deps: V8: cherry-pick 5fe919f78321

Original commit message:

    PPC: clear VXCVI before doing a conversion

    This bit may not get cleared automatically and could show
    results from older executed instructions.

    Change-Id: I5976f9a6c5bf87b1a63ef0f35493b222729e20f6
    Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3812037
    Reviewed-by: Junliang Yan <junyan@redhat.com>
    Commit-Queue: Milad Farazmand <mfarazma@redhat.com>
    Cr-Commit-Position: refs/heads/main@{#82237}

Refs: v8/v8@5fe919f

common.gypi

@@ -36,7 +36,7 @@
 
     # Reset this number to 0 on major V8 upgrades.
     # Increment by one for each non-official patch applied to deps/v8.
-    'v8_embedder_string': '-node.21',
+    'v8_embedder_string': '-node.22',
 
     ##### V8 defaults for Node.js #####
 

deps/v8/src/wasm/baseline/ppc/liftoff-assembler-ppc.h

@@ -1455,6 +1455,7 @@ bool LiftoffAssembler::emit_type_conversion(WasmOpcode opcode,
       fcmpu(src.fp(), kScratchDoubleReg);
       bunordered(trap);
 
+      mtfsb0(VXCVI);  // clear FPSCR:VXCVI bit
       fctiwz(kScratchDoubleReg, src.fp());
       MovDoubleLowToInt(dst.gp(), kScratchDoubleReg);
       mcrfs(cr7, VXCVI);
@@ -1463,6 +1464,7 @@ bool LiftoffAssembler::emit_type_conversion(WasmOpcode opcode,
     }
     case kExprI32UConvertF64:
     case kExprI32UConvertF32: {
+      mtfsb0(VXCVI);  // clear FPSCR:VXCVI bit
       ConvertDoubleToUnsignedInt64(src.fp(), r0, kScratchDoubleReg,
                                    kRoundToZero);
       mcrfs(cr7, VXCVI);  // extract FPSCR field containing VXCVI into cr7
@@ -1478,6 +1480,7 @@ bool LiftoffAssembler::emit_type_conversion(WasmOpcode opcode,
       fcmpu(src.fp(), kScratchDoubleReg);
       bunordered(trap);
 
+      mtfsb0(VXCVI);  // clear FPSCR:VXCVI bit
       fctidz(kScratchDoubleReg, src.fp());
       MovDoubleToInt64(dst.gp(), kScratchDoubleReg);
       mcrfs(cr7, VXCVI);
@@ -1490,6 +1493,7 @@ bool LiftoffAssembler::emit_type_conversion(WasmOpcode opcode,
       fcmpu(src.fp(), kScratchDoubleReg);
       bunordered(trap);
 
+      mtfsb0(VXCVI);  // clear FPSCR:VXCVI bit
       fctiduz(kScratchDoubleReg, src.fp());
       MovDoubleToInt64(dst.gp(), kScratchDoubleReg);
       mcrfs(cr7, VXCVI);