Skip to contentSkip to conversation

Move VSX storage access and compare insns to

PATCH Superseded Chinmay Rath 3 participants 12 messages Jun 13, 2024 – Jun 18, 2024

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover Move VSX storage access and compare insns to

Moving all remaining VSX storage access instructions and all VSX compare instructions of XX3 form with RC field, to decodetree specification.

Change log :

v2:
- Addressed comments by Richard in v1
- Patch 2/4 : Handled proper ea calculation in narrow mode. Also created a new function for ea calculation instead of inlining, for later use by (p){lx,stx}vp insns.
- Patch 4/4 : Unified helper calls.
- Retained Richard's "Reviewed-by" in patches 1, 3 and 4.

v1: https://lore.kernel.org/qemu-devel/[email protected]/

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.c              |  18 ++
 target/ppc/translate/vsx-impl.c.inc | 416 ++++++++++++++--------------
 target/ppc/translate/vsx-ops.c.inc  |  49 ----
 7 files changed, 290 insertions(+), 282 deletions(-)
Signature · 1 line
2.39.3
#1 lore raw

PATCH v2 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-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/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
index 30d6f9f750..88753c75e1 100644
@@ -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
index 3c0a70cb7c..d44cb55836 100644
@@ -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
#2 lore raw

PATCH v2 2/4 target/ppc: Move VSX vector with length storage access insns to decodetree.

In reply to Chinmay Rath

Moving 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.

Also added a new function to calculate the effective address :
EA <- (RA == 0) ? 0 : GPR[RA], which is now used by the above-said insns, and shall be used later by (p){lx, stx}vp insns.

---
 target/ppc/helper.h                 |  8 +--
 target/ppc/insn32.decode            |  6 ++
 target/ppc/mem_helper.c             |  8 +--
 target/ppc/translate.c              | 18 ++++++
 target/ppc/translate/vsx-impl.c.inc | 94 ++++++++++++++++++++---------
 target/ppc/translate/vsx-ops.c.inc  |  8 ---
 6 files changed, 97 insertions(+), 45 deletions(-)
target/ppc/helper.h +4 -4
index 3b4a0c4674..510ce76524 100644
@@ -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
index 88753c75e1..445fdb341f 100644
@@ -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
index ea7e8443a8..dec1b25eb8 100644
@@ -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.c +18 -0
index 93ffec787c..a1f2f4fbda 100644
@@ -3096,6 +3096,7 @@ static inline void gen_align_no_le(DisasContext *ctx)
                       (ctx->opcode & 0x03FF0000) | POWERPC_EXCP_ALIGN_LE);
 }
 
+/* EA <- {(ra == 0) ? 0 : GPR[ra]} + displ */
 static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ)
 {
     TCGv ea = tcg_temp_new();
@@ -3110,6 +3111,23 @@ static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ)
     return ea;
 }
 
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);
+    }
+    return EA;
+}
+
 /***                             Integer load                              ***/
 #define DEF_MEMOP(op) ((op) | ctx->default_tcg_memop_mask)
 #define BSWAP_MEMOP(op) ((op) | (ctx->default_tcg_memop_mask ^ MO_BSWAP))
target/ppc/translate/vsx-impl.c.inc +65 -29

101 lines not on this page · raw message

target/ppc/translate/vsx-ops.c.inc +0 -8
index d44cb55836..7f4326c974 100644
@@ -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
#3 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

target/ppc/translate.c static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ) · line 3126

Quoted 13 lines · Chinmay Rath
target/ppc/translate.c +13 -0
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);

Why are you making a copy, rather than just returning cpu_gpr[ra]? If you need to modify the resulting EA, then you also need to make a copy for 0.

r~

#4 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

Hi Richard,

Quoted 16 lines · Richard Henderson

target/ppc/translate.c static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ) · line 3126

Quoted 13 lines · Chinmay Rath
target/ppc/translate.c +13 -0
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);

Why are you making a copy, rather than just returning cpu_gpr[ra]?

True, this tcg move is redundant. Was carried away to maintain uniformity with the original do_ea_calc function. My bad!

