)]}'
{
  "commit": "1ba7be7d691f6df2557d39c5b1a2e14c32e5dd20",
  "tree": "f6c805c01be475f21de0cdcada8f69c9076ea61e",
  "parents": [
    "21bc4f9b34cc1eab3610955207f72c52495ae8ed"
  ],
  "author": {
    "name": "David Mosberger-Tang",
    "email": "davidm@hpl.hp.com",
    "time": "Wed Apr 27 21:20:51 2005 -0700"
  },
  "committer": {
    "name": "Tony Luck",
    "email": "tony.luck@intel.com",
    "time": "Wed Apr 27 21:20:51 2005 -0700"
  },
  "message": "[IA64] Reschedule fsys_bubble_down().\n\nImprovements come from eliminating srlz.i, not scheduling AR/CR-reads\ntoo early (while there are others still pending), scheduling the\nbacking-store switch as well as possible, splitting the BBB bundle\ninto a MIB/MBB pair.\n\nWhy is it safe to eliminate the srlz.i?  Observe\nthat we used to clear bits ~PSR_PRESERVED_BITS in PSR.L.  Since\nPSR_PRESERVED_BITS\u003d\u003dPSR.{UP,MFL,MFH,PK,DT,PP,SP,RT,IC}, we\nended up clearing PSR.{BE,AC,I,DFL,DFH,DI,DB,SI,TB}.  However,\n\n PSR.BE : already is turned off in __kernel_syscall_via_epc()\n PSR.AC : don\u0027t care (kernel normally turns PSR.AC on)\n PSR.I  : already turned off by the time fsys_bubble_down gets invoked\n PSR.DFL: always 0 (kernel never turns it on)\n PSR.DFH: don\u0027t care --- kernel never touches f32-f127 on its own\n\t  initiative\n PSR.DI : always 0 (kernel never turns it on)\n PSR.SI : always 0 (kernel never turns it on)\n PSR.DB : don\u0027t care --- kernel never enables kernel-level breakpoints\n PSR.TB : must be 0 already; if it wasn\u0027t zero on entry to\n\t  __kernel_syscall_via_epc, the branch to fsys_bubble_down\n\t  will trigger a taken branch; the taken-trap-handler then\n\t  converts the syscall into a break-based system-call.\n\nIn other words: all the bits we\u0027re clearying are either 0 already or\nare don\u0027t cares!  Thus, we don\u0027t have to write PSR.L at all and we\ndon\u0027t have to do a srlz.i either.\n\nGood for another ~20 cycle improvement for EPC-based heavy-weight\nsyscalls.\n\nSigned-off-by: David Mosberger-Tang \u003cdavidm@hpl.hp.com\u003e\nSigned-off-by: Tony Luck \u003ctony.luck@intel.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "0d8650f7fce7561b88ae054b4ca0dff5933f5dcb",
      "old_mode": 33188,
      "old_path": "arch/ia64/kernel/fsys.S",
      "new_id": "57c6556b1e06bdd4e900262dad06dad206d8b190",
      "new_mode": 33188,
      "new_path": "arch/ia64/kernel/fsys.S"
    }
  ]
}