[PATCH] target/riscv: Call probe_write() before atomic operations

[Alistair Francis]

From: Alistair Francis <alistair.francis@wdc.com>

If an atomic operation fails on RISC-V we want to generate a store/amo
fault and not a load fault.

Currently if we have no permissions to access the memory location the
atomic operation will sometimes fail with a load fault (depending on the
path taken in tcg/tcg-op.c) as the atomic helpers perform a load then a
store.

By performing a probe_write() on the memory first, we can ensure we have
permissions to perform the atomic operation. As RISC-V doesn't have
write only pages this should be pretty robust (PMP might be the
exception).

Note that this only fixes the fault for memory regions. I/O and
non-existant regions will still trigger a load fault.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
---
 target/riscv/helper.h                   |  2 +
 target/riscv/op_helper.c                |  6 +++
 target/riscv/insn_trans/trans_rva.c.inc | 59 +++++++++++++++++--------
 3 files changed, 48 insertions(+), 19 deletions(-)

diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 72cc2582f4..afc8dcc1a0 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -106,6 +106,8 @@ DEF_HELPER_1(wfi, void, env)
 DEF_HELPER_1(tlb_flush, void, env)
 #endif
 
+DEF_HELPER_4(atomic_check, void, env, tl, int, int)
+
 /* Hypervisor functions */
 #ifndef CONFIG_USER_ONLY
 DEF_HELPER_1(hyp_tlb_flush, void, env)
diff --git a/target/riscv/op_helper.c b/target/riscv/op_helper.c
index 1a75ba11e6..d883609f4f 100644
--- a/target/riscv/op_helper.c
+++ b/target/riscv/op_helper.c
@@ -114,6 +114,12 @@ target_ulong helper_csrrw_i128(CPURISCVState *env, int csr,
     return int128_getlo(rv);
 }
 
+void helper_atomic_check(CPURISCVState *env, target_ulong address,
+                         int width, int mmu_idx)
+{
+    probe_write(env, address, width, mmu_idx, GETPC());
+}
+
 #ifndef CONFIG_USER_ONLY
 
 target_ulong helper_sret(CPURISCVState *env)
diff --git a/target/riscv/insn_trans/trans_rva.c.inc 
b/target/riscv/insn_trans/trans_rva.c.inc
index 45db82c9be..003a7d0b84 100644
--- a/target/riscv/insn_trans/trans_rva.c.inc
+++ b/target/riscv/insn_trans/trans_rva.c.inc
@@ -78,11 +78,14 @@ static bool gen_sc(DisasContext *ctx, arg_atomic *a, MemOp 
mop)
 
 static bool gen_amo(DisasContext *ctx, arg_atomic *a,
                     void(*func)(TCGv, TCGv, TCGv, TCGArg, MemOp),
-                    MemOp mop)
+                    TCGv_i32 width, MemOp mop)
 {
     TCGv dest = dest_gpr(ctx, a->rd);
     TCGv src1 = get_address(ctx, a->rs1, 0);
     TCGv src2 = get_gpr(ctx, a->rs2, EXT_NONE);
+    TCGv_i32 mem_idx = tcg_constant_i32(ctx->mem_idx);
+
+    gen_helper_atomic_check(cpu_env, src1, width, mem_idx);
 
     func(dest, src1, src2, ctx->mem_idx, mop);
 
@@ -105,55 +108,64 @@ static bool trans_sc_w(DisasContext *ctx, arg_sc_w *a)
 static bool trans_amoswap_w(DisasContext *ctx, arg_amoswap_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_xchg_tl, (MO_ALIGN | MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_xchg_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amoadd_w(DisasContext *ctx, arg_amoadd_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_add_tl, (MO_ALIGN | MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_add_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amoxor_w(DisasContext *ctx, arg_amoxor_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_xor_tl, (MO_ALIGN | MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_xor_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amoand_w(DisasContext *ctx, arg_amoand_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_and_tl, (MO_ALIGN | MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_and_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amoor_w(DisasContext *ctx, arg_amoor_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_or_tl, (MO_ALIGN | MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_or_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amomin_w(DisasContext *ctx, arg_amomin_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smin_tl, (MO_ALIGN | 
MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smin_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amomax_w(DisasContext *ctx, arg_amomax_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smax_tl, (MO_ALIGN | 
MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smax_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amominu_w(DisasContext *ctx, arg_amominu_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umin_tl, (MO_ALIGN | 
MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umin_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_amomaxu_w(DisasContext *ctx, arg_amomaxu_w *a)
 {
     REQUIRE_EXT(ctx, RVA);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umax_tl, (MO_ALIGN | 
MO_TESL));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umax_tl, tcg_constant_i32(4),
+                   (MO_ALIGN | MO_TESL));
 }
 
 static bool trans_lr_d(DisasContext *ctx, arg_lr_d *a)
@@ -171,53 +183,62 @@ static bool trans_sc_d(DisasContext *ctx, arg_sc_d *a)
 static bool trans_amoswap_d(DisasContext *ctx, arg_amoswap_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_xchg_tl, (MO_ALIGN | MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_xchg_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amoadd_d(DisasContext *ctx, arg_amoadd_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_add_tl, (MO_ALIGN | MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_add_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amoxor_d(DisasContext *ctx, arg_amoxor_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_xor_tl, (MO_ALIGN | MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_xor_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amoand_d(DisasContext *ctx, arg_amoand_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_and_tl, (MO_ALIGN | MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_and_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amoor_d(DisasContext *ctx, arg_amoor_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_or_tl, (MO_ALIGN | MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_or_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amomin_d(DisasContext *ctx, arg_amomin_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smin_tl, (MO_ALIGN | 
MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smin_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amomax_d(DisasContext *ctx, arg_amomax_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smax_tl, (MO_ALIGN | 
MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_smax_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amominu_d(DisasContext *ctx, arg_amominu_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umin_tl, (MO_ALIGN | 
MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umin_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
 
 static bool trans_amomaxu_d(DisasContext *ctx, arg_amomaxu_d *a)
 {
     REQUIRE_64BIT(ctx);
-    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umax_tl, (MO_ALIGN | 
MO_TEUQ));
+    return gen_amo(ctx, a, &tcg_gen_atomic_fetch_umax_tl, tcg_constant_i32(8),
+                   (MO_ALIGN | MO_TEUQ));
 }
-- 
2.34.1