Stage 2 VSX Support
cover Stage 2 VSX Support
This patch series continues adding support for the PowerPC Vector Scalar Extension (VSX). Patches are relative to the Stage 1 delivery (see
http://lists.nongnu.org/archive/html/qemu-ppc/2013-09/msg00231.html).
This series adds the following:
a) all remaining load and store instructions defined by the V2.06
Power ISA
(aka Power7).
b) The vector and scalar move instructions.
c) The logical instructions defined by V2.06.
d) Assorted permute and select instructions.
Tom Musta (13):
Abandon GEN_VSX_* macros
Add lxsdx
Add lxvdsx
Add lxvw4x
Add stxsdx
Add stxvw4x
Add VSX Scalar Move Instructions
Add VSX Vector Move Instructions
Add Power7 VSX Logical Instructions
Add xxmrgh/xxmrgl
Add xxsel
Add xxspltw
Add xxsldwi
target-ppc/translate.c | 516 +++++++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 508 insertions(+), 8 deletions(-)
PATCH 1/13 Abandon GEN_VSX_* macros
Abandon GEN_VSX_* macros
This patch eliminates the GEN_VSX_LXVNX/GEN_VSX_STXVNX macros which did not provide significant value above the general GEN_HANDLER_E macro.
--- target-ppc/translate.c | 12 ++---------- 1 files changed, 2 insertions(+), 10 deletions(-)
target-ppc/translate.c +2 -10
@@ -9507,17 +9507,9 @@ GEN_VAFORM_PAIRED(vmsumshm, vmsumshs, 20),GEN_VAFORM_PAIRED (vsel, vperm,21 ),GEN_VAFORM_PAIRED (vmaddfp, vnmsubfp,23 ), -#undef GEN_VSX_LXVNX -#define GEN_VSX_LXVNX (name, opc2, opc3) \ - GEN_HANDLER_E(lxv##name, 0x1F, opc2, opc3, 0x00000000, PPC_NONE, PPC2_VSX) +GEN_HANDLER_E (lxvd2x,0x1F ,0x0C ,0x1A ,0 ,PPC_NONE ,PPC2_VSX ), -GEN_VSX_LXVNX(d2x, 0x0C, 0x1A), - -#undef GEN_VSX_STXVNX -#define GEN_VSX_STXVNX (name, opc2, opc3) \ - GEN_HANDLER_E(stxv##name, 0x1F, opc2, opc3, 0x00000000, PPC_NONE, PPC2_VSX) - -GEN_VSX_STXVNX(d2x, 0x0C, 0x1E), +GEN_HANDLER_E (stxvd2x,0x1F ,0xC ,0x1E ,0 ,PPC_NONE ,PPC2_VSX ), #undef GEN_XX3FORM_DM #define GEN_XX3FORM_DM(name, opc2, opc3) \
Signature · 1 line
1.7.1
PATCH 2/13 Add lxsdx
In reply to Tom MustaThis patch adds the Load VSX Scalar Doubleowrd Indexed (lxsdx) instruction.
The lower 8 bytes of the target register are undefined; this implementation leaves those bytes unaltered.
--- target-ppc/translate.c | 16 ++++++++++++++++ 1 files changed, 16 insertions(+), 0 deletions(-)
target-ppc/translate.c +16 -0
@@ -7007,6 +7007,21 @@ static inline TCGv_i64 cpu_vsrl(int n) } } +static void gen_lxsdx (DisasContext *ctx) +{ +TCGv EA ; +if (unlikely (!ctx->vsx_enabled)) { +gen_exception (ctx,POWERPC_EXCP_VSXU ); +return ; + } +gen_set_access_type (ctx,ACCESS_INT ); +EA =tcg_temp_new (); +gen_addr_reg_index (ctx,EA ); +gen_qemu_ld64 (ctx,cpu_vsrh (xT (ctx->opcode)),EA ); +/* NOTE: cpu_vsrl is undefined */ +tcg_temp_free (EA ); +} +static void gen_lxvd2x (DisasContext *ctx) {TCGv EA ;
@@ -9507,6 +9522,7 @@ GEN_VAFORM_PAIRED(vmsumshm, vmsumshs, 20),GEN_VAFORM_PAIRED (vsel, vperm,21 ),GEN_VAFORM_PAIRED (vmaddfp, vnmsubfp,23 ), +GEN_HANDLER_E (lxsdx,0x1F ,0x0C ,0x12 ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (lxvd2x,0x1F ,0x0C ,0x1A ,0 ,PPC_NONE ,PPC2_VSX ), GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX),
Signature · 1 line
1.7.1
PATCH 3/13 Add lxvdsx
In reply to Tom MustaThis patch adds the Load VSX Vector Doubleword & Splat Indexed (lxvdsx) instruction.
--- target-ppc/translate.c | 16 ++++++++++++++++ 1 files changed, 16 insertions(+), 0 deletions(-)
target-ppc/translate.c +16 -0
@@ -7038,6 +7038,21 @@ static void gen_lxvd2x(DisasContext *ctx)tcg_temp_free (EA ); } +static void gen_lxvdsx (DisasContext *ctx) +{ +TCGv EA ; +if (unlikely (!ctx->vsx_enabled)) { +gen_exception (ctx,POWERPC_EXCP_VSXU ); +return ; + } +gen_set_access_type (ctx,ACCESS_INT ); +EA =tcg_temp_new (); +gen_addr_reg_index (ctx,EA ); +gen_qemu_ld64 (ctx,cpu_vsrh (xT (ctx->opcode)),EA ); +tcg_gen_mov_tl (cpu_vsrl (xT (ctx->opcode)),cpu_vsrh (xT (ctx->opcode))); +tcg_temp_free (EA ); +} +static void gen_stxvd2x (DisasContext *ctx) {TCGv EA ;
@@ -9524,6 +9539,7 @@ GEN_VAFORM_PAIRED(vmaddfp, vnmsubfp, 23),GEN_HANDLER_E (lxsdx,0x1F ,0x0C ,0x12 ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (lxvd2x,0x1F ,0x0C ,0x1A ,0 ,PPC_NONE ,PPC2_VSX ), +GEN_HANDLER_E (lxvdsx,0x1F ,0x0C ,0x0A ,0 ,PPC_NONE ,PPC2_VSX ), GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX),
Signature · 1 line
1.7.1
PATCH 4/13 Add lxvw4x
In reply to Tom MustaThis patch adds the Load VSX Vector Word*4 Indexed (lxvw4x) instruction.
