pa-risc2w.s

mediastreamer2是开源的网络传输媒体流的库

    COPY    %r6,%r24      
    .CALL   ;in=23,24,25,26;out=20,21,22,28,29; (MILLICALL)
    B,L     $$div2U,%r2     
    EXTRD,U %r6,31,32,%r23  
    DEPD    %r28,31,32,%r29 
$D2
    STD     %r29,-272(%r30)                   ; q
    AND     %r5,%r19,%r24                   ; t & 0xffffffff00000000;
    EXTRD,U %r24,31,32,%r24                 ; ??? 
    FLDD    -272(%r30),%fr7                 ; q
    FLDD    -280(%r30),%fr8                 ; d
    XMPYU   %fr8L,%fr7L,%fr10  
    FSTD    %fr10,-256(%r30)   
    XMPYU   %fr8L,%fr7R,%fr22  
    FSTD    %fr22,-264(%r30)   
    XMPYU   %fr8R,%fr7L,%fr11 
    XMPYU   %fr8R,%fr7R,%fr23
    FSTD    %fr11,-232(%r30)
    FSTD    %fr23,-240(%r30)
    LDD     -256(%r30),%r28
    DEPD,Z  %r28,31,32,%r2 
    LDD     -264(%r30),%r20
    ADD,L   %r20,%r2,%r31   
    LDD     -232(%r30),%r22 
    DEPD,Z  %r22,31,32,%r22 
    LDD     -240(%r30),%r21 
    B       $00000024       ; enter loop  
    ADD,L   %r21,%r22,%r23 

$0000002A
    LDO     -1(%r29),%r29   
    SUB     %r23,%r8,%r23   
$00000024
    SUB     %r4,%r31,%r25   
    AND     %r25,%r19,%r26  
    CMPB,*<>,N      %r0,%r26,$00000046  ; (forward)
    DEPD,Z  %r25,31,32,%r20 
    OR      %r20,%r24,%r21  
    CMPB,*<<,N  %r21,%r23,$0000002A ;(backward) 
    SUB     %r31,%r6,%r31   
;-------------Break path---------------------

$00000046
    DEPD,Z  %r23,31,32,%r25              ;tl
    EXTRD,U %r23,31,32,%r26              ;t
    AND     %r25,%r19,%r24               ;tl = (tl<<32)&0xfffffff0000000L
    ADD,L   %r31,%r26,%r31               ;th += t; 
    CMPCLR,*>>=     %r5,%r24,%r0         ;if (l<tl)
    LDO     1(%r31),%r31                 ; th++;
    CMPB,*<<=,N     %r31,%r4,$00000036   ;if (n < th) (forward)
    LDO     -1(%r29),%r29                ;q--; 
    ADD,L   %r4,%r3,%r4                  ;h += d;
$00000036
    ADDIB,=,N       -1,%r9,$D1 ;if (--count == 0) break (forward) 
    SUB     %r5,%r24,%r28                ; l -= tl;
    SUB     %r4,%r31,%r24                ; h -= th;
    SHRPD   %r24,%r28,32,%r4             ; h = ((h<<32)|(l>>32));
    DEPD,Z  %r29,31,32,%r10              ; ret = q<<32
    b      $0000001C
    DEPD,Z  %r28,31,32,%r5               ; l = l << 32 

$D1
    OR      %r10,%r29,%r28           ; ret |= q
$D3
    LDD     -368(%r30),%r2  
$D0
    LDD     -296(%r30),%r10 
    LDD     -304(%r30),%r9  
    LDD     -312(%r30),%r8  
    LDD     -320(%r30),%r7  
    LDD     -328(%r30),%r6  
    LDD     -336(%r30),%r5  
    LDD     -344(%r30),%r4  
    BVE     (%r2)   
        .EXIT
    LDD,MB  -352(%r30),%r3 

bn_div_err_case
    MFIA    %r6     
    ADDIL   L'bn_div_words-bn_div_err_case,%r6,%r1 
    LDO     R'bn_div_words-bn_div_err_case(%r1),%r6  
    ADDIL   LT'__iob,%r27,%r1       
    LDD     RT'__iob(%r1),%r26      
    ADDIL   L'C$4-bn_div_words,%r6,%r1    
    LDO     R'C$4-bn_div_words(%r1),%r25  
    LDO     64(%r26),%r26   
    .CALL           ;in=24,25,26,29;out=28;
    B,L     fprintf,%r2    
    LDO     -48(%r30),%r29 
    LDD     -288(%r30),%r27
    .CALL           ;in=29;
    B,L     abort,%r2      
    LDO     -48(%r30),%r29 
    LDD     -288(%r30),%r27
    B       $D0         
    LDD     -368(%r30),%r2  
	.PROCEND	;in=24,25,26,29;out=28;

