[AMDGPU] Use gcc pragmas to disable -Wtype-limits warning

[vabridgers]

Summary

Use gcc pragmas to disable warning in generated file. This is producing
a warning when using using gcc 9.3.0.

lib/Target/AMDGPU/R600GenSubtargetInfo.inc:282:62: warning: comparison
of unsigned expression < 0 is always false [-Wtype-limits]

if (Bits[R600::FeatureLocalMemorySize0] && LocalMemorySize <0)
   LocalMemorySize = 0;

[dblaikie]

Any chance we could fix AMDGPUGenSubtargetInfo.inc instead of wrapping it in pragmas?

[t-tye]

If LocalMemorySize is unsigned, how can it ever be less than 0? Shouldn't the code be corrected instead of adding pragmas?

[vabridgers]

Agreed it's better to fix in the source. The inc files are generated. Anyone know where to fix it?

[arsenm]

I don't see much reason to have FeatureLocalMemorySize0. It's implied by the absence of the features with a size. I think just removing it should work

[arsenm]

In D108865#2981434, @arsenm wrote:

I don't see much reason to have FeatureLocalMemorySize0. It's implied by the absence of the features with a size. I think just removing it should work

I tired this in 79bcd4a7db29e6b0abd60b44ee825f0d49d4f36a, so that should just avoid emitting this in the first place

llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp

 //===-- AMDGPUSubtarget.cpp - AMDGPU Subtarget Information ----------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //

 
 using namespace llvm;
 
 #define DEBUG_TYPE "amdgpu-subtarget"
 
 #define GET_SUBTARGETINFO_TARGET_DESC
 #define GET_SUBTARGETINFO_CTOR
 #define AMDGPUSubtarget GCNSubtarget
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wtype-limits"
 #include "AMDGPUGenSubtargetInfo.inc"
+#pragma GCC diagnostic pop
 #undef AMDGPUSubtarget
 
 static cl::opt<bool> DisablePowerSched(
   "amdgpu-disable-power-sched",
   cl::desc("Disable scheduling to minimize mAI power bursts"),
   cl::init(false));
 
 static cl::opt<bool> EnableVGPRIndexMode(