--- target-ppc/translate.c | 31 +++++++++++++++++++++++++++++++ 1 files changed, 31 insertions(+), 0 deletions(-)
target-ppc/translate.c +31 -0
@@ -7053,6 +7053,36 @@ static void gen_lxvdsx(DisasContext *ctx)tcg_temp_free (EA ); } +static void gen_lxvw4x (DisasContext *ctx) +{ +TCGv EA , tmp; +TCGv_i64 xth =cpu_vsrh (xT (ctx->opcode)); +TCGv_i64 xtl =cpu_vsrl (xT (ctx->opcode)); +if (unlikely (!ctx->vsx_enabled)) { +gen_exception (ctx,POWERPC_EXCP_VSXU ); +return ; + } +gen_set_access_type (ctx,ACCESS_INT ); +EA =tcg_temp_new (); + tmp =tcg_temp_new (); +gen_addr_reg_index (ctx,EA ); +gen_qemu_ld32u (ctx, xth,EA ); +tcg_gen_shli_tl (xth, xth,32 ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_ld32u (ctx, tmp,EA ); +tcg_gen_or_tl (xth, xth, tmp); + +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_ld32u (ctx, xtl,EA ); +tcg_gen_shli_tl (xtl, xtl,32 ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_ld32u (ctx, tmp,EA ); +tcg_gen_or_tl (xtl, xtl, tmp); + +tcg_temp_free (EA ); +tcg_temp_free (tmp); +} +static void gen_stxvd2x (DisasContext *ctx) {TCGv EA ;
@@ -9540,6 +9570,7 @@ GEN_VAFORM_PAIRED(vmaddfp, vnmsubfp, 23),GEN_HANDLER_E (lxsdx,0x1F ,0x0C ,0x12 ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (lxvd2x,0x1F ,0x0C ,0x1A ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (lxvdsx,0x1F ,0x0C ,0x0A ,0 ,PPC_NONE ,PPC2_VSX ), +GEN_HANDLER_E (lxvw4x,0x1F ,0x0C ,0x18 ,0 ,PPC_NONE ,PPC2_VSX ), GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX),
Signature · 1 line
1.7.1
PATCH 5/13 Add stxsdx
In reply to Tom MustaThis patch adds the Store VSX Scalar Doubleword Indexed (stxsdx) instruction.
--- target-ppc/translate.c | 15 +++++++++++++++ 1 files changed, 15 insertions(+), 0 deletions(-)
target-ppc/translate.c +15 -0
@@ -7083,6 +7083,20 @@ static void gen_lxvw4x(DisasContext *ctx)tcg_temp_free (tmp); } +static void gen_stxsdx (DisasContext *ctx) +{ +TCGv EA ; +if (unlikely (!ctx->vsx_enabled)) { +gen_exception (ctx,POWERPC_EXCP_VSXU ); +return ; + } +gen_set_access_type (ctx,ACCESS_INT ); +EA =tcg_temp_new (); +gen_addr_reg_index (ctx,EA ); +gen_qemu_st64 (ctx,cpu_vsrh (xS (ctx->opcode)),EA ); +tcg_temp_free (EA ); +} +static void gen_stxvd2x (DisasContext *ctx) {TCGv EA ;
@@ -9572,6 +9586,7 @@ GEN_HANDLER_E(lxvd2x, 0x1F, 0x0C, 0x1A, 0, PPC_NONE, PPC2_VSX),GEN_HANDLER_E (lxvdsx,0x1F ,0x0C ,0x0A ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (lxvw4x,0x1F ,0x0C ,0x18 ,0 ,PPC_NONE ,PPC2_VSX ), +GEN_HANDLER_E (stxsdx,0x1F ,0xC ,0x16 ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (stxvd2x,0x1F ,0xC ,0x1E ,0 ,PPC_NONE ,PPC2_VSX ), #undef GEN_XX3FORM_DM
Signature · 1 line
1.7.1
PATCH 6/13 Add stxvw4x
In reply to Tom MustaThis patch adds the Store VSX Vector Word*4 Indexed (stxvw4x) instruction.
--- target-ppc/translate.c | 28 ++++++++++++++++++++++++++++ 1 files changed, 28 insertions(+), 0 deletions(-)
target-ppc/translate.c +28 -0
@@ -7113,6 +7113,33 @@ static void gen_stxvd2x(DisasContext *ctx)tcg_temp_free (EA ); } +static void gen_stxvw4x (DisasContext *ctx) +{ +TCGv EA , tmp; +if (unlikely (!ctx->vsx_enabled)) { +gen_exception (ctx,POWERPC_EXCP_VSXU ); +return ; + } +gen_set_access_type (ctx,ACCESS_INT ); +EA =tcg_temp_new (); +gen_addr_reg_index (ctx,EA ); + tmp =tcg_temp_new (); + +tcg_gen_shri_i64 (tmp,cpu_vsrh (xS (ctx->opcode)),32 ); +gen_qemu_st32 (ctx, tmp,EA ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_st32 (ctx,cpu_vsrh (xS (ctx->opcode)),EA ); + +tcg_gen_shri_i64 (tmp,cpu_vsrl (xS (ctx->opcode)),32 ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_st32 (ctx, tmp,EA ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_st32 (ctx,cpu_vsrl (xS (ctx->opcode)),EA ); + +tcg_temp_free (EA ); +tcg_temp_free (tmp); +} +static void gen_xxpermdi (DisasContext *ctx) {if (unlikely (!ctx->vsx_enabled)) {
@@ -9588,6 +9615,7 @@ GEN_HANDLER_E(lxvw4x, 0x1F, 0x0C, 0x18, 0, PPC_NONE, PPC2_VSX),GEN_HANDLER_E (stxsdx,0x1F ,0xC ,0x16 ,0 ,PPC_NONE ,PPC2_VSX ),GEN_HANDLER_E (stxvd2x,0x1F ,0xC ,0x1E ,0 ,PPC_NONE ,PPC2_VSX ), +GEN_HANDLER_E (stxvw4x,0x1F ,0xC ,0x1C ,0 ,PPC_NONE ,PPC2_VSX ), #undef GEN_XX3FORM_DM #define GEN_XX3FORM_DM(name, opc2, opc3) \
Signature · 1 line
1.7.1
PATCH 7/13 Add VSX Scalar Move Instructions
In reply to Tom MustaThis patch adds the VSX scalar move instructions:
- xsabsdp (Scalar Absolute Value Double-Precision)
- xsnabspd (Scalar Negative Absolute Value Double-Precision)
- xsnegdp (Scalar Negate Double-Precision)
- xscpsgndp (Scalar Copy Sign Double-Precision)
A common generator macro (VSX_SCALAR_MOVE) is added since these instructions vary only slightly from each other.
