From 955b52e4b2caa351b0aa9acebfc73332915d43b0 Mon Sep 17 00:00:00 2001 From: wuyunjie Date: Sat, 27 Mar 2021 10:12:18 +0800 Subject: [PATCH] Description:l0 posix interface Team: OTHERS Feature or Bugfix:Feature Binary Source:No PrivateCode(Yes/No):No Change-Id: I8bcb92a97a98c61077c55704dec03c997d1246da ChangeID:13439291 --- kal/posix/include/bits/alltypes.h | 35 +- kal/posix/include/bits/errno.h | 1 + kal/posix/include/limits.h | 6 +- kal/posix/include/time.h | 3 +- kal/posix/musl_src/COPYRIGHT | 190 ++ kal/posix/musl_src/README.OpenSource | 10 + kal/posix/musl_src/ctype/isalnum.c | 13 + kal/posix/musl_src/ctype/isascii.c | 7 + kal/posix/musl_src/ctype/isdigit.c | 14 + kal/posix/musl_src/ctype/islower.c | 14 + kal/posix/musl_src/ctype/isprint.c | 14 + kal/posix/musl_src/ctype/isspace.c | 14 + kal/posix/musl_src/ctype/isupper.c | 14 + kal/posix/musl_src/ctype/isxdigit.c | 13 + kal/posix/musl_src/ctype/tolower.c | 14 + kal/posix/musl_src/ctype/toupper.c | 14 + kal/posix/musl_src/errno/__strerror.h | 105 + kal/posix/musl_src/errno/strerror.c | 36 + kal/posix/musl_src/exit/abort.c | 8 + kal/posix/musl_src/exit/assert.c | 29 + kal/posix/musl_src/include/features.h | 16 + kal/posix/musl_src/include/string.h | 12 + kal/posix/musl_src/internal/intscan.c | 100 + kal/posix/musl_src/internal/intscan.h | 9 + kal/posix/musl_src/internal/locale_impl.h | 43 + kal/posix/musl_src/internal/shgetc.c | 37 + kal/posix/musl_src/internal/shgetc.h | 33 + kal/posix/musl_src/internal/stdio_impl.h | 87 + kal/posix/musl_src/math/abs.c | 6 + kal/posix/musl_src/math/exp_data.c | 182 ++ kal/posix/musl_src/math/exp_data.h | 26 + kal/posix/musl_src/math/libm.h | 272 ++ kal/posix/musl_src/math/log.c | 127 + kal/posix/musl_src/math/log_data.c | 328 +++ kal/posix/musl_src/math/log_data.h | 28 + kal/posix/musl_src/math/pow.c | 343 +++ kal/posix/musl_src/math/pow_data.c | 180 ++ kal/posix/musl_src/math/pow_data.h | 22 + kal/posix/musl_src/math/round.c | 35 + kal/posix/musl_src/math/sqrt.c | 184 ++ kal/posix/musl_src/misc/dirname.c | 14 + kal/posix/musl_src/prng/random.c | 126 + kal/posix/musl_src/regex/regcomp.c | 2953 +++++++++++++++++++++ kal/posix/musl_src/regex/regexec.c | 1028 +++++++ kal/posix/musl_src/regex/tre-mem.c | 158 ++ kal/posix/musl_src/regex/tre.h | 232 ++ kal/posix/musl_src/stdio/__fdopen.c | 62 + kal/posix/musl_src/stdio/__fmodeflags.c | 16 + kal/posix/musl_src/stdio/__lockfile.c | 12 + kal/posix/musl_src/stdio/__stdio_close.c | 7 + kal/posix/musl_src/stdio/__stdio_read.c | 25 + kal/posix/musl_src/stdio/__stdio_seek.c | 53 + kal/posix/musl_src/stdio/__stdio_write.c | 35 + kal/posix/musl_src/stdio/__stdout_write.c | 11 + kal/posix/musl_src/stdio/__toread.c | 14 + kal/posix/musl_src/stdio/__towrite.c | 18 + kal/posix/musl_src/stdio/__uflow.c | 11 + kal/posix/musl_src/stdio/clearerr.c | 10 + kal/posix/musl_src/stdio/fclose.c | 36 + kal/posix/musl_src/stdio/feof.c | 14 + kal/posix/musl_src/stdio/fflush.c | 47 + kal/posix/musl_src/stdio/fgets.c | 46 + kal/posix/musl_src/stdio/fileno.c | 14 + kal/posix/musl_src/stdio/fopen.c | 42 + kal/posix/musl_src/stdio/fputs.c | 10 + kal/posix/musl_src/stdio/fread.c | 38 + kal/posix/musl_src/stdio/fseek.c | 41 + kal/posix/musl_src/stdio/ftell.c | 39 + kal/posix/musl_src/stdio/fwrite.c | 38 + kal/posix/musl_src/stdio/ofl.c | 28 + kal/posix/musl_src/stdio/ofl_add.c | 11 + kal/posix/musl_src/stdio/perror.c | 30 + kal/posix/musl_src/stdio/rewind.c | 9 + kal/posix/musl_src/stdio/stderr.c | 20 + kal/posix/musl_src/stdio/stdout.c | 20 + kal/posix/musl_src/stdlib/atoi.c | 16 + kal/posix/musl_src/stdlib/atol.c | 17 + kal/posix/musl_src/stdlib/atoll.c | 17 + kal/posix/musl_src/stdlib/strtol.c | 38 + kal/posix/musl_src/string/memchr.c | 27 + kal/posix/musl_src/string/memcmp.c | 8 + kal/posix/musl_src/string/memcpy.c | 124 + kal/posix/musl_src/string/memrchr.c | 10 + kal/posix/musl_src/string/memset.c | 90 + kal/posix/musl_src/string/strcasecmp.c | 16 + kal/posix/musl_src/string/strchr.c | 7 + kal/posix/musl_src/string/strchrnul.c | 28 + kal/posix/musl_src/string/strcmp.c | 7 + kal/posix/musl_src/string/strcspn.c | 17 + kal/posix/musl_src/string/strdup.c | 10 + kal/posix/musl_src/string/strlen.c | 22 + kal/posix/musl_src/string/strncasecmp.c | 10 + kal/posix/musl_src/string/strncmp.c | 9 + kal/posix/musl_src/string/strrchr.c | 6 + kal/posix/musl_src/string/strstr.c | 154 ++ kal/posix/musl_src/string/wcschr.c | 8 + kal/posix/musl_src/string/wcslen.c | 8 + kal/posix/musl_src/time/strftime.c | 281 ++ kal/posix/musl_src/time/strptime.c | 206 ++ kal/posix/musl_src/time/time_impl.h | 14 + kal/posix/src/mqueue.c | 606 +++++ kal/posix/src/mqueue_impl.h | 145 + kal/posix/src/pthread_cond.c | 288 ++ kal/posix/src/semaphore.c | 50 + kal/posix/src/time.c | 90 +- kal/posix/src/time_internal.h | 123 + 106 files changed, 10260 insertions(+), 84 deletions(-) create mode 100644 kal/posix/musl_src/COPYRIGHT create mode 100644 kal/posix/musl_src/README.OpenSource create mode 100644 kal/posix/musl_src/ctype/isalnum.c create mode 100644 kal/posix/musl_src/ctype/isascii.c create mode 100644 kal/posix/musl_src/ctype/isdigit.c create mode 100644 kal/posix/musl_src/ctype/islower.c create mode 100644 kal/posix/musl_src/ctype/isprint.c create mode 100644 kal/posix/musl_src/ctype/isspace.c create mode 100644 kal/posix/musl_src/ctype/isupper.c create mode 100644 kal/posix/musl_src/ctype/isxdigit.c create mode 100644 kal/posix/musl_src/ctype/tolower.c create mode 100644 kal/posix/musl_src/ctype/toupper.c create mode 100644 kal/posix/musl_src/errno/__strerror.h create mode 100644 kal/posix/musl_src/errno/strerror.c create mode 100644 kal/posix/musl_src/exit/abort.c create mode 100644 kal/posix/musl_src/exit/assert.c create mode 100644 kal/posix/musl_src/include/features.h create mode 100644 kal/posix/musl_src/include/string.h create mode 100644 kal/posix/musl_src/internal/intscan.c create mode 100644 kal/posix/musl_src/internal/intscan.h create mode 100644 kal/posix/musl_src/internal/locale_impl.h create mode 100644 kal/posix/musl_src/internal/shgetc.c create mode 100644 kal/posix/musl_src/internal/shgetc.h create mode 100644 kal/posix/musl_src/internal/stdio_impl.h create mode 100644 kal/posix/musl_src/math/abs.c create mode 100644 kal/posix/musl_src/math/exp_data.c create mode 100644 kal/posix/musl_src/math/exp_data.h create mode 100644 kal/posix/musl_src/math/libm.h create mode 100644 kal/posix/musl_src/math/log.c create mode 100644 kal/posix/musl_src/math/log_data.c create mode 100644 kal/posix/musl_src/math/log_data.h create mode 100644 kal/posix/musl_src/math/pow.c create mode 100644 kal/posix/musl_src/math/pow_data.c create mode 100644 kal/posix/musl_src/math/pow_data.h create mode 100644 kal/posix/musl_src/math/round.c create mode 100644 kal/posix/musl_src/math/sqrt.c create mode 100644 kal/posix/musl_src/misc/dirname.c create mode 100644 kal/posix/musl_src/prng/random.c create mode 100644 kal/posix/musl_src/regex/regcomp.c create mode 100644 kal/posix/musl_src/regex/regexec.c create mode 100644 kal/posix/musl_src/regex/tre-mem.c create mode 100644 kal/posix/musl_src/regex/tre.h create mode 100644 kal/posix/musl_src/stdio/__fdopen.c create mode 100644 kal/posix/musl_src/stdio/__fmodeflags.c create mode 100644 kal/posix/musl_src/stdio/__lockfile.c create mode 100644 kal/posix/musl_src/stdio/__stdio_close.c create mode 100644 kal/posix/musl_src/stdio/__stdio_read.c create mode 100644 kal/posix/musl_src/stdio/__stdio_seek.c create mode 100644 kal/posix/musl_src/stdio/__stdio_write.c create mode 100644 kal/posix/musl_src/stdio/__stdout_write.c create mode 100644 kal/posix/musl_src/stdio/__toread.c create mode 100644 kal/posix/musl_src/stdio/__towrite.c create mode 100644 kal/posix/musl_src/stdio/__uflow.c create mode 100644 kal/posix/musl_src/stdio/clearerr.c create mode 100644 kal/posix/musl_src/stdio/fclose.c create mode 100644 kal/posix/musl_src/stdio/feof.c create mode 100644 kal/posix/musl_src/stdio/fflush.c create mode 100644 kal/posix/musl_src/stdio/fgets.c create mode 100644 kal/posix/musl_src/stdio/fileno.c create mode 100644 kal/posix/musl_src/stdio/fopen.c create mode 100644 kal/posix/musl_src/stdio/fputs.c create mode 100644 kal/posix/musl_src/stdio/fread.c create mode 100644 kal/posix/musl_src/stdio/fseek.c create mode 100644 kal/posix/musl_src/stdio/ftell.c create mode 100644 kal/posix/musl_src/stdio/fwrite.c create mode 100644 kal/posix/musl_src/stdio/ofl.c create mode 100644 kal/posix/musl_src/stdio/ofl_add.c create mode 100644 kal/posix/musl_src/stdio/perror.c create mode 100644 kal/posix/musl_src/stdio/rewind.c create mode 100644 kal/posix/musl_src/stdio/stderr.c create mode 100644 kal/posix/musl_src/stdio/stdout.c create mode 100644 kal/posix/musl_src/stdlib/atoi.c create mode 100644 kal/posix/musl_src/stdlib/atol.c create mode 100644 kal/posix/musl_src/stdlib/atoll.c create mode 100644 kal/posix/musl_src/stdlib/strtol.c create mode 100644 kal/posix/musl_src/string/memchr.c create mode 100644 kal/posix/musl_src/string/memcmp.c create mode 100644 kal/posix/musl_src/string/memcpy.c create mode 100644 kal/posix/musl_src/string/memrchr.c create mode 100644 kal/posix/musl_src/string/memset.c create mode 100644 kal/posix/musl_src/string/strcasecmp.c create mode 100644 kal/posix/musl_src/string/strchr.c create mode 100644 kal/posix/musl_src/string/strchrnul.c create mode 100644 kal/posix/musl_src/string/strcmp.c create mode 100644 kal/posix/musl_src/string/strcspn.c create mode 100644 kal/posix/musl_src/string/strdup.c create mode 100644 kal/posix/musl_src/string/strlen.c create mode 100644 kal/posix/musl_src/string/strncasecmp.c create mode 100644 kal/posix/musl_src/string/strncmp.c create mode 100644 kal/posix/musl_src/string/strrchr.c create mode 100644 kal/posix/musl_src/string/strstr.c create mode 100644 kal/posix/musl_src/string/wcschr.c create mode 100644 kal/posix/musl_src/string/wcslen.c create mode 100644 kal/posix/musl_src/time/strftime.c create mode 100644 kal/posix/musl_src/time/strptime.c create mode 100644 kal/posix/musl_src/time/time_impl.h create mode 100644 kal/posix/src/mqueue.c create mode 100644 kal/posix/src/mqueue_impl.h create mode 100644 kal/posix/src/pthread_cond.c create mode 100644 kal/posix/src/time_internal.h diff --git a/kal/posix/include/bits/alltypes.h b/kal/posix/include/bits/alltypes.h index 909d60d5..2ba9a82e 100755 --- a/kal/posix/include/bits/alltypes.h +++ b/kal/posix/include/bits/alltypes.h @@ -437,7 +437,32 @@ typedef struct { unsigned __attr[2]; } pthread_rwlockattr_t; #if defined(__NEED_struct__IO_FILE) && !defined(__DEFINED_struct__IO_FILE) -struct _IO_FILE { char __x; }; +struct _IO_FILE { + unsigned flags; + unsigned char *rpos, *rend; + int (*close)(struct _IO_FILE *); + unsigned char *wend, *wpos; + unsigned char *mustbezero_1; + unsigned char *wbase; + size_t (*read)(struct _IO_FILE *, unsigned char *, size_t); + size_t (*write)(struct _IO_FILE *, const unsigned char *, size_t); + off_t (*seek)(struct _IO_FILE *, off_t, int); + unsigned char *buf; + size_t buf_size; + struct _IO_FILE *prev, *next; + int fd; + int pipe_pid; + int mode; + void *lock; + int lbf; + void *cookie; + off_t off; + char *getln_buf; + void *mustbezero_2; + unsigned char *shend; + off_t shlim, shcnt; + struct __locale_struct *locale; +}; #define __DEFINED_struct__IO_FILE #endif @@ -535,7 +560,13 @@ typedef struct { union { int __i[sizeof(long)==8?10:6]; volatile int __vi[sizeof #endif #if defined(__NEED_pthread_cond_t) && !defined(__DEFINED_pthread_cond_t) -typedef struct { union { int __i[12]; volatile int __vi[12]; void *__p[12*sizeof(int)/sizeof(void*)]; } __u; } pthread_cond_t; +#include "los_event.h" +typedef struct pthread_cond { + volatile int count; /**< The number of tasks blocked by condition */ + EVENT_CB_S event; /**< Event object*/ + pthread_mutex_t* mutex; /**< Mutex locker for condition variable protection */ + volatile int value; /**< Condition variable state value*/ +} pthread_cond_t; #define __DEFINED_pthread_cond_t #endif diff --git a/kal/posix/include/bits/errno.h b/kal/posix/include/bits/errno.h index d2e1eeee..4b1b0954 100644 --- a/kal/posix/include/bits/errno.h +++ b/kal/posix/include/bits/errno.h @@ -1,3 +1,4 @@ +#define ENOERR 0 #define EPERM 1 #define ENOENT 2 #define ESRCH 3 diff --git a/kal/posix/include/limits.h b/kal/posix/include/limits.h index 5b40fda7..b71daa35 100644 --- a/kal/posix/include/limits.h +++ b/kal/posix/include/limits.h @@ -2,6 +2,8 @@ #define _LIMITS_H #ifdef __ICCARM__ /* for iar */ +#define PATH_MAX 256 +#define MQ_PRIO_MAX 1 #define PTHREAD_STACK_MIN LOSCFG_BASE_CORE_TSK_MIN_STACK_SIZE #include_next #else @@ -49,7 +51,7 @@ #ifndef NAME_MAX #define NAME_MAX 255 #endif -#define PATH_MAX 4096 +#define PATH_MAX 256 #define NGROUPS_MAX 32 #define ARG_MAX 131072 #define IOV_MAX 1024 @@ -74,7 +76,7 @@ #define SEM_VALUE_MAX 0x7fffffff #define SEM_NSEMS_MAX 256 #define DELAYTIMER_MAX 0x7fffffff -#define MQ_PRIO_MAX 32768 +#define MQ_PRIO_MAX 1 #define LOGIN_NAME_MAX 256 /* Arbitrary numbers... */ diff --git a/kal/posix/include/time.h b/kal/posix/include/time.h index 14e68b7e..09ecd524 100644 --- a/kal/posix/include/time.h +++ b/kal/posix/include/time.h @@ -83,12 +83,13 @@ extern "C" { #include +#ifndef NULL #ifdef __cplusplus #define NULL 0L #else #define NULL ((void*)0) #endif - +#endif #define __NEED_size_t #define __NEED_time_t diff --git a/kal/posix/musl_src/COPYRIGHT b/kal/posix/musl_src/COPYRIGHT new file mode 100644 index 00000000..e6472371 --- /dev/null +++ b/kal/posix/musl_src/COPYRIGHT @@ -0,0 +1,190 @@ +musl as a whole is licensed under the following standard MIT license: + +---------------------------------------------------------------------- +Copyright © 2005-2020 Rich Felker, et al. