diff mbox series

[RFC] target/loongarch: Fix emulation of float-point disable exception

Message ID 20221103080232.55953-1-wangrui@loongson.cn
State New
Headers show
Series [RFC] target/loongarch: Fix emulation of float-point disable exception | expand

Commit Message

Rui Wang Nov. 3, 2022, 8:02 a.m. UTC
We need to emulate it to generate a floating point disable exception
when CSR.EUEN.FPE is zero.

Signed-off-by: Rui Wang <wangrui@loongson.cn>
---
 target/loongarch/cpu.c                        |  2 ++
 .../loongarch/insn_trans/trans_farith.c.inc   | 36 +++++++++++++++++++
 target/loongarch/insn_trans/trans_fcmp.c.inc  |  5 +++
 .../loongarch/insn_trans/trans_fmemory.c.inc  | 16 +++++++++
 target/loongarch/insn_trans/trans_fmov.c.inc  | 20 +++++++++++
 target/loongarch/translate.c                  |  4 +++
 target/loongarch/translate.h                  |  1 +
 7 files changed, 84 insertions(+)

Comments

Richard Henderson Nov. 3, 2022, 8:47 p.m. UTC | #1
On 11/3/22 19:02, Rui Wang wrote:
> +static void check_fpe(DisasContext *ctx)
> +{
> +#ifndef CONFIG_USER_ONLY
> +    TCGLabel *skip = gen_new_label();
> +    TCGv tmp = tcg_temp_new();
> +
> +    tcg_gen_andi_tl(tmp, cpu_euen, R_CSR_EUEN_FPE_MASK);
> +    tcg_gen_brcond_tl(TCG_COND_NE, tmp, ctx->zero, skip);
> +    tcg_temp_free(tmp);
> +
> +    generate_exception(ctx, EXCCODE_FPD);
> +    ctx->base.is_jmp = DISAS_EXIT_UPDATE;
> +
> +    gen_set_label(skip);
> +#endif
> +}

While this works, it is very inefficient.

You should cache the state of this bit in tb->flags, as set by cpu_get_tb_cpu_state.
Compare, for instance, target/riscv/,

FIELD(TB_FLAGS, MSTATUS_HS_FS, 16, 2)

for which ctx->mstatus_hs_fs controls whether the fpu is clean, dirty, or disabled.

At present, loongarch is using 3 bits of tb->flags for the mmu_idx, re-using that for 
privilege level in check_plv.  Which is nearly correct, except for the case of paging 
disabled in user mode.  This probably never happens in practice, but unless the cpu 
disallows such a state, this should be corrected.


r~
diff mbox series

Patch

diff --git a/target/loongarch/cpu.c b/target/loongarch/cpu.c
index 49393d95d8..abe4a819b2 100644
--- a/target/loongarch/cpu.c
+++ b/target/loongarch/cpu.c
@@ -48,6 +48,7 @@  static const char * const excp_names[] = {
     [EXCCODE_BRK] = "Break",
     [EXCCODE_INE] = "Instruction Non-Existent",
     [EXCCODE_IPE] = "Instruction privilege error",
+    [EXCCODE_FPD] = "Floating Point Disabled",
     [EXCCODE_FPE] = "Floating Point Exception",
     [EXCCODE_DBP] = "Debug breakpoint",
     [EXCCODE_BCE] = "Bound Check Exception",
@@ -184,6 +185,7 @@  static void loongarch_cpu_do_interrupt(CPUState *cs)
     case EXCCODE_BRK:
     case EXCCODE_INE:
     case EXCCODE_IPE:
+    case EXCCODE_FPD:
     case EXCCODE_FPE:
     case EXCCODE_BCE:
         env->CSR_BADV = env->pc;
diff --git a/target/loongarch/insn_trans/trans_farith.c.inc b/target/loongarch/insn_trans/trans_farith.c.inc
index 7bb3f41aee..3574b9b746 100644
--- a/target/loongarch/insn_trans/trans_farith.c.inc
+++ b/target/loongarch/insn_trans/trans_farith.c.inc
@@ -3,9 +3,28 @@ 
  * Copyright (c) 2021 Loongson Technology Corporation Limited
  */
 