Macros to support VSX XX2 and XX3 form opcodes are also added. These macros handle the overloading of "opcode 2" space (instruction bits 26:30) caused by AX and BX bits (29 and 30, respectively).
--- target-ppc/translate.c | 66 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 66 insertions(+), 0 deletions(-)
target-ppc/translate.c +0 -0
@@ -7158,6 +7158,55 @@ static void gen_xxpermdi(DisasContext *ctx) tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)),
cpu_vsrl(xB(ctx->opcode)));
} }
+#define OP_ABS 1
+#define OP_NABS 2
+#define OP_NEG 3
+#define OP_CPSGN 4
+#define SGN_MASK_DP 0x8000000000000000ul
+#define SGN_MASK_SP 0x8000000080000000ul
+
+#define VSX_SCALAR_MOVE(name, op, sgn_mask) \
+static void glue(gen_, name)(DisasContext * ctx) \
+ { \
+ TCGv_i64 xb; \
+ if (unlikely(!ctx->vsx_enabled)) { \
+ gen_exception(ctx, POWERPC_EXCP_VSXU); \
+ return; \
+ } \
+ xb = tcg_temp_new(); \
+ tcg_gen_mov_i64(xb, cpu_vsrh(xB(ctx->opcode))); \
+ switch (op) { \
+ case OP_ABS: { \
+ tcg_gen_andi_i64(xb, xb, ~(sgn_mask)); \
+ break; \
+ } \
+ case OP_NABS: { \
+ tcg_gen_ori_i64(xb, xb, (sgn_mask)); \
+ break; \
+ } \
+ case OP_NEG: { \
+ tcg_gen_xori_i64(xb, xb, (sgn_mask)); \
+ break; \
+ } \
+ case OP_CPSGN: { \
+ TCGv_i64 xa = tcg_temp_new(); \
+ tcg_gen_mov_i64(xa, cpu_vsrh(xA(ctx->opcode))); \
+ tcg_gen_andi_i64(xa, xa, (sgn_mask)); \
+ tcg_gen_andi_i64(xb, xb, ~(sgn_mask)); \
+ tcg_gen_or_i64(xb, xb, xa); \
+ tcg_temp_free(xa); \
+ break; \
+ } \
+ } \
+ tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), xb); \
+ tcg_temp_free(xb); \+ }
+
+VSX_SCALAR_MOVE(xsabsdp, OP_ABS, SGN_MASK_DP) +VSX_SCALAR_MOVE(xsnabsdp, OP_NABS, SGN_MASK_DP) +VSX_SCALAR_MOVE(xsnegdp, OP_NEG, SGN_MASK_DP) +VSX_SCALAR_MOVE(xscpsgndp, OP_CPSGN, SGN_MASK_DP) +
/*** SPE
extension ***/
/* Register moves */
@@ -9617,6 +9666,18 @@ GEN_HANDLER_E(stxsdx, 0x1F, 0xC, 0x16, 0, PPC_NONE, PPC2_VSX),
GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(stxvw4x, 0x1F, 0xC, 0x1C, 0, PPC_NONE, PPC2_VSX),
+#undef GEN_XX2FORM
+#define GEN_XX2FORM(name, opc2, opc3, fl2) \
+GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0, opc3, 0, PPC_NONE, fl2), \
+GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 1, opc3, 0, PPC_NONE, fl2) +
+#undef GEN_XX3FORM
+#define GEN_XX3FORM(name, opc2, opc3, fl2) \
+GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0, opc3, 0, PPC_NONE, fl2), \ +GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 1, opc3, 0, PPC_NONE, fl2), \ +GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 2, opc3, 0, PPC_NONE, fl2), \ +GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 3, opc3, 0, PPC_NONE, fl2)
+
#undef GEN_XX3FORM_DM
#define GEN_XX3FORM_DM(name, opc2, opc3) \
GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x00, opc3|0x00, 0, PPC_NONE,
PPC2_VSX),\
@@ -9636,6 +9697,11 @@ GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x01, opc3|0x0C, 0, PPC_NONE, PPC2_VSX),\
GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x02, opc3|0x0C, 0, PPC_NONE,
PPC2_VSX),\
GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x03, opc3|0x0C, 0, PPC_NONE,PPC2_VSX)
+GEN_XX2FORM(xsabsdp, 0x12, 0x15, PPC2_VSX),
+GEN_XX2FORM(xsnabsdp, 0x12, 0x16, PPC2_VSX),
+GEN_XX2FORM(xsnegdp, 0x12, 0x17, PPC2_VSX),
+GEN_XX3FORM(xscpsgndp, 0x00, 0x16, PPC2_VSX),
+
GEN_XX3FORM_DM(xxpermdi, 0x08, 0x01),
#undef GEN_SPE
Signature · 1 line
1.7.1
PATCH 8/13 Add VSX Vector Move Instructions
In reply to Tom MustaThis patch adds the vector move instructions:
- xvabsdp - Vector Absolute Value Double-Precision
- xvnabsdp - Vector Negative Absolute Value Double-Precision
- xvnegdp - Vector Negate Double-Precision
- xvcpsgndp - Vector Copy Sign Double-Precision
- xvabssp - Vector Absolute Value Single-Precision
- xvnabssp - Vector Negative Absolute Value Single-Precision
- xvnegsp - Vector Negate Single-Precision
- xvcpsgnsp - Vector Copy Sign Single-Precision
--- target-ppc/translate.c | 68 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 68 insertions(+), 0 deletions(-)
target-ppc/translate.c +60 -0