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice shall be +included in all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +---------------------------------------------------------------------- + +Authors/contributors include: + +A. Wilcox +Ada Worcester +Alex Dowad +Alex Suykov +Alexander Monakov +Andre McCurdy +Andrew Kelley +Anthony G. Basile +Aric Belsito +Arvid Picciani +Bartosz Brachaczek +Benjamin Peterson +Bobby Bingham +Boris Brezillon +Brent Cook +Chris Spiegel +Clément Vasseur +Daniel Micay +Daniel Sabogal +Daurnimator +David Carlier +David Edelsohn +Denys Vlasenko +Dmitry Ivanov +Dmitry V. Levin +Drew DeVault +Emil Renner Berthing +Fangrui Song +Felix Fietkau +Felix Janda +Gianluca Anzolin +Hauke Mehrtens +He X +Hiltjo Posthuma +Isaac Dunham +Jaydeep Patil +Jens Gustedt +Jeremy Huntwork +Jo-Philipp Wich +Joakim Sindholt +John Spencer +Julien Ramseier +Justin Cormack +Kaarle Ritvanen +Khem Raj +Kylie McClain +Leah Neukirchen +Luca Barbato +Luka Perkov +M Farkas-Dyck (Strake) +Mahesh Bodapati +Markus Wichmann +Masanori Ogino +Michael Clark +Michael Forney +Mikhail Kremnyov +Natanael Copa +Nicholas J. Kain +orc +Pascal Cuoq +Patrick Oppenlander +Petr Hosek +Petr Skocik +Pierre Carrier +Reini Urban +Rich Felker +Richard Pennington +Ryan Fairfax +Samuel Holland +Segev Finer +Shiz +sin +Solar Designer +Stefan Kristiansson +Stefan O'Rear +Szabolcs Nagy +Timo Teräs +Trutz Behn +Valentin Ochs +Will Dietz +William Haddon +William Pitcock + +Portions of this software are derived from third-party works licensed +under terms compatible with the above MIT license: + +The TRE regular expression implementation (src/regex/reg* and +src/regex/tre*) is Copyright © 2001-2008 Ville Laurikari and licensed +under a 2-clause BSD license (license text in the source files). The +included version has been heavily modified by Rich Felker in 2012, in +the interests of size, simplicity, and namespace cleanliness. + +Much of the math library code (src/math/* and src/complex/*) is +Copyright © 1993,2004 Sun Microsystems or +Copyright © 2003-2011 David Schultz or +Copyright © 2003-2009 Steven G. Kargl or +Copyright © 2003-2009 Bruce D. Evans or +Copyright © 2008 Stephen L. Moshier or +Copyright © 2017-2018 Arm Limited +and labelled as such in comments in the individual source files. All +have been licensed under extremely permissive terms. + +The ARM memcpy code (src/string/arm/memcpy_el.S) is Copyright © 2008 +The Android Open Source Project and is licensed under a two-clause BSD +license. It was taken from Bionic libc, used on Android. + +The implementation of DES for crypt (src/crypt/crypt_des.c) is +Copyright © 1994 David Burren. It is licensed under a BSD license. + +The implementation of blowfish crypt (src/crypt/crypt_blowfish.c) was +originally written by Solar Designer and placed into the public +domain. The code also comes with a fallback permissive license for use +in jurisdictions that may not recognize the public domain. + +The smoothsort implementation (src/stdlib/qsort.c) is Copyright © 2011 +Valentin Ochs and is licensed under an MIT-style license. + +The x86_64 port was written by Nicholas J. Kain and is licensed under +the standard MIT terms. + +The mips and microblaze ports were originally written by Richard +Pennington for use in the ellcc project. The original code was adapted +by Rich Felker for build system and code conventions during upstream +integration. It is licensed under the standard MIT terms. + +The mips64 port was contributed by Imagination Technologies and is +licensed under the standard MIT terms. + +The powerpc port was also originally written by Richard Pennington, +and later supplemented and integrated by John Spencer. It is licensed +under the standard MIT terms. + +All other files which have no copyright comments are original works +produced specifically for use as part of this library, written either +by Rich Felker, the main author of the library, or by one or more +contibutors listed above. Details on authorship of individual files +can be found in the git version control history of the project. The +omission of copyright and license comments in each file is in the +interest of source tree size. + +In addition, permission is hereby granted for all public header files +(include/* and arch/*/bits/*) and crt files intended to be linked into +applications (crt/*, ldso/dlstart.c, and arch/*/crt_arch.h) to omit +the copyright notice and permission notice otherwise required by the +license, and to use these files without any requirement of +attribution. These files include substantial contributions from: + +Bobby Bingham +John Spencer +Nicholas J. Kain +Rich Felker +Richard Pennington +Stefan Kristiansson +Szabolcs Nagy + +all of whom have explicitly granted such permission. + +This file previously contained text expressing a belief that most of +the files covered by the above exception were sufficiently trivial not +to be subject to copyright, resulting in confusion over whether it +negated the permissions granted in the license. In the spirit of +permissive licensing, and of not having licensing issues being an +obstacle to adoption, that text has been removed. diff --git a/kal/posix/musl_src/README.OpenSource b/kal/posix/musl_src/README.OpenSource new file mode 100644 index 00000000..3cc2559a --- /dev/null +++ b/kal/posix/musl_src/README.OpenSource @@ -0,0 +1,10 @@ +[ + { + "Name" : "musl", + "License" : "MIT License", + "License File" : "COPYRIGHT", + "Version Number" : "1.2.0", + "Upstream URL" : "http://www.musl-libc.org/", + "Description" : "musl is an MIT-licensed implementation of the standard C library" + } +] diff --git a/kal/posix/musl_src/ctype/isalnum.c b/kal/posix/musl_src/ctype/isalnum.c new file mode 100644 index 00000000..8018a2bc --- /dev/null +++ b/kal/posix/musl_src/ctype/isalnum.c @@ -0,0 +1,13 @@ +#include + +int isalnum(int c) +{ + return isalpha(c) || isdigit(c); +} + +int __isalnum_l(int c, locale_t l) +{ + return isalnum(c); +} + +weak_alias(__isalnum_l, isalnum_l); diff --git a/kal/posix/musl_src/ctype/isascii.c b/kal/posix/musl_src/ctype/isascii.c new file mode 100644 index 00000000..c8a45c29 --- /dev/null +++ b/kal/posix/musl_src/ctype/isascii.c @@ -0,0 +1,7 @@ +#include +#undef isascii + +int isascii(int c) +{ + return !(c&~0x7f); +} \ No newline at end of file diff --git a/kal/posix/musl_src/ctype/isdigit.c b/kal/posix/musl_src/ctype/isdigit.c new file mode 100644 index 00000000..16beddb4 --- /dev/null +++ b/kal/posix/musl_src/ctype/isdigit.c @@ -0,0 +1,14 @@ +#include +#undef isdigit + +int isdigit(int c) +{ + return (unsigned)c-'0' < 10; +} + +int __isdigit_l(int c, locale_t l) +{ + return isdigit(c); +} + +weak_alias(__isdigit_l, isdigit_l); diff --git a/kal/posix/musl_src/ctype/islower.c b/kal/posix/musl_src/ctype/islower.c new file mode 100644 index 00000000..c3fa74c4 --- /dev/null +++ b/kal/posix/musl_src/ctype/islower.c @@ -0,0 +1,14 @@ +#include +#undef islower + +int islower(int c) +{ + return (unsigned)c-'a' < 26; +} + +int __islower_l(int c, locale_t l) +{ + return islower(c); +} + +weak_alias(__islower_l, islower_l); diff --git a/kal/posix/musl_src/ctype/isprint.c b/kal/posix/musl_src/ctype/isprint.c new file mode 100644 index 00000000..b950816b --- /dev/null +++ b/kal/posix/musl_src/ctype/isprint.c @@ -0,0 +1,14 @@ +#include +#undef isprint + +int isprint(int c) +{ + return (unsigned)c-0x20 < 0x5f; +} + +int __isprint_l(int c, locale_t l) +{ + return isprint(c); +} + +weak_alias(__isprint_l, isprint_l); diff --git a/kal/posix/musl_src/ctype/isspace.c b/kal/posix/musl_src/ctype/isspace.c new file mode 100644 index 00000000..428813e7 --- /dev/null +++ b/kal/posix/musl_src/ctype/isspace.c @@ -0,0 +1,14 @@ +#include +#undef isspace + +int isspace(int c) +{ + return c == ' ' || (unsigned)c-'\t' < 5; +} + +int __isspace_l(int c, locale_t l) +{ + return isspace(c); +} + +weak_alias(__isspace_l, isspace_l); diff --git a/kal/posix/musl_src/ctype/isupper.c b/kal/posix/musl_src/ctype/isupper.c new file mode 100644 index 00000000..bfd15acd --- /dev/null +++ b/kal/posix/musl_src/ctype/isupper.c @@ -0,0 +1,14 @@ +#include +#undef isupper + +int isupper(int c) +{ + return (unsigned)c-'A' < 26; +} + +int __isupper_l(int c, locale_t l) +{ + return isupper(c); +} + +weak_alias(__isupper_l, isupper_l); diff --git a/kal/posix/musl_src/ctype/isxdigit.c b/kal/posix/musl_src/ctype/isxdigit.c new file mode 100644 index 00000000..aab1a745 --- /dev/null +++ b/kal/posix/musl_src/ctype/isxdigit.c @@ -0,0 +1,13 @@ +#include + +int isxdigit(int c) +{ + return isdigit(c) || ((unsigned)c|32)-'a' < 6; +} + +int __isxdigit_l(int c, locale_t l) +{ + return isxdigit(c); +} + +weak_alias(__isxdigit_l, isxdigit_l); diff --git a/kal/posix/musl_src/ctype/tolower.c b/kal/posix/musl_src/ctype/tolower.c new file mode 100644 index 00000000..f10132ec --- /dev/null +++ b/kal/posix/musl_src/ctype/tolower.c @@ -0,0 +1,14 @@ +#include + +int tolower(int c) +{ + if (isupper(c)) return c | 32; + return c; +} + +int __tolower_l(int c, locale_t l) +{ + return tolower(c); +} + +weak_alias(__tolower_l, tolower_l); diff --git a/kal/posix/musl_src/ctype/toupper.c b/kal/posix/musl_src/ctype/toupper.c new file mode 100644 index 00000000..4e74a55c --- /dev/null +++ b/kal/posix/musl_src/ctype/toupper.c @@ -0,0 +1,14 @@ +#include + +int toupper(int c) +{ + if (islower(c)) return c & 0x5f; + return c; +} + +int __toupper_l(int c, locale_t l) +{ + return toupper(c); +} + +weak_alias(__toupper_l, toupper_l); diff --git a/kal/posix/musl_src/errno/__strerror.h b/kal/posix/musl_src/errno/__strerror.h new file mode 100644 index 00000000..2f04d400 --- /dev/null +++ b/kal/posix/musl_src/errno/__strerror.h @@ -0,0 +1,105 @@ +/* This file is sorted such that 'errors' which represent exceptional + * conditions under which a correct program may fail come first, followed + * by messages that indicate an incorrect program or system failure. The + * macro E() along with double-inclusion is used to ensure that ordering + * of the strings remains synchronized. */ + +E(EILSEQ, "Illegal byte sequence") +E(EDOM, "Domain error") +E(ERANGE, "Result not representable") + +E(ENOTTY, "Not a tty") +E(EACCES, "Permission denied") +E(EPERM, "Operation not permitted") +E(ENOENT, "No such file or directory") +E(ESRCH, "No such process") +E(EEXIST, "File exists") + +E(EOVERFLOW, "Value too large for data type") +E(ENOSPC, "No space left on device") +E(ENOMEM, "Out of memory") + +E(EBUSY, "Resource busy") +E(EINTR, "Interrupted system call") +E(EAGAIN, "Resource temporarily unavailable") +E(ESPIPE, "Invalid seek") + +E(EXDEV, "Cross-device link") +E(EROFS, "Read-only file system") +E(ENOTEMPTY, "Directory not empty") + +E(ECONNRESET, "Connection reset by peer") +E(ETIMEDOUT, "Operation timed out") +E(ECONNREFUSED, "Connection refused") +E(EHOSTDOWN, "Host is down") +E(EHOSTUNREACH, "Host is unreachable") +E(EADDRINUSE, "Address in use") + +E(EPIPE, "Broken pipe") +E(EIO, "I/O error") +E(ENXIO, "No such device or address") +E(ENOTBLK, "Block device required") +E(ENODEV, "No such device") +E(ENOTDIR, "Not a directory") +E(EISDIR, "Is a directory") +E(ETXTBSY, "Text file busy") +E(ENOEXEC, "Exec format error") + +E(EINVAL, "Invalid argument") + +E(E2BIG, "Argument list too long") +E(ELOOP, "Symbolic link loop") +E(ENAMETOOLONG, "Filename too long") +E(ENFILE, "Too many open files in system") +E(EMFILE, "No file descriptors available") +E(EBADF, "Bad file descriptor") +E(ECHILD, "No child process") +E(EFAULT, "Bad address") +E(EFBIG, "File too large") +E(EMLINK, "Too many links") +E(ENOLCK, "No locks available") + +E(EDEADLK, "Resource deadlock would occur") +E(ENOTRECOVERABLE, "State not recoverable") +E(EOWNERDEAD, "Previous owner died") +E(ECANCELED, "Operation canceled") +E(ENOSYS, "Function not implemented") +E(ENOMSG, "No message of desired type") +E(EIDRM, "Identifier removed") +E(ENOSTR, "Device not a stream") +E(ENODATA, "No data available") +E(ETIME, "Device timeout") +E(ENOSR, "Out of streams resources") +E(ENOLINK, "Link has been severed") +E(EPROTO, "Protocol error") +E(EBADMSG, "Bad message") +E(EBADFD, "File descriptor in bad state") +E(ENOTSOCK, "Not a socket") +E(EDESTADDRREQ, "Destination address required") +E(EMSGSIZE, "Message too large") +E(EPROTOTYPE, "Protocol wrong type for socket") +E(ENOPROTOOPT, "Protocol not available") +E(EPROTONOSUPPORT,"Protocol not supported") +E(ESOCKTNOSUPPORT,"Socket type not supported") +E(ENOTSUP, "Not supported") +E(EPFNOSUPPORT, "Protocol family not supported") +E(EAFNOSUPPORT, "Address family not supported by protocol") +E(EADDRNOTAVAIL,"Address not available") +E(ENETDOWN, "Network is down") +E(ENETUNREACH, "Network unreachable") +E(ENETRESET, "Connection reset by network") +E(ECONNABORTED, "Connection aborted") +E(ENOBUFS, "No buffer space available") +E(EISCONN, "Socket is connected") +E(ENOTCONN, "Socket not connected") +E(ESHUTDOWN, "Cannot send after socket shutdown") +E(EALREADY, "Operation already in progress") +E(EINPROGRESS, "Operation in progress") +E(ESTALE, "Stale file handle") +E(EREMOTEIO, "Remote I/O error") +E(EDQUOT, "Quota exceeded") +E(ENOMEDIUM, "No medium found") +E(EMEDIUMTYPE, "Wrong medium type") +E(EMULTIHOP, "Multihop attempted") + +E(0, "No error information") diff --git a/kal/posix/musl_src/errno/strerror.c b/kal/posix/musl_src/errno/strerror.c new file mode 100644 index 00000000..52686102 --- /dev/null +++ b/kal/posix/musl_src/errno/strerror.c @@ -0,0 +1,36 @@ +#include +#include +#include "locale_impl.h" + +#define E(a,b) ((unsigned char)a), +static const unsigned char errid[] = { +#include "__strerror.h" +}; + +#undef E +#define E(a,b) b "\0" +static const char errmsg[] = +#include "__strerror.h" +; + +char *__strerror_l(int e, locale_t loc) +{ + const char *s; + int i; + /* mips has one error code outside of the 8-bit range due to a + * historical typo, so we just remap it. */ + if (EDQUOT==1133) { + if (e==109) e=-1; + else if (e==EDQUOT) e=109; + } + for (i=0; errid[i] && errid[i] != e; i++); + for (s=errmsg; i; s++, i--) for (; *s; s++); + return (char *)s; +} + +char *strerror(int e) +{ + return __strerror_l(e, NULL); +} + +weak_alias(__strerror_l, strerror_l); diff --git a/kal/posix/musl_src/exit/abort.c b/kal/posix/musl_src/exit/abort.c new file mode 100644 index 00000000..de863002 --- /dev/null +++ b/kal/posix/musl_src/exit/abort.c @@ -0,0 +1,8 @@ +#include "stdlib.h" +#include "string.h" +#include "los_task.h" + +void abort(void) { + LOS_Panic("System was being aborted\n"); + while (1); +} \ No newline at end of file diff --git a/kal/posix/musl_src/exit/assert.c b/kal/posix/musl_src/exit/assert.c new file mode 100644 index 00000000..828810e4 --- /dev/null +++ b/kal/posix/musl_src/exit/assert.c @@ -0,0 +1,29 @@ +#include "assert.h" +#include "los_context.h" +#include "los_debug.h" + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +void __assert(const char* file, int line, const char* expr) { + PRINT_ERR("__assert error: %s, %d, %s\n", file, line, expr); + LOS_Panic("__assert error:\n"); + while (1); +} + +void __assert2(const char* file, int line, const char* func, const char* expr) { + PRINT_ERR("%s:%d: %s: assertion \"%s\" failed\n", file, line, func, expr); + LOS_Panic("__assert error:\n"); + while (1); +} + +void __assert_fail(const char* expr, const char* file, int line, const char* func) { + PRINT_ERR("%s:%d: %s: assertion \"%s\" failed\n", file, line, func, expr); + LOS_Panic("__assert error:\n"); + while (1); +} + +#ifdef __cplusplus +} +#endif /* __cplusplus */ \ No newline at end of file diff --git a/kal/posix/musl_src/include/features.h b/kal/posix/musl_src/include/features.h new file mode 100644 index 00000000..53ace76c --- /dev/null +++ b/kal/posix/musl_src/include/features.h @@ -0,0 +1,16 @@ +#ifndef FEATURES_H +#define FEATURES_H + +#include "../../include/features.h" + +#define weak __attribute__((__weak__)) +#define hidden __attribute__((__visibility__("hidden"))) +#ifndef weak_alias +#define weak_alias(old, new) \ + extern __typeof(old) new __attribute__((__weak__, __alias__(#old))) +#endif +#ifndef strong_alias +#define strong_alias(old, new) \ + extern __typeof(old) new __attribute__((__alias__(#old))) +#endif +#endif diff --git a/kal/posix/musl_src/include/string.h b/kal/posix/musl_src/include/string.h new file mode 100644 index 00000000..fe8eac90 --- /dev/null +++ b/kal/posix/musl_src/include/string.h @@ -0,0 +1,12 @@ +#ifndef STRING_H +#define STRING_H + +#include "features.h" +#include "../../include/string.h" + +hidden void *__memrchr(const void *, int, size_t); +hidden char *__stpcpy(char *, const char *); +hidden char *__stpncpy(char *, const char *, size_t); +hidden char *__strchrnul(const char *, int); + +#endif diff --git a/kal/posix/musl_src/internal/intscan.c b/kal/posix/musl_src/internal/intscan.c new file mode 100644 index 00000000..a4a5ae86 --- /dev/null +++ b/kal/posix/musl_src/internal/intscan.c @@ -0,0 +1,100 @@ +#include +#include +#include +#include "shgetc.h" + +/* Lookup table for digit values. -1==255>=36 -> invalid */ +static const unsigned char table[] = { -1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,-1,-1,-1,-1,-1,-1, +-1,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24, +25,26,27,28,29,30,31,32,33,34,35,-1,-1,-1,-1,-1, +-1,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24, +25,26,27,28,29,30,31,32,33,34,35,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +}; + +unsigned long long __intscan(FILE *f, unsigned base, int pok, unsigned long long lim) +{ + const unsigned char *val = table+1; + int c, neg=0; + unsigned x; + unsigned long long y; + if (base > 36 || base == 1) { + errno = EINVAL; + return 0; + } + while (isspace((c=shgetc(f)))); + if (c=='+' || c=='-') { + neg = -(c=='-'); + c = shgetc(f); + } + if ((base == 0 || base == 16) && c=='0') { + c = shgetc(f); + if ((c|32)=='x') { + c = shgetc(f); + if (val[c]>=16) { + shunget(f); + if (pok) shunget(f); + else shlim(f, 0); + return 0; + } + base = 16; + } else if (base == 0) { + base = 8; + } + } else { + if (base == 0) base = 10; + if (val[c] >= base) { + shunget(f); + shlim(f, 0); + errno = EINVAL; + return 0; + } + } + if (base == 10) { + for (x=0; c-'0'<10U && x<=UINT_MAX/10-1; c=shgetc(f)) + x = x*10 + (c-'0'); + for (y=x; c-'0'<10U && y<=ULLONG_MAX/10 && 10*y<=ULLONG_MAX-(c-'0'); c=shgetc(f)) + y = y*10 + (c-'0'); + if (c-'0'>=10U) goto done; + } else if (!