)]}'
{
  "commit": "8d4346ecd4950ae08cc76a6de327c264e846758c",
  "tree": "486a9abd91b0200bd3cfe596d44400b5e0f8ee02",
  "parents": [
    "5d6919055dec134de3c40167a490f33c74c12581"
  ],
  "author": {
    "name": "Simon Xue",
    "email": "xxm@rock-chips.com",
    "time": "Tue Apr 28 18:05:31 2026 +0200"
  },
  "committer": {
    "name": "Joerg Roedel",
    "email": "joerg.roedel@amd.com",
    "time": "Tue Jun 02 10:42:54 2026 +0200"
  },
  "message": "iommu/rockchip: disable fetch dte time limit\n\nDisable the Bit 31 of the AUTO_GATING iommu register, as it causes\nhangups with the RGA3 (Raster Graphics Acceleration 3) peripheral.\nThe RGA3 register description of the TRM already states that the bit\nmust be set to 1. The vendor kernel sets the bit unconditionally to\n1 to fix VOP (Video Output Processor) screen black issues. This patch\nsquashes the 2 vendor kernel commits with the following commit messages:\n\nMaster fetch data and cpu update page table may work in parallel, may\nhave the following procedure:\n\n\tmaster                  cpu\n\tfetch dte               update page tabl\n\t        |                       |\n\t(make dte invalid)  \u003c-  zap iotlb entry\n\t        |                       |\n\tfetch dte again\n\t(make dte invalid)  \u003c-  zap iotlb entry\n\t        |                       |\n\tfetch dte again\n\t(make dte invalid)  \u003c-  zap iotlb entry\n\t        |                       |\n\tfetch dte again\n\t(make iommu block)  \u003c-  zap iotlb entry\n\nNew iommu version has the above bug, if fetch dte consecutively four\ntimes, then it will be blocked. Fortunately, we can set bit 31 of\nregister MMU_AUTO_GATING to 1 to make it work as old version which does\nnot have this issue.\n\nThis issue only appears on RV1126 so far, so make a workaround dedicated\nto \"rockchip,rv1126\" machine type.\n\niommu/rockchip: fix vop blocked and screen black on RK356X and RK3588\n\nRK3568 and RK3588 has the same issue as RV1126/RV1109 that caused by\ndte fetch time limit, So we can set BIT(31) of register 0x24 default\nto 1 as a workaround.\n\nSigned-off-by: Simon Xue \u003cxxm@rock-chips.com\u003e\nSigned-off-by: Sven Püschel \u003cs.pueschel@pengutronix.de\u003e\nAcked-by: Heiko Stuebner \u003cheiko@sntech.de\u003e\nSigned-off-by: Joerg Roedel \u003cjoerg.roedel@amd.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "0013cf196c5735effd467bb0050922f042ff6eee",
      "old_mode": 33188,
      "old_path": "drivers/iommu/rockchip-iommu.c",
      "new_id": "87ae036d641455d4a719c851813634a470bf9a56",
      "new_mode": 33188,
      "new_path": "drivers/iommu/rockchip-iommu.c"
    }
  ]
}