l.s

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

/*
 * Memory and machine-specific definitions.  Used in C and assembler.
 */

/*
 * Sizes
 */

#define	BI2BY		8			/* bits per byte */
#define BI2WD		32			/* bits per word */
#define	BY2WD		4			/* bytes per word */
#define	BY2PG		4096			/* bytes per page */
#define	WD2PG		(BY2PG/BY2WD)		/* words per page */
#define	PGSHIFT		12			/* log(BY2PG) */
#define PGROUND(s)	(((s)+(BY2PG-1))&~(BY2PG-1))

#define	MAXMACH		1			/* max # cpus system can run */

/*
 * Time
 */
#define	HZ		20			/* clock frequency */
#define	MS2HZ		(1000/HZ)		/* millisec per clock tick */
#define	TK2SEC(t)	((t)/HZ)		/* ticks to seconds */
#define	TK2MS(t)	((((ulong)(t))*1000)/HZ)	/* ticks to milliseconds */
#define	MS2TK(t)	((((ulong)(t))*HZ)/1000)	/* milliseconds to ticks */

/*
 * PSR bits
 */
#define	PSREC		0x00002000
#define	PSREF		0x00001000
#define PSRSUPER	0x00000080
#define PSRPSUPER	0x00000040
#define	PSRET		0x00000020
#define SPL(n)		(n<<8)

/*
 * Magic registers
 */

#define	MACH		6		/* R6 is m-> */
#define	USER		5		/* R5 is u-> */

/*
 * Fundamental addresses
 */

#define	USERADDR	0xE0000000
#define	UREGADDR	(USERADDR+BY2PG-((32+6)*BY2WD))
#define	BOOTSTACK	(KTZERO-0*BY2PG)
#define	TRAPS		(KTZERO-2*BY2PG)

/*
 * MMU
 */

#define	VAMASK		0x3FFFFFFF
#define	NPMEG		(1<<12)
#define	BY2SEGM		(1<<18)
#define	PG2SEGM		(1<<6)
#define	NTLBPID		(1+NCONTEXT)	/* TLBPID 0 is unallocated */
#define	NCONTEXT	8
#define	CONTEXT		0x30000000	/* in ASI 2 */

/*
 * MMU regions
 */
#define	INVALIDSEGM	0xFFFC0000	/* highest seg of VA reserved as invalid */
#define	INVALIDPMEG	0x7F
#define	SCREENSEGM	0xFFF80000
#define	SCREENPMEG	0x7E
#define	ROMSEGM		0xFFE80000
#define	ROMEND		0xFFEA0000
#define	PG2ROM		((ROMEND-ROMSEGM)/BY2PG)
#define	IOSEGM0		ROMSEGM		/* see mmuinit() */
#define	NIOSEGM		((SCREENSEGM-ROMSEGM)/BY2SEGM)
#define	IOPMEG0		(SCREENPMEG-NIOSEGM)
#define	IOSEGM		ROMEND
#define	IOEND		SCREENSEGM
#define	TOPPMEG		IOPMEG0

/*
 * MMU entries
 */
#define	PTEVALID	(1<<31)
#define	PTERONLY	(0<<30)
#define	PTEWRITE	(1<<30)
#define	PTEKERNEL	(1<<29)
#define	PTENOCACHE	(1<<28)
#define	PTEMAINMEM	(0<<26)
#define	PTEIO		(1<<26)
#define	PTEACCESS	(1<<25)
#define	PTEMODIFY	(1<<24)
#define PTEUNCACHED	0
#define PTEMAPMEM	(1024*1024)	
#define	PTEPERTAB	(PTEMAPMEM/BY2PG)
#define SEGMAPSIZE	16

#define	INVALIDPTE	0
#define	PPN(pa)		((pa>>12)&0xFFFF)

/*
 * Weird addresses in various ASI's
 */
#define	CACHETAGS	0x80000000		/* ASI 2 */
#define	CACHEDATA	0x90000000		/* ASI 2 */
#define	SER		0x60000000		/* ASI 2 */
#define	SEVAR		0x60000004		/* ASI 2 */
#define	ASER		0x60000008		/* ASI 2 */
#define	ASEVAR		0x6000000C		/* ASI 2 */
#define	ENAB		0x40000000		/* ASI 2 */
#define	ENABCACHE	0x10
#define	ENABRESET	0x04

