src/cpu/sparc/vm/copy_sparc.hpp

Wed, 17 Jun 2015 17:48:25 -0700

author
ascarpino
date
Wed, 17 Jun 2015 17:48:25 -0700
changeset 9788
44ef77ad417c
parent 6704
7f1743e1a14e
child 6876
710a3c8b516e
permissions
-rw-r--r--

8073108: Use x86 and SPARC CPU instructions for GHASH acceleration
Reviewed-by: kvn, jrose, phh

     1 /*
     2  * Copyright (c) 2003, 2011, Oracle and/or its affiliates. All rights reserved.
     3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
     4  *
     5  * This code is free software; you can redistribute it and/or modify it
     6  * under the terms of the GNU General Public License version 2 only, as
     7  * published by the Free Software Foundation.
     8  *
     9  * This code is distributed in the hope that it will be useful, but WITHOUT
    10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
    11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
    12  * version 2 for more details (a copy is included in the LICENSE file that
    13  * accompanied this code).
    14  *
    15  * You should have received a copy of the GNU General Public License version
    16  * 2 along with this work; if not, write to the Free Software Foundation,
    17  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
    18  *
    19  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
    20  * or visit www.oracle.com if you need additional information or have any
    21  * questions.
    22  *
    23  */
    25 #ifndef CPU_SPARC_VM_COPY_SPARC_HPP
    26 #define CPU_SPARC_VM_COPY_SPARC_HPP
    28 // Inline functions for memory copy and fill.
    30 static void pd_conjoint_words(HeapWord* from, HeapWord* to, size_t count) {
    31   (void)memmove(to, from, count * HeapWordSize);
    32 }
    34 static void pd_disjoint_words(HeapWord* from, HeapWord* to, size_t count) {
    35   switch (count) {
    36   case 8:  to[7] = from[7];
    37   case 7:  to[6] = from[6];
    38   case 6:  to[5] = from[5];
    39   case 5:  to[4] = from[4];
    40   case 4:  to[3] = from[3];
    41   case 3:  to[2] = from[2];
    42   case 2:  to[1] = from[1];
    43   case 1:  to[0] = from[0];
    44   case 0:  break;
    45   default: (void)memcpy(to, from, count * HeapWordSize);
    46            break;
    47   }
    48 }
    50 static void pd_disjoint_words_atomic(HeapWord* from, HeapWord* to, size_t count) {
    51   switch (count) {
    52   case 8:  to[7] = from[7];
    53   case 7:  to[6] = from[6];
    54   case 6:  to[5] = from[5];
    55   case 5:  to[4] = from[4];
    56   case 4:  to[3] = from[3];
    57   case 3:  to[2] = from[2];
    58   case 2:  to[1] = from[1];
    59   case 1:  to[0] = from[0];
    60   case 0:  break;
    61   default: while (count-- > 0) {
    62              *to++ = *from++;
    63            }
    64            break;
    65   }
    66 }
    68 static void pd_aligned_conjoint_words(HeapWord* from, HeapWord* to, size_t count) {
    69   (void)memmove(to, from, count * HeapWordSize);
    70 }
    72 static void pd_aligned_disjoint_words(HeapWord* from, HeapWord* to, size_t count) {
    73   pd_disjoint_words(from, to, count);
    74 }
    76 static void pd_conjoint_bytes(void* from, void* to, size_t count) {
    77   (void)memmove(to, from, count);
    78 }
    80 static void pd_conjoint_bytes_atomic(void* from, void* to, size_t count) {
    81   (void)memmove(to, from, count);
    82 }
    84 static void pd_conjoint_jshorts_atomic(jshort* from, jshort* to, size_t count) {
    85   if (from > to) {
    86     while (count-- > 0) {
    87       // Copy forwards
    88       *to++ = *from++;
    89     }
    90   } else {
    91     from += count - 1;
    92     to   += count - 1;
    93     while (count-- > 0) {
    94       // Copy backwards
    95       *to-- = *from--;
    96     }
    97   }
    98 }
   100 static void pd_conjoint_jints_atomic(jint* from, jint* to, size_t count) {
   101   if (from > to) {
   102     while (count-- > 0) {
   103       // Copy forwards
   104       *to++ = *from++;
   105     }
   106   } else {
   107     from += count - 1;