+static void check_fpe(DisasContext *ctx)
+{
+#ifndef CONFIG_USER_ONLY
+    TCGLabel *skip = gen_new_label();
+    TCGv tmp = tcg_temp_new();
+
+    tcg_gen_andi_tl(tmp, cpu_euen, R_CSR_EUEN_FPE_MASK);
+    tcg_gen_brcond_tl(TCG_COND_NE, tmp, ctx->zero, skip);
+    tcg_temp_free(tmp);
+
+    generate_exception(ctx, EXCCODE_FPD);
+    ctx->base.is_jmp = DISAS_EXIT_UPDATE;
+
+    gen_set_label(skip);
+#endif
+}
+
 static bool gen_fff(DisasContext *ctx, arg_fff *a,
                     void (*func)(TCGv, TCGv_env, TCGv, TCGv))
 {
+    check_fpe(ctx);
+
     func(cpu_fpr[a->fd], cpu_env, cpu_fpr[a->fj], cpu_fpr[a->fk]);
     return true;
 }
@@ -13,6 +32,8 @@  static bool gen_fff(DisasContext *ctx, arg_fff *a,
 static bool gen_ff(DisasContext *ctx, arg_ff *a,
                    void (*func)(TCGv, TCGv_env, TCGv))
 {
+    check_fpe(ctx);
+
     func(cpu_fpr[a->fd], cpu_env, cpu_fpr[a->fj]);
     return true;
 }
@@ -22,6 +43,9 @@  static bool gen_muladd(DisasContext *ctx, arg_ffff *a,
                        int flag)
 {
     TCGv_i32 tflag = tcg_constant_i32(flag);
+
+    check_fpe(ctx);
+
     func(cpu_fpr[a->fd], cpu_env, cpu_fpr[a->fj],
          cpu_fpr[a->fk], cpu_fpr[a->fa], tflag);
     return true;
@@ -29,18 +53,24 @@  static bool gen_muladd(DisasContext *ctx, arg_ffff *a,
 
 static bool trans_fcopysign_s(DisasContext *ctx, arg_fcopysign_s *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_deposit_i64(cpu_fpr[a->fd], cpu_fpr[a->fk], cpu_fpr[a->fj], 0, 31);
     return true;
 }
 
 static bool trans_fcopysign_d(DisasContext *ctx, arg_fcopysign_d *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_deposit_i64(cpu_fpr[a->fd], cpu_fpr[a->fk], cpu_fpr[a->fj], 0, 63);
     return true;
 }
 
 static bool trans_fabs_s(DisasContext *ctx, arg_fabs_s *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_andi_i64(cpu_fpr[a->fd], cpu_fpr[a->fj], MAKE_64BIT_MASK(0, 31));
     gen_nanbox_s(cpu_fpr[a->fd], cpu_fpr[a->fd]);
     return true;
@@ -48,12 +78,16 @@  static bool trans_fabs_s(DisasContext *ctx, arg_fabs_s *a)
 
 static bool trans_fabs_d(DisasContext *ctx, arg_fabs_d *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_andi_i64(cpu_fpr[a->fd], cpu_fpr[a->fj], MAKE_64BIT_MASK(0, 63));
     return true;
 }
 
 static bool trans_fneg_s(DisasContext *ctx, arg_fneg_s *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_xori_i64(cpu_fpr[a->fd], cpu_fpr[a->fj], 0x80000000);
     gen_nanbox_s(cpu_fpr[a->fd], cpu_fpr[a->fd]);
     return true;
@@ -61,6 +95,8 @@  static bool trans_fneg_s(DisasContext *ctx, arg_fneg_s *a)
 
 static bool trans_fneg_d(DisasContext *ctx, arg_fneg_d *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_xori_i64(cpu_fpr[a->fd], cpu_fpr[a->fj], 0x8000000000000000LL);
     return true;
 }
diff --git a/target/loongarch/insn_trans/trans_fcmp.c.inc b/target/loongarch/insn_trans/trans_fcmp.c.inc
index 93a6a2230f..32b903231b 100644
--- a/target/loongarch/insn_trans/trans_fcmp.c.inc
+++ b/target/loongarch/insn_trans/trans_fcmp.c.inc
@@ -29,6 +29,8 @@  static bool trans_fcmp_cond_s(DisasContext *ctx, arg_fcmp_cond_s *a)
     uint32_t flags;
     void (*fn)(TCGv, TCGv_env, TCGv, TCGv, TCGv_i32);
 
+    check_fpe(ctx);
+
     fn = (a->fcond & 1 ? gen_helper_fcmp_s_s : gen_helper_fcmp_c_s);
     flags = get_fcmp_flags(a->fcond >> 1);
 
