[unix-history] / sys / kern / kern__physio.c

/*
 * Copyright (c) 1989, 1990, 1991, 1992 William F. Jolitz, TeleMuse
 * 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 software is a component of "386BSD" developed by 
	William F. Jolitz, TeleMuse.
 * 4. Neither the name of the developer nor the name "386BSD"
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS A COMPONENT OF 386BSD DEVELOPED BY WILLIAM F. JOLITZ 
 * AND IS INTENDED FOR RESEARCH AND EDUCATIONAL PURPOSES ONLY. THIS 
 * SOFTWARE SHOULD NOT BE CONSIDERED TO BE A COMMERCIAL PRODUCT. 
 * THE DEVELOPER URGES THAT USERS WHO REQUIRE A COMMERCIAL PRODUCT 
 * NOT MAKE USE THIS WORK.
 *
 * FOR USERS WHO WISH TO UNDERSTAND THE 386BSD SYSTEM DEVELOPED
 * BY WILLIAM F. JOLITZ, WE RECOMMEND THE USER STUDY WRITTEN 
 * REFERENCES SUCH AS THE  "PORTING UNIX TO THE 386" SERIES 
 * (BEGINNING JANUARY 1991 "DR. DOBBS JOURNAL", USA AND BEGINNING 
 * JUNE 1991 "UNIX MAGAZIN", GERMANY) BY WILLIAM F. JOLITZ AND 
 * LYNNE GREER JOLITZ, AS WELL AS OTHER BOOKS ON UNIX AND THE 
 * ON-LINE 386BSD USER MANUAL BEFORE USE. A BOOK DISCUSSING THE INTERNALS 
 * OF 386BSD ENTITLED "386BSD FROM THE INSIDE OUT" WILL BE AVAILABLE LATE 1992.
 *
 * THIS SOFTWARE IS PROVIDED BY THE DEVELOPER ``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 DEVELOPER 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.
 *
 *	$Id: kern__physio.c,v 1.6 1994/01/14 16:24:47 davidg Exp $
 */

#include "param.h"
#include "systm.h"
#include "buf.h"
#include "conf.h"
#include "proc.h"
#include "malloc.h"
#include "vnode.h"
#include "vm/vm.h"
#include "specdev.h"

/*
 * Driver interface to do "raw" I/O in the address space of a
 * user process directly for read and write operations..
 */

int
rawread(dev, uio)
	dev_t dev; struct uio *uio;
{
	return (uioapply(physio, (caddr_t) cdevsw[major(dev)].d_strategy,
				 (caddr_t) (u_long) dev, uio));
}

int
rawwrite(dev, uio)
	dev_t dev; struct uio *uio;
{
	return (uioapply(physio, (caddr_t) cdevsw[major(dev)].d_strategy,
				(caddr_t) (u_long) dev, uio));
}


int physio(strat, dev, bp, off, rw, base, len, p)
	d_strategy_t strat; 
	dev_t dev;
	struct buf *bp;
	int rw, off;
	caddr_t base;
	int *len;
	struct proc *p;
{
	int amttodo = *len;
	int error, amtdone;
	vm_prot_t ftype;
	vm_offset_t v, lastv;
	caddr_t adr;
	int oldflags;
	int s;
	
	int bp_alloc = (bp == 0);

/*
 * keep the process from being swapped
 */
	oldflags = p->p_flag;
	p->p_flag |= SPHYSIO;

	rw = rw == UIO_READ ? B_READ : 0;

	/* create and build a buffer header for a transfer */

	if (bp_alloc) {
		bp = (struct buf *)getpbuf();
		bzero((char *)bp, sizeof(*bp));			/* 09 Sep 92*/
	} else {
		s = splbio();
		while (bp->b_flags & B_BUSY) {
			bp->b_flags |= B_WANTED;
			tsleep((caddr_t)bp, PRIBIO, "physbw", 0);
		}
		bp->b_flags |= B_BUSY;
		splx(s);
	}

	bp->b_flags = B_BUSY | B_PHYS | rw;
	bp->b_proc = p;
	bp->b_dev = dev;
	bp->b_error = 0;
	bp->b_blkno = off/DEV_BSIZE;
	amtdone = 0;