@@ -7207,6 +7207,66 @@ VSX_SCALAR_MOVE(xsnabsdp, OP_NABS, SGN_MASK_DP) VSX_SCALAR_MOVE(xsnegdp, OP_NEG, SGN_MASK_DP) VSX_SCALAR_MOVE(xscpsgndp, OP_CPSGN, SGN_MASK_DP) +#define VSX_VECTOR_MOVE (name, op, sgn_mask) \ +static void glue(gen_, name)(DisasContext * ctx) \ + { \ + TCGv_i64 xbh, xbl; \ + if (unlikely(!ctx->vsx_enabled)) { \ + gen_exception(ctx, POWERPC_EXCP_VSXU); \ + return; \ + } \ + xbh = tcg_temp_new(); \ + xbl = tcg_temp_new(); \ + tcg_gen_mov_i64(xbh, cpu_vsrh(xB(ctx->opcode))); \ + tcg_gen_mov_i64(xbl, cpu_vsrl(xB(ctx->opcode))); \ + switch (op) { \ + case OP_ABS: { \ + tcg_gen_andi_i64(xbh, xbh, ~(sgn_mask)); \ + tcg_gen_andi_i64(xbl, xbl, ~(sgn_mask)); \ + break; \ + } \ + case OP_NABS: { \ + tcg_gen_ori_i64(xbh, xbh, (sgn_mask)); \ + tcg_gen_ori_i64(xbl, xbl, (sgn_mask)); \ + break; \ + } \ + case OP_NEG: { \ + tcg_gen_xori_i64(xbh, xbh, (sgn_mask)); \ + tcg_gen_xori_i64(xbl, xbl, (sgn_mask)); \ + break; \ + } \ + case OP_CPSGN: { \ + TCGv_i64 xah = tcg_temp_new(); \ + TCGv_i64 xal = tcg_temp_new(); \ + tcg_gen_mov_i64(xah, cpu_vsrh(xA(ctx->opcode))); \ + tcg_gen_mov_i64(xal, cpu_vsrl(xA(ctx->opcode))); \ + tcg_gen_andi_i64(xah, xah, (sgn_mask)); \ + tcg_gen_andi_i64(xal, xal, (sgn_mask)); \ + tcg_gen_andi_i64(xbh, xbh, ~(sgn_mask)); \ + tcg_gen_andi_i64(xbl, xbl, ~(sgn_mask)); \ + tcg_gen_or_i64(xbh, xbh, xah); \ + tcg_gen_or_i64(xbl, xbl, xal); \ + tcg_temp_free(xah); \ + tcg_temp_free(xal); \ + break; \ + } \ + } \ + tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), xbh); \ + tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)), xbl); \ + tcg_temp_free(xbh); \ + tcg_temp_free(xbl); \ + } + +VSX_VECTOR_MOVE(xvabsdp, OP_ABS, SGN_MASK_DP) +VSX_VECTOR_MOVE(xvnabsdp, OP_NABS, SGN_MASK_DP) +VSX_VECTOR_MOVE(xvnegdp, OP_NEG, SGN_MASK_DP) +VSX_VECTOR_MOVE(xvcpsgndp, OP_CPSGN, SGN_MASK_DP) +VSX_VECTOR_MOVE(xvabssp, OP_ABS, SGN_MASK_SP) +VSX_VECTOR_MOVE(xvnabssp, OP_NABS, SGN_MASK_SP) +VSX_VECTOR_MOVE(xvnegsp, OP_NEG, SGN_MASK_SP) +VSX_VECTOR_MOVE (xvcpsgnsp,OP_CPSGN ,SGN_MASK_SP ) + +/*** SPE
extension ***/
/* Register moves */
@@ -9702,6 +9762,14 @@ GEN_XX2FORM(xsnabsdp, 0x12, 0x16, PPC2_VSX),
GEN_XX2FORM(xsnegdp, 0x12, 0x17, PPC2_VSX), GEN_XX3FORM(xscpsgndp, 0x00, 0x16, PPC2_VSX),
+GEN_XX2FORM(xvabsdp, 0x12, 0x1D, PPC2_VSX),
+GEN_XX2FORM(xvnabsdp, 0x12, 0x1E, PPC2_VSX),
+GEN_XX2FORM(xvnegdp, 0x12, 0x1F, PPC2_VSX),
+GEN_XX3FORM(xvcpsgndp, 0x00, 0x1E, PPC2_VSX),
+GEN_XX2FORM(xvabssp, 0x12, 0x19, PPC2_VSX),
+GEN_XX2FORM(xvnabssp, 0x12, 0x1A, PPC2_VSX),
+GEN_XX2FORM(xvnegsp, 0x12, 0x1B, PPC2_VSX),
+GEN_XX3FORM(xvcpsgnsp, 0x00, 0x1A, PPC2_VSX),
GEN_XX3FORM_DM(xxpermdi, 0x08, 0x01),
#undef GEN_SPE
Signature · 1 line
1.7.1
PATCH 9/13 Add Power7 VSX Logical Instructions
In reply to Tom MustaThis patch adds the VSX logical instructions that are defined by the Version 2.06 Power ISA (aka Power7):
- xxland
- xxlandc
- xxlor
- xxlxor
- xxlnor
--- target-ppc/translate.c | 29 +++++++++++++++++++++++++++++ 1 files changed, 29 insertions(+), 0 deletions(-)
target-ppc/translate.c +18 -0
@@ -7267,6 +7267,24 @@ VSX_VECTOR_MOVE(xvnegsp, OP_NEG, SGN_MASK_SP) VSX_VECTOR_MOVE(xvcpsgnsp, OP_CPSGN, SGN_MASK_SP) +#define VSX_LOGICAL (name, tcg_op) \ +static void glue(gen_, name)(DisasContext * ctx) \ + { \ + if (unlikely(!ctx->vsx_enabled)) { \ + gen_exception(ctx, POWERPC_EXCP_VSXU); \ + return; \ + } \ + tcg_op(cpu_vsrh(xT(ctx->opcode)), cpu_vsrh(xA(ctx->opcode)), \ + cpu_vsrh(xB(ctx->opcode))); \ + tcg_op(cpu_vsrl(xT(ctx->opcode)), cpu_vsrl(xA(ctx->opcode)), \ + cpu_vsrl(xB(ctx->opcode))); \ + } + +VSX_LOGICAL(xxland, tcg_gen_and_tl) +VSX_LOGICAL(xxlandc, tcg_gen_andc_tl) +VSX_LOGICAL(xxlor, tcg_gen_or_tl) +VSX_LOGICAL(xxlxor, tcg_gen_xor_tl) +VSX_LOGICAL (xxlnor, tcg_gen_nor_tl)/*** SPE
extension ***/
/* Register moves */
@@ -9770,6 +9788,17 @@ GEN_XX2FORM(xvabssp, 0x12, 0x19, PPC2_VSX),
GEN_XX2FORM(xvnabssp, 0x12, 0x1A, PPC2_VSX), GEN_XX2FORM(xvnegsp, 0x12, 0x1B, PPC2_VSX), GEN_XX3FORM(xvcpsgnsp, 0x00, 0x1A, PPC2_VSX),
+
+#undef VSX_LOGICAL
+#define VSX_LOGICAL(name, opc2, opc3, fl2) \
+GEN_XX3FORM(name, opc2, opc3, fl2)
+
+VSX_LOGICAL(xxland, 0x8, 0x10, PPC2_VSX), +VSX_LOGICAL(xxlandc, 0x8, 0x11, PPC2_VSX), +VSX_LOGICAL(xxlor, 0x8, 0x12, PPC2_VSX), +VSX_LOGICAL(xxlxor, 0x8, 0x13, PPC2_VSX), +VSX_LOGICAL(xxlnor, 0x8, 0x14, PPC2_VSX),
+
GEN_XX3FORM_DM(xxpermdi, 0x08, 0x01),
#undef GEN_SPE
Signature · 1 line
1.7.1
PATCH 10/13 Add xxmrgh/xxmrgl
In reply to Tom MustaThis patch adds the VSX Merge High Word and VSX Merge Low Word instructions.
