target/ppc: Move VSX storage access and compare
cover target/ppc: Move VSX storage access and compare
Moving all remaining VSX storage access instructions and all VSX compare instructions of XX3 form with RC field, to decodetree specification.
Chinmay Rath (4):
target/ppc: Moving VSX scalar storage access insns to decodetree. target/ppc: Move VSX vector with length storage access insns to
decodetree.
target/ppc: Move VSX vector storage access insns to decodetree. target/ppc: Move VSX fp compare insns to decodetree.
target/ppc/helper.h | 24 +- target/ppc/insn32.decode | 41 +++ target/ppc/fpu_helper.c | 16 +- target/ppc/mem_helper.c | 8 +- target/ppc/translate/vsx-impl.c.inc | 430 ++++++++++++++-------------- target/ppc/translate/vsx-ops.c.inc | 49 ---- 6 files changed, 286 insertions(+), 282 deletions(-)
Signature · 1 line
2.39.3
PATCH 1/4 target/ppc: Moving VSX scalar storage access insns to decodetree.
Moving the following instructions to decodetree specification :
lxs{d, iwa, ibz, ihz, iwz, sp}x : X-form
stxs{d, ib, ih, iw, sp}x : X-formThe changes were verified by validating that the tcg-ops generated by those instructions remain the same, which were captured using the '-d in_asm,op' flag.
--- target/ppc/insn32.decode | 13 +++++ target/ppc/translate/vsx-impl.c.inc | 79 +++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 11 ---- 3 files changed, 49 insertions(+), 54 deletions(-)
target/ppc/insn32.decode +13 -0
@@ -792,6 +792,19 @@ STXVRHX 011111 ..... ..... ..... 0010101101 . @X_TSX STXVRWX 011111 ..... ..... ..... 0011001101 . @X_TSX STXVRDX 011111 ..... ..... ..... 0011101101 . @X_TSX +LXSDX 011111 ..... ..... ..... 1001001100 . @X_TSX +LXSIWAX 011111 ..... ..... ..... 0001001100 . @X_TSX +LXSIBZX 011111 ..... ..... ..... 1100001101 . @X_TSX +LXSIHZX 011111 ..... ..... ..... 1100101101 . @X_TSX +LXSIWZX 011111 ..... ..... ..... 0000001100 . @X_TSX +LXSSPX 011111 ..... ..... ..... 1000001100 . @X_TSX + +STXSDX 011111 ..... ..... ..... 1011001100 . @X_TSX +STXSIBX 011111 ..... ..... ..... 1110001101 . @X_TSX +STXSIHX 011111 ..... ..... ..... 1110101101 . @X_TSX +STXSIWX 011111 ..... ..... ..... 0010001100 . @X_TSX +STXSSPX 011111 ..... ..... ..... 1010001100 . @X_TSX + ## VSX Vector Binary Floating-Point Sign Manipulation Instructions XVABSDP 111100 ..... 00000 ..... 111011001 .. @XX2
target/ppc/translate/vsx-impl.c.inc +36 -43
95 lines not on this page · raw message
target/ppc/translate/vsx-ops.c.inc +0 -11
@@ -1,9 +1,3 @@ -GEN_HANDLER_E(lxsdx, 0x1F, 0x0C, 0x12, 0, PPC_NONE, PPC2_VSX), -GEN_HANDLER_E(lxsiwax, 0x1F, 0x0C, 0x02, 0, PPC_NONE, PPC2_VSX207), -GEN_HANDLER_E(lxsiwzx, 0x1F, 0x0C, 0x00, 0, PPC_NONE, PPC2_VSX207), -GEN_HANDLER_E(lxsibzx, 0x1F, 0x0D, 0x18, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(lxsihzx, 0x1F, 0x0D, 0x19, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(lxsspx, 0x1F, 0x0C, 0x10, 0, PPC_NONE, PPC2_VSX207), GEN_HANDLER_E(lxvd2x, 0x1F, 0x0C, 0x1A, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(lxvwsx, 0x1F, 0x0C, 0x0B, 0, PPC_NONE, PPC2_ISA300), GEN_HANDLER_E(lxvdsx, 0x1F, 0x0C, 0x0A, 0, PPC_NONE, PPC2_VSX),
@@ -15,11 +9,6 @@ GEN_HANDLER_E(lxvl, 0x1F, 0x0D, 0x08, 0, PPC_NONE, PPC2_ISA300), GEN_HANDLER_E(lxvll, 0x1F, 0x0D, 0x09, 0, PPC_NONE, PPC2_ISA300), #endif -GEN_HANDLER_E(stxsdx, 0x1F, 0xC, 0x16, 0, PPC_NONE, PPC2_VSX), -GEN_HANDLER_E(stxsibx, 0x1F, 0xD, 0x1C, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(stxsihx, 0x1F, 0xD, 0x1D, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(stxsiwx, 0x1F, 0xC, 0x04, 0, PPC_NONE, PPC2_VSX207), -GEN_HANDLER_E(stxsspx, 0x1F, 0xC, 0x14, 0, PPC_NONE, PPC2_VSX207), GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(stxvw4x, 0x1F, 0xC, 0x1C, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(stxvh8x, 0x1F, 0x0C, 0x1D, 0, PPC_NONE, PPC2_ISA300),
Signature · 1 line
2.39.3
Re: [PATCH 1/4] target/ppc: Moving VSX scalar storage access insns to decodetree.
