will normally be in an SRUN state
236 * but can also be preempted while it is entering a SZOMB
239 cmpb $SRUN,P_STAT(%ecx)
241 cmpb $SZOMB,P_STAT(%ecx)
246 #if defined(SWTCH_OPTIM_STATS)
247 incl _swtch_optim_stats
250 * Tell the pmap that our cpu is using the VMSPACE now. We cannot
251 * safely test/reload %cr3 until after we have set the bit in the
252 * pmap (remember, we do not hold the MP lock in the switch code).
254 movl P_VMSPACE(%ecx), %ecx /* ECX = vmspace */
255 movl PCPU(cpuid), %esi
256 MPLOCKED btsl %esi, VM_PMAP+PM_ACTIVE(%ecx)
259 * Restore the MMU address space. If it is the same as the last
260 * thread we don't have to invalidate the tlb (i.e. reload cr3).
261 * YYY which naturally also means that the PM_ACTIVE bit had better
262 * already have been set before we set it above, check? YYY
265 movl PCB_CR3(%edx),%ecx
268 #if defined(SWTCH_OPTIM_STATS)
269 decl _swtch_optim_stats
270 incl _tlb_flush_count
275 * Clear TDF_RUNNING flag in old thread only after cleaning up
276 * %cr3. The target thread is already protected by being TDF_RUNQ
277 * so setting TDF_RUNNING isn't as big a deal.
279 andl $~TDF_RUNNING,TD_FLAGS(%ebx)
280 orl $TDF_RUNNING,TD_FLAGS(%eax)
283 * Deal with the PCB extension, restore the private tss
285 movl PCB_EXT(%edx),%edi /* check for a PCB extension */
286 movl $1,%ebx /* maybe mark use of a private tss */
291 * Going back to the common_tss. We may need to update TSS_ESP0
292 * which sets the top of the supervisor stack when entering from
293 * usermode. The PCB is at the top of the stack but we need another
294 * 16 bytes to take vm86 into account.
297 movl %ebx, PCPU(common_tss) + TSS_ESP0
299 cmpl $0,PCPU(private_tss) /* don't have to reload if */
300 je 3f /* already using the common TSS */
302 subl %ebx,%ebx /* unmark use of private tss */
305 * Get the address of the common TSS descriptor for the ltr.
306 * There is no way to get the address of a segment-accessed variable
307 * so we store a self-referential pointer at the base of the per-cpu
308 * data area and add the appropriate offset.
310 movl $gd_common_tssd, %edi
314 * Move the correct TSS descriptor into the GDT slot, then reload
318 movl %ebx,PCPU(private_tss) /* mark/unmark private tss */
319 movl PCPU(tss_gdt), %ebx /* entry in GDT */
324 movl $GPROC0_SEL*8, %esi /* GSEL(entry, SEL_KPL) */
329 * Restore general registers.
331 movl PCB_EBX(%edx),%ebx
332 movl PCB_ESP(%edx),%esp
333 movl PCB_EBP(%edx),%ebp
334 movl PCB_ESI(%edx),%esi
335 movl PCB_EDI(%edx),%edi
336 movl PCB_EIP(%edx),%eax
340 * Restore the user LDT if we have one
342 cmpl $0, PCB_USERLDT(%edx)
344 movl _default_ldt,%eax
345 cmpl PCPU(currentldt),%eax
348 movl %eax,PCPU(currentldt)
355 * Restore the %gs segment register, which must be done after
356 * loading the user LDT. Since user processes can modify the
357 * register via procfs, this may result in a fault which is
358 * detected by checking the fault address against cpu_switch_load_gs
359 * in i386/i386/trap.c
361 .globl cpu_switch_load_gs
363 movl PCB_GS(%edx),%gs
366 * Restore the DEBUG register state if necessary.
368 movb PCB_FLAGS(%edx),%al
370 jz 1f /* no, skip over */
371 movl PCB_DR6(%edx),%eax /* yes, do the restore */
373 movl PCB_DR3(%edx),%eax
375 movl PCB_DR2(%edx),%eax
377 movl PCB_DR1(%edx),%eax
379 movl PCB_DR0(%edx),%eax
381 movl %dr7,%eax /* load dr7 so as not to disturb */
382 andl $0x0000fc00,%eax /* reserved bits */
384 movl PCB_DR7(%edx),%ebx
385 andl $~0x0000fc00,%ebx
397 sw0_2: .asciz "cpu_switch: not SRUN"
402 * Update pcb, saving current processor state.
408 /* caller's return address - child won't execute this routine */
410 movl %eax,PCB_EIP(%ecx)
413 movl %eax,PCB_CR3(%ecx)
415 movl %ebx,PCB_EBX(%ecx)
416 movl %esp,PCB_ESP(%ecx)
417 movl %ebp,PCB_EBP(%ecx)
418 movl %esi,PCB_ESI(%ecx)
419 movl %edi,PCB_EDI(%ecx)
420 movl %gs,PCB_GS(%ecx)
424 * If npxthread == NULL, then the npx h/w state is irrelevant and the
425 * state had better already be in the pcb. This is true for forks
426 * but not for dumps (the old book-keeping with FP flags in the pcb
427 * always lost for dumps because the dump pcb has 0 flags).