--- target-ppc/translate.c | 44 ++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 44 insertions(+), 0 deletions(-)
target-ppc/translate.c +42 -0
@@ -7286,6 +7286,48 @@ VSX_LOGICAL(xxlor, tcg_gen_or_tl) VSX_LOGICAL(xxlxor, tcg_gen_xor_tl) VSX_LOGICAL(xxlnor, tcg_gen_nor_tl) +#define VSX_XXMRG (name, high) \ +static void glue(gen_, name)(DisasContext * ctx) \ + { \ + TCGv_i64 a0, a1, b0, b1; \ + if (unlikely(!ctx->vsx_enabled)) { \ + gen_exception(ctx, POWERPC_EXCP_VSXU); \ + return; \ + } \ + a0 = tcg_temp_new(); \ + a1 = tcg_temp_new(); \ + b0 = tcg_temp_new(); \ + b1 = tcg_temp_new(); \ + if (high) { \ + tcg_gen_mov_i64(a0, cpu_vsrh(xA(ctx->opcode))); \ + tcg_gen_mov_i64(a1, cpu_vsrh(xA(ctx->opcode))); \ + tcg_gen_mov_i64(b0, cpu_vsrh(xB(ctx->opcode))); \ + tcg_gen_mov_i64(b1, cpu_vsrh(xB(ctx->opcode))); \ + } else { \ + tcg_gen_mov_i64(a0, cpu_vsrl(xA(ctx->opcode))); \ + tcg_gen_mov_i64(a1, cpu_vsrl(xA(ctx->opcode))); \ + tcg_gen_mov_i64(b0, cpu_vsrl(xB(ctx->opcode))); \ + tcg_gen_mov_i64(b1, cpu_vsrl(xB(ctx->opcode))); \ + } \ + tcg_gen_andi_i64(a0, a0, 0xFFFFFFFF00000000ul); \ + tcg_gen_shli_i64(a1, a1, 32); \ + tcg_gen_shri_i64(b0, b0, 32); \ + tcg_gen_andi_i64(b0, b0, 0x00000000FFFFFFFFul); \ + tcg_gen_andi_i64(b1, b1, 0x00000000FFFFFFFFul); \ + tcg_gen_or_i64(a0, a0, b0); \ + tcg_gen_or_i64(a1, a1, b1); \ + tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), a0); \ + tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)), a1); \ + tcg_temp_free(a0); \ + tcg_temp_free(a1); \ + tcg_temp_free(b0); \ + tcg_temp_free(b1); \ + } + +VSX_XXMRG(xxmrghw, 1) +VSX_XXMRG (xxmrglw,0 ) + +/*** SPE
extension ***/
/* Register moves */
@@ -9798,6 +9840,8 @@ VSX_LOGICAL(xxlandc, 0x8, 0x11, PPC2_VSX),
VSX_LOGICAL(xxlor, 0x8, 0x12, PPC2_VSX), VSX_LOGICAL(xxlxor, 0x8, 0x13, PPC2_VSX), VSX_LOGICAL(xxlnor, 0x8, 0x14, PPC2_VSX),
+GEN_XX3FORM(xxmrghw, 0x08, 0x02, PPC2_VSX), +GEN_XX3FORM(xxmrglw, 0x08, 0x06, PPC2_VSX),
GEN_XX3FORM_DM(xxpermdi, 0x08, 0x01),
Signature · 1 line
1.7.1
PATCH 11/13 Add xxsel
In reply to Tom MustaThis patch adds the VSX Select (xxsel) instruction.
The xxsel instruction has four VSR operands. Thus the xC instruction decoder is added.
The xxsel instruction is massively overloaded in the opcode table since only bits 26 and 27 are opcode bits. This overloading is done in matrix fashion with two macros (GEN_XXSEL_ROW and GEN_XX_SEL).