Quoted 8 lines · Chinmay Rath
Moving the following instructions to decodetree specification :
lxs{d, iwa, ibz, ihz, iwz, sp}x : X-form stxs{d, ib, ih, iw, sp}x : X-formThe changes were verified by validating that the tcg-ops generated by those instructions remain the same, which were captured using the '-d in_asm,op' flag.
Quoted 6 lines
Signed-off-by Chinmay Rath<[email protected]>--- target/ppc/insn32.decode | 13 +++++ target/ppc/translate/vsx-impl.c.inc | 79 +++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 11 ---- 3 files changed, 49 insertions(+), 54 deletions(-)
r~
PATCH 2/4 target/ppc: Move VSX vector with length storage access insns to decodetree.
In reply to Chinmay RathMoving the following instructions to decodetree specification :
{l, st}xvl(l) : X-form
The changes were verified by validating that the tcg-ops generated by those instructions remain the same, which were captured using the '-d in_asm,op' flag.
--- target/ppc/helper.h | 8 +-- target/ppc/insn32.decode | 6 ++ target/ppc/mem_helper.c | 8 +-- target/ppc/translate/vsx-impl.c.inc | 104 ++++++++++++++++++++-------- target/ppc/translate/vsx-ops.c.inc | 8 --- 5 files changed, 89 insertions(+), 45 deletions(-)
target/ppc/helper.h +4 -4
@@ -274,10 +274,10 @@ DEF_HELPER_3(stvebx, void, env, avr, tl) DEF_HELPER_3(stvehx, void, env, avr, tl) DEF_HELPER_3(stvewx, void, env, avr, tl)#if defined(TARGET_PPC64 ) -DEF_HELPER_4(lxvl, void, env, tl, vsr, tl) -DEF_HELPER_4(lxvll, void, env, tl, vsr, tl) -DEF_HELPER_4(stxvl, void, env, tl, vsr, tl) -DEF_HELPER_4(stxvll, void, env, tl, vsr, tl) +DEF_HELPER_4(LXVL, void, env, tl, vsr, tl) +DEF_HELPER_4(LXVLL, void, env, tl, vsr, tl) +DEF_HELPER_4(STXVL, void, env, tl, vsr, tl) +DEF_HELPER_4(STXVLL, void, env, tl, vsr, tl)#endif DEF_HELPER_4(vsumsws, void, env, avr, avr, avr)DEF_HELPER_4 (vsum2sws, void, env, avr, avr, avr)
target/ppc/insn32.decode +6 -0
@@ -805,6 +805,12 @@ STXSIHX 011111 ..... ..... ..... 1110101101 . @X_TSX STXSIWX 011111 ..... ..... ..... 0010001100 . @X_TSX STXSSPX 011111 ..... ..... ..... 1010001100 . @X_TSX +LXVL 011111 ..... ..... ..... 0100001101 . @X_TSX +LXVLL 011111 ..... ..... ..... 0100101101 . @X_TSX + +STXVL 011111 ..... ..... ..... 0110001101 . @X_TSX +STXVLL 011111 ..... ..... ..... 0110101101 . @X_TSX + ## VSX Vector Binary Floating-Point Sign Manipulation Instructions XVABSDP 111100 ..... 00000 ..... 111011001 .. @XX2
target/ppc/mem_helper.c +4 -4
@@ -467,8 +467,8 @@ void helper_##name(CPUPPCState *env, target_ulong addr, \ *xt = t; \ } -VSX_LXVL(lxvl, 0) -VSX_LXVL(lxvll, 1) +VSX_LXVL(LXVL, 0) +VSX_LXVL(LXVLL, 1)#undef VSX_LXVL#define VSX_STXVL (name, lj) \
@@ -496,8 +496,8 @@ void helper_##name(CPUPPCState *env, target_ulong addr, \ } \ } -VSX_STXVL(stxvl, 0) -VSX_STXVL(stxvll, 1) +VSX_STXVL(STXVL, 0) +VSX_STXVL(STXVLL, 1)#undef VSX_STXVL#undef GET_NB#endif /* TARGET_PPC64 */
target/ppc/translate/vsx-impl.c.inc +75 -29
111 lines not on this page · raw message
target/ppc/translate/vsx-ops.c.inc +0 -8