/*
 * Virtual addresses
 */
#define	VTAG(va)	((va>>22)&0x03F)
#define	VPN(va)		((va>>13)&0x1FF)

#define	PARAM		((char*)0x40500000)
#define	TLBFLUSH_	0x01

/*
 * Address spaces
 */

#define	UZERO	0x00000000		/* base of user address space */
#define	UTZERO	(UZERO+BY2PG)		/* first address in user text */
#define	TSTKTOP	0x10000000		/* end of new stack in sysexec */
#define TSTKSIZ 32
#define	USTKTOP	(TSTKTOP-TSTKSIZ*BY2PG)	/* byte just beyond user stack */
#define	KZERO	0xE0000000		/* base of kernel address space */
#define	KTZERO	(KZERO+4*BY2PG)		/* first address in kernel text */
#define	USTKSIZE	(4*1024*1024)	/* size of user stack */

#define	MACHSIZE	4096

#define isphys(x) (((ulong)(x)&0xF0000000) == KZERO)

#define	SYSPSR	(SPL(0x0)|PSREF|PSRSUPER|0)
#define	NOOP	OR R0, R0; OR R0, R0; OR R0, R0

TEXT	start(SB), $-4

	/* get virtual, fast */
	/* we are executing in segment 0, mapped to pmeg 0. stack is there too */
	/* get virtual by mapping segment(KZERO) to pmeg 0., and next to 1 */
	MOVW	$KZERO, R7
	MOVB	R0, (R7, 3)
	MOVW	$(KZERO+BY2SEGM), R7
	MOVW	$1, R8
	MOVB	R8, (R7, 3)
	/* now mapped correctly.  jmpl to where we want to be */
	MOVW	$setSB(SB), R2
	MOVW	$startvirt(SB), R7
	JMPL	(R7)
	MOVW	$_mul(SB), R0	/* touch _mul etc.; doesn't need to execute */
	RETURN			/* can't get here */

TEXT	startvirt(SB), $-4

	MOVW	$BOOTSTACK, R1

	MOVW	$(SPL(0xF)|PSREF|PSRSUPER), R7
	MOVW	R7, PSR

	MOVW	$(0x35<<22), R7		/* NVM OFM DZM AU */
	MOVW	R7, fsr+0(SB)
	MOVW	fsr+0(SB), FSR
	FMOVD	$0.5, F26		/* 0.5 -> F26 */
	FSUBD	F26, F26, F24		/* 0.0 -> F24 */
	FADDD	F26, F26, F28		/* 1.0 -> F28 */
	FADDD	F28, F28, F30		/* 2.0 -> F30 */

	FMOVD	F24, F0
	FMOVD	F24, F2
	FMOVD	F24, F4
	FMOVD	F24, F6
	FMOVD	F24, F8
	FMOVD	F24, F10
	FMOVD	F24, F12
	FMOVD	F24, F14
	FMOVD	F24, F16
	FMOVD	F24, F18
	FMOVD	F24, F20
	FMOVD	F24, F22

	MOVW	$mach0(SB), R(MACH)
/*	MOVW	$0x8, R7 /**/
	MOVW	R0, WIM
	JMPL	main(SB)
	MOVW	(R0), R0
	RETURN

TEXT	swap1(SB), $0

	TAS	(R7), R7		/* LDSTUB, thank you ken */
	RETURN

TEXT	swap1_should_work(SB), $0

	MOVW	R7, R8
	MOVW	$1, R7
	SWAP	(R8), R7
	RETURN

TEXT	swap1x(SB), $0

	MOVW	PSR, R9
	MOVW	R9, R10
	AND	$~PSRET, R10		/* BUG: book says this is buggy */
	MOVW	R10, PSR
	NOOP
	MOVW	(R7), R7
	CMP	R7, R0
	BNE	was1
	MOVW	$1, R10
	MOVW	R10, (R8)
was1:
	MOVW	R9, PSR
	RETURN

TEXT	spllo(SB), $0

	MOVW	PSR, R7
	MOVW	R7, R10
	OR	$PSRET, R10
	MOVW	R10, PSR
	NOOP
	RETURN

TEXT	splhi(SB), $0

	MOVW	R15, 4(R(MACH))	/* save PC in m->splpc */
	MOVW	PSR, R7
	MOVW	R7, R10
	AND	$~PSRET, R10	/* BUG: book says this is buggy */
	MOVW	R10, PSR
	NOOP
	RETURN

