Mercurial > hg > graal-compiler
view src/share/vm/utilities/utf8.cpp @ 2149:7e37af9d69ef
7011379: G1: overly long concurrent marking cycles
Summary: This changeset introduces filtering of SATB buffers at the point when they are about to be enqueued. If this filtering clears enough entries on each buffer, the buffer can then be re-used and not enqueued. This cuts down the number of SATB buffers that need to be processed by the concurrent marking threads.
Reviewed-by: johnc, ysr
author | tonyp |
---|---|
date | Wed, 19 Jan 2011 09:35:17 -0500 |
parents | f95d63e2154a |
children | 3582bf76420e |
line wrap: on
line source
/* * Copyright (c) 1997, 2010, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #include "precompiled.hpp" #include "utilities/utf8.hpp" // Assume the utf8 string is in legal form and has been // checked in the class file parser/format checker. char* UTF8::next(const char* str, jchar* value) { unsigned const char *ptr = (const unsigned char *)str; unsigned char ch, ch2, ch3; int length = -1; /* bad length */ jchar result; switch ((ch = ptr[0]) >> 4) { default: result = ch; length = 1; break; case 0x8: case 0x9: case 0xA: case 0xB: case 0xF: /* Shouldn't happen. */ break; case 0xC: case 0xD: /* 110xxxxx 10xxxxxx */ if (((ch2 = ptr[1]) & 0xC0) == 0x80) { unsigned char high_five = ch & 0x1F; unsigned char low_six = ch2 & 0x3F; result = (high_five << 6) + low_six; length = 2; break; } break; case 0xE: /* 1110xxxx 10xxxxxx 10xxxxxx */ if (((ch2 = ptr[1]) & 0xC0) == 0x80) { if (((ch3 = ptr[2]) & 0xC0) == 0x80) { unsigned char high_four = ch & 0x0f; unsigned char mid_six = ch2 & 0x3f; unsigned char low_six = ch3 & 0x3f; result = (((high_four << 6) + mid_six) << 6) + low_six; length = 3; } } break; } /* end of switch */ if (length <= 0) { *value = ptr[0]; /* default bad result; */ return (char*)(ptr + 1); // make progress somehow } *value = result; // The assert is correct but the .class file is wrong // assert(UNICODE::utf8_size(result) == length, "checking reverse computation"); return (char *)(ptr + length); } char* UTF8::next_character(const char* str, jint* value) { unsigned const char *ptr = (const unsigned char *)str; /* See if it's legal supplementary character: 11101101 1010xxxx 10xxxxxx 11101101 1011xxxx 10xxxxxx */ if (is_supplementary_character(ptr)) { *value = get_supplementary_character(ptr); return (char *)(ptr + 6); } jchar result; char* next_ch = next(str, &result); *value = result; return next_ch; } // Count bytes of the form 10xxxxxx and deduct this count // from the total byte count. The utf8 string must be in // legal form which has been verified in the format checker. int UTF8::unicode_length(const char* str, int len) { int num_chars = len; for (int i = 0; i < len; i++) { if ((str[i] & 0xC0) == 0x80) { --num_chars; } } return num_chars; } // Count bytes of the utf8 string except those in form // 10xxxxxx which only appear in multibyte characters. // The utf8 string must be in legal form and has been // verified in the format checker. int UTF8::unicode_length(const char* str) { int num_chars = 0; for (const char* p = str; *p; p++) { if (((*p) & 0xC0) != 0x80) { num_chars++; } } return num_chars; } // Writes a jchar a utf8 and returns the end static u_char* utf8_write(u_char* base, jchar ch) { if ((ch != 0) && (ch <=0x7f)) { base[0] = (u_char) ch; return base + 1; } if (ch <= 0x7FF) { /* 11 bits or less. */ unsigned char high_five = ch >> 6; unsigned char low_six = ch & 0x3F; base[0] = high_five | 0xC0; /* 110xxxxx */ base[1] = low_six | 0x80; /* 10xxxxxx */ return base + 2; } /* possibly full 16 bits. */ char high_four = ch >> 12; char mid_six = (ch >> 6) & 0x3F; char low_six = ch & 0x3f; base[0] = high_four | 0xE0; /* 1110xxxx */ base[1] = mid_six | 0x80; /* 10xxxxxx */ base[2] = low_six | 0x80; /* 10xxxxxx */ return base + 3; } void UTF8::convert_to_unicode(const char* utf8_str, jchar* unicode_str, int unicode_length) { unsigned char ch; const char *ptr = (const char *)utf8_str; int index = 0; /* ASCII case loop optimization */ for (; index < unicode_length; index++) { if((ch = ptr[0]) > 0x7F) { break; } unicode_str[index] = ch; ptr = (const char *)(ptr + 1); } for (; index < unicode_length; index++) { ptr = UTF8::next(ptr, &unicode_str[index]); } } // Returns NULL if 'c' it not found. This only works as long // as 'c' is an ASCII character jbyte* UTF8::strrchr(jbyte* base, int length, jbyte c) { assert(length >= 0, "sanity check"); assert(c >= 0, "does not work for non-ASCII characters"); // Skip backwards in string until 'c' is found or end is reached while(--length >= 0 && base[length] != c); return (length < 0) ? NULL : &base[length]; } bool UTF8::equal(jbyte* base1, int length1, jbyte* base2, int length2) { // Length must be the same if (length1 != length2) return false; for (int i = 0; i < length1; i++) { if (base1[i] != base2[i]) return false; } return true; } bool UTF8::is_supplementary_character(const unsigned char* str) { return ((str[0] & 0xFF) == 0xED) && ((str[1] & 0xF0) == 0xA0) && ((str[2] & 0xC0) == 0x80) && ((str[3] & 0xFF) == 0xED) && ((str[4] & 0xF0) == 0xB0) && ((str[5] & 0xC0) == 0x80); } jint UTF8::get_supplementary_character(const unsigned char* str) { return 0x10000 + ((str[1] & 0x0f) << 16) + ((str[2] & 0x3f) << 10) + ((str[4] & 0x0f) << 6) + (str[5] & 0x3f); } //------------------------------------------------------------------------------------- int UNICODE::utf8_size(jchar c) { if ((0x0001 <= c) && (c <= 0x007F)) return 1; if (c <= 0x07FF) return 2; return 3; } int UNICODE::utf8_length(jchar* base, int length) { int result = 0; for (int index = 0; index < length; index++) { jchar c = base[index]; if ((0x0001 <= c) && (c <= 0x007F)) result += 1; else if (c <= 0x07FF) result += 2; else result += 3; } return result; } char* UNICODE::as_utf8(jchar* base, int length) { int utf8_len = utf8_length(base, length); u_char* result = NEW_RESOURCE_ARRAY(u_char, utf8_len + 1); u_char* p = result; for (int index = 0; index < length; index++) { p = utf8_write(p, base[index]); } *p = '\0'; assert(p == &result[utf8_len], "length prediction must be correct"); return (char*) result; } char* UNICODE::as_utf8(jchar* base, int length, char* buf, int buflen) { u_char* p = (u_char*)buf; u_char* end = (u_char*)buf + buflen; for (int index = 0; index < length; index++) { jchar c = base[index]; if (p + utf8_size(c) >= end) break; // string is truncated p = utf8_write(p, base[index]); } *p = '\0'; return buf; } void UNICODE::convert_to_utf8(const jchar* base, int length, char* utf8_buffer) { for(int index = 0; index < length; index++) { utf8_buffer = (char*)utf8_write((u_char*)utf8_buffer, base[index]); } *utf8_buffer = '\0'; }