	/* iteratively do I/O on as large a chunk as possible */
	do {
		bp->b_flags &= ~B_DONE;
		bp->b_un.b_addr = base;
		/* XXX limit */
		bp->b_bcount = min (256*1024, amttodo);

		/* first, check if accessible */
		if (rw == B_READ && !useracc(base, bp->b_bcount, B_WRITE)) {
			error = EFAULT;
			goto errrtn;
		}
		if (rw == B_WRITE && !useracc(base, bp->b_bcount, B_READ)) {
			error = EFAULT;
			goto errrtn;
		}

		/* update referenced and dirty bits, handle copy objects */
		if (rw == B_READ)
			ftype = VM_PROT_READ | VM_PROT_WRITE;
		else
			ftype = VM_PROT_READ;

		lastv = 0;
		for (adr = (caddr_t)trunc_page(base); adr < base + bp->b_bcount;
			adr += NBPG) {
	
/*
 * make sure that the pde is valid and wired
 */
			v = trunc_page(((vm_offset_t)vtopte(adr)));
			if (v != lastv) {
				vm_map_pageable(&p->p_vmspace->vm_map, v,
					round_page(v+1), FALSE);
				lastv = v;
			}

/*
 * do the vm_fault if needed, do the copy-on-write thing when
 * reading stuff off device into memory.
 */
			if (ftype & VM_PROT_WRITE) {
				/*
				 * properly handle copy-on-write
				 */
				*(volatile int *) adr += 0;
			}
#if 0
			else {
				/*
				 * this clause is not really necessary because 
				 * vslock does a vm_map_pageable FALSE.
				 * It is not optimally efficient to reference the
				 * page with the possiblity of it being paged out, but
				 * if this page is faulted here, it will be placed on the
				 * active queue, with the probability of it being paged
				 * out being very very low.  This is here primarily for
				 * "symmetry".
				 */
				*(volatile int *) adr;
			}
#endif
		}
/*
 * if the process has been blocked by the wiring of the page table pages
 * above or faults of other pages, then the vm_map_pageable contained in the
 * vslock will fault the pages back in if they have been paged out since
 * being referenced in the loop above. (vm_map_pageable calls vm_fault_wire
 * which calls vm_fault to get the pages if needed.)
 */

		/* lock in core (perform vm_map_pageable, FALSE) */
		vslock (base, bp->b_bcount);

		/* perform transfer */
		physstrat(bp, strat, PRIBIO);

		/* unlock (perform vm_map_pageable, TRUE) */
		vsunlock (base, bp->b_bcount, 0);

		lastv = 0;

/*
 * unwire the pde
 */
		for (adr = (caddr_t)trunc_page(base); adr < base + bp->b_bcount;
			adr += NBPG) {
			v = trunc_page(((vm_offset_t)vtopte(adr)));
			if (v != lastv) {
				vm_map_pageable(&p->p_vmspace->vm_map, v, round_page(v+1), TRUE);
				lastv = v;
			}
		}
			

		amtdone = bp->b_bcount - bp->b_resid;
		amttodo -= amtdone;
		base += amtdone;
		bp->b_blkno += amtdone/DEV_BSIZE;
	} while (amttodo && (bp->b_flags & B_ERROR) == 0 && amtdone > 0);

	error = bp->b_error;
errrtn:
	if (bp_alloc) {
		relpbuf(bp);
	} else {
		bp->b_flags &= ~B_BUSY;
		wakeup((caddr_t)bp);
	}
	*len = amttodo;

