[AMDGPU] Fix error handling in asm constraint syntax

[rampitec]

Summary

I believe this is unexploitable because in either case the result
will be 'couldn't allocate register for constraint' error message,
but error code checking is clearly wrong.

[arsenm]

Testcase?

[rampitec]

In D117189#3241074, @arsenm wrote:

Testcase?

It gives exactly the same final error, finally dropping to the base implementation. The effect is unobservable.

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

if (RegName.consume_front("v")) {
       RC = &AMDGPU::AGPR_32RegClass;
     }
 
     if (RC) {
       uint32_t Idx;
       if (RegName.consume_front("[")) {
         uint32_t End;
         bool Failed = RegName.consumeInteger(10, Idx);
-        Failed &= !RegName.consume_front(":");
-        Failed &= RegName.consumeInteger(10, End);
-        Failed &= !RegName.consume_back("]");
+        Failed |= !RegName.consume_front(":");
+        Failed |= RegName.consumeInteger(10, End);
+        Failed |= !RegName.consume_back("]");
         if (!Failed) {
           uint32_t Width = (End - Idx + 1) * 32;
           MCRegister Reg = RC->getRegister(Idx);
           if (SIRegisterInfo::isVGPRClass(RC))
             RC = TRI->getVGPRClassForBitWidth(Width);
           else if (SIRegisterInfo::isSGPRClass(RC))
             RC = TRI->getSGPRClassForBitWidth(Width);
           else if (SIRegisterInfo::isAGPRClass(RC))