(base & base-1)) { + int bs = "\0\1\2\4\7\3\6\5"[(0x17*base)>>5&7]; + for (x=0; val[c]>bs; c=shgetc(f)) + y = y<=lim) { + if (!(lim&1) && !neg) { + errno = ERANGE; + return lim-1; + } else if (y>lim) { + errno = ERANGE; + return lim; + } + } + return (y^neg)-neg; +} diff --git a/kal/posix/musl_src/internal/intscan.h b/kal/posix/musl_src/internal/intscan.h new file mode 100644 index 00000000..fa9561fb --- /dev/null +++ b/kal/posix/musl_src/internal/intscan.h @@ -0,0 +1,9 @@ +#ifndef INTSCAN_H +#define INTSCAN_H + +#include +#include "../include/features.h" + +hidden unsigned long long __intscan(FILE *, unsigned, int, unsigned long long); + +#endif diff --git a/kal/posix/musl_src/internal/locale_impl.h b/kal/posix/musl_src/internal/locale_impl.h new file mode 100644 index 00000000..c8425658 --- /dev/null +++ b/kal/posix/musl_src/internal/locale_impl.h @@ -0,0 +1,43 @@ +#ifndef _LOCALE_IMPL_H +#define _LOCALE_IMPL_H + +#include +#include +#include "libc.h" +#include "../include/features.h" + +#define LOCALE_NAME_MAX 23 + +struct __locale_map { + const void *map; + size_t map_size; + char name[LOCALE_NAME_MAX+1]; + const struct __locale_map *next; +}; + +extern hidden const struct __locale_map __c_dot_utf8; +extern hidden const struct __locale_struct __c_locale; +extern hidden const struct __locale_struct __c_dot_utf8_locale; + +hidden const struct __locale_map *__get_locale(int, const char *); +hidden const char *__mo_lookup(const void *, size_t, const char *); +hidden const char *__lctrans(const char *, const struct __locale_map *); +hidden const char *__lctrans_cur(const char *); +hidden const char *__lctrans_impl(const char *, const struct __locale_map *); +hidden int __loc_is_allocated(locale_t); +hidden char *__gettextdomain(void); + + +#define LOC_MAP_FAILED ((const struct __locale_map *)-1) + +#define LCTRANS(msg, lc, loc) __lctrans(msg, (loc)->cat[(lc)]) +#define LCTRANS_CUR(msg) __lctrans_cur(msg) + +#define C_LOCALE ((locale_t)&__c_locale) +#define UTF8_LOCALE ((locale_t)&__c_dot_utf8_locale) + + +#undef MB_CUR_MAX +#define MB_CUR_MAX (CURRENT_UTF8 ? 4 : 1) + +#endif diff --git a/kal/posix/musl_src/internal/shgetc.c b/kal/posix/musl_src/internal/shgetc.c new file mode 100644 index 00000000..a4a9c633 --- /dev/null +++ b/kal/posix/musl_src/internal/shgetc.c @@ -0,0 +1,37 @@ +#include "shgetc.h" + +/* The shcnt field stores the number of bytes read so far, offset by + * the value of buf-rpos at the last function call (__shlim or __shgetc), + * so that between calls the inline shcnt macro can add rpos-buf to get + * the actual count. */ + +void __shlim(FILE *f, off_t lim) +{ + f->shlim = lim; + f->shcnt = f->buf - f->rpos; + /* If lim is nonzero, rend must be a valid pointer. */ + if (lim && f->rend - f->rpos > lim) + f->shend = f->rpos + lim; + else + f->shend = f->rend; +} + +int __shgetc(FILE *f) +{ + int c; + off_t cnt = shcnt(f); + if (f->shlim && cnt >= f->shlim || (c=__uflow(f)) < 0) { + f->shcnt = f->buf - f->rpos + cnt; + f->shend = f->rpos; + f->shlim = -1; + return EOF; + } + cnt++; + if (f->shlim && f->rend - f->rpos > f->shlim - cnt) + f->shend = f->rpos + (f->shlim - cnt); + else + f->shend = f->rend; + f->shcnt = f->buf - f->rpos + cnt; + if (f->rpos[-1] != c) f->rpos[-1] = c; + return c; +} diff --git a/kal/posix/musl_src/internal/shgetc.h b/kal/posix/musl_src/internal/shgetc.h new file mode 100644 index 00000000..e5d9b70c --- /dev/null +++ b/kal/posix/musl_src/internal/shgetc.h @@ -0,0 +1,33 @@ +#include "stdio_impl.h" +#include "../include/features.h" + +/* Scan helper "stdio" functions for use by scanf-family and strto*-family + * functions. These accept either a valid stdio FILE, or a minimal pseudo + * FILE whose buffer pointers point into a null-terminated string. In the + * latter case, the sh_fromstring macro should be used to setup the FILE; + * the rest of the structure can be left uninitialized. + * + * To begin using these functions, shlim must first be called on the FILE + * to set a field width limit, or 0 for no limit. For string pseudo-FILEs, + * a nonzero limit is not valid and produces undefined behavior. After that, + * shgetc, shunget, and shcnt are valid as long as no other stdio functions + * are called on the stream. + * + * When used with a real FILE object, shunget has only one byte of pushback + * available. Further shunget (up to a limit of the stdio UNGET buffer size) + * will adjust the position but will not restore the data to be read again. + * This functionality is needed for the wcsto*-family functions, where it's + * okay because the FILE will be discarded immediately anyway. When used + * with string pseudo-FILEs, shunget has unlimited pushback, back to the + * beginning of the string. */ + +hidden void __shlim(FILE *, off_t); +hidden int __shgetc(FILE *); + +#define shcnt(f) ((f)->shcnt + ((f)->rpos - (f)->buf)) +#define shlim(f, lim) __shlim((f), (lim)) +#define shgetc(f) (((f)->rpos != (f)->shend) ? *(f)->rpos++ : __shgetc(f)) +#define shunget(f) ((f)->shlim>=0 ? (void)(f)->rpos-- : (void)0) + +#define sh_fromstring(f, s) \ + ((f)->buf = (f)->rpos = (void *)(s), (f)->rend = (void*)-1) diff --git a/kal/posix/musl_src/internal/stdio_impl.h b/kal/posix/musl_src/internal/stdio_impl.h new file mode 100644 index 00000000..8b32cbd3 --- /dev/null +++ b/kal/posix/musl_src/internal/stdio_impl.h @@ -0,0 +1,87 @@ +#ifndef _STDIO_IMPL_H +#define _STDIO_IMPL_H + +#define __NEED_struct__IO_FILE + +#include +#include "../include/features.h" + +#define UNGET 8 + +#define FLOCK(f) __lockfile(f) +#define FUNLOCK(f) __unlockfile(f) + +#define F_PERM 1 +#define F_NORD 4 +#define F_NOWR 8 +#define F_EOF 16 +#define F_ERR 32 +#define F_SVB 64 +#define F_APP 128 + +#define PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP PTHREAD_MUTEX_INITIALIZER +extern hidden FILE *volatile __stdin_used; +extern hidden FILE *volatile __stdout_used; +extern hidden FILE *volatile __stderr_used; + +hidden int __lockfile(FILE *); +hidden int __unlockfile(FILE *); + +hidden size_t __stdio_read(FILE *, unsigned char *, size_t); +hidden size_t __stdio_write(FILE *, const unsigned char *, size_t); +hidden size_t __stdout_write(FILE *, const unsigned char *, size_t); +hidden off_t __stdio_seek(FILE *, off_t, int); +hidden int __stdio_close(FILE *); + +hidden size_t __string_read(FILE *, unsigned char *, size_t); + +hidden int __toread(FILE *); +hidden int __towrite(FILE *); + +hidden void __stdio_exit(void); +hidden void __stdio_exit_needed(void); + +#if defined(__PIC__) && (100*__GNUC__+__GNUC_MINOR__ >= 303) +__attribute__((visibility("protected"))) +#endif +int __overflow(FILE *, int), __uflow(FILE *); + +hidden int __fseeko(FILE *, off_t, int); +hidden int __fseeko_unlocked(FILE *, off_t, int); +hidden off_t __ftello(FILE *); +hidden off_t __ftello_unlocked(FILE *); +hidden size_t __fwritex(const unsigned char *, size_t, FILE *); +hidden int __putc_unlocked(int, FILE *); + +hidden FILE *__fdopen(int, const char *); +hidden int __fmodeflags(const char *); + +hidden FILE *__ofl_add(FILE *f); +hidden FILE **__ofl_lock(void); +hidden void __ofl_unlock(void); + +struct __pthread; +hidden void __register_locked_file(FILE *, struct __pthread *); +hidden void __unlist_locked_file(FILE *); +hidden void __do_orphaned_stdio_locks(void); + +#define MAYBE_WAITERS 0x40000000 + +hidden void __getopt_msg(const char *, const char *, const char *, size_t); + +#define feof(f) ((f)->flags & F_EOF) +#define ferror(f) ((f)->flags & F_ERR) + +#define getc_unlocked(f) \ + ( ((f)->rpos != (f)->rend) ? *(f)->rpos++ : __uflow((f)) ) + +#define putc_unlocked(c, f) \ + ( (((unsigned char)(c)!=(f)->lbf && (f)->wpos!=(f)->wend)) \ + ? *(f)->wpos++ = (unsigned char)(c) \ + : __overflow((f),(unsigned char)(c)) ) + +/* Caller-allocated FILE * operations */ +hidden FILE *__fopen_rb_ca(const char *, FILE *, unsigned char *, size_t); +hidden int __fclose_ca(FILE *); + +#endif diff --git a/kal/posix/musl_src/math/abs.c b/kal/posix/musl_src/math/abs.c new file mode 100644 index 00000000..e721fdc2 --- /dev/null +++ b/kal/posix/musl_src/math/abs.c @@ -0,0 +1,6 @@ +#include + +int abs(int a) +{ + return a>0 ? a : -a; +} diff --git a/kal/posix/musl_src/math/exp_data.c b/kal/posix/musl_src/math/exp_data.c new file mode 100644 index 00000000..2d7faf9f --- /dev/null +++ b/kal/posix/musl_src/math/exp_data.c @@ -0,0 +1,182 @@ +/* + * Shared data between exp, exp2 and pow. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ + +#include "exp_data.h" + +#define N (1 << EXP_TABLE_BITS) + +const struct exp_data __exp_data1 = { +// N/ln2 +.invln2N = 0x1.71547652b82fep0 * N, +// -ln2/N +.negln2hiN = -0x1.62e42fefa0000p-8, +.negln2loN = -0x1.cf79abc9e3b3ap-47, +// Used for rounding when !TOINT_INTRINSICS +#if EXP_USE_TOINT_NARROW +.shift = 0x1800000000.8p0, +#else +.shift = 0x1.8p52, +#endif +// exp polynomial coefficients. +.poly = { +// abs error: 1.555*2^-66 +// ulp error: 0.509 (0.511 without fma) +// if |x| < ln2/256+eps +// abs error if |x| < ln2/256+0x1p-15: 1.09*2^-65 +// abs error if |x| < ln2/128: 1.7145*2^-56 +0x1.ffffffffffdbdp-2, +0x1.555555555543cp-3, +0x1.55555cf172b91p-5, +0x1.1111167a4d017p-7, +}, +.exp2_shift = 0x1.8p52 / N, +// exp2 polynomial coefficients. +.exp2_poly = { +// abs error: 1.2195*2^-65 +// ulp error: 0.507 (0.511 without fma) +// if |x| < 1/256 +// abs error if |x| < 1/128: 1.9941*2^-56 +0x1.62e42fefa39efp-1, +0x1.ebfbdff82c424p-3, +0x1.c6b08d70cf4b5p-5, +0x1.3b2abd24650ccp-7, +0x1.5d7e09b4e3a84p-10, +}, +// 2^(k/N) ~= H[k]*(1 + T[k]) for int k in [0,N) +// tab[2*k] = asuint64(T[k]) +// tab[2*k+1] = asuint64(H[k]) - (k << 52)/N +.tab = { +0x0, 0x3ff0000000000000, +0x3c9b3b4f1a88bf6e, 0x3feff63da9fb3335, +0xbc7160139cd8dc5d, 0x3fefec9a3e778061, +0xbc905e7a108766d1, 0x3fefe315e86e7f85, +0x3c8cd2523567f613, 0x3fefd9b0d3158574, +0xbc8bce8023f98efa, 0x3fefd06b29ddf6de, +0x3c60f74e61e6c861, 0x3fefc74518759bc8, +0x3c90a3e45b33d399, 0x3fefbe3ecac6f383, +0x3c979aa65d837b6d, 0x3fefb5586cf9890f, +0x3c8eb51a92fdeffc, 0x3fefac922b7247f7, +0x3c3ebe3d702f9cd1, 0x3fefa3ec32d3d1a2, +0xbc6a033489906e0b, 0x3fef9b66affed31b, +0xbc9556522a2fbd0e, 0x3fef9301d0125b51, +0xbc5080ef8c4eea55, 0x3fef8abdc06c31cc, +0xbc91c923b9d5f416, 0x3fef829aaea92de0, +0x3c80d3e3e95c55af, 0x3fef7a98c8a58e51, +0xbc801b15eaa59348, 0x3fef72b83c7d517b, +0xbc8f1ff055de323d, 0x3fef6af9388c8dea, +0x3c8b898c3f1353bf, 0x3fef635beb6fcb75, +0xbc96d99c7611eb26, 0x3fef5be084045cd4, +0x3c9aecf73e3a2f60, 0x3fef54873168b9aa, +0xbc8fe782cb86389d, 0x3fef4d5022fcd91d, +0x3c8a6f4144a6c38d, 0x3fef463b88628cd6, +0x3c807a05b0e4047d, 0x3fef3f49917ddc96, +0x3c968efde3a8a894, 0x3fef387a6e756238, +0x3c875e18f274487d, 0x3fef31ce4fb2a63f, +0x3c80472b981fe7f2, 0x3fef2b4565e27cdd, +0xbc96b87b3f71085e, 0x3fef24dfe1f56381, +0x3c82f7e16d09ab31, 0x3fef1e9df51fdee1, +0xbc3d219b1a6fbffa, 0x3fef187fd0dad990, +0x3c8b3782720c0ab4, 0x3fef1285a6e4030b, +0x3c6e149289cecb8f, 0x3fef0cafa93e2f56, +0x3c834d754db0abb6, 0x3fef06fe0a31b715, +0x3c864201e2ac744c, 0x3fef0170fc4cd831, +0x3c8fdd395dd3f84a, 0x3feefc08b26416ff, +0xbc86a3803b8e5b04, 0x3feef6c55f929ff1, +0xbc924aedcc4b5068, 0x3feef1a7373aa9cb, +0xbc9907f81b512d8e, 0x3feeecae6d05d866, +0xbc71d1e83e9436d2, 0x3feee7db34e59ff7, +0xbc991919b3ce1b15, 0x3feee32dc313a8e5, +0x3c859f48a72a4c6d, 0x3feedea64c123422, +0xbc9312607a28698a, 0x3feeda4504ac801c, +0xbc58a78f4817895b, 0x3feed60a21f72e2a, +0xbc7c2c9b67499a1b, 0x3feed1f5d950a897, +0x3c4363ed60c2ac11, 0x3feece086061892d, +0x3c9666093b0664ef, 0x3feeca41ed1d0057, +0x3c6ecce1daa10379, 0x3feec6a2b5c13cd0, +0x3c93ff8e3f0f1230, 0x3feec32af0d7d3de, +0x3c7690cebb7aafb0, 0x3feebfdad5362a27, +0x3c931dbdeb54e077, 0x3feebcb299fddd0d, +0xbc8f94340071a38e, 0x3feeb9b2769d2ca7, +0xbc87deccdc93a349, 0x3feeb6daa2cf6642, +0xbc78dec6bd0f385f, 0x3feeb42b569d4f82, +0xbc861246ec7b5cf6, 0x3feeb1a4ca5d920f, +0x3c93350518fdd78e, 0x3feeaf4736b527da, +0x3c7b98b72f8a9b05, 0x3feead12d497c7fd, +0x3c9063e1e21c5409, 0x3feeab07dd485429, +0x3c34c7855019c6ea, 0x3feea9268a5946b7, +0x3c9432e62b64c035, 0x3feea76f15ad2148, +0xbc8ce44a6199769f, 0x3feea5e1b976dc09, +0xbc8c33c53bef4da8, 0x3feea47eb03a5585, +0xbc845378892be9ae, 0x3feea34634ccc320, +0xbc93cedd78565858, 0x3feea23882552225, +0x3c5710aa807e1964, 0x3feea155d44ca973, +0xbc93b3efbf5e2228, 0x3feea09e667f3bcd, +0xbc6a12ad8734b982, 0x3feea012750bdabf, +0xbc6367efb86da9ee, 0x3fee9fb23c651a2f, +0xbc80dc3d54e08851, 0x3fee9f7df9519484, +0xbc781f647e5a3ecf, 0x3fee9f75e8ec5f74, +0xbc86ee4ac08b7db0, 0x3fee9f9a48a58174, +0xbc8619321e55e68a, 0x3fee9feb564267c9, +0x3c909ccb5e09d4d3, 0x3feea0694fde5d3f, +0xbc7b32dcb94da51d, 0x3feea11473eb0187, +0x3c94ecfd5467c06b, 0x3feea1ed0130c132, +0x3c65ebe1abd66c55, 0x3feea2f336cf4e62, +0xbc88a1c52fb3cf42, 0x3feea427543e1a12, +0xbc9369b6f13b3734, 0x3feea589994cce13, +0xbc805e843a19ff1e, 0x3feea71a4623c7ad, +0xbc94d450d872576e, 0x3feea8d99b4492ed, +0x3c90ad675b0e8a00, 0x3feeaac7d98a6699, +0x3c8db72fc1f0eab4, 0x3feeace5422aa0db, +0xbc65b6609cc5e7ff, 0x3feeaf3216b5448c, +0x3c7bf68359f35f44, 0x3feeb1ae99157736, +0xbc93091fa71e3d83, 0x3feeb45b0b91ffc6, +0xbc5da9b88b6c1e29, 0x3feeb737b0cdc5e5, +0xbc6c23f97c90b959, 0x3feeba44cbc8520f, +0xbc92434322f4f9aa, 0x3feebd829fde4e50, +0xbc85ca6cd7668e4b, 0x3feec0f170ca07ba, +0x3c71affc2b91ce27, 0x3feec49182a3f090, +0x3c6dd235e10a73bb, 0x3feec86319e32323, +0xbc87c50422622263, 0x3feecc667b5de565, +0x3c8b1c86e3e231d5, 0x3feed09bec4a2d33, +0xbc91bbd1d3bcbb15, 0x3feed503b23e255d, +0x3c90cc319cee31d2, 0x3feed99e1330b358, +0x3c8469846e735ab3, 0x3feede6b5579fdbf, +0xbc82dfcd978e9db4, 0x3feee36bbfd3f37a, +0x3c8c1a7792cb3387, 0x3feee89f995ad3ad, +0xbc907b8f4ad1d9fa, 0x3feeee07298db666, +0xbc55c3d956dcaeba, 0x3feef3a2b84f15fb, +0xbc90a40e3da6f640, 0x3feef9728de5593a, +0xbc68d6f438ad9334, 0x3feeff76f2fb5e47, +0xbc91eee26b588a35, 