TEXT	splx(SB), $0

	MOVW	R15, 4(R(MACH))	/* save PC in m->splpc */
	MOVW	R7, PSR		/* BUG: book says this is buggy */
	NOOP
	RETURN

TEXT	spldone(SB), $0

	RETURN

TEXT	touser(SB), $0
	MOVW	$(SYSPSR&~PSREF), R8
	MOVW	R8, PSR
	NOOP

	MOVW	R7, R1
	SAVE	R0, R0			/* RETT is implicit RESTORE */
	MOVW	$(UTZERO+32), R7	/* PC; header appears in text */
	MOVW	$(UTZERO+32+4), R8	/* nPC */
	RETT	R7, R8

TEXT	rfnote(SB), $0

	MOVW	R7, R1			/* 1st arg is &uregpointer */
	ADD	$4, R1			/* point at ureg */
	JMP	restore

TEXT	traplink(SB), $-4

	/* R8 to R23 are free to play with */
	/* R17 contains PC, R18 contains nPC */
	/* R19 has PSR loaded from vector code */

	ANDCC	$PSRPSUPER, R19, R0
	BE	usertrap

kerneltrap:
	/*
	 * Interrupt or fault from kernel
	 */
	ANDN	$7, R1, R20			/* dbl aligned */
	MOVW	R1, (0-(4*(32+6))+(4*1))(R20)	/* save R1=SP */
	/* really clumsy: store these in Ureg so can be restored below */
	MOVW	R2, (0-(4*(32+6))+(4*2))(R20)	/* SB */
	MOVW	R5, (0-(4*(32+6))+(4*5))(R20)	/* USER */
	MOVW	R6, (0-(4*(32+6))+(4*6))(R20)	/* MACH */
	SUB	$(4*(32+6)), R20, R1

trap1:
	MOVW	Y, R20
	MOVW	R20, (4*(32+0))(R1)		/* Y */
	MOVW	TBR, R20
	MOVW	R20, (4*(32+1))(R1)		/* TBR */
	AND	$~0x1F, R19			/* force CWP=0 */
	MOVW	R19, (4*(32+2))(R1)		/* PSR */
	MOVW	R18, (4*(32+3))(R1)		/* nPC */
	MOVW	R17, (4*(32+4))(R1)		/* PC */
	MOVW	R0, (4*0)(R1)
	MOVW	R3, (4*3)(R1)
	MOVW	R4, (4*4)(R1)
	MOVW	R7, (4*7)(R1)
	RESTORE	R0, R0
	/* now our registers R8-R31 are same as before trap */
	/* save registers two at a time */
	MOVD	R8, (4*8)(R1)
	MOVD	R10, (4*10)(R1)
	MOVD	R12, (4*12)(R1)
	MOVD	R14, (4*14)(R1)
	MOVD	R16, (4*16)(R1)
	MOVD	R18, (4*18)(R1)
	MOVD	R20, (4*20)(R1)
	MOVD	R22, (4*22)(R1)
	MOVD	R24, (4*24)(R1)
	MOVD	R26, (4*26)(R1)
	MOVD	R28, (4*28)(R1)
	MOVD	R30, (4*30)(R1)
	/* SP and SB and u and m are already set; away we go */
	MOVW	R1, R7		/* pointer to Ureg */
	SUB	$8, R1
	MOVW	$SYSPSR, R8
	MOVW	R8, PSR
	NOOP
	JMPL	trap(SB)

	ADD	$8, R1
restore:
	MOVW	(4*(32+2))(R1), R8		/* PSR */
	MOVW	R8, PSR
	NOOP

	MOVD	(4*30)(R1), R30
	MOVD	(4*28)(R1), R28
	MOVD	(4*26)(R1), R26
	MOVD	(4*24)(R1), R24
	MOVD	(4*22)(R1), R22
	MOVD	(4*20)(R1), R20
	MOVD	(4*18)(R1), R18
	MOVD	(4*16)(R1), R16
	MOVD	(4*14)(R1), R14
	MOVD	(4*12)(R1), R12
	MOVD	(4*10)(R1), R10
	MOVD	(4*8)(R1), R8
	SAVE	R0, R0
	MOVD	(4*6)(R1), R6
	MOVD	(4*4)(R1), R4
	MOVD	(4*2)(R1), R2
	MOVW	(4*(32+0))(R1), R20		/* Y */
	MOVW	R20, Y
	MOVW	(4*(32+4))(R1), R17		/* PC */
	MOVW	(4*(32+3))(R1), R18		/* nPC */
	MOVW	(4*1)(R1), R1	/* restore R1=SP */
	RETT	R17, R18
	