   108     to   += count - 1;
   109     while (count-- > 0) {
   110       // Copy backwards
   111       *to-- = *from--;
   112     }
   113   }
   114 }
   116 static void pd_conjoint_jlongs_atomic(jlong* from, jlong* to, size_t count) {
   117 #ifdef _LP64
   118   assert(BytesPerLong == BytesPerOop, "jlongs and oops must be the same size");
   119   pd_conjoint_oops_atomic((oop*)from, (oop*)to, count);
   120 #else
   121   // Guarantee use of ldd/std via some asm code, because compiler won't.
   122   // See solaris_sparc.il.
   123   _Copy_conjoint_jlongs_atomic(from, to, count);
   124 #endif
   125 }
   127 static void pd_conjoint_oops_atomic(oop* from, oop* to, size_t count) {
   128   // Do better than this: inline memmove body  NEEDS CLEANUP
   129   if (from > to) {
   130     while (count-- > 0) {
   131       // Copy forwards
   132       *to++ = *from++;
   133     }
   134   } else {
   135     from += count - 1;
   136     to   += count - 1;
   137     while (count-- > 0) {
   138       // Copy backwards
   139       *to-- = *from--;
   140     }
   141   }
   142 }
   144 static void pd_arrayof_conjoint_bytes(HeapWord* from, HeapWord* to, size_t count) {
   145   pd_conjoint_bytes_atomic(from, to, count);
   146 }
   148 static void pd_arrayof_conjoint_jshorts(HeapWord* from, HeapWord* to, size_t count) {
   149   pd_conjoint_jshorts_atomic((jshort*)from, (jshort*)to, count);
   150 }
   152 static void pd_arrayof_conjoint_jints(HeapWord* from, HeapWord* to, size_t count) {
   153   pd_conjoint_jints_atomic((jint*)from, (jint*)to, count);
   154 }
   156 static void pd_arrayof_conjoint_jlongs(HeapWord* from, HeapWord* to, size_t count) {
   157   pd_conjoint_jlongs_atomic((jlong*)from, (jlong*)to, count);
   158 }
   160 static void pd_arrayof_conjoint_oops(HeapWord* from, HeapWord* to, size_t count) {
   161   pd_conjoint_oops_atomic((oop*)from, (oop*)to, count);
   162 }
   164 static void pd_fill_to_words(HeapWord* tohw, size_t count, juint value) {
   165 #ifdef _LP64
   166   guarantee(mask_bits((uintptr_t)tohw, right_n_bits(LogBytesPerLong)) == 0,
   167          "unaligned fill words");
   168   julong* to = (julong*)tohw;
   169   julong  v  = ((julong)value << 32) | value;
   170   while (count-- > 0) {
   171     *to++ = v;
   172   }
   173 #else // _LP64
   174   juint* to = (juint*)tohw;
   175   while (count-- > 0) {
   176     *to++ = value;
   177   }
   178 #endif // _LP64
   179 }
   181 typedef void (*_zero_Fn)(HeapWord* to, size_t count);
   183 static void pd_fill_to_aligned_words(HeapWord* tohw, size_t count, juint value) {
   184   assert(MinObjAlignmentInBytes >= BytesPerLong, "need alternate implementation");
   186   if (value == 0 && UseBlockZeroing &&
   187       (count > (size_t)(BlockZeroingLowLimit >> LogHeapWordSize))) {
   188    // Call it only when block zeroing is used
   189    ((_zero_Fn)StubRoutines::zero_aligned_words())(tohw, count);
   190   } else {
   191    julong* to = (julong*)tohw;
   192    julong  v  = ((julong)value << 32) | value;
   193    // If count is odd, odd will be equal to 1 on 32-bit platform
   194    // and be equal to 0 on 64-bit platform.
   195    size_t odd = count % (BytesPerLong / HeapWordSize) ;
   197    size_t aligned_count = align_object_offset(count - odd) / HeapWordsPerLong;
   198    julong* end = ((julong*)tohw) + aligned_count - 1;
   199    while (to <= end) {
   200      DEBUG_ONLY(count -= BytesPerLong / HeapWordSize ;)
   201      *to++ = v;
   202    }
   203    assert(count == odd, "bad bounds on loop filling to aligned words");
   204    if (odd) {
   205      *((juint*)to) = value;
   207    }
   208   }
   209 }
   211 static void pd_fill_to_bytes(void* to, size_t count, jubyte value) {
   212   (void)memset(to, value, count);
   213 }
   215 static void pd_zero_to_words(HeapWord* tohw, size_t count) {
   216   pd_fill_to_words(tohw, count, 0);
   217 }
   219 static void pd_zero_to_bytes(void* to, size_t count) {
   220   (void)memset(to, 0, count);
   221 }
   223 #endif // CPU_SPARC_VM_COPY_SPARC_HPP

mercurial