)]}'
{
  "commit": "1c919724d3f887b184f5194a8709c8ae652e4b7f",
  "tree": "70f215849733bcfc8028cfc6306a824939e2400f",
  "parents": [
    "5cf02188feba518aecfda7a05c72b1586c5c4080"
  ],
  "author": {
    "name": "David Benjamin",
    "email": "davidben@google.com",
    "time": "Wed Jan 27 16:35:25 2021 -0500"
  },
  "committer": {
    "name": "Adam Langley",
    "email": "agl@google.com",
    "time": "Wed Jan 27 22:51:57 2021 +0000"
  },
  "message": "Support MOVLPS and MOVHPS in delocate.\n\nGCC 10.2.1 seems to be emitting code like this:\n\n    movq    gcm_gmult_clmul@GOTPCREL(%rip), %xmm0\n    movhps  gcm_ghash_clmul@GOTPCREL(%rip), %xmm0\n    movaps  %xmm0, (%rsp)\n\nThis is assembling a pair of function pointers in %xmm0 and writing the\ntwo out together. I\u0027ve not observed the compiler output movlps, but\nsupporting movhps and movlps are about as tricky. The main complication\nis that these instructions preserve the unwritten half of the\ndestination, and they do not support register sources, only memory.\n\nThis CL supports them by loading in a general-purpose register as we\nusually do, pushing the register on the stack, and then running the\ninstruction on (%rsp). Some alternatives I considered:\n\n- Save/restore a temporary XMM register and then use MOVHLPS and\n  MOVLHPS. This would work but require another saveRegister-like\n  wrapper.\n\n- Take advantage of loadFromGOT ending in a memory mov and swap out\n  the final instruction. This would be more efficient, but we downgrade\n  GOT-based accesses to local symbols to a plain LEA. The compiler will\n  only do this when we write a pair of function pointers in a row, so\n  trying to optimize the non-local symbols seems not worth the trouble.\n  (Really the compiler should not be emitting GOT-relative loads at all,\n  but the compiler doesn\u0027t know these symbols will be private and in the\n  same module, so it has a habit of pessimally using GOT-based loads.)\n\nThis option seemed the simplest.\n\nChange-Id: I8c4915a6a0d72aa4c5f4d581081b99b3a6ab64c2\nReviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45244\nReviewed-by: Adam Langley \u003cagl@google.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "dc34c6883b6557dd66ee52202f10509fd0472672",
      "old_mode": 33188,
      "old_path": "util/fipstools/delocate/delocate.go",
      "new_id": "2d92520dc7d87f49ee8a25f6c8d9a40d52850bfb",
      "new_mode": 33188,
      "new_path": "util/fipstools/delocate/delocate.go"
    },
    {
      "type": "modify",
      "old_id": "398032f846c45274d0e14468df77a79790453cbc",
      "old_mode": 33188,
      "old_path": "util/fipstools/delocate/testdata/x86_64-GOTRewrite/in.s",
      "new_id": "a00a69147c2c1d4eeede50fe4300c41d7e309190",
      "new_mode": 33188,
      "new_path": "util/fipstools/delocate/testdata/x86_64-GOTRewrite/in.s"
    },
    {
      "type": "modify",
      "old_id": "e14a71e60fab3765f8d5d545b1d81e9c176981fe",
      "old_mode": 33188,
      "old_path": "util/fipstools/delocate/testdata/x86_64-GOTRewrite/out.s",
      "new_id": "467db65b4f5c7b6bdbdeb78e6a87499debfec976",
      "new_mode": 33188,
      "new_path": "util/fipstools/delocate/testdata/x86_64-GOTRewrite/out.s"
    }
  ]
}