@@ -44,6 +46,9 @@  static bool trans_fcmp_cond_d(DisasContext *ctx, arg_fcmp_cond_d *a)
     TCGv var = tcg_temp_new();
     uint32_t flags;
     void (*fn)(TCGv, TCGv_env, TCGv, TCGv, TCGv_i32);
+
+    check_fpe(ctx);
+
     fn = (a->fcond & 1 ? gen_helper_fcmp_s_d : gen_helper_fcmp_c_d);
     flags = get_fcmp_flags(a->fcond >> 1);
 
diff --git a/target/loongarch/insn_trans/trans_fmemory.c.inc b/target/loongarch/insn_trans/trans_fmemory.c.inc
index 74ee98f63a..508af75adf 100644
--- a/target/loongarch/insn_trans/trans_fmemory.c.inc
+++ b/target/loongarch/insn_trans/trans_fmemory.c.inc
@@ -15,6 +15,8 @@  static bool gen_fload_i(DisasContext *ctx, arg_fr_i *a, MemOp mop)
     TCGv addr = gpr_src(ctx, a->rj, EXT_NONE);
     TCGv temp = NULL;
 
+    check_fpe(ctx);
+
     if (a->imm) {
         temp = tcg_temp_new();
         tcg_gen_addi_tl(temp, addr, a->imm);
@@ -36,6 +38,8 @@  static bool gen_fstore_i(DisasContext *ctx, arg_fr_i *a, MemOp mop)
     TCGv addr = gpr_src(ctx, a->rj, EXT_NONE);
     TCGv temp = NULL;
 
+    check_fpe(ctx);
+
     if (a->imm) {
         temp = tcg_temp_new();
         tcg_gen_addi_tl(temp, addr, a->imm);
@@ -56,6 +60,8 @@  static bool gen_floadx(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_ld_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
     maybe_nanbox_load(cpu_fpr[a->fd], mop);
@@ -70,6 +76,8 @@  static bool gen_fstorex(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_st_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
     tcg_temp_free(addr);
@@ -83,6 +91,8 @@  static bool gen_fload_gt(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     gen_helper_asrtgt_d(cpu_env, src1, src2);
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_ld_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
@@ -98,6 +108,8 @@  static bool gen_fstore_gt(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     gen_helper_asrtgt_d(cpu_env, src1, src2);
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_st_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
@@ -112,6 +124,8 @@  static bool gen_fload_le(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     gen_helper_asrtle_d(cpu_env, src1, src2);
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_ld_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
@@ -127,6 +141,8 @@  static bool gen_fstore_le(DisasContext *ctx, arg_frr *a, MemOp mop)
     TCGv src2 = gpr_src(ctx, a->rk, EXT_NONE);
     TCGv addr = tcg_temp_new();
 
+    check_fpe(ctx);
+
     gen_helper_asrtle_d(cpu_env, src1, src2);
     tcg_gen_add_tl(addr, src1, src2);
     tcg_gen_qemu_st_tl(cpu_fpr[a->fd], addr, ctx->mem_idx, mop);
diff --git a/target/loongarch/insn_trans/trans_fmov.c.inc b/target/loongarch/insn_trans/trans_fmov.c.inc
index 5537e3dd35..a1d87719f8 100644
--- a/target/loongarch/insn_trans/trans_fmov.c.inc
+++ b/target/loongarch/insn_trans/trans_fmov.c.inc
@@ -12,6 +12,8 @@  static bool trans_fsel(DisasContext *ctx, arg_fsel *a)
     TCGv zero = tcg_constant_tl(0);
     TCGv cond = tcg_temp_new();
 
+    check_fpe(ctx);
+
     tcg_gen_ld8u_tl(cond, cpu_env, offsetof(CPULoongArchState, cf[a->ca]));
     tcg_gen_movcond_tl(TCG_COND_EQ, cpu_fpr[a->fd], cond, zero,
                        cpu_fpr[a->fj], cpu_fpr[a->fk]);
@@ -26,6 +28,8 @@  static bool gen_f2f(DisasContext *ctx, arg_ff *a,
     TCGv dest = cpu_fpr[a->fd];
     TCGv src = cpu_fpr[a->fj];
 
+    check_fpe(ctx);
+
     func(dest, src);
     if (nanbox) {
         gen_nanbox_s(cpu_fpr[a->fd], cpu_fpr[a->fd]);
@@ -39,6 +43,8 @@  static bool gen_r2f(DisasContext *ctx, arg_fr *a,
 {
     TCGv src = gpr_src(ctx, a->rj, EXT_NONE);
 
+    check_fpe(ctx);
+
     func(cpu_fpr[a->fd], src);
     return true;
 }
@@ -48,6 +54,8 @@  static bool gen_f2r(DisasContext *ctx, arg_rf *a,
 {
     TCGv dest = gpr_dst(ctx, a->rd, EXT_NONE);
 
+    check_fpe(ctx);
+
     func(dest, cpu_fpr[a->fj]);
     gen_set_gpr(a->rd, dest, EXT_NONE);
 
