Orange Pi5 kernel

^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   1) /* SPDX-License-Identifier: GPL-2.0 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   3)  * arch/alpha/lib/divide.S
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   4)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   5)  * (C) 1995 Linus Torvalds
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   6)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   7)  * Alpha division..
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   8)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   9) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  10) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  11)  * The alpha chip doesn't provide hardware division, so we have to do it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  12)  * by hand.  The compiler expects the functions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  13)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  14)  *	__divqu: 64-bit unsigned long divide
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  15)  *	__remqu: 64-bit unsigned long remainder
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  16)  *	__divqs/__remqs: signed 64-bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  17)  *	__divlu/__remlu: unsigned 32-bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  18)  *	__divls/__remls: signed 32-bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  19)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  20)  * These are not normal C functions: instead of the normal
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  21)  * calling sequence, these expect their arguments in registers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  22)  * $24 and $25, and return the result in $27. Register $28 may
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  23)  * be clobbered (assembly temporary), anything else must be saved. 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  24)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  25)  * In short: painful.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  26)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  27)  * This is a rather simple bit-at-a-time algorithm: it's very good
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  28)  * at dividing random 64-bit numbers, but the more usual case where
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  29)  * the divisor is small is handled better by the DEC algorithm
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  30)  * using lookup tables. This uses much less memory, though, and is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  31)  * nicer on the cache.. Besides, I don't know the copyright status
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  32)  * of the DEC code.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  33)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  34) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  35) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  36)  * My temporaries:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  37)  *	$0 - current bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  38)  *	$1 - shifted divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  39)  *	$2 - modulus/quotient
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  40)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  41)  *	$23 - return address
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  42)  *	$24 - dividend
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  43)  *	$25 - divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  44)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  45)  *	$27 - quotient/modulus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  46)  *	$28 - compare status
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  47)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  48) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  49) #include <asm/export.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  50) #define halt .long 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  51) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  52) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  53)  * Select function type and registers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  54)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  55) #define mask	$0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  56) #define divisor	$1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  57) #define compare $28
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  58) #define tmp1	$3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  59) #define tmp2	$4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  60) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  61) #ifdef DIV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  62) #define DIV_ONLY(x,y...) x,##y
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  63) #define MOD_ONLY(x,y...)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  64) #define func(x) __div##x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  65) #define modulus $2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  66) #define quotient $27
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  67) #define GETSIGN(x) xor $24,$25,x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  68) #define STACK 48
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  69) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  70) #define DIV_ONLY(x,y...)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  71) #define MOD_ONLY(x,y...) x,##y
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  72) #define func(x) __rem##x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  73) #define modulus $27
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  74) #define quotient $2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  75) #define GETSIGN(x) bis $24,$24,x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  76) #define STACK 32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  77) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  78) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  79) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  80)  * For 32-bit operations, we need to extend to 64-bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  81)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  82) #ifdef INTSIZE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  83) #define ufunction func(lu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  84) #define sfunction func(l)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  85) #define LONGIFY(x) zapnot x,15,x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  86) #define SLONGIFY(x) addl x,0,x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  87) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  88) #define ufunction func(qu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  89) #define sfunction func(q)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  90) #define LONGIFY(x)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  91) #define SLONGIFY(x)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  92) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  93) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  94) .set noat
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  95) .align	3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  96) .globl	ufunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  97) .ent	ufunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  98) ufunction:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  99) 	subq	$30,STACK,$30
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) 	.frame	$30,STACK,$23
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) 	.prologue 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) 7:	stq	$1, 0($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) 	bis	$25,$25,divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) 	stq	$2, 8($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) 	bis	$24,$24,modulus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) 	stq	$0,16($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) 	bis	$31,$31,quotient
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) 	LONGIFY(divisor)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) 	stq	tmp1,24($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) 	LONGIFY(modulus)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) 	bis	$31,1,mask
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) 	DIV_ONLY(stq tmp2,32($30))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) 	beq	divisor, 9f			/* div by zero */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) #ifdef INTSIZE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) 	/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) 	 * shift divisor left, using 3-bit shifts for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) 	 * 32-bit divides as we can't overflow. Three-bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) 	 * shifts will result in looping three times less
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) 	 * here, but can result in two loops more later.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) 	 * Thus using a large shift isn't worth it (and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) 	 * s8add pairs better than a sll..)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) 1:	cmpult	divisor,modulus,compare
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) 	s8addq	divisor,$31,divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) 	s8addq	mask,$31,mask
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) 	bne	compare,1b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) 1:	cmpult	divisor,modulus,compare
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) 	blt     divisor, 2f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) 	addq	divisor,divisor,divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) 	addq	mask,mask,mask
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) 	bne	compare,1b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) 	unop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) 	/* ok, start to go right again.. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) 2:	DIV_ONLY(addq quotient,mask,tmp2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) 	srl	mask,1,mask
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) 	cmpule	divisor,modulus,compare
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) 	subq	modulus,divisor,tmp1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) 	DIV_ONLY(cmovne compare,tmp2,quotient)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) 	srl	divisor,1,divisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) 	cmovne	compare,tmp1,modulus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) 	bne	mask,2b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) 9:	ldq	$1, 0($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) 	ldq	$2, 8($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) 	ldq	$0,16($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) 	ldq	tmp1,24($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) 	DIV_ONLY(ldq tmp2,32($30))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) 	addq	$30,STACK,$30
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) 	ret	$31,($23),1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) 	.end	ufunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) EXPORT_SYMBOL(ufunction)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159)  * Uhh.. Ugly signed division. I'd rather not have it at all, but
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160)  * it's needed in some circumstances. There are different ways to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161)  * handle this, really. This does:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162)  * 	-a / b = a / -b = -(a / b)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163)  *	-a % b = -(a % b)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164)  *	a % -b = a % b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165)  * which is probably not the best solution, but at least should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166)  * have the property that (x/y)*y + (x%y) = x.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) .align 3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) .globl	sfunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) .ent	sfunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) sfunction:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) 	subq	$30,STACK,$30
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) 	.frame	$30,STACK,$23
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) 	.prologue 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) 	bis	$24,$25,$28
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) 	SLONGIFY($28)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) 	bge	$28,7b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) 	stq	$24,0($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) 	subq	$31,$24,$28
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) 	stq	$25,8($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) 	cmovlt	$24,$28,$24	/* abs($24) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) 	stq	$23,16($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) 	subq	$31,$25,$28
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) 	stq	tmp1,24($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) 	cmovlt	$25,$28,$25	/* abs($25) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) 	unop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) 	bsr	$23,ufunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) 	ldq	$24,0($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) 	ldq	$25,8($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) 	GETSIGN($28)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) 	subq	$31,$27,tmp1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) 	SLONGIFY($28)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) 	ldq	$23,16($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) 	cmovlt	$28,tmp1,$27
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) 	ldq	tmp1,24($30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) 	addq	$30,STACK,$30
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) 	ret	$31,($23),1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) 	.end	sfunction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) EXPORT_SYMBOL(sfunction)