)]}'
{
  "commit": "5560a4e8777f44f390899426365b0ef3118d6d0e",
  "tree": "fd6d0eb8e7b399a3c53f9275543b1d695d2228e8",
  "parents": [
    "c7818b2302df620c024c3e8b78e522742e0cc907"
  ],
  "author": {
    "name": "Lucas Ramirez",
    "email": "11032120+lucas-rami@users.noreply.github.com",
    "time": "Mon May 11 19:29:40 2026 +0200"
  },
  "committer": {
    "name": "GitHub",
    "email": "noreply@github.com",
    "time": "Mon May 11 19:29:40 2026 +0200"
  },
  "message": "[AMDGPU] Fix inconsistencies in RP tracking advance/reset behavior (#196595)\n\nSome of the variants of `advance` and `reset` in the `GCNRPTracker` and\n`GCNDownwardRPTracker` had unclear/inconsistent semantics on their\nreturn value. This aims to clarify that through improved documentation\nand light functional changes.\n\nThese inconsistencies ultimately triggered an assert in\n`GCNRPTaget::saveRP` on a complex kernel during scheduling.\n`GCNScheduleDAGMILive::getRealRegPressure` would incorrectly return a\nnull pressure for a non-empty region which only had debug values. Such\nregions can arise if the `PreRARematStage` rematerializes all non-debug\ninstructions out of their original region, leaving only debug values.\nAttempting to rematerialize registers across that same region afterwards\nwould trigger the assert.",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "f6fe48bf5bb49577424364a5317e191d784a3900",
      "old_mode": 33188,
      "old_path": "llvm/lib/Target/AMDGPU/AMDGPUNextUseAnalysis.cpp",
      "new_id": "cad1513f280fb9fe52cf70762e77c95a16946bb1",
      "new_mode": 33188,
      "new_path": "llvm/lib/Target/AMDGPU/AMDGPUNextUseAnalysis.cpp"
    },
    {
      "type": "modify",
      "old_id": "683e658aa4fb4fda7644fe478deaf869c2e1f6c7",
      "old_mode": 33188,
      "old_path": "llvm/lib/Target/AMDGPU/GCNRegPressure.cpp",
      "new_id": "42a430da0acb8853ff716c47966c4ed086ed08a9",
      "new_mode": 33188,
      "new_path": "llvm/lib/Target/AMDGPU/GCNRegPressure.cpp"
    },
    {
      "type": "modify",
      "old_id": "00cb617a55fa79aeb29fcc57ff0f968d8e86caba",
      "old_mode": 33188,
      "old_path": "llvm/lib/Target/AMDGPU/GCNRegPressure.h",
      "new_id": "85757eb718d46ad0376030c489f0ef89a8ddda19",
      "new_mode": 33188,
      "new_path": "llvm/lib/Target/AMDGPU/GCNRegPressure.h"
    },
    {
      "type": "modify",
      "old_id": "727ae68e88bfbb72e092e6c5df60476d15f760db",
      "old_mode": 33188,
      "old_path": "llvm/lib/Target/AMDGPU/GCNSchedStrategy.cpp",
      "new_id": "8a6c1d5e21e85f01847e7a4a126a035ea0d5d62b",
      "new_mode": 33188,
      "new_path": "llvm/lib/Target/AMDGPU/GCNSchedStrategy.cpp"
    },
    {
      "type": "modify",
      "old_id": "982034189892cb0582f09774eb874bc7298064b5",
      "old_mode": 33188,
      "old_path": "llvm/lib/Target/AMDGPU/SIFormMemoryClauses.cpp",
      "new_id": "905c2afa9fcd89acef9257a3da43dd46cbdcac56",
      "new_mode": 33188,
      "new_path": "llvm/lib/Target/AMDGPU/SIFormMemoryClauses.cpp"
    },
    {
      "type": "modify",
      "old_id": "ad84f4df65288c9ed3e118848e6dc250b32a8442",
      "old_mode": 33188,
      "old_path": "llvm/unittests/Target/AMDGPU/GCNRegPressureTest.cpp",
      "new_id": "d907ee269a4487b40ba16fb71813ca5390800c69",
      "new_mode": 33188,
      "new_path": "llvm/unittests/Target/AMDGPU/GCNRegPressureTest.cpp"
    }
  ]
}