usertrap:
	/*
	 * Interrupt or fault from user
	 */
	MOVW	R1, R8
	MOVW	R2, R9
	MOVW	$setSB(SB), R2
	MOVW	$(USERADDR+BY2PG), R1
	MOVW	R8, (0-(4*(32+6))+(4*1))(R1)	/* save R1=SP */
	MOVW	R9, (0-(4*(32+6))+(4*2))(R1)	/* save R2=SB */
	MOVW	R5, (0-(4*(32+6))+(4*5))(R1)	/* save R5=USER */
	MOVW	R6, (0-(4*(32+6))+(4*6))(R1)	/* save R6=MACH */
	MOVW	$USERADDR, R(USER)
	MOVW	$mach0(SB), R(MACH)
	SUB	$(4*(32+6)), R1
	JMP	trap1

TEXT	syslink(SB), $-4

	/* R8 to R23 are free to play with */
	/* R17 contains PC, R18 contains nPC */
	/* R19 has PSR loaded from vector code */
	/* assume user did it; syscall checks */

	MOVW	R1, R8
	MOVW	R2, R9
	MOVW	$setSB(SB), R2
	MOVW	$(USERADDR+BY2PG), R1
	MOVW	R8, (0-(4*(32+6))+4)(R1)	/* save R1=SP */
	SUB	$(4*(32+6)), R1
	MOVW	R9, (4*2)(R1)			/* save R2=SB */
	MOVW	R3, (4*3)(R1)			/* global register */
	MOVD	R4, (4*4)(R1)			/* global register, R5=USER */
	MOVD	R6, (4*6)(R1)			/* save R6=MACH, R7=syscall# */
	MOVW	$USERADDR, R(USER)
	MOVW	$mach0(SB), R(MACH)
	MOVW	TBR, R20
	MOVW	R20, (4*(32+1))(R1)		/* TBR */
	AND	$~0x1F, R19
	MOVW	R19, (4*(32+2))(R1)		/* PSR */
	MOVW	R18, (4*(32+3))(R1)		/* nPC */
	MOVW	R17, (4*(32+4))(R1)		/* PC */
	RESTORE	R0, R0
	/* now our registers R8-R31 are same as before trap */
	MOVW	R15, (4*15)(R1)
	/* SP and SB and u and m are already set; away we go */
	MOVW	R1, R7			/* pointer to Ureg */
	SUB	$8, R1
	MOVW	$SYSPSR, R8
	MOVW	R8, PSR
	JMPL	syscall(SB)
	/* R7 contains return value from syscall */

	ADD	$8, R1
	MOVW	(4*(32+2))(R1), R8		/* PSR */
	MOVW	R8, PSR
	NOOP

	MOVW	(4*15)(R1), R15
	SAVE	R0, R0
	MOVW	(4*6)(R1), R6
	MOVD	(4*4)(R1), R4
	MOVD	(4*2)(R1), R2
	MOVW	(4*(32+4))(R1), R17		/* PC */
	MOVW	(4*(32+3))(R1), R18		/* nPC */
	MOVW	(4*1)(R1), R1	/* restore R1=SP */
	RETT	R17, R18

TEXT	puttbr(SB), $0

	MOVW	R7, TBR
	NOOP
	RETURN

TEXT	gettbr(SB), $0

	MOVW	TBR, R7
	RETURN

TEXT	r1(SB), $0

	MOVW	R1, R7
	RETURN

TEXT	getwim(SB), $0

	MOVW	WIM, R7
	RETURN

TEXT	setlabel(SB), $0

	MOVW	R1, (R7)
	MOVW	R15, 4(R7)
	MOVW	$0, R7
	RETURN

TEXT	gotolabel(SB), $0

	MOVW	(R7), R1
	MOVW	4(R7), R15
	MOVW	$1, R7
	RETURN

TEXT	putcxsegm(SB), $0

	MOVW	R7, R8			/* context */
	MOVW	4(FP), R9		/* segment addr */
	MOVW	8(FP), R10		/* segment value */
	MOVW	$0xFFE80118, R7
	JMPL	(R7)
	RETURN