;----------------------------------------------------------------------------
;
; Registers to hold 64-bit values to manipulate.  The "L" part
; of the register corresponds to the upper 32-bits, while the "R"
; part corresponds to the lower 32-bits
; 
; Note, that when using b6 and b7, the code must save these before
; using them because they are callee save registers 
; 
;
; Floating point registers to use to save values that
; are manipulated.  These don't collide with ftemp1-6 and
; are all caller save registers
;
a0        .reg %fr22
a0L       .reg %fr22L
a0R       .reg %fr22R

a1        .reg %fr23
a1L       .reg %fr23L
a1R       .reg %fr23R

a2        .reg %fr24
a2L       .reg %fr24L
a2R       .reg %fr24R

a3        .reg %fr25
a3L       .reg %fr25L
a3R       .reg %fr25R

a4        .reg %fr26
a4L       .reg %fr26L
a4R       .reg %fr26R

a5        .reg %fr27
a5L       .reg %fr27L
a5R       .reg %fr27R

a6        .reg %fr28
a6L       .reg %fr28L
a6R       .reg %fr28R

a7        .reg %fr29
a7L       .reg %fr29L
a7R       .reg %fr29R

b0        .reg %fr30
b0L       .reg %fr30L
b0R       .reg %fr30R

b1        .reg %fr31
b1L       .reg %fr31L
b1R       .reg %fr31R

;
; Temporary floating point variables, these are all caller save
; registers
;
ftemp1    .reg %fr4
ftemp2    .reg %fr5
ftemp3    .reg %fr6
ftemp4    .reg %fr7

;
; The B set of registers when used.
;

b2        .reg %fr8
b2L       .reg %fr8L
b2R       .reg %fr8R

b3        .reg %fr9
b3L       .reg %fr9L
b3R       .reg %fr9R

b4        .reg %fr10
b4L       .reg %fr10L
b4R       .reg %fr10R

b5        .reg %fr11
b5L       .reg %fr11L
b5R       .reg %fr11R

b6        .reg %fr12
b6L       .reg %fr12L
b6R       .reg %fr12R

b7        .reg %fr13
b7L       .reg %fr13L
b7R       .reg %fr13R

c1           .reg %r21   ; only reg
temp1        .reg %r20   ; only reg
temp2        .reg %r19   ; only reg
temp3        .reg %r31   ; only reg

m1           .reg %r28   
c2           .reg %r23   
high_one     .reg %r1
ht           .reg %r6
lt           .reg %r5
m            .reg %r4
c3           .reg %r3

SQR_ADD_C  .macro  A0L,A0R,C1,C2,C3
    XMPYU   A0L,A0R,ftemp1       ; m
    FSTD    ftemp1,-24(%sp)      ; store m

    XMPYU   A0R,A0R,ftemp2       ; lt
    FSTD    ftemp2,-16(%sp)      ; store lt

    XMPYU   A0L,A0L,ftemp3       ; ht
    FSTD    ftemp3,-8(%sp)       ; store ht

    LDD     -24(%sp),m           ; load m
    AND     m,high_mask,temp2    ; m & Mask
    DEPD,Z  m,30,31,temp3        ; m << 32+1
    LDD     -16(%sp),lt          ; lt

    LDD     -8(%sp),ht           ; ht
    EXTRD,U temp2,32,33,temp1    ; temp1 = m&Mask >> 32-1
    ADD     temp3,lt,lt          ; lt = lt+m
    ADD,L   ht,temp1,ht          ; ht += temp1
    ADD,DC  ht,%r0,ht            ; ht++