This can rather just be :
/* ea <- (ra == 0) ? 0 : GPR[ra] */
static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
    TCGv EA;
    if (!ra) {
        return tcg_constant_tl(0);
    }
    if (NARROW_MODE(ctx)) {
        EA = tcg_temp_new(); tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
        return cpu_gpr[ra];
    }
    return EA;
}

Quoted 2 lines

If you need to modify the resulting EA, then you also need to make a copy for 0.

Hey, didn't properly get what you meant here.
Did you mean : Since I'm using a tcg_constant for 0, if the EA is to be modified later, this constant would be an issue, in which case, I should make a copy for it ??

Considering that, there are no tcg level modifications with this EA. However, the underlying helper method, which considers this EA as a target_ulong type does modify it, which I don't think should be an issue.

Please let me know if I missed something.

Thanks & Regards,
Chinmay

Quoted 2 lines

r~

#5 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

Quoted 15 lines · Chinmay Rath

static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
    TCGv EA;
    if (!ra) {
        return tcg_constant_tl(0);
    }
    if (NARROW_MODE(ctx)) {
        EA = tcg_temp_new(); tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
        return cpu_gpr[ra];
    }
    return EA;
}

Quoted 4 lines
Quoted 2 lines

If you need to modify the resulting EA, then you also need to make a copy for 0.

Hey, didn't properly get what you meant here.
Did you mean : Since I'm using a tcg_constant for 0, if the EA is to be modified later, this constant would be an issue, in which case, I should make a copy for it ??

Yes.

Quoted 1 line

Considering that, there are no tcg level modifications with this EA.

Ok, good.

Quoted 1 line

However, the underlying helper method, which considers this EA as a target_ulong type does modify it, which I don't think should be an issue.

Correct, that's fine.

r~

#6 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

Quoted 23 lines · Richard Henderson
Quoted 15 lines · Chinmay Rath

static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
     TCGv EA;
     if (!ra) {
         return tcg_constant_tl(0);
     }
     if (NARROW_MODE(ctx)) {
         EA = tcg_temp_new(); tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
         return cpu_gpr[ra];
     }
     return EA;
}

Quoted 4 lines
Quoted 2 lines

If you need to modify the resulting EA, then you also need to make a copy for 0.

Hey, didn't properly get what you meant here.
Did you mean : Since I'm using a tcg_constant for 0, if the EA is to be modified later, this constant would be an issue, in which case, I should make a copy for it ??

Yes.

Quoted 8 lines
Quoted 1 line

Considering that, there are no tcg level modifications with this EA.

Ok, good.

Quoted 1 line

However, the underlying helper method, which considers this EA as a target_ulong type does modify it, which I don't think should be an issue.

Correct, that's fine.

Awesome ! Thanks for the clarification.

Regards,
Chinmay

Quoted 4 lines

r~

#7 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

In reply to Richard Henderson

Hi Richard,

Quoted 18 lines · Richard Henderson

target/ppc/translate.c static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ) · line 3126

Quoted 13 lines · Chinmay Rath
target/ppc/translate.c +13 -0
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);

Why are you making a copy, rather than just returning cpu_gpr[ra]? If you need to modify the resulting EA, then you also need to make a copy for 0.

Please ignore my previous response.
I think do_ea_calc_ra should allow modification to the resulting EA, hence below change appears more appropriate to me :

/* EA <- (ra == 0) ? 0 : GPR[ra] */
static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
    TCGv EA = tcg_temp_new();
    if (!ra) {
        tcg_gen_movi_tl(EA, 0);
        return EA;
    }
    if (NARROW_MODE(ctx)) { tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
        tcg_gen_mov_tl(EA, cpu_gpr[ra]);
    }
    return EA;
}

Let me know your thoughts.

Thanks & Regards,
Chinmay

Quoted 3 lines

r~

#8 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

Quoted 25 lines · Chinmay Rath

Hi Richard,

Quoted 18 lines · Richard Henderson

target/ppc/translate.c static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ) · line 3126

Quoted 13 lines · Chinmay Rath
target/ppc/translate.c +13 -0
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);

Why are you making a copy, rather than just returning cpu_gpr[ra]? If you need to modify the resulting EA, then you also need to make a copy for 0.