@@ -4,19 +4,11 @@ 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(lxvh8x, 0x1F, 0x0C, 0x19, 0, PPC_NONE, PPC2_ISA300), GEN_HANDLER_E(lxvb16x, 0x1F, 0x0C, 0x1B, 0, PPC_NONE, PPC2_ISA300), -#if defined(TARGET_PPC64) -GEN_HANDLER_E(lxvl, 0x1F, 0x0D, 0x08, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(lxvll, 0x1F, 0x0D, 0x09, 0, PPC_NONE, PPC2_ISA300), -#endif GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(stxvw4x, 0x1F, 0xC, 0x1C, 0, PPC_NONE, PPC2_VSX), GEN_HANDLER_E(stxvh8x, 0x1F, 0x0C, 0x1D, 0, PPC_NONE, PPC2_ISA300), GEN_HANDLER_E(stxvb16x, 0x1F, 0x0C, 0x1F, 0, PPC_NONE, PPC2_ISA300), -#if defined(TARGET_PPC64) -GEN_HANDLER_E(stxvl, 0x1F, 0x0D, 0x0C, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(stxvll, 0x1F, 0x0D, 0x0D, 0, PPC_NONE, PPC2_ISA300), -#endif GEN_HANDLER_E(mfvsrwz, 0x1F, 0x13, 0x03, 0x0000F800, PPC_NONE, PPC2_VSX207), GEN_HANDLER_E(mtvsrwa, 0x1F, 0x13, 0x06, 0x0000F800, PPC_NONE, PPC2_VSX207),
Signature · 1 line
2.39.3
Re: [PATCH 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.
Quoted 11 lines · Chinmay Rath
target/ppc/translate/vsx-impl.c.inc +11
-0+static bool do_ld_st_vl(DisasContext *ctx, arg_X *a, + void (*helper)(TCGv_ptr, TCGv, TCGv_ptr, TCGv)) +{ + TCGv EA; + TCGv_ptr xt; + if (a->rt < 32) { + REQUIRE_VSX(ctx); + } else { + REQUIRE_VECTOR(ctx); + } + xt = gen_vsr_ptr(a->rt);
target/ppc/translate/vsx-impl.c.inc static void gen_lxvb16x(DisasContext *ctx) · line 256
Quoted 10 lines
target/ppc/translate/vsx-impl.c.inc +10
-0+ gen_set_access_type(ctx, ACCESS_INT); + + if (a->ra) { + EA = tcg_temp_new(); + tcg_gen_mov_tl(EA, cpu_gpr[a->ra]); + } else { + EA = tcg_constant_tl(0); + } + if (NARROW_MODE(ctx)) { + tcg_gen_ext32u_tl(EA, EA);
ra == 0, narrow mode, will crash, due to write into constant 0. Obviously 0 does not need extending, so this could be
if (!a->ra) {
ea = constant 0;
} else if (narrow mode) {
ea = tcg_temp_new();
tcg_gen_ext32u_tl(ea, cpu_gpr[a->ra]);
} else {
ra = cpu_gpr[a->ra];
}Aren't there existing helper functions for computing this address? And if not, better to create one.
r~
Re: [PATCH 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.
Hi Richard,
Quoted 26 lines · Richard Henderson
Quoted 11 lines · Chinmay Rath
target/ppc/translate/vsx-impl.c.inc +11
-0+static bool do_ld_st_vl(DisasContext *ctx, arg_X *a, + void (*helper)(TCGv_ptr, TCGv, TCGv_ptr, TCGv)) +{ + TCGv EA; + TCGv_ptr xt; + if (a->rt < 32) { + REQUIRE_VSX(ctx); + } else { + REQUIRE_VECTOR(ctx); + } + xt = gen_vsr_ptr(a->rt);target/ppc/translate/vsx-impl.c.inc static void gen_lxvb16x(DisasContext *ctx) · line 256
Quoted 10 lines
target/ppc/translate/vsx-impl.c.inc +10
-0+ gen_set_access_type(ctx, ACCESS_INT); + + if (a->ra) { + EA = tcg_temp_new(); + tcg_gen_mov_tl(EA, cpu_gpr[a->ra]); + } else { + EA = tcg_constant_tl(0); + } + if (NARROW_MODE(ctx)) { + tcg_gen_ext32u_tl(EA, EA);ra == 0, narrow mode, will crash, due to write into constant 0. Obviously 0 does not need extending, so this could be
Quoted 9 lines
if (!a->ra) {
ea = constant 0;
} else if (narrow mode) {
ea =tcg_temp_new();
tcg_gen_ext32u_tl(ea, cpu_gpr[a->ra]);
} else {
ra = cpu_gpr[a->ra];
}
^ Thank you Richard, will take care in v2.
Quoted 3 lines
Aren't there existing helper functions for computing this address? And if not, better to create one.