    ADD     C1,lt,C1             ; c1=c1+lt
    ADD,DC  ht,%r0,ht            ; ht++

    ADD     C2,ht,C2             ; c2=c2+ht
    ADD,DC  C3,%r0,C3            ; c3++
.endm

SQR_ADD_C2 .macro  A0L,A0R,A1L,A1R,C1,C2,C3
    XMPYU   A0L,A1R,ftemp1          ; m1 = bl*ht
    FSTD    ftemp1,-16(%sp)         ;
    XMPYU   A0R,A1L,ftemp2          ; m = bh*lt
    FSTD    ftemp2,-8(%sp)          ;
    XMPYU   A0R,A1R,ftemp3          ; lt = bl*lt
    FSTD    ftemp3,-32(%sp)
    XMPYU   A0L,A1L,ftemp4          ; ht = bh*ht
    FSTD    ftemp4,-24(%sp)         ;

    LDD     -8(%sp),m               ; r21 = m
    LDD     -16(%sp),m1             ; r19 = m1
    ADD,L   m,m1,m                  ; m+m1

    DEPD,Z  m,31,32,temp3           ; (m+m1<<32)
    LDD     -24(%sp),ht             ; r24 = ht

    CMPCLR,*>>= m,m1,%r0            ; if (m < m1)
    ADD,L   ht,high_one,ht          ; ht+=high_one

    EXTRD,U m,31,32,temp1           ; m >> 32
    LDD     -32(%sp),lt             ; lt
    ADD,L   ht,temp1,ht             ; ht+= m>>32
    ADD     lt,temp3,lt             ; lt = lt+m1
    ADD,DC  ht,%r0,ht               ; ht++

    ADD     ht,ht,ht                ; ht=ht+ht;
    ADD,DC  C3,%r0,C3               ; add in carry (c3++)

    ADD     lt,lt,lt                ; lt=lt+lt;
    ADD,DC  ht,%r0,ht               ; add in carry (ht++)

    ADD     C1,lt,C1                ; c1=c1+lt
    ADD,DC,*NUV ht,%r0,ht           ; add in carry (ht++)
    LDO     1(C3),C3              ; bump c3 if overflow,nullify otherwise

    ADD     C2,ht,C2                ; c2 = c2 + ht
    ADD,DC  C3,%r0,C3             ; add in carry (c3++)
.endm

;
;void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
; arg0 = r_ptr
; arg1 = a_ptr
;

bn_sqr_comba8
	.PROC
	.CALLINFO FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
	.EXPORT	bn_sqr_comba8,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
    .ENTRY
	.align 64

    STD     %r3,0(%sp)          ; save r3
    STD     %r4,8(%sp)          ; save r4
    STD     %r5,16(%sp)         ; save r5
    STD     %r6,24(%sp)         ; save r6

	;
	; Zero out carries
	;
	COPY     %r0,c1
	COPY     %r0,c2
	COPY     %r0,c3

	LDO      128(%sp),%sp       ; bump stack
    DEPDI,Z -1,32,33,high_mask   ; Create Mask 0xffffffff80000000L
    DEPDI,Z  1,31,1,high_one     ; Create Value  1 << 32

	;
	; Load up all of the values we are going to use
	;
    FLDD     0(a_ptr),a0       
    FLDD     8(a_ptr),a1       
    FLDD    16(a_ptr),a2       
    FLDD    24(a_ptr),a3       
    FLDD    32(a_ptr),a4       
    FLDD    40(a_ptr),a5       
    FLDD    48(a_ptr),a6       
    FLDD    56(a_ptr),a7       

	SQR_ADD_C a0L,a0R,c1,c2,c3
	STD     c1,0(r_ptr)          ; r[0] = c1;
	COPY    %r0,c1

	SQR_ADD_C2 a1L,a1R,a0L,a0R,c2,c3,c1
	STD     c2,8(r_ptr)          ; r[1] = c2;
	COPY    %r0,c2

	SQR_ADD_C a1L,a1R,c3,c1,c2
	SQR_ADD_C2 a2L,a2R,a0L,a0R,c3,c1,c2
	STD     c3,16(r_ptr)            ; r[2] = c3;
	COPY    %r0,c3

	SQR_ADD_C2 a3L,a3R,a0L,a0R,c1,c2,c3
	SQR_ADD_C2 a2L,a2R,a1L,a1R,c1,c2,c3
	STD     c1,24(r_ptr)           ; r[3] = c1;
	COPY    %r0,c1