--- target-ppc/translate.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 83 insertions(+), 0 deletions(-)
target-ppc/translate.c +1 -0
@@ -501,6 +501,7 @@ EXTRACT_HELPER_SPLIT(xT, 0, 1, 21, 5);EXTRACT_HELPER_SPLIT (xS,0 ,1 ,21 ,5 );EXTRACT_HELPER_SPLIT (xA,2 ,1 ,16 ,5 );EXTRACT_HELPER_SPLIT (xB,1 ,1 ,11 ,5 ); +EXTRACT_HELPER_SPLIT (xC,3 ,1 ,6 ,5 );EXTRACT_HELPER (DM ,8 ,2 );/*****************************************************************************/ /* PowerPC instructions
table */
@@ -7327,6 +7328,42 @@ static void glue(gen_, name)(DisasContext * ctx) \
VSX_XXMRG(xxmrghw, 1)
VSX_XXMRG(xxmrglw, 0)
+static void gen_xxsel(DisasContext * ctx)
+{
+ TCGv_i64 a, b, c;
+ if (unlikely(!ctx->vsx_enabled)) {
+ gen_exception(ctx, POWERPC_EXCP_VSXU);
+ return;
+ } + a = tcg_temp_new(); + b = tcg_temp_new(); + c = tcg_temp_new(); + + tcg_gen_mov_i64(a, cpu_vsrh(xA(ctx->opcode))); + tcg_gen_mov_i64(b, cpu_vsrh(xB(ctx->opcode))); + tcg_gen_mov_i64(c, cpu_vsrh(xC(ctx->opcode))); + + tcg_gen_and_i64(b, b, c); + tcg_gen_not_i64(c, c); + tcg_gen_and_i64(a, a, c); + tcg_gen_or_i64(a, a, b); + tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), a); + + tcg_gen_mov_i64(a, cpu_vsrl(xA(ctx->opcode))); + tcg_gen_mov_i64(b, cpu_vsrl(xB(ctx->opcode))); + tcg_gen_mov_i64(c, cpu_vsrl(xC(ctx->opcode))); + + tcg_gen_and_i64(b, b, c); + tcg_gen_not_i64(c, c); + tcg_gen_and_i64(a, a, c); + tcg_gen_or_i64(a, a, b); + tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)), a); + + tcg_temp_free(a); + tcg_temp_free(b); + tcg_temp_free(c);
+}
+
/*** SPE
extension ***/
/* Register moves */
@@ -9843,6 +9880,52 @@ VSX_LOGICAL(xxlnor, 0x8, 0x14, PPC2_VSX),
GEN_XX3FORM(xxmrghw, 0x08, 0x02, PPC2_VSX), GEN_XX3FORM(xxmrglw, 0x08, 0x06, PPC2_VSX),
+#define GEN_XXSEL_ROW(opc3) \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x18, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x19, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1A, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1B, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1C, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1D, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1E, opc3, 0, PPC_NONE, PPC2_VSX), \
+GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x1F, opc3, 0, PPC_NONE, PPC2_VSX), \
+
+#define GEN_XXSEL() \
+GEN_XXSEL_ROW(0x00) \
+GEN_XXSEL_ROW(0x01) \
+GEN_XXSEL_ROW(0x02) \
+GEN_XXSEL_ROW(0x03) \
+GEN_XXSEL_ROW(0x04) \
+GEN_XXSEL_ROW(0x05) \
+GEN_XXSEL_ROW(0x06) \
+GEN_XXSEL_ROW(0x07) \
+GEN_XXSEL_ROW(0x08) \
+GEN_XXSEL_ROW(0x09) \
+GEN_XXSEL_ROW(0x0A) \
+GEN_XXSEL_ROW(0x0B) \
+GEN_XXSEL_ROW(0x0C) \
+GEN_XXSEL_ROW(0x0D) \
+GEN_XXSEL_ROW(0x0E) \
+GEN_XXSEL_ROW(0x0F) \
+GEN_XXSEL_ROW(0x10) \
+GEN_XXSEL_ROW(0x11) \
+GEN_XXSEL_ROW(0x12) \
+GEN_XXSEL_ROW(0x13) \
+GEN_XXSEL_ROW(0x14) \
+GEN_XXSEL_ROW(0x15) \
+GEN_XXSEL_ROW(0x16) \
+GEN_XXSEL_ROW(0x17) \
+GEN_XXSEL_ROW(0x18) \
+GEN_XXSEL_ROW(0x19) \
+GEN_XXSEL_ROW(0x1A) \
+GEN_XXSEL_ROW(0x1B) \
+GEN_XXSEL_ROW(0x1C) \
+GEN_XXSEL_ROW(0x1D) \
+GEN_XXSEL_ROW(0x1E) \
+GEN_XXSEL_ROW(0x1F)
+
+GEN_XXSEL()
+
GEN_XX3FORM_DM(xxpermdi, 0x08, 0x01),
#undef GEN_SPE
Signature · 1 line
1.7.1
PATCH 12/13 Add xxspltw
In reply to Tom MustaThis patch adds the VSX Splat Word (xxsplatw) instruction.
This is the first instruction to use the UIM immediate field and consequently a decoder is also added.
--- target-ppc/translate.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 50 insertions(+), 0 deletions(-)
target-ppc/translate.c +1 -0
@@ -503,6 +503,7 @@ EXTRACT_HELPER_SPLIT(xA, 2, 1, 16, 5);EXTRACT_HELPER_SPLIT (xB,1 ,1 ,11 ,5 );EXTRACT_HELPER_SPLIT (xC,3 ,1 ,6 ,5 );EXTRACT_HELPER (DM ,8 ,2 ); +EXTRACT_HELPER (UIM ,16 ,2 );/*****************************************************************************/ /* PowerPC instructions
table */
@@ -7364,6 +7365,54 @@ static void gen_xxsel(DisasContext * ctx)
tcg_temp_free(c);
}
+static void gen_xxspltw(DisasContext *ctx)
+{
+ TCGv_i64 b, b2;
+ if (unlikely(!ctx->vsx_enabled)) {
+ gen_exception(ctx, POWERPC_EXCP_VSXU);
+ return;
+ }
+
+ b = tcg_temp_new();
+ b2 = tcg_temp_new();
+
+ tcg_gen_mov_i64(b, cpu_vsrh(xB(ctx->opcode)));
+
+ switch (UIM(ctx->opcode)) {
+ case 0: {
+ tcg_gen_mov_i64(b, cpu_vsrh(xB(ctx->opcode)));
+ tcg_gen_andi_i64(b, b, 0xFFFFFFFF00000000ul);
+ tcg_gen_shri_i64(b, b, 32);
+ break;
+ }
+ case 1: {
+ tcg_gen_mov_i64(b, cpu_vsrh(xB(ctx->opcode)));
+ tcg_gen_andi_i64(b, b, 0x00000000FFFFFFFFul);
+ break;
+ }
+ case 2: {
+ tcg_gen_mov_i64(b, cpu_vsrl(xB(ctx->opcode)));
+ tcg_gen_andi_i64(b, b, 0xFFFFFFFF00000000ul);
+ tcg_gen_shri_i64(b, b, 32);
+ break;
+ }
+ case 3: {
+ tcg_gen_mov_i64(b, cpu_vsrl(xB(ctx->opcode)));
+ tcg_gen_andi_i64(b, b, 0x00000000FFFFFFFFul);
+ break;+ }
+ }
+
+ tcg_gen_shli_i64(b2, b, 32);
+ tcg_gen_or_i64(b, b, b2);
+
+ tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), b);
+ tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)), b);
+
+ tcg_temp_free(b);
+ tcg_temp_free(b2);
+}
+
/*** SPE
extension ***/
/* Register moves */
@@ -9879,6 +9928,7 @@ VSX_LOGICAL(xxlxor, 0x8, 0x13, PPC2_VSX),
VSX_LOGICAL(xxlnor, 0x8, 0x14, PPC2_VSX), GEN_XX3FORM(xxmrghw, 0x08, 0x02, PPC2_VSX), GEN_XX3FORM(xxmrglw, 0x08, 0x06, PPC2_VSX),
+GEN_XX2FORM(xxspltw, 0x08, 0x0A, PPC2_VSX),
#define GEN_XXSEL_ROW(opc3) \
GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x18, opc3, 0, PPC_NONE, PPC2_VSX), \
Signature · 1 line
1.7.1
PATCH 13/13 Add xxsldwi
In reply to Tom MustaThis patch adds the VSX Shift Left Double by Word Immediate (xxsldwi) instruction.