^
The calculation of effective address in these instructions is slightly different than the others, for which helper function exist :
EA for these insns : EA ← (RA=0) ? 0 : GPR[RA]
EA for rest storage access insns : EA ← ((RA=0) ? 0 : GPR[RA]) + GPR[RB]
This is why I could not reuse that function. Also, this calculation of EA is limited to these 4 insns above, and only 2 others (prefixed insns), which is why I did not create a new function for this, considering it won't be reused for any other insn.
Please let me know if I should create a new function in this case as well.
Thanks and Regards,
Chinmay
Quoted 4 lines
r~
Re: [PATCH 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.
Quoted 11 lines · Chinmay Rath
The calculation of effective address in these instructions is slightly different than the others, for which helper function exist :
EA for these insns : EA ← (RA=0) ? 0 : GPR[RA]
EA for rest storage access insns : EA ← ((RA=0) ? 0 : GPR[RA]) + GPR[RB]This is why I could not reuse that function. Also, this calculation of EA is limited to these 4 insns above, and only 2 others (prefixed insns), which is why I did not create a new function for this, considering it won't be reused for any other insn.
Please let me know if I should create a new function in this case as well.
If you expect this to be used just once, then leaving it inline is perfectly fine.
r~
PATCH 3/4 target/ppc: Move VSX vector storage access insns to decodetree.
In reply to Chinmay RathMoving the following instructions to decodetree specification:
lxv{b16, d2, h8, w4, ds, ws}x : X-form
stxv{b16, d2, h8, w4}x : X-form
The changes were verified by validating that the tcg-ops generated for those instructions remain the same, which were captured using the '-d in_asm,op' flag.
--- target/ppc/insn32.decode | 10 ++ target/ppc/translate/vsx-impl.c.inc | 199 ++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 12 -- 3 files changed, 97 insertions(+), 124 deletions(-)
target/ppc/insn32.decode +10 -0
@@ -805,9 +805,19 @@ STXSIHX 011111 ..... ..... ..... 1110101101 . @X_TSX STXSIWX 011111 ..... ..... ..... 0010001100 . @X_TSX STXSSPX 011111 ..... ..... ..... 1010001100 . @X_TSX +LXVB16X 011111 ..... ..... ..... 1101101100 . @X_TSX +LXVD2X 011111 ..... ..... ..... 1101001100 . @X_TSX +LXVH8X 011111 ..... ..... ..... 1100101100 . @X_TSX +LXVW4X 011111 ..... ..... ..... 1100001100 . @X_TSX +LXVDSX 011111 ..... ..... ..... 0101001100 . @X_TSX +LXVWSX 011111 ..... ..... ..... 0101101100 . @X_TSX LXVL 011111 ..... ..... ..... 0100001101 . @X_TSX LXVLL 011111 ..... ..... ..... 0100101101 . @X_TSX +STXVB16X 011111 ..... ..... ..... 1111101100 . @X_TSX +STXVD2X 011111 ..... ..... ..... 1111001100 . @X_TSX +STXVH8X 011111 ..... ..... ..... 1110101100 . @X_TSX +STXVW4X 011111 ..... ..... ..... 1110001100 . @X_TSX STXVL 011111 ..... ..... ..... 0110001101 . @X_TSX STXVLL 011111 ..... ..... ..... 0110101101 . @X_TSX
target/ppc/translate/vsx-impl.c.inc +87 -112
330 lines not on this page · raw message
target/ppc/translate/vsx-ops.c.inc +0 -12
@@ -1,15 +1,3 @@ -GEN_HANDLER_E(lxvd2x, 0x1F, 0x0C, 0x1A, 0, PPC_NONE, PPC2_VSX), -GEN_HANDLER_E(lxvwsx, 0x1F, 0x0C, 0x0B, 0, PPC_NONE, PPC2_ISA300), -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(lxvh8x, 0x1F, 0x0C, 0x19, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(lxvb16x, 0x1F, 0x0C, 0x1B, 0, PPC_NONE, PPC2_ISA300), - -GEN_HANDLER_E(stxvd2x, 0x1F, 0xC, 0x1E, 0, PPC_NONE, PPC2_VSX), -GEN_HANDLER_E(stxvw4x, 0x1F, 0xC, 0x1C, 0, PPC_NONE, PPC2_VSX), -GEN_HANDLER_E(stxvh8x, 0x1F, 0x0C, 0x1D, 0, PPC_NONE, PPC2_ISA300), -GEN_HANDLER_E(stxvb16x, 0x1F, 0x0C, 0x1F, 0, PPC_NONE, PPC2_ISA300), - GEN_HANDLER_E(mfvsrwz, 0x1F, 0x13, 0x03, 0x0000F800, PPC_NONE, PPC2_VSX207), GEN_HANDLER_E(mtvsrwa, 0x1F, 0x13, 0x06, 0x0000F800, PPC_NONE, PPC2_VSX207), GEN_HANDLER_E(mtvsrwz, 0x1F, 0x13, 0x07, 0x0000F800, PPC_NONE, PPC2_VSX207),
Signature · 1 line
2.39.3
Re: [PATCH 3/4] target/ppc: Move VSX vector storage access insns to decodetree.