/*
 * allow the process to be swapped
 */
	p->p_flag &= ~SPHYSIO;
	p->p_flag |= (oldflags & SPHYSIO);

	return (error);
}
Commit	Line	Data
15637ed4 RG	1	/*
	2	* Copyright (c) 1989, 1990, 1991, 1992 William F. Jolitz, TeleMuse
	3	* All rights reserved.
	4	*
	5	* Redistribution and use in source and binary forms, with or without
	6	* modification, are permitted provided that the following conditions
	7	* are met:
	8	* 1. Redistributions of source code must retain the above copyright
	9	* notice, this list of conditions and the following disclaimer.
	10	* 2. Redistributions in binary form must reproduce the above copyright
	11	* notice, this list of conditions and the following disclaimer in the
	12	* documentation and/or other materials provided with the distribution.
	13	* 3. All advertising materials mentioning features or use of this software
	14	* must display the following acknowledgement:
	15	* This software is a component of "386BSD" developed by
	16	William F. Jolitz, TeleMuse.
	17	* 4. Neither the name of the developer nor the name "386BSD"
	18	* may be used to endorse or promote products derived from this software
	19	* without specific prior written permission.
	20	*
	21	* THIS SOFTWARE IS A COMPONENT OF 386BSD DEVELOPED BY WILLIAM F. JOLITZ
	22	* AND IS INTENDED FOR RESEARCH AND EDUCATIONAL PURPOSES ONLY. THIS
	23	* SOFTWARE SHOULD NOT BE CONSIDERED TO BE A COMMERCIAL PRODUCT.
	24	* THE DEVELOPER URGES THAT USERS WHO REQUIRE A COMMERCIAL PRODUCT
	25	* NOT MAKE USE THIS WORK.
	26	*
	27	* FOR USERS WHO WISH TO UNDERSTAND THE 386BSD SYSTEM DEVELOPED
	28	* BY WILLIAM F. JOLITZ, WE RECOMMEND THE USER STUDY WRITTEN
	29	* REFERENCES SUCH AS THE "PORTING UNIX TO THE 386" SERIES
	30	* (BEGINNING JANUARY 1991 "DR. DOBBS JOURNAL", USA AND BEGINNING
	31	* JUNE 1991 "UNIX MAGAZIN", GERMANY) BY WILLIAM F. JOLITZ AND
	32	* LYNNE GREER JOLITZ, AS WELL AS OTHER BOOKS ON UNIX AND THE
	33	* ON-LINE 386BSD USER MANUAL BEFORE USE. A BOOK DISCUSSING THE INTERNALS
	34	* OF 386BSD ENTITLED "386BSD FROM THE INSIDE OUT" WILL BE AVAILABLE LATE 1992.
	35	*
	36	* THIS SOFTWARE IS PROVIDED BY THE DEVELOPER ``AS IS'' AND
	37	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	38	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	39	* ARE DISCLAIMED. IN NO EVENT SHALL THE DEVELOPER BE LIABLE
	40	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	41	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	42	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	43	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	44	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	45	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	46	* SUCH DAMAGE.
	47	*
ce619eaa	48	* $Id: kern__physio.c,v 1.6 1994/01/14 16:24:47 davidg Exp $
15637ed4	49	*/
15637ed4 RG	50
	51	#include "param.h"
	52	#include "systm.h"
	53	#include "buf.h"
	54	#include "conf.h"