Please ignore my previous response.
I think do_ea_calc_ra should allow modification to the resulting EA, hence below change appears more appropriate to me :

Quoted 15 lines

/* EA <- (ra == 0) ? 0 : GPR[ra] */
static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
    TCGv EA = tcg_temp_new();
    if (!ra) {
        tcg_gen_movi_tl(EA, 0);
        return EA;
    }
    if (NARROW_MODE(ctx)) { tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
        tcg_gen_mov_tl(EA, cpu_gpr[ra]);
    }
    return EA;
}

If that's what's needed by the callers of do_ea_calc_ra, then yes. You can drop the first return EA and use else if instead.

r~

#9 lore raw

Re: [PATCH v2 2/4] target/ppc: Move VSX vector with length storage access insns to decodetree.

Quoted 42 lines · Richard Henderson
Quoted 25 lines · Chinmay Rath

Hi Richard,

Quoted 18 lines · Richard Henderson

target/ppc/translate.c static TCGv do_ea_calc(DisasContext *ctx, int ra, TCGv displ) · line 3126

Quoted 13 lines · Chinmay Rath
target/ppc/translate.c +13 -0
+/* EA <- (ra == 0) ? 0 : GPR[ra] */
+static TCGv do_ea_calc_ra(DisasContext *ctx, int ra)
+{
+    TCGv EA;
+    if (!ra) {
+        EA = tcg_constant_tl(0);
+        return EA;
+    }
+    EA = tcg_temp_new();
+    if (NARROW_MODE(ctx)) {
+        tcg_gen_ext32u_tl(EA, cpu_gpr[ra]);
+    } else {
+        tcg_gen_mov_tl(EA, cpu_gpr[ra]);

Why are you making a copy, rather than just returning cpu_gpr[ra]? If you need to modify the resulting EA, then you also need to make a copy for 0.

Please ignore my previous response.
I think do_ea_calc_ra should allow modification to the resulting EA, hence below change appears more appropriate to me :

Quoted 15 lines

/* EA <- (ra == 0) ? 0 : GPR[ra] */
static TCGv do_ea_calc_ra(DisasContext *ctx, int ra) {
     TCGv EA = tcg_temp_new();
     if (!ra) {
         tcg_gen_movi_tl(EA, 0);
         return EA;
     }
     if (NARROW_MODE(ctx)) { tcg_gen_ext32u_tl(EA, cpu_gpr[ra]); } else {
         tcg_gen_mov_tl(EA, cpu_gpr[ra]);
     }
     return EA;
}

Quoted 2 lines

If that's what's needed by the callers of do_ea_calc_ra, then yes. You can drop the first return EA and use else if instead.

Sure.
I shall stick to keeping EA modifiable, (even though it is not modified by the callers in this patch), to allow its proper usage by (p){lx, stx}vp insns in future.

Thanks & Regards,
Chinmay

Quoted 3 lines

r~

#10 lore raw

PATCH v2 3/4 target/ppc: Move VSX vector storage access insns to decodetree.

In reply to 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-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
index 445fdb341f..3d31ef52f8 100644
@@ -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
index 7f4326c974..91cde088bc 100644
@@ -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
#11 lore raw

PATCH v2 4/4 target/ppc: Move VSX fp compare insns to decodetree.

In reply to Chinmay Rath

Moving the following instructions to decodetree specification:

	xvcmp{eq, gt, ge, ne}{s, d}p	: XX3-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/helper.h                 | 16 +++++-----
 target/ppc/insn32.decode            | 12 ++++++++
 target/ppc/fpu_helper.c             | 16 +++++-----
 target/ppc/translate/vsx-impl.c.inc | 46 +++++++++++++----------------
 target/ppc/translate/vsx-ops.c.inc  | 18 -----------
 5 files changed, 48 insertions(+), 60 deletions(-)
target/ppc/helper.h +8 -8
index 510ce76524..3fd849628a 100644
@@ -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
index 3d31ef52f8..bcaf03f24c 100644
@@ -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
index a013160644..5a300a3c86 100644
@@ -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 +20 -26

54 lines not on this page · raw message

target/ppc/translate/vsx-ops.c.inc +0 -18
index 91cde088bc..e553b5b8fa 100644
@@ -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