Quoted 8 lines · Chinmay Rath
Moving the following instructions to decodetree specification:
lxv{b16, d2, h8, w4, ds, ws}x : X-form stxv{b16, d2, h8, w4}x : X-formThe changes were verified by validating that the tcg-ops generated for those instructions remain the same, which were captured using the '-d in_asm,op' flag.
Quoted 6 lines
Signed-off-by Chinmay Rath <[email protected]>--- target/ppc/insn32.decode | 10 ++ target/ppc/translate/vsx-impl.c.inc | 199 ++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 12 -- 3 files changed, 97 insertions(+), 124 deletions(-)
Because the ops are identical,
But you really should update these to use tcg_gen_qemu_ld/st_i128 with the proper atomicity flags. This will fix an existing bug...
target/ppc/translate/vsx-impl.c.inc TRANS_FLAGS2(ISA300, LXSIHZX, do_lxs, gen_qemu_ld16u_i64); · line 61
Quoted 13 lines
target/ppc/translate/vsx-impl.c.inc +7
-0+static bool trans_LXVD2X(DisasContext *ctx, arg_LXVD2X *a) { TCGv EA; TCGv_i64 t0; + + REQUIRE_VSX(ctx); + REQUIRE_INSNS_FLAGS2(ctx, VSX); + t0 = tcg_temp_new_i64(); gen_set_access_type(ctx, ACCESS_INT); + EA = do_ea_calc(ctx, a->ra, cpu_gpr[a->rb]); gen_qemu_ld64_i64(ctx, t0, EA); + set_cpu_vsr(a->rt, t0, true);
where the vector register is partially modified ...
Quoted 2 lines
tcg_gen_addi_tl(EA, EA, 8);
gen_qemu_ld64_i64(ctx, t0, EA);before a fault from the second load is recognized. Similarly for stores leaving memory partially modified.
r~
Re: [PATCH 3/4] target/ppc: Move VSX vector storage access insns to decodetree.
Hi Richard,
Quoted 16 lines · Richard Henderson
Quoted 8 lines · Chinmay Rath
Moving the following instructions to decodetree specification:
lxv{b16, d2, h8, w4, ds, ws}x : X-form
stxv{b16, d2, h8, w4}x : X-formThe changes were verified by validating that the tcg-ops generated for those instructions remain the same, which were captured using the '-d in_asm,op' flag.
Quoted 6 lines
Signed-off-by Chinmay Rath <[email protected]>---
target/ppc/insn32.decode | 10 ++ target/ppc/translate/vsx-impl.c.inc | 199 ++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 12 --3 files changed, 97 insertions(+), 124 deletions(-)
Quoted 5 lines
Because the ops are identical,
Reviewed-by Richard Henderson <[email protected]>But you really should update these to use tcg_gen_qemu_ld/st_i128 with the proper atomicity flags. This will fix an existing bug...
^ Surely Richard, I have noted this suggestion of yours from an earlier patch, and plan to do this change and implement a few of your other suggestions, which I couldn't implement earlier, along with some clean-ups this week.
I refrained from doing it with the decodetree movement, to take proper time to understand and test.
Should send out those patches soon.
Thanks & Regards,
Chinmay
Quoted 26 lines
Quoted 8 lines
+static bool
trans_LXVD2X(DisasContext *ctx, arg_LXVD2X *a){
TCGv EA;
TCGv_i64t0;target/ppc/translate/vsx-impl.c.inc +4
-0+ + REQUIRE_VSX(ctx); + REQUIRE_INSNS_FLAGS2(ctx, VSX); +Quoted 5 lines
t0 =
tcg_temp_new_i64();
gen_set_access_type(ctx, ACCESS_INT);
+ EA =do_ea_calc(ctx, a->ra, cpu_gpr[a->rb]);
gen_qemu_ld64_i64(ctx, t0, EA);
+set_cpu_vsr(a->rt, t0, true);where the vector register is partially modified ...
Quoted 2 lines
tcg_gen_addi_tl(EA, EA, 8);
gen_qemu_ld64_i64(ctx, t0, EA);before a fault from the second load is recognized. Similarly for stores leaving memory partially modified.
r~
Re: [PATCH 3/4] target/ppc: Move VSX vector storage access insns to decodetree.
In reply to Richard HendersonHi Richard,
My apologies for the ill formatted reply in this patch series. Just realized it now. The cliched 'Tab' issue with the mail client XD.