	55	#include "proc.h"
	56	#include "malloc.h"
	57	#include "vnode.h"
	58	#include "vm/vm.h"
	59	#include "specdev.h"
	60
15637ed4 RG	61	/*
	62	* Driver interface to do "raw" I/O in the address space of a
	63	* user process directly for read and write operations..
	64	*/
	65
4c45483e	66	int
15637ed4	67	rawread(dev, uio)
1d7f5b47	68	dev_t dev; struct uio *uio;
15637ed4	69	{
1d7f5b47 DG	70	return (uioapply(physio, (caddr_t) cdevsw[major(dev)].d_strategy,
1d7f5b47 DG	71	(caddr_t) (u_long) dev, uio));
15637ed4 RG	72	}
15637ed4 RG	73
4c45483e	74	int
15637ed4	75	rawwrite(dev, uio)
1d7f5b47	76	dev_t dev; struct uio *uio;
15637ed4	77	{
1d7f5b47 DG	78	return (uioapply(physio, (caddr_t) cdevsw[major(dev)].d_strategy,
1d7f5b47 DG	79	(caddr_t) (u_long) dev, uio));
15637ed4 RG	80	}
15637ed4 RG	81
1d7f5b47 DG	82
	83	int physio(strat, dev, bp, off, rw, base, len, p)
	84	d_strategy_t strat;
15637ed4	85	dev_t dev;
519fb2b7	86	struct buf *bp;
15637ed4 RG	87	int rw, off;
	88	caddr_t base;
	89	int *len;
	90	struct proc *p;
	91	{
1d7f5b47 DG	92	int amttodo = *len;
1d7f5b47 DG	93	int error, amtdone;
15637ed4	94	vm_prot_t ftype;
1d7f5b47	95	vm_offset_t v, lastv;
15637ed4	96	caddr_t adr;
1d7f5b47 DG	97	int oldflags;
	98	int s;
	99
519fb2b7	100	int bp_alloc = (bp == 0);
15637ed4	101
1d7f5b47 DG	102	/*
	103	* keep the process from being swapped
	104	*/
	105	oldflags = p->p_flag;
	106	p->p_flag \|= SPHYSIO;
	107
15637ed4 RG	108	rw = rw == UIO_READ ? B_READ : 0;
	109
	110	/* create and build a buffer header for a transfer */
519fb2b7 RG	111
519fb2b7 RG	112	if (bp_alloc) {
1d7f5b47	113	bp = (struct buf *)getpbuf();
519fb2b7	114	bzero((char )bp, sizeof(bp)); /* 09 Sep 92*/
1d7f5b47 DG	115	} else {
1d7f5b47 DG	116	s = splbio();
ce619eaa	117	while (bp->b_flags & B_BUSY) {
1d7f5b47 DG	118	bp->b_flags \|= B_WANTED;
	119	tsleep((caddr_t)bp, PRIBIO, "physbw", 0);
	120	}
	121	bp->b_flags \|= B_BUSY;
	122	splx(s);
519fb2b7	123	}
1d7f5b47	124
15637ed4 RG	125	bp->b_flags = B_BUSY \| B_PHYS \| rw;
	126	bp->b_proc = p;
	127	bp->b_dev = dev;
	128	bp->b_error = 0;
	129	bp->b_blkno = off/DEV_BSIZE;
	130	amtdone = 0;
	131
	132	/* iteratively do I/O on as large a chunk as possible */
	133	do {
	134	bp->b_flags &= ~B_DONE;
	135	bp->b_un.b_addr = base;
	136	/* XXX limit */
	137	bp->b_bcount = min (256*1024, amttodo);
	138
	139	/* first, check if accessible */
	140	if (rw == B_READ && !useracc(base, bp->b_bcount, B_WRITE)) {
1d7f5b47 DG	141	error = EFAULT;
1d7f5b47 DG	142	goto errrtn;
15637ed4 RG	143	}
15637ed4 RG	144	if (rw == B_WRITE && !useracc(base, bp->b_bcount, B_READ)) {
1d7f5b47 DG	145	error = EFAULT;
1d7f5b47 DG	146	goto errrtn;
15637ed4 RG	147	}
	148
	149	/* update referenced and dirty bits, handle copy objects */
	150	if (rw == B_READ)
	151	ftype = VM_PROT_READ \| VM_PROT_WRITE;
	152	else
	153	ftype = VM_PROT_READ;
1d7f5b47 DG	154
	155	lastv = 0;
	156	for (adr = (caddr_t)trunc_page(base); adr < base + bp->b_bcount;