0x3fef05b030a1064a, +0x3c74ffd70a5fddcd, 0x3fef0c1e904bc1d2, +0xbc91bdfbfa9298ac, 0x3fef12c25bd71e09, +0x3c736eae30af0cb3, 0x3fef199bdd85529c, +0x3c8ee3325c9ffd94, 0x3fef20ab5fffd07a, +0x3c84e08fd10959ac, 0x3fef27f12e57d14b, +0x3c63cdaf384e1a67, 0x3fef2f6d9406e7b5, +0x3c676b2c6c921968, 0x3fef3720dcef9069, +0xbc808a1883ccb5d2, 0x3fef3f0b555dc3fa, +0xbc8fad5d3ffffa6f, 0x3fef472d4a07897c, +0xbc900dae3875a949, 0x3fef4f87080d89f2, +0x3c74a385a63d07a7, 0x3fef5818dcfba487, +0xbc82919e2040220f, 0x3fef60e316c98398, +0x3c8e5a50d5c192ac, 0x3fef69e603db3285, +0x3c843a59ac016b4b, 0x3fef7321f301b460, +0xbc82d52107b43e1f, 0x3fef7c97337b9b5f, +0xbc892ab93b470dc9, 0x3fef864614f5a129, +0x3c74b604603a88d3, 0x3fef902ee78b3ff6, +0x3c83c5ec519d7271, 0x3fef9a51fbc74c83, +0xbc8ff7128fd391f0, 0x3fefa4afa2a490da, +0xbc8dae98e223747d, 0x3fefaf482d8e67f1, +0x3c8ec3bc41aa2008, 0x3fefba1bee615a27, +0x3c842b94c3a9eb32, 0x3fefc52b376bba97, +0x3c8a64a931d185ee, 0x3fefd0765b6e4540, +0xbc8e37bae43be3ed, 0x3fefdbfdad9cbe14, +0x3c77893b4d91cd9d, 0x3fefe7c1819e90d8, +0x3c5305c14160cc89, 0x3feff3c22b8f71f1, +}, +}; diff --git a/kal/posix/musl_src/math/exp_data.h b/kal/posix/musl_src/math/exp_data.h new file mode 100644 index 00000000..968ad760 --- /dev/null +++ b/kal/posix/musl_src/math/exp_data.h @@ -0,0 +1,26 @@ +/* + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ +#ifndef _EXP_DATA_H +#define _EXP_DATA_H + +#include +#include + +#define EXP_TABLE_BITS 7 +#define EXP_POLY_ORDER 5 +#define EXP_USE_TOINT_NARROW 0 +#define EXP2_POLY_ORDER 5 +extern const struct exp_data { + double invln2N; + double shift; + double negln2hiN; + double negln2loN; + double poly[4]; /* Last four coefficients. */ + double exp2_shift; + double exp2_poly[EXP2_POLY_ORDER]; + uint64_t tab[2*(1 << EXP_TABLE_BITS)]; +} __exp_data1; + +#endif diff --git a/kal/posix/musl_src/math/libm.h b/kal/posix/musl_src/math/libm.h new file mode 100644 index 00000000..a0983612 --- /dev/null +++ b/kal/posix/musl_src/math/libm.h @@ -0,0 +1,272 @@ +#ifndef _LIBM_H +#define _LIBM_H + +#include +#include +#include +#include +#include "../include/features.h" + + +#if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024 +#elif LDBL_MANT_DIG == 64 && LDBL_MAX_EXP == 16384 && __BYTE_ORDER == __LITTLE_ENDIAN +union ldshape { + long double f; + struct { + uint64_t m; + uint16_t se; + } i; +}; +#elif LDBL_MANT_DIG == 64 && LDBL_MAX_EXP == 16384 && __BYTE_ORDER == __BIG_ENDIAN +/* This is the m68k variant of 80-bit long double, and this definition only works + * on archs where the alignment requirement of uint64_t is <= 4. */ +union ldshape { + long double f; + struct { + uint16_t se; + uint16_t pad; + uint64_t m; + } i; +}; +#elif LDBL_MANT_DIG == 113 && LDBL_MAX_EXP == 16384 && __BYTE_ORDER == __LITTLE_ENDIAN +union ldshape { + long double f; + struct { + uint64_t lo; + uint32_t mid; + uint16_t top; + uint16_t se; + } i; + struct { + uint64_t lo; + uint64_t hi; + } i2; +}; +#elif LDBL_MANT_DIG == 113 && LDBL_MAX_EXP == 16384 && __BYTE_ORDER == __BIG_ENDIAN +union ldshape { + long double f; + struct { + uint16_t se; + uint16_t top; + uint32_t mid; + uint64_t lo; + } i; + struct { + uint64_t hi; + uint64_t lo; + } i2; +}; +#else +#error Unsupported long double representation +#endif + +/* Support non-nearest rounding mode. */ +#define WANT_ROUNDING 1 +/* Support signaling NaNs. */ +#define WANT_SNAN 0 + +#if WANT_SNAN +#error SNaN is unsupported +#else +#define issignalingf_inline(x) 0 +#define issignaling_inline(x) 0 +#endif + +#ifndef TOINT_INTRINSICS +#define TOINT_INTRINSICS 0 +#endif + +#if TOINT_INTRINSICS +/* Round x to nearest int in all rounding modes, ties have to be rounded + consistently with converttoint so the results match. If the result + would be outside of [-2^31, 2^31-1] then the semantics is unspecified. */ +static double_t roundtoint(double_t); + +/* Convert x to nearest int in all rounding modes, ties have to be rounded + consistently with roundtoint. If the result is not representible in an + int32_t then the semantics is unspecified. */ +static int32_t converttoint(double_t); +#endif + +/* Helps static branch prediction so hot path can be better optimized. */ +#ifdef __GNUC__ +#define predict_true(x) __builtin_expect(!!(x), 1) +#define predict_false(x) __builtin_expect(x, 0) +#else +#define predict_true(x) (x) +#define predict_false(x) (x) +#endif + +/* Evaluate an expression as the specified type. With standard excess + precision handling a type cast or assignment is enough (with + -ffloat-store an assignment is required, in old compilers argument + passing and return statement may not drop excess precision). */ + +static inline float eval_as_float(float x) +{ + float y = x; + return y; +} + +static inline double eval_as_double(double x) +{ + double y = x; + return y; +} + +/* fp_barrier returns its input, but limits code transformations + as if it had a side-effect (e.g. observable io) and returned + an arbitrary value. */ + +#ifndef fp_barrierf +#define fp_barrierf fp_barrierf +static inline float fp_barrierf(float x) +{ + volatile float y = x; + return y; +} +#endif + +#ifndef fp_barrier +#define fp_barrier fp_barrier +static inline double fp_barrier(double x) +{ + volatile double y = x; + return y; +} +#endif + +#ifndef fp_barrierl +#define fp_barrierl fp_barrierl +static inline long double fp_barrierl(long double x) +{ + volatile long double y = x; + return y; +} +#endif + +/* fp_force_eval ensures that the input value is computed when that's + otherwise unused. To prevent the constant folding of the input + expression, an additional fp_barrier may be needed or a compilation + mode that does so (e.g. -frounding-math in gcc). Then it can be + used to evaluate an expression for its fenv side-effects only. */ + +#ifndef fp_force_evalf +#define fp_force_evalf fp_force_evalf +static inline void fp_force_evalf(float x) +{ + volatile float y; + y = x; +} +#endif + +#ifndef fp_force_eval +#define fp_force_eval fp_force_eval +static inline void fp_force_eval(double x) +{ + volatile double y; + y = x; +} +#endif + +#ifndef fp_force_evall +#define fp_force_evall fp_force_evall +static inline void fp_force_evall(long double x) +{ + volatile long double y; + y = x; +} +#endif + +#define FORCE_EVAL(x) do { \ + if (sizeof(x) == sizeof(float)) { \ + fp_force_evalf(x); \ + } else if (sizeof(x) == sizeof(double)) { \ + fp_force_eval(x); \ + } else { \ + fp_force_evall(x); \ + } \ +} while(0) + +#define asuint(f) ((union{float _f; uint32_t _i;}){f})._i +#define asfloat(i) ((union{uint32_t _i; float _f;}){i})._f +#define asuint64(f) ((union{double _f; uint64_t _i;}){f})._i +#define asdouble(i) ((union{uint64_t _i; double _f;}){i})._f + +#define EXTRACT_WORDS(hi,lo,d) \ +do { \ + uint64_t __u = asuint64(d); \ + (hi) = __u >> 32; \ + (lo) = (uint32_t)__u; \ +} while (0) + +#define GET_HIGH_WORD(hi,d) \ +do { \ + (hi) = asuint64(d) >> 32; \ +} while (0) + +#define GET_LOW_WORD(lo,d) \ +do { \ + (lo) = (uint32_t)asuint64(d); \ +} while (0) + +#define INSERT_WORDS(d,hi,lo) \ +do { \ + (d) = asdouble(((uint64_t)(hi)<<32) | (uint32_t)(lo)); \ +} while (0) + +#define SET_HIGH_WORD(d,hi) \ + INSERT_WORDS(d, hi, (uint32_t)asuint64(d)) + +#define SET_LOW_WORD(d,lo) \ + INSERT_WORDS(d, asuint64(d)>>32, lo) + +#define GET_FLOAT_WORD(w,d) \ +do { \ + (w) = asuint(d); \ +} while (0) + +#define SET_FLOAT_WORD(d,w) \ +do { \ + (d) = asfloat(w); \ +} while (0) + +hidden int __rem_pio2_large(double*,double*,int,int,int); + +hidden int __rem_pio2(double,double*); +hidden double __sin(double,double,int); +hidden double __cos(double,double); +hidden double __tan(double,double,int); +hidden double __expo2(double); + +hidden int __rem_pio2f(float,double*); +hidden float __sindf(double); +hidden float __cosdf(double); +hidden float __tandf(double,int); +hidden float __expo2f(float); + +hidden int __rem_pio2l(long double, long double *); +hidden long double __sinl(long double, long double, int); +hidden long double __cosl(long double, long double); +hidden long double __tanl(long double, long double, int); + +hidden long double __polevll(long double, const long double *, int); +hidden long double __p1evll(long double, const long double *, int); + +extern int __signgam; +hidden double __lgamma_r(double, int *); +hidden float __lgammaf_r(float, int *); + +/* error handling functions */ +hidden float __math_xflowf(uint32_t, float); +hidden float __math_uflowf(uint32_t); +hidden float __math_oflowf(uint32_t); +hidden float __math_divzerof(uint32_t); +hidden float __math_invalidf(float); +hidden double __math_xflow(uint32_t, double); +hidden double __math_uflow(uint32_t); +hidden double __math_oflow(uint32_t); +hidden double __math_divzero(uint32_t); +hidden double __math_invalid(double); + +#endif diff --git a/kal/posix/musl_src/math/log.c b/kal/posix/musl_src/math/log.c new file mode 100644 index 00000000..4024f828 --- /dev/null +++ b/kal/posix/musl_src/math/log.c @@ -0,0 +1,127 @@ +/* + * Double-precision log(x) function. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ + +#include +#include +#include "libm.h" +#include "log_data.h" + +#define T __log_data.tab +#define T2 __log_data.tab2 +#define B __log_data.poly1 +#define A __log_data.poly +#define Ln2hi __log_data.ln2hi +#define Ln2lo __log_data.ln2lo +#define N (1 << LOG_TABLE_BITS) +#define OFF 0x3fe6000000000000 + +#ifdef NEED_MATH_DIVZERO +/* base math internal func */ +double __math_divzero(uint32_t sign) +{ + return fp_barrier(sign ? -1.0 : 1.0) / 0.0; +} +#endif + +#ifdef NEED_MATH_INVALID +double __math_invalid(double x) +{ + return (x - x) / (x - x); +} +#endif + +/* Top 16 bits of a double. */ +static inline uint32_t top16(double x) +{ + return asuint64(x) >> 48; +} + +double log(double x) +{ + double_t w, z, r, r2, r3, y, invc, logc, kd, hi, lo; + uint64_t ix, iz, tmp; + uint32_t top; + int k, i; + + ix = asuint64(x); + top = top16(x); +#define LO asuint64(1.0 - 0x1p-4) +#define HI asuint64(1.0 + 0x1.09p-4) + if (predict_false(ix - LO < HI - LO)) { + /* Handle close to 1.0 inputs separately. */ + /* Fix sign of zero with downward rounding when x==1. */ + if (WANT_ROUNDING && predict_false(ix == asuint64(1.0))) + return 0; + r = x - 1.0; + r2 = r * r; + r3 = r * r2; + y = r3 * + (B[1] + r * B[2] + r2 * B[3] + + r3 * (B[4] + r * B[5] + r2 * B[6] + + r3 * (B[7] + r * B[8] + r2 * B[9] + r3 * B[10]))); + /* Worst-case error is around 0.507 ULP. */ + w = r * 0x1p27; + double_t rhi = r + w - w; + double_t rlo = r - rhi; + w = rhi * rhi * B[0]; /* B[0] == -0.5. */ + hi = r + w; + lo = r - hi + w; + lo += B[0] * rlo * (rhi + r); + y += lo; + y += hi; + return eval_as_double(y); + } + if (predict_false(top - 0x0010 >= 0x7ff0 - 0x0010)) { + /* x < 0x1p-1022 or inf or nan. */ + if (ix * 2 == 0) + return __math_divzero(1); + if (ix == asuint64(INFINITY)) /* log(inf) == inf. */ + return x; + if ((top & 0x8000) || (top & 0x7ff0) == 0x7ff0) + return __math_invalid(x); + /* x is subnormal, normalize it. */ + ix = asuint64(x * 0x1p52); + ix -= 52ULL << 52; + } + + /* x = 2^k z; where z is in range [OFF,2*OFF) and exact. + The range is split into N subintervals. + The ith subinterval contains z and c is near its center. */ + tmp = ix - OFF; + i = (tmp >> (52 - LOG_TABLE_BITS)) % N; + k = (int64_t)tmp >> 52; /* arithmetic shift */ + iz = ix - (tmp & 0xfffULL << 52); + invc = T[i].invc; + logc = T[i].logc; + z = asdouble(iz); + + /* log(x) = log1p(z/c-1) + log(c) + k*Ln2. */ + /* r ~= z/c - 1, |r| < 1/(2*N). */ +#if __FP_FAST_FMA + /* rounding error: 0x1p-55/N. */ + r = __builtin_fma(z, invc, -1.0); +#else + /* rounding error: 0x1p-55/N + 0x1p-66. */ + r = (z - T2[i].chi - T2[i].clo) * invc; +#endif + kd = (double_t)k; + + /* hi + lo = r + log(c) + k*Ln2. */ + w = kd * Ln2hi + logc; + hi = w + r; + lo = w - hi + r + kd * Ln2lo; + + /* log(x) = lo + (log1p(r) - r) + hi. */ + r2 = r * r; /* rounding error: 0x1p-54/N^2. */ + /* Worst case error if |y| > 0x1p-5: + 0.5 + 4.13/N + abs-poly-error*2^57 ULP (+ 0.002 ULP without fma) + Worst case error if |y| > 0x1p-4: + 0.5 + 2.06/N + abs-poly-error*2^56 ULP (+ 0.001 ULP without fma). */ + y = lo + r2 * A[0] + + r * r2 * (A[1] + r * A[2] + r2 * (A[3] + r * A[4])) + hi; + return eval_as_double(y); +} diff --git a/kal/posix/musl_src/math/log_data.c b/kal/posix/musl_src/math/log_data.c new file mode 100644 index 00000000..1a6ec712 --- /dev/null +++ b/kal/posix/musl_src/math/log_data.c @@ -0,0 +1,328 @@ +/* + * Data for log. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ + +#include "log_data.h" + +#define N (1 << LOG_TABLE_BITS) + +const struct log_data __log_data = { +.ln2hi = 0x1.62e42fefa3800p-1, +.ln2lo = 0x1.ef35793c76730p-45, +.poly1 = { +// relative error: 0x1.c04d76cp-63 +// in -0x1p-4 0x1.09p-4 (|log(1+x)| > 0x1p-4 outside the interval) +-0x1p-1, +0x1.5555555555577p-2, +-0x1.ffffffffffdcbp-3, +0x1.999999995dd0cp-3, +-0x1.55555556745a7p-3, +0x1.24924a344de3p-3, +-0x1.fffffa4423d65p-4, +0x1.c7184282ad6cap-4, +-0x1.999eb43b068ffp-4, +0x1.78182f7afd085p-4, +-0x1.5521375d145cdp-4, +}, +.poly = { +// relative error: 0x1.926199e8p-56 +// abs error: 0x1.882ff33p-65 +// in -0x1.fp-9 0x1.fp-9 +-0x1.0000000000001p-1, +0x1.555555551305bp-2, +-0x1.fffffffeb459p-3, +0x1.999b324f10111p-3, +-0x1.55575e506c89fp-3, +}, +/* Algorithm: + + x = 2^k z + log(x) = k ln2 + log(c) + log(z/c) + log(z/c) = poly(z/c - 1) + +where z is in [1.6p-1; 1.6p0] which is split into N subintervals and z falls +into the ith one, then table entries are computed as + + tab[i].invc = 1/c + tab[i].logc = (double)log(c) + tab2[i].chi = (double)c + tab2[i].clo = (double)(c - (double)c) + +where c is near the center of the subinterval and is chosen by trying +-2^29 +floating point invc candidates around 1/center and selecting one for which + + 1) the rounding error in 0x1.8p9 + logc is 0, + 2) the rounding error in z - chi - clo is < 0x1p-66 and + 3) the rounding error in (double)log(c) is minimized (< 0x1p-66). + +Note: 1) ensures that k*ln2hi + logc can be computed without rounding error, +2) ensures that z/c - 1 can be computed as (z - chi - clo)*invc with close to +a single rounding error when there is no fast fma for z*invc - 1, 3) ensures +that logc + poly(z/c - 1) has small error, however near x == 1 when +|log(x)| < 0x1p-4, this is not enough so that is special cased. */ +.tab = { +{0x1.734f0c3e0de9fp+0, -0x1.7cc7f79e69000p-2}, +{0x1.713786a2ce91fp+0, -0x1.76feec20d0000p-2}, +{0x1.6f26008fab5a0p+0, -0x1.713e31351e000p-2}, +{0x1.6d1a61f138c7dp+0, -0x1.6b85b38287800p-2}, +{0x1.6b1490bc5b4d1p+0, -0x1.65d5590807800p-2}, +{0x1.69147332f0cbap+0, -0x1.602d076180000p-2}, +{0x1.6719f18224223p+0, -0x1.5a8ca86909000p-2}, +{0x1.6524f99a51ed9p+0, -0x1.54f4356035000p-2}, +{0x1.63356aa8f24c4p+0, -0x1.4f637c36b4000p-2}, +{0x1.614b36b9ddc14p+0, -0x1.49da7fda85000p-2}, +{0x1.5f66452c65c4cp+0, -0x1.445923989a800p-2}, +{0x1.5d867b5912c4fp+0, -0x1.3edf439b0b800p-2}, +{0x1.5babccb5b90dep+0, -0x1.396ce448f7000p-2}, +{0x1.59d61f2d91a78p+0, -0x1.3401e17bda000p-2}, +{0x1.5805612465687p+0, -0x1.2e9e2ef468000p-2}, +{0x1.56397cee76bd3p+0, -0x1.2941b3830e000p-2}, +{0x1.54725e2a77f93p+0, -0x1.23ec58cda8800p-2}, +{0x1.52aff42064583p+0, -0x1.1e9e129279000p-2}, +{0x1.50f22dbb2bddfp+0, -0x1.1956d2b48f800p-2}, +{0x1.4f38f4734ded7p+0, -0x1.141679ab9f800p-2}, +{0x1.4d843cfde2840p+0, -0x1.0edd094ef9800p-2}, +{0x1.4bd3ec078a3c8p+0, -0x1.09aa518db1000p-2}, +{0x1.4a27fc3e0258ap+0, -0x1.047e65263b800p-2}, +{0x1.4880524d48434p+0, -0x1.feb224586f000p-3}, +{0x1.46dce1b192d0bp+0, -0x1.f474a7517b000p-3}, +{0x1.453d9d3391854p+0, -0x1.ea4443d103000p-3}, +{0x1.43a2744b4845ap+0, -0x1.e020d44e9b000p-3}, +{0x1.420b54115f8fbp+0, -0x1.d60a22977f000p-3}, +{0x1.40782da3ef4b1p+0, -0x1.cc00104959000p-3}, +{0x1.3ee8f5d57fe8fp+0, -0x1.c202956891000p-3}, +{0x1.3d5d9a00b4ce9p+0, -0x1.b81178d811000p-3}, +{0x1.3bd60c010c12bp+0, -0x1.ae2c9ccd3d000p-3}, +{0x1.3a5242b75dab8p+0, -0x1.a45402e129000p-3}, +{0x1.38d22cd9fd002p+0, -0x1.9a877681df000p-3}, +{0x1.3755bc5847a1cp+0, -0x1.90c6d69483000p-3}, +{0x1.35dce49ad36e2p+0, -0x1.87120a645c000p-3}, +{0x1.34679984dd440p+0, -0x1.7d68fb4143000p-3}, +{0x1.32f5cceffcb24p+0, -0x1.73cb83c627000p-3}, +{0x1.3187775a10d49p+0, -0x1.6a39a9b376000p-3}, +{0x1.301c8373e3990p+0, -0x1.60b3154b7a000p-3}, +{0x1.2eb4ebb95f841p+0, -0x1.5737d76243000p-3}, +{0x1.2d50a0219a9d1p+0, -0x1.4dc7b8fc23000p-3}, +{0x1.2bef9a8b7fd2ap+0, -0x1.4462c51d20000p-3}, +{0x1.2a91c7a0c1babp+0, -0x1.3b08abc830000p-3}, +{0x1.293726014b530p+0, -0x1.31b996b490000p-3}, +{0x1.27dfa5757a1f5p+0, -0x1.2875490a44000p-3}, +{0x1.268b39b1d3bbfp+0, -0x1.1f3b9f879a000p-3}, +{0x1.2539d838ff5bdp+0, -0x1.160c8252ca000p-3}, +{0x1.23eb7aac9083bp+0, -0x1.0ce7f57f72000p-3}, +{0x1.22a012ba940b6p+0, -0x1.03cdc49fea000p-3}, +{0x1.2157996cc4132p+0, -0x1.f57bdbc4b8000p-4}, +{0x1.201201dd2fc9bp+0, -0x1.e370896404000p-4}, +{0x1.1ecf4494d480bp+0, -0x1.d17983ef94000p-4}, +{0x1.1d8f5528f6569p+0, -0x1.bf9674ed8a000p-4}, +{0x1.1c52311577e7cp+0, -0x1.adc79202f6000p-4}, +{0x1.1b17c74cb26e9p+0, -0x1.9c0c3e7288000p-4}, +{0x1.19e010c2c1ab6p+0, -0x1.8a646b372c000p-4}, +{0x1.18ab07bb670bdp+0, -0x1.78d01b3ac0000p-4}, +{0x1.1778a25efbcb6p+0, -0x1.674f145380000p-4}, +{0x1.1648d354c31dap+0, -0x1.55e0e6d878000p-4}, +{0x1.151b990275fddp+0, -0x1.4485cdea1e000p-4}, +{0x1.13f0ea432d24cp+0, -0x1.333d94d6aa000p-4}, +{0x1.12c8b7210f9dap+0, -0x1.22079f8c56000p-4}, +{0x1.11a3028ecb531p+0, -0x1.10e4698622000p-4}, +{0x1.107fbda8434afp+0, -0x1.ffa6c6ad20000p-5}, +{0x1.0f5ee0f4e6bb3p+0, -0x1.dda8d4a774000p-5}, +{0x1.0e4065d2a9fcep+0, -0x1.bbcece4850000p-5}, +{0x1.0d244632ca521p+0, -0x1.9a1894012c000p-5}, +{0x1.0c0a77ce2981ap+0, -0x1.788583302c000p-5}, +{0x1.0af2f83c636d1p+0, -0x1.5715e67d68000p-5}, +{0x1.09ddb98a01339p+0, -0x1.35c8a49658000p-5}, +{0x1.08cabaf52e7dfp+0, -0x1.149e364154000p-5}, +{0x1.07b9f2f4e28fbp+0, -0x1.e72c082eb8000p-6}, +{0x1.06ab58c358f19p+0, -0x1.a55f152528000p-6}, +{0x1.059eea5ecf92cp+0, -0x1.63d62cf818000p-6}, +{0x1.04949cdd12c90p+0, -0x1.228fb8caa0000p-6}, +{0x1.038c6c6f0ada9p+0, -0x1.c317b20f90000p-7}, +{0x1.02865137932a9p+0, -0x1.419355daa0000p-7}, +{0x1.0182427ea7348p+0, -0x1.81203c2ec0000p-8}, +{0x1.008040614b195p+0, -0x1.0040979240000p-9}, +{0x1.fe01ff726fa1ap-1, 0x1.feff384900000p-9}, +{0x1.fa11cc261ea74p-1, 0x1.7dc41353d0000p-7}, +{0x1.f6310b081992ep-1, 0x1.3cea3c4c28000p-6}, +{0x1.f25f63ceeadcdp-1, 0x1.b9fc114890000p-6}, +{0x1.ee9c8039113e7p-1, 0x1.1b0d8ce110000p-5}, +{0x1.eae8078cbb1abp-1, 0x1.58a5bd001c000p-5}, +{0x1.e741aa29d0c9bp-1, 0x1.95c8340d88000p-5}, +{0x1.e3a91830a99b5p-1, 0x1.d276aef578000p-5}, +{0x1.e01e009609a56p-1, 0x1.07598e598c000p-4}, +{0x1.dca01e577bb98p-1, 0x1.253f5e30d2000p-4}, +{0x1.d92f20b7c9103p-1, 0x1.42edd8b380000p-4}, +{0x1.d5cac66fb5ccep-1, 0x1.606598757c000p-4}, +{0x1.d272caa5ede9dp-1, 0x1.7da76356a0000p-4}, +{0x1.cf26e3e6b2ccdp-1, 0x1.9ab434e1c6000p-4}, +{0x1.cbe6da2a77902p-1, 0x1.b78c7bb0d6000p-4}, +{0x1.c8b266d37086dp-1, 0x1.d431332e72000p-4}, +{0x1.c5894bd5d5804p-1, 0x1.f0a3171de6000p-4}, +{0x1.c26b533bb9f8cp-1, 0x1.067152b914000p-3}, +{0x1.bf583eeece73fp-1, 0x1.147858292b000p-3}, +{0x1.bc4fd75db96c1p-1, 0x1.2266ecdca3000p-3}, +{0x1.b951e0c864a28p-1, 0x1.303d7a6c55000p-3}, +{0x1.b65e2c5ef3e2cp-1, 0x1.3dfc33c331000p-3}, +{0x1.b374867c9888bp-1, 0x1.4ba366b7a8000p-3}, +{0x1.b094b211d304ap-1, 0x1.5933928d1f000p-3}, +{0x1.adbe885f2ef7ep-1, 0x1.66acd2418f000p-3}, +{0x1.aaf1d31603da2p-1, 0x1.740f8ec669000p-3}, +{0x1.a82e63fd358a7p-1, 0x1.815c0f51af000p-3}, +{0x1.a5740ef09738bp-1, 0x1.8e92954f68000p-3}, +{0x1.a2c2a90ab4b27p-1, 0x1.9bb3602f84000p-3}, +{0x1.a01a01393f2d1p-1, 0x1.a8bed1c2c0000p-3}, +{0x1.9d79f24db3c1bp-1, 0x1.b5b515c01d000p-3}, +{0x1.9ae2505c7b190p-1, 0x1.c2967ccbcc000p-3}, +{0x1.9852ef297ce2fp-1, 0x1.cf635d5486000p-3}, +{0x1.95cbaeea44b75p-1, 0x1.dc1bd3446c000p-3}, +{0x1.934c69de74838p-1, 0x1.e8c01b8cfe000p-3}, +{0x1.90d4f2f6752e6p-1, 0x1.f5509c0179000p-3}, +{0x1.8e6528effd79dp-1, 0x1.00e6c121fb800p-2}, +{0x1.8bfce9fcc007cp-1, 0x1.071b80e93d000p-2}, +{0x1.899c0dabec30ep-1, 0x1.0d46b9e867000p-2}, +{0x1.87427aa2317fbp-1, 0x1.13687334bd000p-2}, +{0x1.84f00acb39a08p-1, 0x1.1980d67234800p-2}, +{0x1.82a49e8653e55p-1, 0x1.1f8ffe0cc8000p-2}, +{0x1.8060195f40260p-1, 0x1.2595fd7636800p-2}, +{0x1.7e22563e0a329p-1, 0x1.2b9300914a800p-2}, +{0x1.7beb377dcb5adp-1, 0x1.3187210436000p-2}, +{0x1.79baa679725c2p-1, 0x1.377266dec1800p-2}, +{0x1.77907f2170657p-1, 0x1.3d54ffbaf3000p-2}, +{0x1.756cadbd6130cp-1, 0x1.432eee32fe000p-2}, +}, +#if !__FP_FAST_FMA +.tab2 = { +{0x1.61000014fb66bp-1, 0x1.e026c91425b3cp-56}, +{0x1.63000034db495p-1, 0x1.dbfea48005d41p-55}, +{0x1.650000d94d478p-1, 0x1.e7fa786d6a5b7p-55}, +{0x1.67000074e6fadp-1, 0x1.1fcea6b54254cp-57}, +{0x1.68ffffedf0faep-1, -0x1.c7e274c590efdp-56}, +{0x1.6b0000763c5bcp-1, -0x1.ac16848dcda01p-55}, +{0x1.6d0001e5cc1f6p-1, 0x1.33f1c9d499311p-55}, +{0x1.6efffeb05f63ep-1, -0x1.e80041ae22d53p-56}, +{0x1.710000e86978p-1, 0x1.bff6671097952p-56}, +{0x1.72ffffc67e912p-1, 0x1.c00e226bd8724p-55}, +{0x1.74fffdf81116ap-1, -0x1.e02916ef101d2p-57}, +{0x1.770000f679c9p-1, -0x1.7fc71cd549c74p-57}, +{0x1.78ffffa7ec835p-1, 0x1.1bec19ef50483p-55}, +{0x1.7affffe20c2e6p-1, -0x1.07e1729cc6465p-56}, +{0x1.7cfffed3fc9p-1, -0x1.08072087b8b1cp-55}, +{0x1.7efffe9261a76p-1, 0x1.dc0286d9df9aep-55}, +{0x1.81000049ca3e8p-1, 0x1.97fd251e54c33p-55}, +{0x1.8300017932c8fp-1, -0x1.afee9b630f381p-55}, +{0x1.850000633739cp-1, 0x1.9bfbf6b6535bcp-55}, +{0x1.87000204289c6p-1, -0x1.bbf65f3117b75p-55}, +{0x1.88fffebf57904p-1, -0x1.9006ea23dcb57p-55}, +{0x1.8b00022bc04dfp-1, -0x1.d00df38e04b0ap-56}, +{0x1.8cfffe50c1b8ap-1, -0x1.8007146ff9f05p-55}, +{0x1.8effffc918e43p-1, 0x1.3817bd07a7038p-55}, +{0x1.910001efa5fc7p-1, 0x1.93e9176dfb403p-55}, +{0x1.9300013467bb9p-1, 0x1.f804e4b980276p-56}, +{0x1.94fffe6ee076fp-1, -0x1.f7ef0d9ff622ep-55}, +{0x1.96fffde3c12d1p-1, -0x1.082aa962638bap-56}, +{0x1.98ffff4458a0dp-1, -0x1.7801b9164a8efp-55}, +{0x1.9afffdd982e3ep-1, -0x1.740e08a5a9337p-55}, +{0x1.9cfffed49fb66p-1, 0x1.fce08c19bep-60}, +{0x1.9f00020f19c51p-1, -0x1.a3faa27885b0ap-55}, +{0x1.a10001145b006p-1, 0x1.4ff489958da56p-56}, +{0x1.a300007bbf6fap-1, 0x1.cbeab8a2b6d18p-55}, +{0x1.a500010971d79p-1, 0x1.8fecadd78793p-55}, +{0x1.a70001df52e48p-1, -0x1.f41763dd8abdbp-55}, +{0x1.a90001c593352p-1, -0x1.ebf0284c27612p-55}, +{0x1.ab0002a4f3e4bp-1, -0x1.9fd043cff3f5fp-57}, +{0x1.acfffd7ae1ed1p-1, -0x1.23ee7129070b4p-55}, +{0x1.aefffee510478p-1, 0x1.a063ee00edea3p-57}, +{0x1.b0fffdb650d5bp-1, 0x1.a06c8381f0ab9p-58}, +{0x1.b2ffffeaaca57p-1, -0x1.9011e74233c1dp-56}, +{0x1.b4fffd995badcp-1, -0x1.9ff1068862a9fp-56}, +{0x1.b7000249e659cp-1, 0x1.aff45d0864f3ep-55}, +{0x1.b8ffff987164p-1, 0x1.cfe7796c2c3f9p-56}, +{0x1.bafffd204cb4fp-1, -0x1.3ff27eef22bc4p-57}, +{0x1.bcfffd2415c45p-1, -0x1.cffb7ee3bea21p-57}, +{0x1.beffff86309dfp-1, -0x1.14103972e0b5cp-55}, +{0x1.c0fffe1b57653p-1, 0x1.bc16494b76a19p-55}, +{0x1.c2ffff1fa57e3p-1, -0x1.4feef8d30c6edp-57}, +{0x1.c4fffdcbfe424p-1, -0x1.43f68bcec4775p-55}, +{0x1.c6fffed54b9f7p-1, 0x1.47ea3f053e0ecp-55}, +{0x1.c8fffeb998fd5p-1, 0x1.383068df992f1p-56}, +{0x1.cb0002125219ap-1, -0x1.8fd8e64180e04p-57}, +{0x1.ccfffdd94469cp-1, 0x1.e7ebe1cc7ea72p-55}, +{0x1.cefffeafdc476p-1, 0x1.ebe39ad9f88fep-55}, +{0x1.d1000169af82bp-1, 0x1.57d91a8b95a71p-56}, +{0x1.d30000d0ff71dp-1, 0x1.9c1906970c7dap-55}, +{0x1.d4fffea790fc4p-1, -0x1.80e37c558fe0cp-58}, +{0x1.d70002edc87e5p-1, -0x1.f80d64dc10f44p-56}, +{0x1.d900021dc82aap-1, -0x1.47c8f94fd5c5cp-56}, +{0x1.dafffd86b0283p-1, 0x1.c7f1dc521617ep-55}, +{0x1.dd000296c4739p-1, 0x1.8019eb2ffb153p-55}, +{0x1.defffe54490f5p-1, 0x1.e00d2c652cc89p-57}, +{0x1.e0fffcdabf694p-1, -0x1.f8340202d69d2p-56}, +{0x1.e2fffdb52c8ddp-1, 0x1.b00c1ca1b0864p-56}, +{0x1.e4ffff24216efp-1, 0x1.2ffa8b094ab51p-56}, +{0x1.e6fffe88a5e11p-1, -0x1.7f673b1efbe59p-58}, +{0x1.e9000119eff0dp-1, -0x1.4808d5e0bc801p-55}, +{0x1.eafffdfa51744p-1, 0x1.80006d54320b5p-56}, +{0x1.ed0001a127fa1p-1, -0x1.002f860565c92p-58}, +{0x1.ef00007babcc4p-1, -0x1.540445d35e611p-55}, +{0x1.f0ffff57a8d02p-1, -0x1.ffb3139ef9105p-59}, +{0x1.f30001ee58ac7p-1, 0x1.a81acf2731155p-55}, +{0x1.f4ffff5823494p-1, 0x1.a3f41d4d7c743p-55}, +{0x1.f6ffffca94c6bp-1, -0x1.202f41c987875p-57}, +{0x1.f8fffe1f9c441p-1, 0x1.77dd1f477e74bp-56}, +{0x1.fafffd2e0e37ep-1, -0x1.f01199a7ca331p-57}, +{0x1.fd0001c77e49ep-1, 0x1.181ee4bceacb1p-56}, +{0x1.feffff7e0c331p-1, -0x1.e05370170875ap-57}, +{0x1.00ffff465606ep+0, -0x1.a7ead491c0adap-55}, +{0x1.02ffff3867a58p+0, -0x1.77f69c3fcb2ep-54}, +{0x1.04ffffdfc0d17p+0, 0x1.7bffe34cb945bp-54}, +{0x1.0700003cd4d82p+0, 0x1.20083c0e456cbp-55}, +{0x1.08ffff9f2cbe8p+0, -0x1.dffdfbe37751ap-57}, +{0x1.0b000010cda65p+0, -0x1.13f7faee626ebp-54}, +{0x1.0d00001a4d338p+0, 0x1.07dfa79489ff7p-55}, +{0x1.0effffadafdfdp+0, -0x1.7040570d66bcp-56}, +{0x1.110000bbafd96p+0, 0x1.e80d4846d0b62p-55}, +{0x1.12ffffae5f45dp+0, 0x1.dbffa64fd36efp-54}, +{0x1.150000dd59ad9p+0, 0x1.a0077701250aep-54}, +{0x1.170000f21559ap+0, 0x1.dfdf9e2e3deeep-55}, +{0x1.18ffffc275426p+0, 0x1.10030dc3b7273p-54}, +{0x1.1b000123d3c59p+0, 0x1.97f7980030188p-54}, +{0x1.1cffff8299eb7p+0, -0x1.5f932ab9f8c67p-57}, +{0x1.1effff48ad4p+0, 0x1.37fbf9da75bebp-54}, +{0x1.210000c8b86a4p+0, 0x1.f806b91fd5b22p-54}, +{0x1.2300003854303p+0, 0x1.3ffc2eb9fbf33p-54}, +{0x1.24fffffbcf684p+0, 0x1.601e77e2e2e72p-56}, +{0x1.26ffff52921d9p+0, 0x1.ffcbb767f0c61p-56}, +{0x1.2900014933a3cp+0, -0x1.202ca3c02412bp-56}, +{0x1.2b00014556313p+0, -0x1.2808233f21f02p-54}, +{0x1.2cfffebfe523bp+0, -0x1.8ff7e384fdcf2p-55}, +{0x1.2f0000bb8ad96p+0, -0x1.5ff51503041c5p-55}, +{0x1.30ffffb7ae2afp+0, -0x1.10071885e289dp-55}, +{0x1.32ffffeac5f7fp+0, -0x1.1ff5d3fb7b715p-54}, +{0x1.350000ca66756p+0, 0x1.57f82228b82bdp-54}, +{0x1.3700011fbf721p+0, 0x1.000bac40dd5ccp-55}, +{0x1.38ffff9592fb9p+0, -0x1.43f9d2db2a751p-54}, +{0x1.3b00004ddd242p+0, 0x1.57f6b707638e1p-55}, +{0x1.3cffff5b2c957p+0, 0x1.a023a10bf1231p-56}, +{0x1.3efffeab0b418p+0, 0x1.87f6d66b152bp-54}, +{0x1.410001532aff4p+0, 0x1.7f8375f198524p-57}, +{0x1.4300017478b29p+0, 0x1.301e672dc5143p-55}, +{0x1.44fffe795b463p+0, 0x1.9ff69b8b2895ap-55}, +{0x1.46fffe80475ep+0, -0x1.5c0b19bc2f254p-54}, +{0x1.48fffef6fc1e7p+0, 0x1.b4009f23a2a72p-54}, +{0x1.4afffe5bea704p+0, -0x1.4ffb7bf0d7d45p-54}, +{0x1.4d000171027dep+0, -0x1.9c06471dc6a3dp-54}, +{0x1.4f0000ff03ee2p+0, 0x1.77f890b85531cp-54}, +{0x1.5100012dc4bd1p+0, 0x1.004657166a436p-57}, +{0x1.530001605277ap+0, -0x1.6bfcece233209p-54}, +{0x1.54fffecdb704cp+0, -0x1.902720505a1d7p-55}, +{0x1.56fffef5f54a9p+0, 0x1.bbfe60ec96412p-54}, +{0x1.5900017e61012p+0, 0x1.87ec581afef9p-55}, +{0x1.5b00003c93e92p+0, -0x1.f41080abf0ccp-54}, +{0x1.5d0001d4919bcp+0, -0x1.8812afb254729p-54}, +{0x1.5efffe7b87a89p+0, -0x1.47eb780ed6904p-54}, +}, +#endif +}; diff --git a/kal/posix/musl_src/math/log_data.h b/kal/posix/musl_src/math/log_data.h new file mode 100644 index 00000000..d8675236 --- /dev/null +++ b/kal/posix/musl_src/math/log_data.h @@ -0,0 +1,28 @@ +/* + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ +#ifndef _LOG_DATA_H +#define _LOG_DATA_H + +#include + +#define LOG_TABLE_BITS 7 +#define LOG_POLY_ORDER 6 +#define LOG_POLY1_ORDER 12 +extern const struct log_data { + double ln2hi; + double ln2lo; + double poly[LOG_POLY_ORDER - 1]; /* First coefficient is 1. */ + double poly1[LOG_POLY1_ORDER - 1]; + struct { + double invc, logc; + } tab[1 << LOG_TABLE_BITS]; +#if !__FP_FAST_FMA + struct { + double chi, clo; + } tab2[1 << LOG_TABLE_BITS]; +#endif +} __log_data; + +#endif diff --git a/kal/posix/musl_src/math/pow.c b/kal/posix/musl_src/math/pow.c new file mode 100644 index 00000000..434aea3a --- /dev/null +++ b/kal/posix/musl_src/math/pow.c @@ -0,0 +1,343 @@ +/* + * Double-precision x^y function. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ + +#include +#include +#include "libm.h" +#include "exp_data.h" +#include "pow_data.h" + +/* +Worst-case error: 0.54 ULP (~= ulperr_exp + 1024*Ln2*relerr_log*2^53) +relerr_log: 1.3 * 2^-68 (Relative error of log, 1.5 * 2^-68 without fma) +ulperr_exp: 0.509 ULP (ULP error of exp, 0.511 ULP without fma) +*/ + +#define T __pow_log_data1.tab +#define A __pow_log_data1.poly +#define Ln2hi __pow_log_data1.ln2hi +#define Ln2lo __pow_log_data1.ln2lo +#define N (1 << POW_LOG_TABLE_BITS) +#define OFF 0x3fe6955500000000 + +/* Top 12 bits of a double (sign and exponent bits). */ +static inline uint32_t top12(double x) +{ + return asuint64(x) >> 52; +} + +/* Compute y+TAIL = log(x) where the rounded result is y and TAIL has about + additional 15 bits precision. IX is the bit representation of x, but + normalized in the subnormal range using the sign bit for the exponent. */ +static inline double_t log_inline(uint64_t ix, double_t *tail) +{ + /* double_t for better performance on targets with FLT_EVAL_METHOD==2. */ + double_t z, r, y, invc, logc, logctail, kd, hi, t1, t2, lo, lo1, lo2, p; + uint64_t iz, tmp; + int k, i; + + /* x = 2^k z; where z is in range [OFF,2*OFF) and exact. + The range is split into N subintervals. + The ith subinterval contains z and c is near its center. */ + tmp = ix - OFF; + i = (tmp >> (52 - POW_LOG_TABLE_BITS)) % N; + k = (int64_t)tmp >> 52; /* arithmetic shift */ + iz = ix - (tmp & 0xfffULL << 52); + z = asdouble(iz); + kd = (double_t)k; + + /* log(x) = k*Ln2 + log(c) + log1p(z/c-1). */ + invc = T[i].invc; + logc = T[i].logc; + logctail = T[i].logctail; + + /* Note: 1/c is j/N or j/N/2 where j is an integer in [N,2N) and + |z/c - 1| < 1/N, so r = z/c - 1 is exactly representible. */ +#if __FP_FAST_FMA + r = __builtin_fma(z, invc, -1.0); +#else + /* Split z such that rhi, rlo and rhi*rhi are exact and |rlo| <= |r|. */ + double_t zhi = asdouble((iz + (1ULL << 31)) & (-1ULL << 32)); + double_t zlo = z - zhi; + double_t rhi = zhi * invc - 1.0; + double_t rlo = zlo * invc; + r = rhi + rlo; +#endif + + /* k*Ln2 + log(c) + r. */ + t1 = kd * Ln2hi + logc; + t2 = t1 + r; + lo1 = kd * Ln2lo + logctail; + lo2 = t1 - t2 + r; + + /* Evaluation is optimized assuming superscalar pipelined execution. */ + double_t ar, ar2, ar3, lo3, lo4; + ar = A[0] * r; /* A[0] = -0.5. */ + ar2 = r * ar; + ar3 = r * ar2; + /* k*Ln2 + log(c) + r + A[0]*r*r. */ +#if __FP_FAST_FMA + hi = t2 + ar2; + lo3 = __builtin_fma(ar, r, -ar2); + lo4 = t2 - hi + ar2; +#else + double_t arhi = A[0] * rhi; + double_t arhi2 = rhi * arhi; + hi = t2 + arhi2; + lo3 = rlo * (ar + arhi); + lo4 = t2 - hi + arhi2; +#endif + /* p = log1p(r) - r - A[0]*r*r. */ + p = (ar3 * (A[1] + r * A[2] + + ar2 * (A[3] + r * A[4] + ar2 * (A[5] + r * A[6])))); + lo = lo1 + lo2 + lo3 + lo4 + p; + y = hi + lo; + *tail = hi - y + lo; + return y; +} + +#undef N +#undef T +#define N (1 << EXP_TABLE_BITS) +#define InvLn2N __exp_data1.invln2N +#define NegLn2hiN __exp_data1.negln2hiN +#define NegLn2loN __exp_data1.negln2loN +#define Shift __exp_data1.shift +#define T __exp_data1.tab +#define C2 __exp_data1.poly[5 - EXP_POLY_ORDER] +#define C3 __exp_data1.poly[6 - EXP_POLY_ORDER] +#define C4 __exp_data1.poly[7 - EXP_POLY_ORDER] +#define C5 __exp_data1.poly[8 - EXP_POLY_ORDER] +#define C6 __exp_data1.poly[9 - EXP_POLY_ORDER] + +/* Handle cases that may overflow or underflow when computing the result that + is scale*(1+TMP) without intermediate rounding. The bit representation of + scale is in SBITS, however it has a computed exponent that may have + overflown into the sign bit so that needs to be adjusted before using it as + a double. (int32_t)KI is the k used in the argument reduction and exponent + adjustment of scale, positive k here means the result may overflow and + negative k means the result may underflow. */ +static inline double specialcase(double_t tmp, uint64_t sbits, uint64_t ki) +{ + double_t scale, y; + + if ((ki & 0x80000000) == 0) { + /* k > 0, the exponent of scale might have overflowed by <= 460. */ + sbits -= 1009ull << 52; + scale = asdouble(sbits); + y = 0x1p1009 * (scale + scale * tmp); + return eval_as_double(y); + } + /* k < 0, need special care in the subnormal range. */ + sbits += 1022ull << 52; + /* Note: sbits is signed scale. */ + scale = asdouble(sbits); + y = scale + scale * tmp; + if (fabs(y) < 1.0) { + /* Round y to the right precision before scaling it into the subnormal + range to avoid double rounding that can cause 0.5+E/2 ulp error where + E is the worst-case ulp error outside the subnormal range. So this + is only useful if the goal is better than 1 ulp worst-case error. */ + double_t hi, lo, one = 1.0; + if (y < 0.0) + one = -1.0; + lo = scale - y + scale * tmp; + hi = one + y; + lo = one - hi + y + lo; + y = eval_as_double(hi + lo) - one; + /* Fix the sign of 0. */ + if (y == 0.0) + y = asdouble(sbits & 0x8000000000000000); + /* The underflow exception needs to be signaled explicitly. */ + fp_force_eval(fp_barrier(0x1p-1022) * 0x1p-1022); + } + y = 0x1p-1022 * y; + return eval_as_double(y); +} + +#define SIGN_BIAS (0x800 << EXP_TABLE_BITS) + +/* Computes sign*exp(x+xtail) where |xtail| < 2^-8/N and |xtail| <= |x|. + The sign_bias argument is SIGN_BIAS or 0 and sets the sign to -1 or 1. */ +static inline double exp_inline(double_t x, double_t xtail, uint32_t sign_bias) +{ + uint32_t abstop; + uint64_t ki, idx, top, sbits; + /* double_t for better performance on targets with FLT_EVAL_METHOD==2. */ + double_t kd, z, r, r2, scale, tail, tmp; + + abstop = top12(x) & 0x7ff; + if (predict_false(abstop - top12(0x1p-54) >= + top12(512.0) - top12(0x1p-54))) { + if (abstop - top12(0x1p-54) >= 0x80000000) { + /* Avoid spurious underflow for tiny x. */ + /* Note: 0 is common input. */ + double_t one = WANT_ROUNDING ? 1.0 + x : 1.0; + return sign_bias ? -one : one; + } + if (abstop >= top12(1024.0)) { + /* Note: inf and nan are already handled. */ + if (asuint64(x) >> 63) + return __math_uflow(sign_bias); + else + return __math_oflow(sign_bias); + } + /* Large x is special cased below. */ + abstop = 0; + } + + /* exp(x) = 2^(k/N) * exp(r), with exp(r) in [2^(-1/2N),2^(1/2N)]. */ + /* x = ln2/N*k + r, with int k and r in [-ln2/2N, ln2/2N]. */ + z = InvLn2N * x; +#if TOINT_INTRINSICS + kd = roundtoint(z); + ki = converttoint(z); +#elif EXP_USE_TOINT_NARROW + /* z - kd is in [-0.5-2^-16, 0.5] in all rounding modes. */ + kd = eval_as_double(z + Shift); + ki = asuint64(kd) >> 16; + kd = (double_t)(int32_t)ki; +#else + /* z - kd is in [-1, 1] in non-nearest rounding modes. */ + kd = eval_as_double(z + Shift); + ki = asuint64(kd); + kd -= Shift; +#endif + r = x + kd * NegLn2hiN + kd * NegLn2loN; + /* The code assumes 2^-200 < |xtail| < 2^-8/N. */ + r += xtail; + /* 2^(k/N) ~= scale * (1 + tail). */ + idx = 2 * (ki % N); + top = (ki + sign_bias) << (52 - EXP_TABLE_BITS); + tail = asdouble(T[idx]); + /* This is only a valid scale when -1023*N < k < 1024*N. */ + sbits = T[idx + 1] + top; + /* exp(x) = 2^(k/N) * exp(r) ~= scale + scale * (tail + exp(r) - 1). */ + /* Evaluation is optimized assuming superscalar pipelined execution. */ + r2 = r * r; + /* Without fma the worst case error is 0.25/N ulp larger. */ + /* Worst case error is less than 0.5+1.11/N+(abs poly error * 2^53) ulp. */ + tmp = tail + r + r2 * (C2 + r * C3) + r2 * r2 * (C4 + r * C5); + if (predict_false(abstop == 0)) + return specialcase(tmp, sbits, ki); + scale = asdouble(sbits); + /* Note: tmp == 0 or |tmp| > 2^-200 and scale > 2^-739, so there + is no spurious underflow here even without fma. */ + return eval_as_double(scale + scale * tmp); +} + +/* Returns 0 if not int, 1 if odd int, 2 if even int. The argument is + the bit representation of a non-zero finite floating-point value. */ +static inline int checkint(uint64_t iy) +{ + int e = iy >> 52 & 0x7ff; + if (e < 0x3ff) + return 0; + if (e > 0x3ff + 52) + return 2; + if (iy & ((1ULL << (0x3ff + 52 - e)) - 1)) + return 0; + if (iy & (1ULL << (0x3ff + 52 - e))) + return 1; + return 2; +} + +/* Returns 1 if input is the bit representation of 0, infinity or nan. */ +static inline int zeroinfnan(uint64_t i) +{ + return 2 * i - 1 >= 2 * asuint64(INFINITY) - 1; +} + +double pow(double x, double y) +{ + uint32_t sign_bias = 0; + uint64_t ix, iy; + uint32_t topx, topy; + + ix = asuint64(x); + iy = asuint64(y); + topx = top12(x); + topy = top12(y); + if (predict_false(topx - 0x001 >= 0x7ff - 0x001 || + (topy & 0x7ff) - 0x3be >= 0x43e - 0x3be)) { + /* Note: if |y| > 1075 * ln2 * 2^53 ~= 0x1.749p62 then pow(x,y) = inf/0 + and if |y| < 2^-54 / 1075 ~= 0x1.e7b6p-65 then pow(x,y) = +-1. */ + /* Special cases: (x < 0x1p-126 or inf or nan) or + (|y| < 0x1p-65 or |y| >= 0x1p63 or nan). */ + if (predict_false(zeroinfnan(iy))) { + if (2 * iy == 0) + return issignaling_inline(x) ? x + y : 1.0; + if (ix == asuint64(1.0)) + return issignaling_inline(y) ? x + y : 1.0; + if (2 * ix > 2 * asuint64(INFINITY) || + 2 * iy > 2 * asuint64(INFINITY)) + return x + y; + if (2 * ix == 2 * asuint64(1.0)) + return 1.0; + if ((2 * ix < 2 * asuint64(1.0)) == !(iy >> 63)) + return 0.0; /* |x|<1 && y==inf or |x|>1 && y==-inf. */ + return y * y; + } + if (predict_false(zeroinfnan(ix))) { + double_t x2 = x * x; + if (ix >> 63 && checkint(iy) == 1) + x2 = -x2; + /* Without the barrier some versions of clang hoist the 1/x2 and + thus division by zero exception can be signaled spuriously. */ + return iy >> 63 ? fp_barrier(1 / x2) : x2; + } + /* Here x and y are non-zero finite. */ + if (ix >> 63) { + /* Finite x < 0. */ + int yint = checkint(iy); + if (yint == 0) + return __math_invalid(x); + if (yint == 1) + sign_bias = SIGN_BIAS; + ix &= 0x7fffffffffffffff; + topx &= 0x7ff; + } + if ((topy & 0x7ff) - 0x3be >= 0x43e - 0x3be) { + /* Note: sign_bias == 0 here because y is not odd. */ + if (ix == asuint64(1.0)) + return 1.0; + if ((topy & 0x7ff) < 0x3be) { + /* |y| < 2^-65, x^y ~= 1 + y*log(x). */ + if (WANT_ROUNDING) + return ix > asuint64(1.0) ? 1.0 + y : + 1.0 - y; + else + return 1.0; + } + return (ix > asuint64(1.0)) == (topy < 0x800) ? + __math_oflow(0) : + __math_uflow(0); + } + if (topx == 0) { + /* Normalize subnormal x so exponent becomes negative. */ + ix = asuint64(x * 0x1p52); + ix &= 0x7fffffffffffffff; + ix -= 52ULL << 52; + } + } + + double_t lo; + double_t hi = log_inline(ix, &lo); + double_t ehi, elo; +#if __FP_FAST_FMA + ehi = y * hi; + elo = y * lo + __builtin_fma(y, hi, -ehi); +#else + double_t yhi = asdouble(iy & -1ULL << 27); + double_t ylo = y - yhi; + double_t lhi = asdouble(asuint64(hi) & -1ULL << 27); + double_t llo = hi - lhi + lo; + ehi = yhi * lhi; + elo = ylo * lhi + y * llo; /* |elo| < |ehi| * 2^-25. */ +#endif + return exp_inline(ehi, elo, sign_bias); +} diff --git a/kal/posix/musl_src/math/pow_data.c b/kal/posix/musl_src/math/pow_data.c new file mode 100644 index 00000000..0c9d07de --- /dev/null +++ b/kal/posix/musl_src/math/pow_data.c @@ -0,0 +1,180 @@ +/* + * Data for the log part of pow. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ + +#include "pow_data.h" + +#define N (1 << POW_LOG_TABLE_BITS) + +const struct pow_log_data __pow_log_data1 = { +.ln2hi = 0x1.62e42fefa3800p-1, +.ln2lo = 0x1.ef35793c76730p-45, +.poly = { +// relative error: 0x1.11922ap-70 +// in -0x1.6bp-8 0x1.6bp-8 +// Coefficients are scaled to match the scaling during evaluation. +-0x1p-1, +0x1.555555555556p-2 * -2, +-0x1.0000000000006p-2 * -2, +0x1.999999959554ep-3 * 4, +-0x1.555555529a47ap-3 * 4, +0x1.2495b9b4845e9p-3 * -8, +-0x1.0002b8b263fc3p-3 * -8, +}, +/* Algorithm: + + x = 2^k z + log(x) = k ln2 + log(c) + log(z/c) + log(z/c) = poly(z/c - 1) + +where z is in [0x1.69555p-1; 0x1.69555p0] which is split into N subintervals +and z falls into the ith one, then table entries are computed as + + tab[i].invc = 1/c + tab[i].logc = round(0x1p43*log(c))/0x1p43 + tab[i].logctail = (double)(log(c) - logc) + +where c is chosen near the center of the subinterval such that 1/c has only a +few precision bits so z/c - 1 is exactly representible as double: + + 1/c = center < 1 ? round(N/center)/N : round(2*N/center)/N/2 + +Note: |z/c - 1| < 1/N for the chosen c, |log(c) - logc - logctail| < 0x1p-97, +the last few bits of logc are rounded away so k*ln2hi + logc has no rounding +error and the interval for z is selected such that near x == 1, where log(x) +is tiny, large cancellation error is avoided in logc + poly(z/c - 1). */ +.tab = { +#define A(a, b, c) {a, 0, b, c}, +A(0x1.6a00000000000p+0, -0x1.62c82f2b9c800p-2, 0x1.ab42428375680p-48) +A(0x1.6800000000000p+0, -0x1.5d1bdbf580800p-2, -0x1.ca508d8e0f720p-46) +A(0x1.6600000000000p+0, -0x1.5767717455800p-2, -0x1.362a4d5b6506dp-45) +A(0x1.6400000000000p+0, -0x1.51aad872df800p-2, -0x1.684e49eb067d5p-49) +A(0x1.6200000000000p+0, -0x1.4be5f95777800p-2, -0x1.41b6993293ee0p-47) +A(0x1.6000000000000p+0, -0x1.4618bc21c6000p-2, 0x1.3d82f484c84ccp-46) +A(0x1.5e00000000000p+0, -0x1.404308686a800p-2, 0x1.c42f3ed820b3ap-50) +A(0x1.5c00000000000p+0, -0x1.3a64c55694800p-2, 0x1.0b1c686519460p-45) +A(0x1.5a00000000000p+0, -0x1.347dd9a988000p-2, 0x1.5594dd4c58092p-45) +A(0x1.5800000000000p+0, -0x1.2e8e2bae12000p-2, 0x1.67b1e99b72bd8p-45) +A(0x1.5600000000000p+0, -0x1.2895a13de8800p-2, 0x1.5ca14b6cfb03fp-46) +A(0x1.5600000000000p+0, -0x1.2895a13de8800p-2, 0x1.5ca14b6cfb03fp-46) +A(0x1.5400000000000p+0, -0x1.22941fbcf7800p-2, -0x1.65a242853da76p-46) +A(0x1.5200000000000p+0, -0x1.1c898c1699800p-2, -0x1.fafbc68e75404p-46) +A(0x1.5000000000000p+0, -0x1.1675cababa800p-2, 0x1.f1fc63382a8f0p-46) +A(0x1.4e00000000000p+0, -0x1.1058bf9ae4800p-2, -0x1.6a8c4fd055a66p-45) +A(0x1.4c00000000000p+0, -0x1.0a324e2739000p-2, -0x1.c6bee7ef4030ep-47) +A(0x1.4a00000000000p+0, -0x1.0402594b4d000p-2, -0x1.036b89ef42d7fp-48) +A(0x1.4a00000000000p+0, -0x1.0402594b4d000p-2, -0x1.036b89ef42d7fp-48) +A(0x1.4800000000000p+0, -0x1.fb9186d5e4000p-3, 0x1.d572aab993c87p-47) +A(0x1.4600000000000p+0, -0x1.ef0adcbdc6000p-3, 0x1.b26b79c86af24p-45) +A(0x1.4400000000000p+0, -0x1.e27076e2af000p-3, -0x1.72f4f543fff10p-46) +A(0x1.4200000000000p+0, -0x1.d5c216b4fc000p-3, 0x1.1ba91bbca681bp-45) +A(0x1.4000000000000p+0, -0x1.c8ff7c79aa000p-3, 0x1.7794f689f8434p-45) +A(0x1.4000000000000p+0, -0x1.c8ff7c79aa000p-3, 0x1.7794f689f8434p-45) +A(0x1.3e00000000000p+0, -0x1.bc286742d9000p-3, 0x1.94eb0318bb78fp-46) +A(0x1.3c00000000000p+0, -0x1.af3c94e80c000p-3, 0x1.a4e633fcd9066p-52) +A(0x1.3a00000000000p+0, -0x1.a23bc1fe2b000p-3, -0x1.58c64dc46c1eap-45) +A(0x1.3a00000000000p+0, -0x1.a23bc1fe2b000p-3, -0x1.58c64dc46c1eap-45) +A(0x1.3800000000000p+0, -0x1.9525a9cf45000p-3, -0x1.ad1d904c1d4e3p-45) +A(0x1.3600000000000p+0, -0x1.87fa06520d000p-3, 0x1.bbdbf7fdbfa09p-45) +A(0x1.3400000000000p+0, -0x1.7ab890210e000p-3, 0x1.bdb9072534a58p-45) +A(0x1.3400000000000p+0, -0x1.7ab890210e000p-3, 0x1.bdb9072534a58p-45) +A(0x1.3200000000000p+0, -0x1.6d60fe719d000p-3, -0x1.0e46aa3b2e266p-46) +A(0x1.3000000000000p+0, -0x1.5ff3070a79000p-3, -0x1.e9e439f105039p-46) +A(0x1.3000000000000p+0, -0x1.5ff3070a79000p-3, -0x1.e9e439f105039p-46) +A(0x1.2e00000000000p+0, -0x1.526e5e3a1b000p-3, -0x1.0de8b90075b8fp-45) +A(0x1.2c00000000000p+0, -0x1.44d2b6ccb8000p-3, 0x1.70cc16135783cp-46) +A(0x1.2c00000000000p+0, -0x1.44d2b6ccb8000p-3, 0x1.70cc16135783cp-46) +A(0x1.2a00000000000p+0, -0x1.371fc201e9000p-3, 0x1.178864d27543ap-48) +A(0x1.2800000000000p+0, -0x1.29552f81ff000p-3, -0x1.48d301771c408p-45) +A(0x1.2600000000000p+0, -0x1.1b72ad52f6000p-3, -0x1.e80a41811a396p-45) +A(0x1.2600000000000p+0, -0x1.1b72ad52f6000p-3, -0x1.e80a41811a396p-45) +A(0x1.2400000000000p+0, -0x1.0d77e7cd09000p-3, 0x1.a699688e85bf4p-47) +A(0x1.2400000000000p+0, -0x1.0d77e7cd09000p-3, 0x1.a699688e85bf4p-47) +A(0x1.2200000000000p+0, -0x1.fec9131dbe000p-4, -0x1.575545ca333f2p-45) +A(0x1.2000000000000p+0, -0x1.e27076e2b0000p-4, 0x1.a342c2af0003cp-45) +A(0x1.2000000000000p+0, -0x1.e27076e2b0000p-4, 0x1.a342c2af0003cp-45) +A(0x1.1e00000000000p+0, -0x1.c5e548f5bc000p-4, -0x1.d0c57585fbe06p-46) +A(0x1.1c00000000000p+0, -0x1.a926d3a4ae000p-4, 0x1.53935e85baac8p-45) +A(0x1.1c00000000000p+0, -0x1.a926d3a4ae000p-4, 0x1.53935e85baac8p-45) +A(0x1.1a00000000000p+0, -0x1.8c345d631a000p-4, 0x1.37c294d2f5668p-46) +A(0x1.1a00000000000p+0, -0x1.8c345d631a000p-4, 0x1.37c294d2f5668p-46) +A(0x1.1800000000000p+0, -0x1.6f0d28ae56000p-4, -0x1.69737c93373dap-45) +A(0x1.1600000000000p+0, -0x1.51b073f062000p-4, 0x1.f025b61c65e57p-46) +A(0x1.1600000000000p+0, -0x1.51b073f062000p-4, 0x1.f025b61c65e57p-46) +A(0x1.1400000000000p+0, -0x1.341d7961be000p-4, 0x1.c5edaccf913dfp-45) +A(0x1.1400000000000p+0, -0x1.341d7961be000p-4, 0x1.c5edaccf913dfp-45) +A(0x1.1200000000000p+0, -0x1.16536eea38000p-4, 0x1.47c5e768fa309p-46) +A(0x1.1000000000000p+0, -0x1.f0a30c0118000p-5, 0x1.d599e83368e91p-45) +A(0x1.1000000000000p+0, -0x1.f0a30c0118000p-5, 0x1.d599e83368e91p-45) +A(0x1.0e00000000000p+0, -0x1.b42dd71198000p-5, 0x1.c827ae5d6704cp-46) +A(0x1.0e00000000000p+0, -0x1.b42dd71198000p-5, 0x1.c827ae5d6704cp-46) +A(0x1.0c00000000000p+0, -0x1.77458f632c000p-5, -0x1.cfc4634f2a1eep-45) +A(0x1.0c00000000000p+0, -0x1.77458f632c000p-5, -0x1.cfc4634f2a1eep-45) +A(0x1.0a00000000000p+0, -0x1.39e87b9fec000p-5, 0x1.502b7f526feaap-48) +A(0x1.0a00000000000p+0, -0x1.39e87b9fec000p-5, 0x1.502b7f526feaap-48) +A(0x1.0800000000000p+0, -0x1.f829b0e780000p-6, -0x1.980267c7e09e4p-45) +A(0x1.0800000000000p+0, -0x1.f829b0e780000p-6, -0x1.980267c7e09e4p-45) +A(0x1.0600000000000p+0, -0x1.7b91b07d58000p-6, -0x1.88d5493faa639p-45) +A(0x1.0400000000000p+0, -0x1.fc0a8b0fc0000p-7, -0x1.f1e7cf6d3a69cp-50) +A(0x1.0400000000000p+0, -0x1.fc0a8b0fc0000p-7, -0x1.f1e7cf6d3a69cp-50) +A(0x1.0200000000000p+0, -0x1.fe02a6b100000p-8, -0x1.9e23f0dda40e4p-46) +A(0x1.0200000000000p+0, -0x1.fe02a6b100000p-8, -0x1.9e23f0dda40e4p-46) +A(0x1.0000000000000p+0, 0x0.0000000000000p+0, 0x0.0000000000000p+0) +A(0x1.0000000000000p+0, 0x0.0000000000000p+0, 0x0.0000000000000p+0) +A(0x1.fc00000000000p-1, 0x1.0101575890000p-7, -0x1.0c76b999d2be8p-46) +A(0x1.f800000000000p-1, 0x1.0205658938000p-6, -0x1.3dc5b06e2f7d2p-45) +A(0x1.f400000000000p-1, 0x1.8492528c90000p-6, -0x1.aa0ba325a0c34p-45) +A(0x1.f000000000000p-1, 0x1.0415d89e74000p-5, 0x1.111c05cf1d753p-47) +A(0x1.ec00000000000p-1, 0x1.466aed42e0000p-5, -0x1.c167375bdfd28p-45) +A(0x1.e800000000000p-1, 