--- target-ppc/translate.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 62 insertions(+), 0 deletions(-)
target-ppc/translate.c +1 -0
@@ -504,6 +504,7 @@ EXTRACT_HELPER_SPLIT(xB, 1, 1, 11, 5);EXTRACT_HELPER_SPLIT (xC,3 ,1 ,6 ,5 );EXTRACT_HELPER (DM ,8 ,2 );EXTRACT_HELPER (UIM ,16 ,2 ); +EXTRACT_HELPER (SHW ,8 ,2 );/*****************************************************************************/ /* PowerPC instructions
table */
@@ -7413,6 +7414,66 @@ static void gen_xxspltw(DisasContext *ctx)
tcg_temp_free(b2);
}
+static void gen_xxsldwi(DisasContext *ctx)
+{
+ TCGv_i64 xth, xtl;
+ if (unlikely(!ctx->vsx_enabled)) {
+ gen_exception(ctx, POWERPC_EXCP_VSXU);
+ return;
+ }
+ xth = tcg_temp_new();
+ xtl = tcg_temp_new();
+
+ switch (SHW(ctx->opcode)) {
+ case 0: {
+ tcg_gen_mov_i64(xth, cpu_vsrh(xA(ctx->opcode)));
+ tcg_gen_mov_i64(xtl, cpu_vsrl(xA(ctx->opcode)));
+ break;
+ }
+ case 1: {
+ TCGv_i64 t0 = tcg_temp_new();
+ tcg_gen_mov_i64(xth, cpu_vsrh(xA(ctx->opcode)));
+ tcg_gen_shli_i64(xth, xth, 32);
+ tcg_gen_mov_i64(t0, cpu_vsrl(xA(ctx->opcode)));
+ tcg_gen_shri_i64(t0, t0, 32);
+ tcg_gen_or_i64(xth, xth, t0);
+ tcg_gen_mov_i64(xtl, cpu_vsrl(xA(ctx->opcode)));
+ tcg_gen_shli_i64(xtl, xtl, 32);
+ tcg_gen_mov_i64(t0, cpu_vsrh(xB(ctx->opcode)));
+ tcg_gen_shri_i64(t0, t0, 32);
+ tcg_gen_or_i64(xtl, xtl, t0);
+ tcg_temp_free(t0);
+ break;
+ }
+ case 2: {
+ tcg_gen_mov_i64(xth, cpu_vsrl(xA(ctx->opcode)));
+ tcg_gen_mov_i64(xtl, cpu_vsrh(xB(ctx->opcode)));
+ break;
+ }
+ case 3: {
+ TCGv_i64 t0 = tcg_temp_new();
+ tcg_gen_mov_i64(xth, cpu_vsrl(xA(ctx->opcode)));
+ tcg_gen_shli_i64(xth, xth, 32);
+ tcg_gen_mov_i64(t0, cpu_vsrh(xB(ctx->opcode)));
+ tcg_gen_shri_i64(t0, t0, 32);
+ tcg_gen_or_i64(xth, xth, t0);
+ tcg_gen_mov_i64(xtl, cpu_vsrh(xB(ctx->opcode)));
+ tcg_gen_shli_i64(xtl, xtl, 32);
+ tcg_gen_mov_i64(t0, cpu_vsrl(xB(ctx->opcode)));
+ tcg_gen_shri_i64(t0, t0, 32);
+ tcg_gen_or_i64(xtl, xtl, t0);
+ tcg_temp_free(t0);
+ break;+ }
+ }
+
+ tcg_gen_mov_i64(cpu_vsrh(xT(ctx->opcode)), xth);
+ tcg_gen_mov_i64(cpu_vsrl(xT(ctx->opcode)), xtl);
+
+ tcg_temp_free(xth);
+ tcg_temp_free(xtl);
+}
+
/*** SPE
extension ***/
/* Register moves */
@@ -9929,6 +9990,7 @@ VSX_LOGICAL(xxlnor, 0x8, 0x14, PPC2_VSX),
GEN_XX3FORM(xxmrghw, 0x08, 0x02, PPC2_VSX), GEN_XX3FORM(xxmrglw, 0x08, 0x06, PPC2_VSX), GEN_XX2FORM(xxspltw, 0x08, 0x0A, PPC2_VSX),
+GEN_XX3FORM_DM(xxsldwi, 0x08, 0x00),
#define GEN_XXSEL_ROW(opc3) \
GEN_HANDLER2_E(xxsel, "xxsel", 0x3C, 0x18, opc3, 0, PPC_NONE, PPC2_VSX), \
Signature · 1 line
1.7.1
Re: [Qemu-devel] [PATCH 04/13] Add lxvw4x
In reply to Tom Mustatarget-ppc/translate.c static void gen_lxvdsx(DisasContext *ctx) · line 7073
Quoted 4 lines · Tom Musta
target-ppc/translate.c +4
-0+tcg_gen_shli_tl (xth, xth,32 ); +tcg_gen_addi_tl (EA ,EA ,4 ); +gen_qemu_ld32u (ctx, tmp,EA ); +tcg_gen_or_tl (xth, xth, tmp);
Better with deposit_i64.