@@ -59,6 +67,8 @@  static bool trans_movgr2fcsr(DisasContext *ctx, arg_movgr2fcsr *a)
     uint32_t mask = fcsr_mask[a->fcsrd];
     TCGv Rj = gpr_src(ctx, a->rj, EXT_NONE);
 
+    check_fpe(ctx);
+
     if (mask == UINT32_MAX) {
         tcg_gen_st32_i64(Rj, cpu_env, offsetof(CPULoongArchState, fcsr0));
     } else {
@@ -90,6 +100,8 @@  static bool trans_movfcsr2gr(DisasContext *ctx, arg_movfcsr2gr *a)
 {
     TCGv dest = gpr_dst(ctx, a->rd, EXT_NONE);
 
+    check_fpe(ctx);
+
     tcg_gen_ld32u_i64(dest, cpu_env, offsetof(CPULoongArchState, fcsr0));
     tcg_gen_andi_i64(dest, dest, fcsr_mask[a->fcsrs]);
     gen_set_gpr(a->rd, dest, EXT_NONE);
@@ -116,6 +128,8 @@  static bool trans_movfr2cf(DisasContext *ctx, arg_movfr2cf *a)
 {
     TCGv t0 = tcg_temp_new();
 
+    check_fpe(ctx);
+
     tcg_gen_andi_tl(t0, cpu_fpr[a->fj], 0x1);
     tcg_gen_st8_tl(t0, cpu_env, offsetof(CPULoongArchState, cf[a->cd & 0x7]));
     tcg_temp_free(t0);
@@ -125,6 +139,8 @@  static bool trans_movfr2cf(DisasContext *ctx, arg_movfr2cf *a)
 
 static bool trans_movcf2fr(DisasContext *ctx, arg_movcf2fr *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_ld8u_tl(cpu_fpr[a->fd], cpu_env,
                     offsetof(CPULoongArchState, cf[a->cj & 0x7]));
     return true;
@@ -134,6 +150,8 @@  static bool trans_movgr2cf(DisasContext *ctx, arg_movgr2cf *a)
 {
     TCGv t0 = tcg_temp_new();
 
+    check_fpe(ctx);
+
     tcg_gen_andi_tl(t0, gpr_src(ctx, a->rj, EXT_NONE), 0x1);
     tcg_gen_st8_tl(t0, cpu_env, offsetof(CPULoongArchState, cf[a->cd & 0x7]));
     tcg_temp_free(t0);
@@ -143,6 +161,8 @@  static bool trans_movgr2cf(DisasContext *ctx, arg_movgr2cf *a)
 
 static bool trans_movcf2gr(DisasContext *ctx, arg_movcf2gr *a)
 {
+    check_fpe(ctx);
+
     tcg_gen_ld8u_tl(gpr_dst(ctx, a->rd, EXT_NONE), cpu_env,
                     offsetof(CPULoongArchState, cf[a->cj & 0x7]));
     return true;
diff --git a/target/loongarch/translate.c b/target/loongarch/translate.c
index 6091772349..5204653701 100644
--- a/target/loongarch/translate.c
+++ b/target/loongarch/translate.c
@@ -18,11 +18,13 @@ 
 #include "fpu/softfloat.h"
 #include "translate.h"
 #include "internals.h"
+#include "cpu-csr.h"
 
 /* Global register indices */
 TCGv cpu_gpr[32], cpu_pc;
 static TCGv cpu_lladdr, cpu_llval;
 TCGv_i64 cpu_fpr[32];
+TCGv_i64 cpu_euen;
 
 #include "exec/gen-icount.h"
 
@@ -267,6 +269,8 @@  void loongarch_translate_init(void)
     }
 
     cpu_pc = tcg_global_mem_new(cpu_env, offsetof(CPULoongArchState, pc), "pc");
+    cpu_euen = tcg_global_mem_new(cpu_env,
+                    offsetof(CPULoongArchState, CSR_EUEN), "cpu_euen");
     cpu_lladdr = tcg_global_mem_new(cpu_env,
                     offsetof(CPULoongArchState, lladdr), "lladdr");
     cpu_llval = tcg_global_mem_new(cpu_env,
diff --git a/target/loongarch/translate.h b/target/loongarch/translate.h
index 9cc12512d1..3bf68867f5 100644
--- a/target/loongarch/translate.h
+++ b/target/loongarch/translate.h
@@ -41,5 +41,6 @@  void generate_exception(DisasContext *ctx, int excp);
 extern TCGv cpu_gpr[32], cpu_pc;
 extern TCGv_i32 cpu_fscr0;
 extern TCGv_i64 cpu_fpr[32];
+extern TCGv_i64 cpu_euen;
 
 #endif