429 * If npxthread != NULL, then we have to save the npx h/w state to
430 * npxthread's pcb and copy it to the requested pcb, or save to the
431 * requested pcb and reload. Copying is easier because we would
432 * have to handle h/w bugs for reloading. We used to lose the
433 * parent's npx state for forks by forgetting to reload.
435 movl PCPU(npxthread),%eax
439 pushl %ecx /* target pcb */
440 movl TD_SAVEFPU(%eax),%eax /* originating savefpu area */
450 pushl $PCB_SAVEFPU_SIZE
451 leal PCB_SAVEFPU(%ecx),%ecx
456 #endif /* NNPX > 0 */
462 * cpu_idle_restore() (current thread in %eax on entry) (one-time execution)
464 * Don't bother setting up any regs other then %ebp so backtraces
465 * don't die. This restore function is used to bootstrap into the
466 * cpu_idle() LWKT only, after that cpu_lwkt_*() will be used for
469 * Clear TDF_RUNNING in old thread only after we've cleaned up %cr3.
471 * If we are an AP we have to call ap_init() before jumping to
472 * cpu_idle(). ap_init() will synchronize with the BP and finish
473 * setting up various ncpu-dependant globaldata fields. This may
474 * happen on UP as well as SMP if we happen to be simulating multiple
477 ENTRY(cpu_idle_restore)
483 andl $~TDF_RUNNING,TD_FLAGS(%ebx)
484 orl $TDF_RUNNING,TD_FLAGS(%eax)
495 * cpu_kthread_restore() (current thread is %eax on entry) (one-time execution)
497 * Don't bother setting up any regs other then %ebp so backtraces
498 * don't die. This restore function is used to bootstrap into an
499 * LWKT based kernel thread only. cpu_lwkt_switch() will be used
502 * Since all of our context is on the stack we are reentrant and
503 * we can release our critical section and enable interrupts early.
505 ENTRY(cpu_kthread_restore)
508 movl TD_PCB(%eax),%edx
511 andl $~TDF_RUNNING,TD_FLAGS(%ebx)
512 orl $TDF_RUNNING,TD_FLAGS(%eax)
513 subl $TDPRI_CRIT,TD_PRI(%eax)
514 popl %eax /* kthread exit function */
515 pushl PCB_EBX(%edx) /* argument to ESI function */
516 pushl %eax /* set exit func as return address */
517 movl PCB_ESI(%edx),%eax
523 * Standard LWKT switching function. Only non-scratch registers are
524 * saved and we don't bother with the MMU state or anything else.
526 * This function is always called while in a critical section.
528 * There is a one-instruction window where curthread is the new
529 * thread but %esp still points to the old thread's stack, but
530 * we are protected by a critical section so it is ok.
534 ENTRY(cpu_lwkt_switch)
535 pushl %ebp /* note: GDB hacked to locate ebp relative to td_sp */
537 movl PCPU(curthread),%ebx
541 /* warning: adjust movl into %eax below if you change the pushes */
545 * Save the FP state if we have used the FP. Note that calling
546 * npxsave will NULL out PCPU(npxthread).
548 * We have to deal with the FP state for LWKT threads in case they
549 * happen to get preempted or block while doing an optimized
550 * bzero/bcopy/memcpy.
552 cmpl %ebx,PCPU(npxthread)
554 pushl TD_SAVEFPU(%ebx)
555 call npxsave /* do it in a big C function */
556 addl $4,%esp /* EAX, ECX, EDX trashed */
558 #endif /* NNPX > 0 */
560 movl 4+20(%esp),%eax /* switch to this thread */
561 pushl $cpu_lwkt_restore
562 movl %esp,TD_SP(%ebx)
563 movl %eax,PCPU(curthread)
564 movl TD_SP(%eax),%esp
567 * eax contains new thread, ebx contains old thread.
572 * cpu_lwkt_restore() (current thread in %eax on entry)
574 * Standard LWKT restore function. This function is always called
575 * while in a critical section.
577 * Warning: due to preemption the restore function can be used to
578 * 'return' to the original thread. Interrupt disablement must be
579 * protected through the switch so we cannot run splz here.
581 * YYY we theoretically do not need to load IdlePTD into cr3, but if
582 * so we need a way to detect when the PTD we are using is being
583 * deleted due to a process exiting.
585 ENTRY(cpu_lwkt_restore)
586 movl IdlePTD,%ecx /* YYY borrow but beware desched/cpuchg/exit */
592 andl $~TDF_RUNNING,TD_FLAGS(%ebx)
593 orl $TDF_RUNNING,TD_FLAGS(%eax)