TEXT	getpsr(SB), $0

	MOVW	PSR, R7
	RETURN

TEXT	putcxreg(SB), $0

	MOVW	$CONTEXT, R8
	MOVB	R7, (R8, 2)
	RETURN

TEXT	putb2(SB), $0

	MOVW	4(FP), R8
	MOVB	R8, (R7, 2)
	RETURN

TEXT	getb2(SB), $0

	MOVB	(R7, 2), R7
	RETURN

TEXT	getw2(SB), $0

	MOVW	(R7, 2), R7
	RETURN

TEXT	putw2(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 2)
	RETURN

TEXT	putw4(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 4)
	RETURN

TEXT	getw4(SB), $0

	MOVW	(R7, 4), R7
	RETURN

TEXT	putwC(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 0xC)
	RETURN

TEXT	putwD(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 0xD)
	RETURN

TEXT	putwD16(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xD)
	RETURN

TEXT	putwE(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 0xE)
	RETURN

TEXT	putwE16(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	ADD	$(1<<4), R7
	MOVW	R8, (R7, 0xE)
	RETURN

TEXT	putsegm(SB), $0

	MOVW	4(FP), R8
	MOVW	R8, (R7, 3)
	RETURN

/*
 * in savefpregs and restfpregs, incoming R7 points to doubleword
 * below where F0 will go; doubleword align in and backfill FSR
 */
TEXT	savefpregs(SB), $0

	ADD	$8, R7
	ANDN	$7, R7		/* now MOVD-aligned */
	MOVW	FSR, -4(R7)

	MOVD	F0, (0*4)(R7)
	MOVD	F2, (2*4)(R7)
	MOVD	F4, (4*4)(R7)
	MOVD	F6, (6*4)(R7)
	MOVD	F8, (8*4)(R7)
	MOVD	F10, (10*4)(R7)
	MOVD	F12, (12*4)(R7)
	MOVD	F14, (14*4)(R7)
	MOVD	F16, (16*4)(R7)
	MOVD	F18, (18*4)(R7)
	MOVD	F20, (20*4)(R7)
	MOVD	F22, (22*4)(R7)
	MOVD	F24, (24*4)(R7)
	MOVD	F26, (26*4)(R7)
	MOVD	F28, (28*4)(R7)
	MOVD	F30, (30*4)(R7)

	MOVW	PSR, R8
	ANDN	$PSREF, R8
	MOVW	R8, PSR
	RETURN

TEXT	restfpregs(SB), $0

	MOVW	PSR, R8
	OR	$PSREF, R8
	MOVW	R8, PSR

	ADD	$8, R7
	ANDN	$7, R7		/* now MOVD-aligned */
	OR	R0, R0

	MOVW	-4(R7), FSR

	MOVD	(0*4)(R7), F0
	MOVD	(2*4)(R7), F2
	MOVD	(4*4)(R7), F4
	MOVD	(6*4)(R7), F6
	MOVD	(8*4)(R7), F8
	MOVD	(10*4)(R7), F10
	MOVD	(12*4)(R7), F12
	MOVD	(14*4)(R7), F14
	MOVD	(16*4)(R7), F16
	MOVD	(18*4)(R7), F18
	MOVD	(20*4)(R7), F20
	MOVD	(22*4)(R7), F22
	MOVD	(24*4)(R7), F24
	MOVD	(26*4)(R7), F26
	MOVD	(28*4)(R7), F28
	MOVD	(30*4)(R7), F30

	ANDN	$PSREF, R8
	MOVW	R8, PSR
	RETURN

TEXT	clearfpintr(SB), $0

	MOVW	$fpq+BY2WD(SB), R7
	ANDN	$0x7, R7		/* must be D aligned */
	MOVW	$fsr+0(SB), R9
clrq:
	MOVD	FQ, (R7)
	MOVW	FSR, (R9)
	MOVW	(R9), R8
	AND	$(1<<13), R8		/* queue not empty? */
	BNE	clrq
	RETURN

TEXT	getfsr(SB), $0
	MOVW	$fsr+0(SB), R7
	MOVW	FSR, (R7)
	MOVW	(R7), R7
	RETURN

GLOBL	mach0+0(SB), $MACHSIZE
GLOBL	fpq+0(SB), $(3*BY2WD)
GLOBL	fsr+0(SB), $BY2WD