)]}'
{
  "log": [
    {
      "commit": "9f9c0382cda2334b35b40b00f4ed9d6f89f37a7b",
      "tree": "8d862fd0883343e8788f86b8f7d37823db0e19a5",
      "parents": [
        "fb702b942bf638baa6cbbbda9f76794db62921ef"
      ],
      "author": {
        "name": "Chris Metcalf",
        "email": "cmetcalf@tilera.com",
        "time": "Fri Jun 25 17:00:56 2010 -0400"
      },
      "committer": {
        "name": "Chris Metcalf",
        "email": "cmetcalf@tilera.com",
        "time": "Tue Jul 06 13:34:15 2010 -0400"
      },
      "message": "arch/tile: Add driver to enable access to the user dynamic network.\n\nThis network (the \"UDN\") connects all the cpus on the chip in a\nwormhole-routed dynamic network.  Subrectangles of the chip can\nbe allocated by a \"create\" ioctl on /dev/hardwall, and then to access the\nUDN in that rectangle, tasks must perform an \"activate\" ioctl on that\nsame file object after affinitizing themselves to a single cpu in\nthe region.  Sending a wormhole-routed message that tries to leave\nthat subrectangle causes all activated tasks to receive a SIGILL\n(just as they would if they tried to access the UDN without first\nactivating themselves to a hardwall rectangle).\n\nThe original submission of this code to LKML had the driver\ninstantiated under /proc/tile/hardwall.  Now we just use a character\ndevice for this, conventionally /dev/hardwall.  Some futures planning\nfor the TILE-Gx chip suggests that we may want to have other types of\ndevices that share the general model of \"bind a task to a cpu, then\n\u0027activate\u0027 a file descriptor on a pseudo-device that gives access to\nsome hardware resource\".  As such, we are using a device rather\nthan, for example, a syscall, to set up and activate this code.\n\nAs part of this change, the compat_ptr() declaration was fixed and used\nto pass the compat_ioctl argument to the normal ioctl.  So far we limit\ncompat code to 2GB, so the difference between zero-extend and sign-extend\n(the latter being correct, eventually) had been overlooked.\n\nSigned-off-by: Chris Metcalf \u003ccmetcalf@tilera.com\u003e\nAcked-by: Arnd Bergmann \u003carnd@arndb.de\u003e\n"
    },
    {
      "commit": "867e359b97c970a60626d5d76bbe2a8fadbf38fb",
      "tree": "c5ccbb7f5172e8555977119608ecb1eee3cc37e3",
      "parents": [
        "5360bd776f73d0a7da571d72a09a03f237e99900"
      ],
      "author": {
        "name": "Chris Metcalf",
        "email": "cmetcalf@tilera.com",
        "time": "Fri May 28 23:09:12 2010 -0400"
      },
      "committer": {
        "name": "Chris Metcalf",
        "email": "cmetcalf@tilera.com",
        "time": "Fri Jun 04 17:11:18 2010 -0400"
      },
      "message": "arch/tile: core support for Tilera 32-bit chips.\n\nThis change is the core kernel support for TILEPro and TILE64 chips.\nNo driver support (except the console driver) is included yet.\n\nThis includes the relevant Linux headers in asm/; the low-level\nlow-level \"Tile architecture\" headers in arch/, which are\nshared with the hypervisor, etc., and are build-system agnostic;\nand the relevant hypervisor headers in hv/.\n\nSigned-off-by: Chris Metcalf \u003ccmetcalf@tilera.com\u003e\nAcked-by: Arnd Bergmann \u003carnd@arndb.de\u003e\nAcked-by: FUJITA Tomonori \u003cfujita.tomonori@lab.ntt.co.jp\u003e\nReviewed-by: Paul Mundt \u003clethal@linux-sh.org\u003e\n"
    }
  ]
}