Quoted 16 lines · Richard Henderson
Quoted 8 lines · Chinmay Rath
Moving the following instructions to decodetree specification:
lxv{b16, d2, h8, w4, ds, ws}x : X-form
stxv{b16, d2, h8, w4}x : X-formThe changes were verified by validating that the tcg-ops generated for those instructions remain the same, which were captured using the '-d in_asm,op' flag.
Quoted 6 lines
Signed-off-by Chinmay Rath <[email protected]>---
target/ppc/insn32.decode | 10 ++ target/ppc/translate/vsx-impl.c.inc | 199 ++++++++++++---------------- target/ppc/translate/vsx-ops.c.inc | 12 --3 files changed, 97 insertions(+), 124 deletions(-)
Quoted 4 lines
Because the ops are identical,
Reviewed-by Richard Henderson <[email protected]>But you really should update these to use tcg_gen_qemu_ld/st_i128 with the proper atomicity flags. This will fix an existing bug...
^
Surely Richard, I have noted this suggestion from earlier patch and plan to do this, and a few others which I couldn't implement earlier, along with some clean-ups this week.
I refrained from doing it with the decodetree movement, to take time to properly understand and test. Should send out those patches soon.
Thanks & Regards,
Chinmay
Quoted 27 lines
Quoted 9 lines
+static bool
trans_LXVD2X(DisasContext *ctx, arg_LXVD2X *a){
TCGv EA;
TCGv_i64t0;target/ppc/translate/vsx-impl.c.inc +4
-0+ + REQUIRE_VSX(ctx); + REQUIRE_INSNS_FLAGS2(ctx, VSX); +Quoted 5 lines
t0 =
tcg_temp_new_i64();
gen_set_access_type(ctx, ACCESS_INT);
+ EA =do_ea_calc(ctx, a->ra, cpu_gpr[a->rb]);
gen_qemu_ld64_i64(ctx, t0, EA);
+set_cpu_vsr(a->rt, t0, true);where the vector register is partially modified ...
Quoted 2 lines
tcg_gen_addi_tl(EA, EA, 8);
gen_qemu_ld64_i64(ctx, t0, EA);before a fault from the second load is recognized. Similarly for stores leaving memory partially modified.
r~
PATCH 4/4 target/ppc: Move VSX fp compare insns to decodetree.
In reply to Chinmay RathMoving the following instructions to decodetree specification:
xvcmp{eq, gt, ge, ne}{s, d}p : XX3-formThe changes were verified by validating that the tcg-ops generated for those instructions remain the same which were captured using the '-d in_asm,op' flag.
--- target/ppc/helper.h | 16 ++++----- target/ppc/insn32.decode | 12 +++++++ target/ppc/fpu_helper.c | 16 ++++----- target/ppc/translate/vsx-impl.c.inc | 50 ++++++++++++++--------------- target/ppc/translate/vsx-ops.c.inc | 18 ----------- 5 files changed, 52 insertions(+), 60 deletions(-)
target/ppc/helper.h +8 -8
@@ -473,10 +473,10 @@ DEF_HELPER_5(xvnmadddp, void, env, vsr, vsr, vsr, vsr) DEF_HELPER_5(xvnmsubdp, void, env, vsr, vsr, vsr, vsr) DEF_HELPER_4(XVMAXDP, void, env, vsr, vsr, vsr) DEF_HELPER_4(XVMINDP, void, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpeqdp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpgedp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpgtdp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpnedp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPEQDP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPGEDP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPGTDP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPNEDP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) DEF_HELPER_3(xvcvdpsp, void, env, vsr, vsr) DEF_HELPER_3(xvcvdpsxds, void, env, vsr, vsr)DEF_HELPER_3 (xvcvdpsxws, void, env, vsr, vsr)
