mirror/gmsm
1
0
Fork 0
mirror of https://github.com/emmansun/gmsm.git synced 2025-04-26 04:06:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

sm3: ppc64x, transpose matrix fix

Browse Source
This commit is contained in:
Sun Yimin 2024-09-05 13:46:41 +08:00 committed by GitHub
parent bc4b8e691f
commit 379dbdd3fb
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 35 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
sm3/sm3blocks_ppc64x.s
View File

@ -6,26 +6,37 @@
#include "textflag.h" #include "textflag.h"
DATA mask<>+0x00(SB)/8, $0x0001020310111213 // For P9 instruction emulation
DATA mask<>+0x08(SB)/8, $0x0405060714151617 #define ESPERM V21 // Endian swapping permute into BE
DATA mask<>+0x10(SB)/8, $0x08090a0b18191a1b #define TMP2 V22 // Temporary for P8_STXVB16X/P8_STXVB16X
DATA mask<>+0x18(SB)/8, $0x0c0d0e0f1c1d1e1f
DATA mask<>+0x20(SB)/8, $0x0001020304050607 DATA ·mask+0x00(SB)/8, $0x0f0e0d0c0b0a0908 // Permute for vector doubleword endian swap
DATA mask<>+0x28(SB)/8, $0x1011121314151617 DATA ·mask+0x08(SB)/8, $0x0706050403020100
DATA mask<>+0x30(SB)/8, $0x08090a0b0c0d0e0f DATA ·mask+0x10(SB)/8, $0x0001020310111213 // Permute for transpose matrix
DATA mask<>+0x38(SB)/8, $0x18191a1b1c1d1e1f DATA ·mask+0x18(SB)/8, $0x0405060714151617
GLOBL mask<>(SB), 8, $64 DATA ·mask+0x20(SB)/8, $0x08090a0b18191a1b
DATA ·mask+0x28(SB)/8, $0x0c0d0e0f1c1d1e1f
DATA ·mask+0x30(SB)/8, $0x0001020304050607
DATA ·mask+0x38(SB)/8, $0x1011121314151617
DATA ·mask+0x40(SB)/8, $0x08090a0b0c0d0e0f
DATA ·mask+0x48(SB)/8, $0x18191a1b1c1d1e1f
GLOBL ·mask(SB), RODATA, $80
#ifdef GOARCH_ppc64le #ifdef GOARCH_ppc64le
# ifdef GOPPC64_power9 # ifdef GOPPC64_power9
#define P8_LXVB16X(RA,RB,VT) LXVB16X (RA+RB), VT #define P8_LXVB16X(RA,RB,VT) LXVB16X (RA+RB), VT
#define P8_STXVB16X(VS,RA,RB) STXVB16X VS, (RA+RB) #define P8_STXVB16X(VS,RA,RB) STXVB16X VS, (RA+RB)
# else # else
// On POWER8/ppc64le, emulate the POWER9 instructions by loading unaligned
// doublewords and byte-swapping each doubleword to emulate BE load/stores.
#define NEEDS_ESPERM
#define P8_LXVB16X(RA,RB,VT) \ #define P8_LXVB16X(RA,RB,VT) \
LXVD2X (RA+RB), VT LXVD2X (RA+RB), VT \
VPERM VT, VT, ESPERM, VT
#define P8_STXVB16X(VS,RA,RB) \ #define P8_STXVB16X(VS,RA,RB) \
STXVD2X VS, (RA+RB) VPERM VS, VS, ESPERM, TMP2 \
STXVD2X TMP2, (RA+RB)
# endif // defined(GOPPC64_power9) # endif // defined(GOPPC64_power9)
#else #else
@ -52,6 +63,19 @@ TEXT ·transposeMatrix(SB),NOSPLIT,$0
MOVD $32, R8 MOVD $32, R8
MOVD $48, R9 MOVD $48, R9
#ifdef NEEDS_ESPERM
MOVD $·mask(SB), R4
LVX (R4), ESPERM
ADD $0x10, R4
#else
MOVD $·mask+0x10(SB), R4
#endif
LVX (R0)(R4), V8
LVX (R6)(R4), V9
LVX (R8)(R4), V10
LVX (R9)(R4), V11
MOVD (R0)(R3), R4 MOVD (R0)(R3), R4
P8_LXVB16X(R4, R0, V0) P8_LXVB16X(R4, R0, V0)
P8_LXVB16X(R4, R6, V4) P8_LXVB16X(R4, R6, V4)
@ -65,12 +89,6 @@ TEXT ·transposeMatrix(SB),NOSPLIT,$0
P8_LXVB16X(R4, R0, V3) P8_LXVB16X(R4, R0, V3)
P8_LXVB16X(R4, R6, V7) P8_LXVB16X(R4, R6, V7)
MOVD $mask<>+0x00(SB), R4
LVX (R0)(R4), V8
LVX (R6)(R4), V9
LVX (R8)(R4), V10
LVX (R9)(R4), V11
TRANSPOSE_MATRIX(V0, V1, V2, V3, V8, V9, V10, V11, V12, V13, V14, V15) TRANSPOSE_MATRIX(V0, V1, V2, V3, V8, V9, V10, V11, V12, V13, V14, V15)
TRANSPOSE_MATRIX(V4, V5, V6, V7, V8, V9, V10, V11, V12, V13, V14, V15) TRANSPOSE_MATRIX(V4, V5, V6, V7, V8, V9, V10, V11, V12, V13, V14, V15)