MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion correctly on arm

Port de369129

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32

Review-Url: https://codereview.chromium.org/2130763002
Cr-Commit-Position: refs/heads/master@{#37586}

src/compiler/mips/code-generator-mips.cc

@@ -1328,6 +1328,11 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       FPURegister scratch = kScratchDoubleReg;
       __ trunc_w_s(scratch, i.InputDoubleRegister(0));
       __ mfc1(i.OutputRegister(), scratch);
+      // Avoid INT32_MAX as an overflow indicator and use INT32_MIN instead,
+      // because INT32_MIN allows easier out-of-bounds detection.
+      __ addiu(kScratchReg, i.OutputRegister(), 1);
+      __ slt(kScratchReg2, kScratchReg, i.OutputRegister());
+      __ Movn(i.OutputRegister(), kScratchReg, kScratchReg2);
       break;
     }
     case kMipsTruncUwD: {
@@ -1340,6 +1345,10 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       FPURegister scratch = kScratchDoubleReg;
       // TODO(plind): Fix wrong param order of Trunc_uw_s() macro-asm function.
       __ Trunc_uw_s(i.InputDoubleRegister(0), i.OutputRegister(), scratch);
+      // Avoid UINT32_MAX as an overflow indicator and use 0 instead,
+      // because 0 allows easier out-of-bounds detection.
+      __ addiu(kScratchReg, i.OutputRegister(), 1);
+      __ Movz(i.OutputRegister(), zero_reg, kScratchReg);
       break;
     }
     case kMipsFloat64ExtractLowWord32:

src/compiler/mips64/code-generator-mips64.cc

@@ -1494,6 +1494,11 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       FPURegister scratch = kScratchDoubleReg;
       __ trunc_w_s(scratch, i.InputDoubleRegister(0));
       __ mfc1(i.OutputRegister(), scratch);
+      // Avoid INT32_MAX as an overflow indicator and use INT32_MIN instead,
+      // because INT32_MIN allows easier out-of-bounds detection.
+      __ addiu(kScratchReg, i.OutputRegister(), 1);
+      __ slt(kScratchReg2, kScratchReg, i.OutputRegister());
+      __ Movn(i.OutputRegister(), kScratchReg, kScratchReg2);
       break;
     }
     case kMips64TruncLS: {
@@ -1562,6 +1567,10 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       FPURegister scratch = kScratchDoubleReg;
       // TODO(plind): Fix wrong param order of Trunc_uw_d() macro-asm function.
       __ Trunc_uw_s(i.InputDoubleRegister(0), i.OutputRegister(), scratch);
+      // Avoid UINT32_MAX as an overflow indicator and use 0 instead,
+      // because 0 allows easier out-of-bounds detection.
+      __ addiu(kScratchReg, i.OutputRegister(), 1);
+      __ Movz(i.OutputRegister(), zero_reg, kScratchReg);
       break;
     }
     case kMips64TruncUlS: {