@@ -507,10 +507,10 @@ DEF_HELPER_5(xvnmaddsp, void, env, vsr, vsr, vsr, vsr) DEF_HELPER_5(xvnmsubsp, void, env, vsr, vsr, vsr, vsr) DEF_HELPER_4(XVMAXSP, void, env, vsr, vsr, vsr) DEF_HELPER_4(XVMINSP, void, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpeqsp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpgesp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpgtsp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) -DEF_HELPER_FLAGS_4(xvcmpnesp, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPEQSP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPGESP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPGTSP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) +DEF_HELPER_FLAGS_4(XVCMPNESP, TCG_CALL_NO_RWG, i32, env, vsr, vsr, vsr) DEF_HELPER_3(xvcvspdp, void, env, vsr, vsr) DEF_HELPER_3(xvcvsphp, void, env, vsr, vsr)DEF_HELPER_3 (xvcvhpsp, void, env, vsr, vsr)
target/ppc/insn32.decode +12 -0
@@ -217,6 +217,9 @@ &XX3 xt xa xb @XX3 ...... ..... ..... ..... ........ ... &XX3 xt=%xx_xt xa=%xx_xa xb=%xx_xb +&XX3_rc xt xa xb rc:bool +@XX3_rc ...... ..... ..... ..... rc:1 ....... ... &XX3_rc xt=%xx_xt xa=%xx_xa xb=%xx_xb + # 32 bit GER instructions have all mask bits considered 1 &MMIRR_XX3 xa xb xt pmsk xmsk ymsk %xx_at 23:3
@@ -923,6 +926,15 @@ XSCMPEQQP 111111 ..... ..... ..... 0001000100 - @X XSCMPGEQP 111111 ..... ..... ..... 0011000100 - @X XSCMPGTQP 111111 ..... ..... ..... 0011100100 - @X +XVCMPEQSP 111100 ..... ..... ..... . 1000011 ... @XX3_rc +XVCMPGTSP 111100 ..... ..... ..... . 1001011 ... @XX3_rc +XVCMPGESP 111100 ..... ..... ..... . 1010011 ... @XX3_rc +XVCMPNESP 111100 ..... ..... ..... . 1011011 ... @XX3_rc +XVCMPEQDP 111100 ..... ..... ..... . 1100011 ... @XX3_rc +XVCMPGTDP 111100 ..... ..... ..... . 1101011 ... @XX3_rc +XVCMPGEDP 111100 ..... ..... ..... . 1110011 ... @XX3_rc +XVCMPNEDP 111100 ..... ..... ..... . 1111011 ... @XX3_rc + XSMAXDP 111100 ..... ..... ..... 10100000 ... @XX3 XSMINDP 111100 ..... ..... ..... 10101000 ... @XX3
target/ppc/fpu_helper.c +8 -8
@@ -2624,14 +2624,14 @@ uint32_t helper_##op(CPUPPCState *env, ppc_vsr_t *xt, \return crf6; \ } -VSX_CMP(xvcmpeqdp, 2, float64, VsrD(i), eq, 0, 1) -VSX_CMP(xvcmpgedp, 2, float64, VsrD(i), le, 1, 1) -VSX_CMP(xvcmpgtdp, 2, float64, VsrD(i), lt, 1, 1) -VSX_CMP(xvcmpnedp, 2, float64, VsrD(i), eq, 0, 0) -VSX_CMP(xvcmpeqsp, 4, float32, VsrW(i), eq, 0, 1) -VSX_CMP(xvcmpgesp, 4, float32, VsrW(i), le, 1, 1) -VSX_CMP(xvcmpgtsp, 4, float32, VsrW(i), lt, 1, 1) -VSX_CMP (xvcmpnesp,4 , float32,VsrW (i), eq,0 ,0 ) +VSX_CMP(XVCMPEQDP, 2, float64, VsrD(i), eq, 0, 1) +VSX_CMP(XVCMPGEDP, 2, float64, VsrD(i), le, 1, 1) +VSX_CMP(XVCMPGTDP, 2, float64, VsrD(i), lt, 1, 1) +VSX_CMP(XVCMPNEDP, 2, float64, VsrD(i), eq, 0, 0) +VSX_CMP(XVCMPEQSP, 4, float32, VsrW(i), eq, 0, 1) +VSX_CMP(XVCMPGESP, 4, float32, VsrW(i), le, 1, 1) +VSX_CMP(XVCMPGTSP, 4, float32, VsrW(i), lt, 1, 1) +VSX_CMP (XVCMPNESP ,4 , float32,VsrW (i), eq,0 ,0 )/* * VSX_CVT_FP_TO_FP - VSX floating point/floating point conversion
target/ppc/translate/vsx-impl.c.inc +24 -26
58 lines not on this page · raw message
target/ppc/translate/vsx-ops.c.inc +0 -18
@@ -43,16 +43,6 @@ GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 1, opc3, 1, PPC_NONE, fl2), \ GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 2, opc3, 1, PPC_NONE, fl2), \ GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 3, opc3, 1, PPC_NONE, fl2) -#define GEN_XX3_RC_FORM(name, opc2, opc3, fl2) \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x00, opc3 | 0x00, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x01, opc3 | 0x00, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x02, opc3 | 0x00, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x03, opc3 | 0x00, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x00, opc3 | 0x10, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x01, opc3 | 0x10, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x02, opc3 | 0x10, 0, PPC_NONE, fl2), \ -GEN_HANDLER2_E(name, #name, 0x3C, opc2 | 0x03, opc3 | 0x10, 0, PPC_NONE, fl2) - #define GEN_XX3FORM_DM(name, opc2, opc3) \ GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x00, opc3|0x00, 0, PPC_NONE, PPC2_VSX),\ GEN_HANDLER2_E(name, #name, 0x3C, opc2|0x01, opc3|0x00, 0, PPC_NONE, PPC2_VSX),\