15637ed4	157	adr += NBPG) {
1d7f5b47 DG	158
	159	/*
	160	* make sure that the pde is valid and wired
	161	*/
ce619eaa DG	162	v = trunc_page(((vm_offset_t)vtopte(adr)));
ce619eaa DG	163	if (v != lastv) {
1d7f5b47 DG	164	vm_map_pageable(&p->p_vmspace->vm_map, v,
	165	round_page(v+1), FALSE);
	166	lastv = v;
	167	}
	168
	169	/*
ce619eaa DG	170	* do the vm_fault if needed, do the copy-on-write thing when
ce619eaa DG	171	* reading stuff off device into memory.
1d7f5b47	172	*/
ce619eaa DG	173	if (ftype & VM_PROT_WRITE) {
	174	/*
	175	* properly handle copy-on-write
	176	*/
1d7f5b47	177	(volatile int ) adr += 0;
ce619eaa DG	178	}
	179	#if 0
	180	else {
	181	/*
	182	* this clause is not really necessary because
	183	* vslock does a vm_map_pageable FALSE.
	184	* It is not optimally efficient to reference the
	185	* page with the possiblity of it being paged out, but
	186	* if this page is faulted here, it will be placed on the
	187	* active queue, with the probability of it being paged
	188	* out being very very low. This is here primarily for
	189	* "symmetry".
	190	*/
1d7f5b47	191	(volatile int ) adr;
ce619eaa DG	192	}
ce619eaa DG	193	#endif
15637ed4	194	}
ce619eaa DG	195	/*
	196	* if the process has been blocked by the wiring of the page table pages
	197	* above or faults of other pages, then the vm_map_pageable contained in the
	198	* vslock will fault the pages back in if they have been paged out since
	199	* being referenced in the loop above. (vm_map_pageable calls vm_fault_wire
	200	* which calls vm_fault to get the pages if needed.)
	201	*/
15637ed4	202
ce619eaa	203	/* lock in core (perform vm_map_pageable, FALSE) */
15637ed4 RG	204	vslock (base, bp->b_bcount);
	205
	206	/* perform transfer */
	207	physstrat(bp, strat, PRIBIO);
	208
ce619eaa	209	/* unlock (perform vm_map_pageable, TRUE) */
15637ed4	210	vsunlock (base, bp->b_bcount, 0);
1d7f5b47 DG	211
	212	lastv = 0;
	213
	214	/*
	215	* unwire the pde
	216	*/
	217	for (adr = (caddr_t)trunc_page(base); adr < base + bp->b_bcount;
	218	adr += NBPG) {
ce619eaa DG	219	v = trunc_page(((vm_offset_t)vtopte(adr)));
ce619eaa DG	220	if (v != lastv) {
1d7f5b47 DG	221	vm_map_pageable(&p->p_vmspace->vm_map, v, round_page(v+1), TRUE);
	222	lastv = v;
	223	}
	224	}
	225
	226
15637ed4 RG	227	amtdone = bp->b_bcount - bp->b_resid;
	228	amttodo -= amtdone;
	229	base += amtdone;
	230	bp->b_blkno += amtdone/DEV_BSIZE;
	231	} while (amttodo && (bp->b_flags & B_ERROR) == 0 && amtdone > 0);
	232
	233	error = bp->b_error;
1d7f5b47 DG	234	errrtn:
	235	if (bp_alloc) {
	236	relpbuf(bp);
	237	} else {
	238	bp->b_flags &= ~B_BUSY;
	239	wakeup((caddr_t)bp);
	240	}
15637ed4	241	*len = amttodo;
1d7f5b47 DG	242
	243	/*
	244	* allow the process to be swapped
	245	*/
	246	p->p_flag &= ~SPHYSIO;
	247	p->p_flag \|= (oldflags & SPHYSIO);
	248
15637ed4 RG	249	return (error);
15637ed4 RG	250	}