[llvm] [AMDGPU] Folding imm offset in more cases for scratch access (PR #70634)

[Jay Foad via llvm-commits]

================
@@ -1860,7 +1905,7 @@ bool AMDGPUDAGToDAGISel::SelectScratchSVAddr(SDNode *N, SDValue Addr,
           CurDAG->getTargetConstant(RemainderOffset, SDLoc(), MVT::i32));
         VAddr = SDValue(VMov, 0);
----------------
jayfoad wrote:

`isUInt<32>` allows all 32-bit values, including values that have bit 31 set so they are negative 32-bit values.

https://github.com/llvm/llvm-project/pull/70634