@@ -175,10 +165,6 @@ GEN_XX3FORM_NAME(xvnmadddp, "xvnmaddadp", 0x04, 0x1C, PPC2_VSX), GEN_XX3FORM_NAME(xvnmadddp, "xvnmaddmdp", 0x04, 0x1D, PPC2_VSX), GEN_XX3FORM_NAME(xvnmsubdp, "xvnmsubadp", 0x04, 0x1E, PPC2_VSX), GEN_XX3FORM_NAME(xvnmsubdp, "xvnmsubmdp", 0x04, 0x1F, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpeqdp, 0x0C, 0x0C, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpgtdp, 0x0C, 0x0D, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpgedp, 0x0C, 0x0E, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpnedp, 0x0C, 0x0F, PPC2_ISA300), GEN_XX2FORM(xvcvdpsp, 0x12, 0x18, PPC2_VSX), GEN_XX2FORM(xvcvdpsxds, 0x10, 0x1D, PPC2_VSX), GEN_XX2FORM(xvcvdpsxws, 0x10, 0x0D, PPC2_VSX),
@@ -207,10 +193,6 @@ GEN_XX3FORM_NAME(xvnmaddsp, "xvnmaddasp", 0x04, 0x18, PPC2_VSX), GEN_XX3FORM_NAME(xvnmaddsp, "xvnmaddmsp", 0x04, 0x19, PPC2_VSX), GEN_XX3FORM_NAME(xvnmsubsp, "xvnmsubasp", 0x04, 0x1A, PPC2_VSX), GEN_XX3FORM_NAME(xvnmsubsp, "xvnmsubmsp", 0x04, 0x1B, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpeqsp, 0x0C, 0x08, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpgtsp, 0x0C, 0x09, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpgesp, 0x0C, 0x0A, PPC2_VSX), -GEN_XX3_RC_FORM(xvcmpnesp, 0x0C, 0x0B, PPC2_ISA300), GEN_XX2FORM(xvcvspdp, 0x12, 0x1C, PPC2_VSX), GEN_XX2FORM(xvcvspsxds, 0x10, 0x19, PPC2_VSX), GEN_XX2FORM(xvcvspsxws, 0x10, 0x09, PPC2_VSX),
Signature · 1 line
2.39.3
Re: [PATCH 4/4] target/ppc: Move VSX fp compare insns to decodetree.
Quoted 5 lines · Chinmay Rath
target/ppc/translate/vsx-impl.c.inc +5
-0+static bool do_cmp(DisasContext *ctx, arg_XX3_rc *a, + void (*helper)(TCGv_i32, TCGv_ptr, TCGv_ptr, TCGv_ptr, TCGv_ptr)) +{ + TCGv_i32 ignored; + TCGv_ptr xt, xa, xb;
target/ppc/translate/vsx-impl.c.inc static bool do_xvcpsgn(DisasContext *ctx, arg_XX3 *a, unsigned vece) · line 819
Quoted 10 lines
target/ppc/translate/vsx-impl.c.inc +10
-0+ REQUIRE_VSX(ctx); + xt = gen_vsr_ptr(a->xt); + xa = gen_vsr_ptr(a->xa); + xb = gen_vsr_ptr(a->xb); + if (a->rc) { + helper(cpu_crf[6], tcg_env, xt, xa, xb); + } else { + ignored = tcg_temp_new_i32(); + helper(ignored, tcg_env, xt, xa, xb); + }
Better to unify the helper call. E.g.
dest = a->rc ? cpu_crf[6] : tcg_temp_new_i32();
helper(dest, ...)Anyway,
r~
Re: [PATCH 4/4] target/ppc: Move VSX fp compare insns to decodetree.
Quoted 20 lines · Richard Henderson
Quoted 8 lines · Chinmay Rath
+static bool do_cmp(DisasContext *ctx,
arg_XX3_rc*a,
+ void (*helper)(TCGv_i32,TCGv_ptr,TCGv_ptr,TCGv_ptr,TCGv_ptr))
+{
+TCGv_i32ignored;
+TCGv_ptrxt, xa, xb;
+REQUIRE_VSX(ctx);
+ xt =gen_vsr_ptr(a->xt);Quoted 8 lines
+ xa =
gen_vsr_ptr(a->xa);
+ xb =gen_vsr_ptr(a->xb);
+ if (a->rc) {
+helper(cpu_crf[6], tcg_env, xt, xa, xb);
+ } else {
+ ignored =tcg_temp_new_i32();
+helper(ignored, tcg_env, xt, xa, xb);
+ }Better to unify the helper call. E.g.
Quoted 3 lines
dest =
a->rc? cpu_crf[6] :tcg_temp_new_i32();helper(dest, ...)
^
Sure Richard, will do in v2.
Thanks & Regards,
Chinmay
Quoted 6 lines
Anyway,
Reviewed-by Richard Henderson <[email protected]>r~