r~
Re: [Qemu-devel] [PATCH 10/13] Add xxmrgh/xxmrgl
In reply to Tom Mustatarget-ppc/translate.c VSX_LOGICAL(xxlor, tcg_gen_or_tl) · line 7320
Quoted 9 lines · Tom Musta
target-ppc/translate.c +9
-0+tcg_gen_andi_i64 (a0, a0,0xFFFFFFFF00000000ul ); \ +tcg_gen_shli_i64 (a1, a1,32 ); \ +tcg_gen_shri_i64 (b0, b0,32 ); \ +tcg_gen_andi_i64 (b0, b0,0x00000000FFFFFFFFul ); \ +tcg_gen_andi_i64 (b1, b1,0x00000000FFFFFFFFul ); \ +tcg_gen_or_i64 (a0, a0, b0); \ +tcg_gen_or_i64 (a1, a1, b1); \ +tcg_gen_mov_i64 (cpu_vsrh (xT (ctx->opcode)), a0); \ +tcg_gen_mov_i64 (cpu_vsrl (xT (ctx->opcode)), a1); \
Two deposit operations.
r~
Re: [Qemu-devel] [PATCH 11/13] Add xxsel
In reply to Tom MustaReview
Quoted 3 lines · Tom Musta
(quoted diff) +3
-0+ tcg_gen_and_i64(b, b, c); + tcg_gen_not_i64(c, c); + tcg_gen_and_i64(a, a, c);
tcg_gen_andc_i64.
Review
Quoted 3 lines
(quoted diff) +3
-0+#define GEN_XXSEL() \ +GEN_XXSEL_ROW(0x00) \ +GEN_XXSEL_ROW(0x01) \
Why bother with defining GEN_XXSEL when its only used once? Surely just put the rows there.
OTOH, this does suggest that we could do with a better way to decode the instructions, because this is ugly...
r~
Re: [Qemu-devel] [PATCH 12/13] Add xxspltw
In reply to Tom MustaQuoted 5 lines · Tom Musta
(quoted diff) +5
-0+ case 0: { + tcg_gen_mov_i64(b, cpu_vsrh(xB(ctx->opcode))); + tcg_gen_andi_i64(b, b, 0xFFFFFFFF00000000ul); + tcg_gen_shri_i64(b, b, 32); + break;
...
Review
Quoted 5 lines
(quoted diff) +5
-0+ case 2: { + tcg_gen_mov_i64(b, cpu_vsrl(xB(ctx->opcode))); + tcg_gen_andi_i64(b, b, 0xFFFFFFFF00000000ul); + tcg_gen_shri_i64(b, b, 32); + break;
No need for the and.
Perhaps better as
TCGv_i64 vsr = (uim & 2 ? cpu_vrsl(xb) : cpu_vrsh(xb));
if (uim & 1) {
tcg_gen_ext32u_i64(b, vsr);
} else {
tcg_gen_shri_i32(b, vsr, 32);
}Review
Quoted 2 lines
(quoted diff) +2
-0+ tcg_gen_shli_i64(b2, b, 32); + tcg_gen_or_i64(b, b, b2);
deposit.
r~
Re: [Qemu-devel] [PATCH 10/13] Add xxmrgh/xxmrgl
In reply to Richard HendersonQuoted 10 lines · Richard Henderson
target-ppc/translate.c VSX_LOGICAL(xxlor, tcg_gen_or_tl) · line 7320
Quoted 9 lines · Tom Musta
target-ppc/translate.c +9
-0+tcg_gen_andi_i64 (a0, a0,0xFFFFFFFF00000000ul ); \ +tcg_gen_shli_i64 (a1, a1,32 ); \ +tcg_gen_shri_i64 (b0, b0,32 ); \ +tcg_gen_andi_i64 (b0, b0,0x00000000FFFFFFFFul ); \ +tcg_gen_andi_i64 (b1, b1,0x00000000FFFFFFFFul ); \ +tcg_gen_or_i64 (a0, a0, b0); \ +tcg_gen_or_i64 (a1, a1, b1); \ +tcg_gen_mov_i64 (cpu_vsrh (xT (ctx->opcode)), a0); \ +tcg_gen_mov_i64 (cpu_vsrl (xT (ctx->opcode)), a1); \
Quoted 4 lines
Two deposit operations.
r~
Richard: Thanks for the comments. I will rework this to use deposit (and also lxvw4x and xxspltw).
Re: [Qemu-devel] [PATCH 11/13] Add xxsel
In reply to Richard HendersonQuoted 12 lines · Richard Henderson
Review
Quoted 3 lines · Tom Musta
(quoted diff) +3
-0+ tcg_gen_and_i64(b, b, c); + tcg_gen_not_i64(c, c); + tcg_gen_and_i64(a, a, c);
tcg_gen_andc_i64.Review
Quoted 3 lines
(quoted diff) +3
-0+#define GEN_XXSEL() \ +GEN_XXSEL_ROW(0x00) \ +GEN_XXSEL_ROW(0x01) \Why bother with defining
GEN_XXSELwhen its only used once? Surely just put the rows there.
Quoted 5 lines
OTOH, this does suggest that we could do with a better way to decode the instructions, because this is ugly...
r~
Yeah ... it isn't very pretty. There was precedent for this (see, for example, rldcl). And the decoding logic very much wants to use instruction bits 26:30 and 21:25 as opc2 and opc3 respectively. Perhaps I could inject a handler for opcode 60 that would handle the VSX map a little more gracefully.
Is your concern aesthetic? Memory consumption? And do you feel this is a showstopper or something that could be addressed later?
Re: [Qemu-devel] [PATCH 11/13] Add xxsel
Quoted 2 lines · Tom Musta
Is your concern aesthetic? Memory consumption? And do you feel this is a showstopper or something that could be addressed later?
Aesthetic, and it definitely ought to be addressed outside this patch set.
r~