Mercurial > hg > graal-compiler
annotate src/share/vm/utilities/copy.cpp @ 1711:a6bff45449bc
6973570: OrderAccess::storestore() scales poorly on multi-socket x64 and sparc: cache-line ping-ponging
Summary: volatile store to static variable removed in favour of a volatile store to stack to avoid excessive cache coherency traffic; verified that the volatile store is not elided by any of our current compilers.
Reviewed-by: dholmes, dice, jcoomes, kvn
author | ysr |
---|---|
date | Tue, 10 Aug 2010 14:53:35 -0700 |
parents | d93949c5bdcc |
children | f95d63e2154a |
rev | line source |
---|---|
0 | 1 /* |
1552
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
0
diff
changeset
|
2 * Copyright (c) 2006, 2007, Oracle and/or its affiliates. All rights reserved. |
0 | 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 * | |
1552
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
0
diff
changeset
|
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
0
diff
changeset
|
20 * or visit www.oracle.com if you need additional information or have any |
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
0
diff
changeset
|
21 * questions. |
0 | 22 * |
23 */ | |
24 | |
25 # include "incls/_precompiled.incl" | |
26 # include "incls/_copy.cpp.incl" | |
27 | |
28 | |
29 // Copy bytes; larger units are filled atomically if everything is aligned. | |
30 void Copy::conjoint_memory_atomic(void* from, void* to, size_t size) { | |
31 address src = (address) from; | |
32 address dst = (address) to; | |
33 uintptr_t bits = (uintptr_t) src | (uintptr_t) dst | (uintptr_t) size; | |
34 | |
35 // (Note: We could improve performance by ignoring the low bits of size, | |
36 // and putting a short cleanup loop after each bulk copy loop. | |
37 // There are plenty of other ways to make this faster also, | |
38 // and it's a slippery slope. For now, let's keep this code simple | |
39 // since the simplicity helps clarify the atomicity semantics of | |
40 // this operation. There are also CPU-specific assembly versions | |
41 // which may or may not want to include such optimizations.) | |
42 | |
43 if (bits % sizeof(jlong) == 0) { | |
44 Copy::conjoint_jlongs_atomic((jlong*) src, (jlong*) dst, size / sizeof(jlong)); | |
45 } else if (bits % sizeof(jint) == 0) { | |
46 Copy::conjoint_jints_atomic((jint*) src, (jint*) dst, size / sizeof(jint)); | |
47 } else if (bits % sizeof(jshort) == 0) { | |
48 Copy::conjoint_jshorts_atomic((jshort*) src, (jshort*) dst, size / sizeof(jshort)); | |
49 } else { | |
50 // Not aligned, so no need to be atomic. | |
1603
d93949c5bdcc
6730276: JDI_REGRESSION tests fail with "Error: count must be non-zero" error on x86
kvn
parents:
1552
diff
changeset
|
51 Copy::conjoint_jbytes((void*) src, (void*) dst, size); |
0 | 52 } |
53 } | |
54 | |
55 | |
56 // Fill bytes; larger units are filled atomically if everything is aligned. | |
57 void Copy::fill_to_memory_atomic(void* to, size_t size, jubyte value) { | |
58 address dst = (address) to; | |
59 uintptr_t bits = (uintptr_t) to | (uintptr_t) size; | |
60 if (bits % sizeof(jlong) == 0) { | |
61 jlong fill = (julong)( (jubyte)value ); // zero-extend | |
62 if (fill != 0) { | |
63 fill += fill << 8; | |
64 fill += fill << 16; | |
65 fill += fill << 32; | |
66 } | |
67 //Copy::fill_to_jlongs_atomic((jlong*) dst, size / sizeof(jlong)); | |
68 for (uintptr_t off = 0; off < size; off += sizeof(jlong)) { | |
69 *(jlong*)(dst + off) = fill; | |
70 } | |
71 } else if (bits % sizeof(jint) == 0) { | |
72 jint fill = (juint)( (jubyte)value ); // zero-extend | |
73 if (fill != 0) { | |
74 fill += fill << 8; | |
75 fill += fill << 16; | |
76 } | |
77 //Copy::fill_to_jints_atomic((jint*) dst, size / sizeof(jint)); | |
78 for (uintptr_t off = 0; off < size; off += sizeof(jint)) { | |
79 *(jint*)(dst + off) = fill; | |
80 } | |
81 } else if (bits % sizeof(jshort) == 0) { | |
82 jshort fill = (jushort)( (jubyte)value ); // zero-extend | |
83 fill += fill << 8; | |
84 //Copy::fill_to_jshorts_atomic((jshort*) dst, size / sizeof(jshort)); | |
85 for (uintptr_t off = 0; off < size; off += sizeof(jshort)) { | |
86 *(jshort*)(dst + off) = fill; | |
87 } | |
88 } else { | |
89 // Not aligned, so no need to be atomic. | |
90 Copy::fill_to_bytes(dst, size, value); | |
91 } | |
92 } |