	SQR_ADD_C a2L,a2R,c2,c3,c1
	SQR_ADD_C2 a3L,a3R,a1L,a1R,c2,c3,c1
	SQR_ADD_C2 a4L,a4R,a0L,a0R,c2,c3,c1
	STD     c2,32(r_ptr)          ; r[4] = c2;
	COPY    %r0,c2

	SQR_ADD_C2 a5L,a5R,a0L,a0R,c3,c1,c2
	SQR_ADD_C2 a4L,a4R,a1L,a1R,c3,c1,c2
	SQR_ADD_C2 a3L,a3R,a2L,a2R,c3,c1,c2
	STD     c3,40(r_ptr)          ; r[5] = c3;
	COPY    %r0,c3

	SQR_ADD_C a3L,a3R,c1,c2,c3
	SQR_ADD_C2 a4L,a4R,a2L,a2R,c1,c2,c3
	SQR_ADD_C2 a5L,a5R,a1L,a1R,c1,c2,c3
	SQR_ADD_C2 a6L,a6R,a0L,a0R,c1,c2,c3
	STD     c1,48(r_ptr)          ; r[6] = c1;
	COPY    %r0,c1

	SQR_ADD_C2 a7L,a7R,a0L,a0R,c2,c3,c1
	SQR_ADD_C2 a6L,a6R,a1L,a1R,c2,c3,c1
	SQR_ADD_C2 a5L,a5R,a2L,a2R,c2,c3,c1
	SQR_ADD_C2 a4L,a4R,a3L,a3R,c2,c3,c1
	STD     c2,56(r_ptr)          ; r[7] = c2;
	COPY    %r0,c2

	SQR_ADD_C a4L,a4R,c3,c1,c2
	SQR_ADD_C2 a5L,a5R,a3L,a3R,c3,c1,c2
	SQR_ADD_C2 a6L,a6R,a2L,a2R,c3,c1,c2
	SQR_ADD_C2 a7L,a7R,a1L,a1R,c3,c1,c2
	STD     c3,64(r_ptr)          ; r[8] = c3;
	COPY    %r0,c3

	SQR_ADD_C2 a7L,a7R,a2L,a2R,c1,c2,c3
	SQR_ADD_C2 a6L,a6R,a3L,a3R,c1,c2,c3
	SQR_ADD_C2 a5L,a5R,a4L,a4R,c1,c2,c3
	STD     c1,72(r_ptr)          ; r[9] = c1;
	COPY    %r0,c1

	SQR_ADD_C a5L,a5R,c2,c3,c1
	SQR_ADD_C2 a6L,a6R,a4L,a4R,c2,c3,c1
	SQR_ADD_C2 a7L,a7R,a3L,a3R,c2,c3,c1
	STD     c2,80(r_ptr)          ; r[10] = c2;
	COPY    %r0,c2

	SQR_ADD_C2 a7L,a7R,a4L,a4R,c3,c1,c2
	SQR_ADD_C2 a6L,a6R,a5L,a5R,c3,c1,c2
	STD     c3,88(r_ptr)          ; r[11] = c3;
	COPY    %r0,c3
	
	SQR_ADD_C a6L,a6R,c1,c2,c3
	SQR_ADD_C2 a7L,a7R,a5L,a5R,c1,c2,c3
	STD     c1,96(r_ptr)          ; r[12] = c1;
	COPY    %r0,c1

	SQR_ADD_C2 a7L,a7R,a6L,a6R,c2,c3,c1
	STD     c2,104(r_ptr)         ; r[13] = c2;
	COPY    %r0,c2

	SQR_ADD_C a7L,a7R,c3,c1,c2
	STD     c3, 112(r_ptr)       ; r[14] = c3
	STD     c1, 120(r_ptr)       ; r[15] = c1

    .EXIT
    LDD     -104(%sp),%r6        ; restore r6
    LDD     -112(%sp),%r5        ; restore r5
    LDD     -120(%sp),%r4        ; restore r4
    BVE     (%rp)
    LDD,MB  -128(%sp),%r3

	.PROCEND	

;-----------------------------------------------------------------------------