0x1.894aa149fc000p-5, -0x1.97995d05a267dp-46) +A(0x1.e400000000000p-1, 0x1.ccb73cdddc000p-5, -0x1.a68f247d82807p-46) +A(0x1.e200000000000p-1, 0x1.eea31c006c000p-5, -0x1.e113e4fc93b7bp-47) +A(0x1.de00000000000p-1, 0x1.1973bd1466000p-4, -0x1.5325d560d9e9bp-45) +A(0x1.da00000000000p-1, 0x1.3bdf5a7d1e000p-4, 0x1.cc85ea5db4ed7p-45) +A(0x1.d600000000000p-1, 0x1.5e95a4d97a000p-4, -0x1.c69063c5d1d1ep-45) +A(0x1.d400000000000p-1, 0x1.700d30aeac000p-4, 0x1.c1e8da99ded32p-49) +A(0x1.d000000000000p-1, 0x1.9335e5d594000p-4, 0x1.3115c3abd47dap-45) +A(0x1.cc00000000000p-1, 0x1.b6ac88dad6000p-4, -0x1.390802bf768e5p-46) +A(0x1.ca00000000000p-1, 0x1.c885801bc4000p-4, 0x1.646d1c65aacd3p-45) +A(0x1.c600000000000p-1, 0x1.ec739830a2000p-4, -0x1.dc068afe645e0p-45) +A(0x1.c400000000000p-1, 0x1.fe89139dbe000p-4, -0x1.534d64fa10afdp-45) +A(0x1.c000000000000p-1, 0x1.1178e8227e000p-3, 0x1.1ef78ce2d07f2p-45) +A(0x1.be00000000000p-1, 0x1.1aa2b7e23f000p-3, 0x1.ca78e44389934p-45) +A(0x1.ba00000000000p-1, 0x1.2d1610c868000p-3, 0x1.39d6ccb81b4a1p-47) +A(0x1.b800000000000p-1, 0x1.365fcb0159000p-3, 0x1.62fa8234b7289p-51) +A(0x1.b400000000000p-1, 0x1.4913d8333b000p-3, 0x1.5837954fdb678p-45) +A(0x1.b200000000000p-1, 0x1.527e5e4a1b000p-3, 0x1.633e8e5697dc7p-45) +A(0x1.ae00000000000p-1, 0x1.6574ebe8c1000p-3, 0x1.9cf8b2c3c2e78p-46) +A(0x1.ac00000000000p-1, 0x1.6f0128b757000p-3, -0x1.5118de59c21e1p-45) +A(0x1.aa00000000000p-1, 0x1.7898d85445000p-3, -0x1.c661070914305p-46) +A(0x1.a600000000000p-1, 0x1.8beafeb390000p-3, -0x1.73d54aae92cd1p-47) +A(0x1.a400000000000p-1, 0x1.95a5adcf70000p-3, 0x1.7f22858a0ff6fp-47) +A(0x1.a000000000000p-1, 0x1.a93ed3c8ae000p-3, -0x1.8724350562169p-45) +A(0x1.9e00000000000p-1, 0x1.b31d8575bd000p-3, -0x1.c358d4eace1aap-47) +A(0x1.9c00000000000p-1, 0x1.bd087383be000p-3, -0x1.d4bc4595412b6p-45) +A(0x1.9a00000000000p-1, 0x1.c6ffbc6f01000p-3, -0x1.1ec72c5962bd2p-48) +A(0x1.9600000000000p-1, 0x1.db13db0d49000p-3, -0x1.aff2af715b035p-45) +A(0x1.9400000000000p-1, 0x1.e530effe71000p-3, 0x1.212276041f430p-51) +A(0x1.9200000000000p-1, 0x1.ef5ade4dd0000p-3, -0x1.a211565bb8e11p-51) +A(0x1.9000000000000p-1, 0x1.f991c6cb3b000p-3, 0x1.bcbecca0cdf30p-46) +A(0x1.8c00000000000p-1, 0x1.07138604d5800p-2, 0x1.89cdb16ed4e91p-48) +A(0x1.8a00000000000p-1, 0x1.0c42d67616000p-2, 0x1.7188b163ceae9p-45) +A(0x1.8800000000000p-1, 0x1.1178e8227e800p-2, -0x1.c210e63a5f01cp-45) +A(0x1.8600000000000p-1, 0x1.16b5ccbacf800p-2, 0x1.b9acdf7a51681p-45) +A(0x1.8400000000000p-1, 0x1.1bf99635a6800p-2, 0x1.ca6ed5147bdb7p-45) +A(0x1.8200000000000p-1, 0x1.214456d0eb800p-2, 0x1.a87deba46baeap-47) +A(0x1.7e00000000000p-1, 0x1.2bef07cdc9000p-2, 0x1.a9cfa4a5004f4p-45) +A(0x1.7c00000000000p-1, 0x1.314f1e1d36000p-2, -0x1.8e27ad3213cb8p-45) +A(0x1.7a00000000000p-1, 0x1.36b6776be1000p-2, 0x1.16ecdb0f177c8p-46) +A(0x1.7800000000000p-1, 0x1.3c25277333000p-2, 0x1.83b54b606bd5cp-46) +A(0x1.7600000000000p-1, 0x1.419b423d5e800p-2, 0x1.8e436ec90e09dp-47) +A(0x1.7400000000000p-1, 0x1.4718dc271c800p-2, -0x1.f27ce0967d675p-45) +A(0x1.7200000000000p-1, 0x1.4c9e09e173000p-2, -0x1.e20891b0ad8a4p-45) +A(0x1.7000000000000p-1, 0x1.522ae0738a000p-2, 0x1.ebe708164c759p-45) +A(0x1.6e00000000000p-1, 0x1.57bf753c8d000p-2, 0x1.fadedee5d40efp-46) +A(0x1.6c00000000000p-1, 0x1.5d5bddf596000p-2, -0x1.a0b2a08a465dcp-47) +}, +}; diff --git a/kal/posix/musl_src/math/pow_data.h b/kal/posix/musl_src/math/pow_data.h new file mode 100644 index 00000000..18d2cb2e --- /dev/null +++ b/kal/posix/musl_src/math/pow_data.h @@ -0,0 +1,22 @@ +/* + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: MIT + */ +#ifndef _POW_DATA_H +#define _POW_DATA_H + +#include + +#define POW_LOG_TABLE_BITS 7 +#define POW_LOG_POLY_ORDER 8 +extern const struct pow_log_data { + double ln2hi; + double ln2lo; + double poly[POW_LOG_POLY_ORDER - 1]; /* First coefficient is 1. */ + /* Note: the pad field is unused, but allows slightly faster indexing. */ + struct { + double invc, pad, logc, logctail; + } tab[1 << POW_LOG_TABLE_BITS]; +} __pow_log_data1; + +#endif diff --git a/kal/posix/musl_src/math/round.c b/kal/posix/musl_src/math/round.c new file mode 100644 index 00000000..130d58d2 --- /dev/null +++ b/kal/posix/musl_src/math/round.c @@ -0,0 +1,35 @@ +#include "libm.h" + +#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1 +#define EPS DBL_EPSILON +#elif FLT_EVAL_METHOD==2 +#define EPS LDBL_EPSILON +#endif +static const double_t toint = 1/EPS; + +double round(double x) +{ + union {double f; uint64_t i;} u = {x}; + int e = u.i >> 52 & 0x7ff; + double_t y; + + if (e >= 0x3ff+52) + return x; + if (u.i >> 63) + x = -x; + if (e < 0x3ff-1) { + /* raise inexact if x!=0 */ + FORCE_EVAL(x + toint); + return 0*u.f; + } + y = x + toint - toint - x; + if (y > 0.5) + y = y + x - 1; + else if (y <= -0.5) + y = y + x + 1; + else + y = y + x; + if (u.i >> 63) + y = -y; + return y; +} diff --git a/kal/posix/musl_src/math/sqrt.c b/kal/posix/musl_src/math/sqrt.c new file mode 100644 index 00000000..f1f6d76c --- /dev/null +++ b/kal/posix/musl_src/math/sqrt.c @@ -0,0 +1,184 @@ +/* origin: FreeBSD /usr/src/lib/msun/src/e_sqrt.c */ +/* + * ==================================================== + * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved. + * + * Developed at SunSoft, a Sun Microsystems, Inc. business. + * Permission to use, copy, modify, and distribute this + * software is freely granted, provided that this notice + * is preserved. + * ==================================================== + */ +/* sqrt(x) + * Return correctly rounded sqrt. + * ------------------------------------------ + * | Use the hardware sqrt if you have one | + * ------------------------------------------ + * Method: + * Bit by bit method using integer arithmetic. (Slow, but portable) + * 1. Normalization + * Scale x to y in [1,4) with even powers of 2: + * find an integer k such that 1 <= (y=x*2^(2k)) < 4, then + * sqrt(x) = 2^k * sqrt(y) + * 2. Bit by bit computation + * Let q = sqrt(y) truncated to i bit after binary point (q = 1), + * i 0 + * i+1 2 + * s = 2*q , and y = 2 * ( y - q ). (1) + * i i i i + * + * To compute q from q , one checks whether + * i+1 i + * + * -(i+1) 2 + * (q + 2 ) <= y. (2) + * i + * -(i+1) + * If (2) is false, then q = q ; otherwise q = q + 2 . + * i+1 i i+1 i + * + * With some algebric manipulation, it is not difficult to see + * that (2) is equivalent to + * -(i+1) + * s + 2 <= y (3) + * i i + * + * The advantage of (3) is that s and y can be computed by + * i i + * the following recurrence formula: + * if (3) is false + * + * s = s , y = y ; (4) + * i+1 i i+1 i + * + * otherwise, + * -i -(i+1) + * s = s + 2 , y = y - s - 2 (5) + * i+1 i i+1 i i + * + * One may easily use induction to prove (4) and (5). + * Note. Since the left hand side of (3) contain only i+2 bits, + * it does not necessary to do a full (53-bit) comparison + * in (3). + * 3. Final rounding + * After generating the 53 bits result, we compute one more bit. + * Together with the remainder, we can decide whether the + * result is exact, bigger than 1/2ulp, or less than 1/2ulp + * (it will never equal to 1/2ulp). + * The rounding mode can be detected by checking whether + * huge + tiny is equal to huge, and whether huge - tiny is + * equal to huge for some floating point number "huge" and "tiny". + * + * Special cases: + * sqrt(+-0) = +-0 ... exact + * sqrt(inf) = inf + * sqrt(-ve) = NaN ... with invalid signal + * sqrt(NaN) = NaN ... with invalid signal for signaling NaN + */ + +#include "libm.h" + +static const double tiny = 1.0e-300; + +double sqrt(double x) +{ + double z; + int32_t sign = (int)0x80000000; + int32_t ix0,s0,q,m,t,i; + uint32_t r,t1,s1,ix1,q1; + + EXTRACT_WORDS(ix0, ix1, x); + + /* take care of Inf and NaN */ + if ((ix0&0x7ff00000) == 0x7ff00000) { + return x*x + x; /* sqrt(NaN)=NaN, sqrt(+inf)=+inf, sqrt(-inf)=sNaN */ + } + /* take care of zero */ + if (ix0 <= 0) { + if (((ix0&~sign)|ix1) == 0) + return x; /* sqrt(+-0) = +-0 */ + if (ix0 < 0) + return (x-x)/(x-x); /* sqrt(-ve) = sNaN */ + } + /* normalize x */ + m = ix0>>20; + if (m == 0) { /* subnormal x */ + while (ix0 == 0) { + m -= 21; + ix0 |= (ix1>>11); + ix1 <<= 21; + } + for (i=0; (ix0&0x00100000) == 0; i++) + ix0<<=1; + m -= i - 1; + ix0 |= ix1>>(32-i); + ix1 <<= i; + } + m -= 1023; /* unbias exponent */ + ix0 = (ix0&0x000fffff)|0x00100000; + if (m & 1) { /* odd m, double x to make it even */ + ix0 += ix0 + ((ix1&sign)>>31); + ix1 += ix1; + } + m >>= 1; /* m = [m/2] */ + + /* generate sqrt(x) bit by bit */ + ix0 += ix0 + ((ix1&sign)>>31); + ix1 += ix1; + q = q1 = s0 = s1 = 0; /* [q,q1] = sqrt(x) */ + r = 0x00200000; /* r = moving bit from right to left */ + + while (r != 0) { + t = s0 + r; + if (t <= ix0) { + s0 = t + r; + ix0 -= t; + q += r; + } + ix0 += ix0 + ((ix1&sign)>>31); + ix1 += ix1; + r >>= 1; + } + + r = sign; + while (r != 0) { + t1 = s1 + r; + t = s0; + if (t < ix0 || (t == ix0 && t1 <= ix1)) { + s1 = t1 + r; + if ((t1&sign) == sign && (s1&sign) == 0) + s0++; + ix0 -= t; + if (ix1 < t1) + ix0--; + ix1 -= t1; + q1 += r; + } + ix0 += ix0 + ((ix1&sign)>>31); + ix1 += ix1; + r >>= 1; + } + + /* use floating add to find out rounding direction */ + if ((ix0|ix1) != 0) { + z = 1.0 - tiny; /* raise inexact flag */ + if (z >= 1.0) { + z = 1.0 + tiny; + if (q1 == (uint32_t)0xffffffff) { + q1 = 0; + q++; + } else if (z > 1.0) { + if (q1 == (uint32_t)0xfffffffe) + q++; + q1 += 2; + } else + q1 += q1 & 1; + } + } + ix0 = (q>>1) + 0x3fe00000; + ix1 = q1>>1; + if (q&1) + ix1 |= sign; + INSERT_WORDS(z, ix0 + ((uint32_t)m << 20), ix1); + return z; +} diff --git a/kal/posix/musl_src/misc/dirname.c b/kal/posix/musl_src/misc/dirname.c new file mode 100644 index 00000000..dd570883 --- /dev/null +++ b/kal/posix/musl_src/misc/dirname.c @@ -0,0 +1,14 @@ +#include +#include + +char *dirname(char *s) +{ + size_t i; + if (!s || !*s) return "."; + i = strlen(s)-1; + for (; s[i]=='/'; i--) if (!i) return "/"; + for (; s[i]!='/'; i--) if (!i) return "."; + for (; s[i]=='/'; i--) if (!i) return "/"; + s[i+1] = 0; + return s; +} diff --git a/kal/posix/musl_src/prng/random.c b/kal/posix/musl_src/prng/random.c new file mode 100644 index 00000000..dafd3098 --- /dev/null +++ b/kal/posix/musl_src/prng/random.c @@ -0,0 +1,126 @@ +#include +#include +#include "los_context.h" +/* +this code uses the same lagged fibonacci generator as the +original bsd random implementation except for the seeding +which was broken in the original +*/ + +static uint32_t init[] = { +0x00000000,0x5851f42d,0xc0b18ccf,0xcbb5f646, +0xc7033129,0x30705b04,0x20fd5db4,0x9a8b7f78, +0x502959d8,0xab894868,0x6c0356a7,0x88cdb7ff, +0xb477d43f,0x70a3a52b,0xa8e4baf1,0xfd8341fc, +0x8ae16fd9,0x742d2f7a,0x0d1f0796,0x76035e09, +0x40f7702c,0x6fa72ca5,0xaaa84157,0x58a0df74, +0xc74a0364,0xae533cc4,0x04185faf,0x6de3b115, +0x0cab8628,0xf043bfa4,0x398150e9,0x37521657}; + +static int n = 31; +static int i = 3; +static int j = 0; +static uint32_t *x = init+1; + +static uint32_t lcg31(uint32_t x) { + return (1103515245*x + 12345) & 0x7fffffff; +} + +static uint64_t lcg64(uint64_t x) { + return 6364136223846793005ull*x + 1; +} + +static void *savestate(void) { + x[-1] = (n<<16)|(i<<8)|j; + return x-1; +} + +static void loadstate(uint32_t *state) { + x = state+1; + n = x[-1]>>16; + i = (x[-1]>>8)&0xff; + j = x[-1]&0xff; +} + +static void __srandom(unsigned seed) { + int k; + uint64_t s = seed; + + if (n == 0) { + x[0] = s; + return; + } + i = n == 31 || n == 7 ? 3 : 1; + j = 0; + for (k = 0; k < n; k++) { + s = lcg64(s); + x[k] = s>>32; + } + /* make sure x contains at least one odd number */ + x[0] |= 1; +} + +void srandom(unsigned seed) { + unsigned int intSave; + + intSave = LOS_IntLock(); + __srandom(seed); + LOS_IntRestore(intSave); +} + +char *initstate(unsigned seed, char *state, size_t size) { + void *old; + unsigned int intSave; + + if (size < 8) + return 0; + + intSave = LOS_IntLock(); + old = savestate(); + if (size < 32) + n = 0; + else if (size < 64) + n = 7; + else if (size < 128) + n = 15; + else if (size < 256) + n = 31; + else + n = 63; + x = (uint32_t*)state + 1; + __srandom(seed); + savestate(); + LOS_IntRestore(intSave); + return old; +} + +char *setstate(char *state) { + void *old; + unsigned int intSave; + + intSave = LOS_IntLock(); + old = savestate(); + loadstate((uint32_t*)state); + LOS_IntRestore(intSave); + return old; +} + +long random(void) { + long k; + unsigned int intSave; + + intSave = LOS_IntLock(); + if (n == 0) { + k = x[0] = lcg31(x[0]); + goto end; + } + x[i] += x[j]; + k = x[i]>>1; + if (++i == n) + i = 0; + if (++j == n) + j = 0; +end: + LOS_IntRestore(intSave); + return k; +} diff --git a/kal/posix/musl_src/regex/regcomp.c b/kal/posix/musl_src/regex/regcomp.c new file mode 100644 index 00000000..fb24556e --- /dev/null +++ b/kal/posix/musl_src/regex/regcomp.c @@ -0,0 +1,2953 @@ +/* + regcomp.c - TRE POSIX compatible regex compilation functions. + + Copyright (c) 2001-2009 Ville Laurikari + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + 1. Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + + 2. Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS + ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +*/ + +#include +#include +#include +#include +#include +#include + +#include "tre.h" + +#include + +/*********************************************************************** + from tre-compile.h +***********************************************************************/ + +typedef struct { + int position; + int code_min; + int code_max; + int *tags; + int assertions; + tre_ctype_t class; + tre_ctype_t *neg_classes; + int backref; +} tre_pos_and_tags_t; + + +/*********************************************************************** + from tre-ast.c and tre-ast.h +***********************************************************************/ + +/* The different AST node types. */ +typedef enum { + LITERAL, + CATENATION, + ITERATION, + UNION +} tre_ast_type_t; + +/* Special subtypes of TRE_LITERAL. */ +#define EMPTY -1 /* Empty leaf (denotes empty string). */ +#define ASSERTION -2 /* Assertion leaf. */ +#define TAG -3 /* Tag leaf. */ +#define BACKREF -4 /* Back reference leaf. */ + +#define IS_SPECIAL(x) ((x)->code_min < 0) +#define IS_EMPTY(x) ((x)->code_min == EMPTY) +#define IS_ASSERTION(x) ((x)->code_min == ASSERTION) +#define IS_TAG(x) ((x)->code_min == TAG) +#define IS_BACKREF(x) ((x)->code_min == BACKREF) + + +/* A generic AST node. All AST nodes consist of this node on the top + level with `obj' pointing to the actual content. */ +typedef struct { + tre_ast_type_t type; /* Type of the node. */ + void *obj; /* Pointer to actual node. */ + int nullable; + int submatch_id; + int num_submatches; + int num_tags; + tre_pos_and_tags_t *firstpos; + tre_pos_and_tags_t *lastpos; +} tre_ast_node_t; + + +/* A "literal" node. These are created for assertions, back references, + tags, matching parameter settings, and all expressions that match one + character. */ +typedef struct { + long code_min; + long code_max; + int position; + tre_ctype_t class; + tre_ctype_t *neg_classes; +} tre_literal_t; + +/* A "catenation" node. These are created when two regexps are concatenated. + If there are more than one subexpressions in sequence, the `left' part + holds all but the last, and `right' part holds the last subexpression + (catenation is left associative). */ +typedef struct { + tre_ast_node_t *left; + tre_ast_node_t *right; +} tre_catenation_t; + +/* An "iteration" node. These are created for the "*", "+", "?", and "{m,n}" + operators. */ +typedef struct { + /* Subexpression to match. */ + tre_ast_node_t *arg; + /* Minimum number of consecutive matches. */ + int min; + /* Maximum number of consecutive matches. */ + int max; + /* If 0, match as many characters as possible, if 1 match as few as + possible. Note that this does not always mean the same thing as + matching as many/few repetitions as possible. */ + unsigned int minimal:1; +} tre_iteration_t; + +/* An "union" node. These are created for the "|" operator. */ +typedef struct { + tre_ast_node_t *left; + tre_ast_node_t *right; +} tre_union_t; + + +static tre_ast_node_t * +tre_ast_new_node(tre_mem_t mem, int type, void *obj) +{ + tre_ast_node_t *node = tre_mem_calloc(mem, sizeof *node); + if (!node || !obj) + return 0; + node->obj = obj; + node->type = type; + node->nullable = -1; + node->submatch_id = -1; + return node; +} + +static tre_ast_node_t * +tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position) +{ + tre_ast_node_t *node; + tre_literal_t *lit; + + lit = tre_mem_calloc(mem, sizeof *lit); + node = tre_ast_new_node(mem, LITERAL, lit); + if (!node) + return 0; + lit->code_min = code_min; + lit->code_max = code_max; + lit->position = position; + return node; +} + +static tre_ast_node_t * +tre_ast_new_iter(tre_mem_t mem, tre_ast_node_t *arg, int min, int max, int minimal) +{ + tre_ast_node_t *node; + tre_iteration_t *iter; + + iter = tre_mem_calloc(mem, sizeof *iter); + node = tre_ast_new_node(mem, ITERATION, iter); + if (!node) + return 0; + iter->arg = arg; + iter->min = min; + iter->max = max; + iter->minimal = minimal; + node->num_submatches = arg->num_submatches; + return node; +} + +static tre_ast_node_t * +tre_ast_new_union(tre_mem_t mem, tre_ast_node_t *left, tre_ast_node_t *right) +{ + tre_ast_node_t *node; + tre_union_t *un; + + if (!left) + return right; + un = tre_mem_calloc(mem, sizeof *un); + node = tre_ast_new_node(mem, UNION, un); + if (!node || !right) + return 0; + un->left = left; + un->right = right; + node->num_submatches = left->num_submatches + right->num_submatches; + return node; +} + +static tre_ast_node_t * +tre_ast_new_catenation(tre_mem_t mem, tre_ast_node_t *left, tre_ast_node_t *right) +{ + tre_ast_node_t *node; + tre_catenation_t *cat; + + if (!left) + return right; + cat = tre_mem_calloc(mem, sizeof *cat); + node = tre_ast_new_node(mem, CATENATION, cat); + if (!node) + return 0; + cat->left = left; + cat->right = right; + node->num_submatches = left->num_submatches + right->num_submatches; + return node; +} + + +/*********************************************************************** + from tre-stack.c and tre-stack.h +***********************************************************************/ + +typedef struct tre_stack_rec tre_stack_t; + +/* Creates a new stack object. `size' is initial size in bytes, `max_size' + is maximum size, and `increment' specifies how much more space will be + allocated with realloc() if all space gets used up. Returns the stack + object or NULL if out of memory. */ +static tre_stack_t * +tre_stack_new(int size, int max_size, int increment); + +/* Frees the stack object. */ +static void +tre_stack_destroy(tre_stack_t *s); + +/* Returns the current number of objects in the stack. */ +static int +tre_stack_num_objects(tre_stack_t *s); + +/* Each tre_stack_push_*(tre_stack_t *s, value) function pushes + `value' on top of stack `s'. Returns REG_ESPACE if out of memory. + This tries to realloc() more space before failing if maximum size + has not yet been reached. Returns REG_OK if successful. */ +#define declare_pushf(typetag, type) \ + static reg_errcode_t tre_stack_push_ ## typetag(tre_stack_t *s, type value) + +declare_pushf(voidptr, void *); +declare_pushf(int, int); + +/* Each tre_stack_pop_*(tre_stack_t *s) function pops the topmost + element off of stack `s' and returns it. The stack must not be + empty. */ +#define declare_popf(typetag, type) \ + static type tre_stack_pop_ ## typetag(tre_stack_t *s) + +declare_popf(voidptr, void *); +declare_popf(int, int); + +/* Just to save some typing. */ +#define STACK_PUSH(s, typetag, value) \ + do \ + { \ + status = tre_stack_push_ ## typetag(s, value); \ + } \ + while (/*CONSTCOND*/0) + +#define STACK_PUSHX(s, typetag, value) \ + { \ + status = tre_stack_push_ ## typetag(s, value); \ + if (status != REG_OK) \ + break; \ + } + +#define STACK_PUSHR(s, typetag, value) \ + { \ + reg_errcode_t _status; \ + _status = tre_stack_push_ ## typetag(s, value); \ + if (_status != REG_OK) \ + return _status; \ + } + +union tre_stack_item { + void *voidptr_value; + int int_value; +}; + +struct tre_stack_rec { + int size; + int max_size; + int increment; + int ptr; + union tre_stack_item *stack; +}; + + +static tre_stack_t * +tre_stack_new(int size, int max_size, int increment) +{ + tre_stack_t *s; + + s = xmalloc(sizeof(*s)); + if (s != NULL) + { + s->stack = xmalloc(sizeof(*s->stack) * size); + if (s->stack == NULL) + { + xfree(s); + return NULL; + } + s->size = size; + s->max_size = max_size; + s->increment = increment; + s->ptr = 0; + } + return s; +} + +static void +tre_stack_destroy(tre_stack_t *s) +{ + xfree(s->stack); + xfree(s); +} + +static int +tre_stack_num_objects(tre_stack_t *s) +{ + return s->ptr; +} + +static reg_errcode_t +tre_stack_push(tre_stack_t *s, union tre_stack_item value) +{ + if (s->ptr < s->size) + { + s->stack[s->ptr] = value; + s->ptr++; + } + else + { + if (s->size >= s->max_size) + { + return REG_ESPACE; + } + else + { + union tre_stack_item *new_buffer; + int new_size; + new_size = s->size + s->increment; + if (new_size > s->max_size) + new_size = s->max_size; + new_buffer = xrealloc(s->stack, sizeof(*new_buffer) * new_size); + if (new_buffer == NULL) + { + return REG_ESPACE; + } + assert(new_size > s->size); + s->size = new_size; + s->stack = new_buffer; + tre_stack_push(s, value); + } + } + return REG_OK; +} + +#define define_pushf(typetag, type) \ + declare_pushf(typetag, type) { \ + union tre_stack_item item; \ + item.typetag ## _value = value; \ + return tre_stack_push(s, item); \ +} + +define_pushf(int, int) +define_pushf(voidptr, void *) + +#define define_popf(typetag, type) \ + declare_popf(typetag, type) { \ + return s->stack[--s->ptr].typetag ## _value; \ + } + +define_popf(int, int) +define_popf(voidptr, void *) + + +/*********************************************************************** + from tre-parse.c and tre-parse.h +***********************************************************************/ + +/* Parse context. */ +typedef struct { + /* Memory allocator. The AST is allocated using this. */ + tre_mem_t mem; + /* Stack used for keeping track of regexp syntax. */ + tre_stack_t *stack; + /* The parsed node after a parse function returns. */ + tre_ast_node_t *n; + /* Position in the regexp pattern after a parse function returns. */ + const char *s; + /* The first character of the last subexpression parsed. */ + const char *start; + /* Current submatch ID. */ + int submatch_id; + /* Current position (number of literal). */ + int position; + /* The highest back reference or -1 if none seen so far. */ + int max_backref; + /* Compilation flags. */ + int cflags; +} tre_parse_ctx_t; + +/* Some macros for expanding \w, \s, etc. */ +static const struct { + char c; + const char *expansion; +} tre_macros[] = { + {'t', "\t"}, {'n', "\n"}, {'r', "\r"}, + {'f', "\f"}, {'a', "\a"}, {'e', "\033"}, + {'w', "[[:alnum:]_]"}, {'W', "[^[:alnum:]_]"}, {'s', "[[:space:]]"}, + {'S', "[^[:space:]]"}, {'d', "[[:digit:]]"}, {'D', "[^[:digit:]]"}, + { 0, 0 } +}; + +/* Expands a macro delimited by `regex' and `regex_end' to `buf', which + must have at least `len' items. Sets buf[0] to zero if the there + is no match in `tre_macros'. */ +static const char *tre_expand_macro(const char *s) +{ + int i; + for (i = 0; tre_macros[i].c && tre_macros[i].c != *s; i++); + return tre_macros[i].expansion; +} + +static int +tre_compare_lit(const void *a, const void *b) +{ + const tre_literal_t *const *la = a; + const tre_literal_t *const *lb = b; + /* assumes the range of valid code_min is < INT_MAX */ + return la[0]->code_min - lb[0]->code_min; +} + +struct literals { + tre_mem_t mem; + tre_literal_t **a; + int len; + int cap; +}; + +static tre_literal_t *tre_new_lit(struct literals *p) +{ + tre_literal_t **a; + if (p->len >= p->cap) { + if (p->cap >= 1<<15) + return 0; + p->cap *= 2; + a = xrealloc(p->a, p->cap * sizeof *p->a); + if (!a) + return 0; + p->a = a; + } + a = p->a + p->len++; + *a = tre_mem_calloc(p->mem, sizeof **a); + return *a; +} + +static int add_icase_literals(struct literals *ls, int min, int max) +{ + tre_literal_t *lit; + int b, e, c; + for (c=min; c<=max; ) { + /* assumes islower(c) and isupper(c) are exclusive + and toupper(c)!=c if islower(c). + multiple opposite case characters are not supported */ + if (tre_islower(c)) { + b = e = tre_toupper(c); + for (c++, e++; c<=max; c++, e++) + if (tre_toupper(c) != e) break; + } else if (tre_isupper(c)) { + b = e = tre_tolower(c); + for (c++, e++; c<=max; c++, e++) + if (tre_tolower(c) != e) break; + } else { + c++; + continue; + } + lit = tre_new_lit(ls); + if (!lit) + return -1; + lit->code_min = b; + lit->code_max = e-1; + lit->position = -1; + } + return 0; +} + + +/* Maximum number of character classes in a negated bracket expression. */ +#define MAX_NEG_CLASSES 64 + +struct neg { + int negate; + int len; + tre_ctype_t a[MAX_NEG_CLASSES]; +}; + +// TODO: parse bracket into a set of non-overlapping [lo,hi] ranges + +/* +bracket grammar: +Bracket = '[' List ']' | '[^' List ']' +List = Term | List Term +Term = Char | Range | Chclass | Eqclass +Range = Char '-' Char | Char '-' '-' +Char = Coll | coll_single +Meta = ']' | '-' +Coll = '[.' coll_single '.]' | '[.' coll_multi '.]' | '[.' Meta '.]' +Eqclass = '[=' coll_single '=]' | '[=' coll_multi '=]' +Chclass = '[:' class ':]' + +coll_single is a single char collating element but it can be + '-' only at the beginning or end of a List and + ']' only at the beginning of a List and + '^' anywhere except after the openning '[' +*/ + +static reg_errcode_t parse_bracket_terms(tre_parse_ctx_t *ctx, const char *s, struct literals *ls, struct neg *neg) +{ + const char *start = s; + tre_ctype_t class; + int min, max; + wchar_t wc; + int len; + + for (;;) { + class = 0; + len = mbtowc(&wc, s, -1); + if (len <= 0) + return *s ? REG_BADPAT : REG_EBRACK; + if (*s == ']' && s != start) { + ctx->s = s+1; + return REG_OK; + } + if (*s == '-' && s != start && s[1] != ']' && + /* extension: [a-z--@] is accepted as [a-z]|[--@] */ + (s[1] != '-' || s[2] == ']')) + return REG_ERANGE; + if (*s == '[' && (s[1] == '.' || s[1] == '=')) + /* collating symbols and equivalence classes are not supported */ + return REG_ECOLLATE; + if (*s == '[' && s[1] == ':') { + char tmp[CHARCLASS_NAME_MAX+1]; + s += 2; + for (len=0; len < CHARCLASS_NAME_MAX && s[len]; len++) { + if (s[len] == ':') { + memcpy(tmp, s, len); + tmp[len] = 0; + class = tre_ctype(tmp); + break; + } + } + if (!class || s[len+1] != ']') + return REG_ECTYPE; + min = 0; + max = TRE_CHAR_MAX; + s += len+2; + } else { + min = max = wc; + s += len; + if (*s == '-' && s[1] != ']') { + s++; + len = mbtowc(&wc, s, -1); + max = wc; + /* XXX - Should use collation order instead of + encoding values in character ranges. */ + if (len <= 0 || min > max) + return REG_ERANGE; + s += len; + } + } + + if (class && neg->negate) { + if (neg->len >= MAX_NEG_CLASSES) + return REG_ESPACE; + neg->a[neg->len++] = class; + } else { + tre_literal_t *lit = tre_new_lit(ls); + if (!lit) + return REG_ESPACE; + lit->code_min = min; + lit->code_max = max; + lit->class = class; + lit->position = -1; + + /* Add opposite-case codepoints if REG_ICASE is present. + It seems that POSIX requires that bracket negation + should happen before case-folding, but most practical + implementations do it the other way around. Changing + the order would need efficient representation of + case-fold ranges and bracket range sets even with + simple patterns so this is ok for now. */ + if (ctx->cflags & REG_ICASE && !class) + if (add_icase_literals(ls, min, max)) + return REG_ESPACE; + } + } +} + +static reg_errcode_t parse_bracket(tre_parse_ctx_t *ctx, const char *s) +{ + int i, max, min, negmax, negmin; + tre_ast_node_t *node = 0, *n; + tre_ctype_t *nc = 0; + tre_literal_t *lit; + struct literals ls; + struct neg neg; + reg_errcode_t err; + + ls.mem = ctx->mem; + ls.len = 0; + ls.cap = 32; + ls.a = xmalloc(ls.cap * sizeof *ls.a); + if (!ls.a) + return REG_ESPACE; + neg.len = 0; + neg.negate = *s == '^'; + if (neg.negate) + s++; + + err = parse_bracket_terms(ctx, s, &ls, &neg); + if (err != REG_OK) + goto parse_bracket_done; + + if (neg.negate) { + /* + * With REG_NEWLINE, POSIX requires that newlines are not matched by + * any form of a non-matching list. + */ + if (ctx->cflags & REG_NEWLINE) { + lit = tre_new_lit(&ls); + if (!lit) { + err = REG_ESPACE; + goto parse_bracket_done; + } + lit->code_min = '\n'; + lit->code_max = '\n'; + lit->position = -1; + } + /* Sort the array if we need to negate it. */ + qsort(ls.a, ls.len, sizeof *ls.a, tre_compare_lit); + /* extra lit for the last negated range */ + lit = tre_new_lit(&ls); + if (!lit) { + err = REG_ESPACE; + goto parse_bracket_done; + } + lit->code_min = TRE_CHAR_MAX+1; + lit->code_max = TRE_CHAR_MAX+1; + lit->position = -1; + /* negated classes */ + if (neg.len) { + nc = tre_mem_alloc(ctx->mem, (neg.len+1)*sizeof *neg.a); + if (!nc) { + err = REG_ESPACE; + goto parse_bracket_done; + } + memcpy(nc, neg.a, neg.len*sizeof *neg.a); + nc[neg.len] = 0; + } + } + + /* Build a union of the items in the array, negated if necessary. */ + negmax = negmin = 0; + for (i = 0; i < ls.len; i++) { + lit = ls.a[i]; + min = lit->code_min; + max = lit->code_max; + if (neg.negate) { + if (min <= negmin) { + /* Overlap. */ + negmin = MAX(max + 1, negmin); + continue; + } + negmax = min - 1; + lit->code_min = negmin; + lit->code_max = negmax; + negmin = max + 1; + } + lit->position = ctx->position; + lit->neg_classes = nc; + n = tre_ast_new_node(ctx->mem, LITERAL, lit); + node = tre_ast_new_union(ctx->mem, node, n); + if (!node) { + err = REG_ESPACE; + break; + } + } + +parse_bracket_done: + xfree(ls.a); + ctx->position++; + ctx->n = node; + return err; +} + +static const char *parse_dup_count(const char *s, int *n) +{ + *n = -1; + if (!isdigit(*s)) + return s; + *n = 0; + for (;;) { + *n = 10 * *n + (*s - '0'); + s++; + if (!isdigit(*s) || *n > RE_DUP_MAX) + break; + } + return s; +} + +static const char *parse_dup(const char *s, int ere, int *pmin, int *pmax) +{ + int min, max; + + s = parse_dup_count(s, &min); + if (*s == ',') + s = parse_dup_count(s+1, &max); + else + max = min; + + if ( + (max < min && max >= 0) || + max > RE_DUP_MAX || + min > RE_DUP_MAX || + min < 0 || + (!ere && *s++ != '\\') || + *s++ != '}' + ) + return 0; + *pmin = min; + *pmax = max; + return s; +} + +static int hexval(unsigned c) +{ + if (c-'0'<10) return c-'0'; + c |= 32; + if (c-'a'<6) return c-'a'+10; + return -1; +} + +static reg_errcode_t marksub(tre_parse_ctx_t *ctx, tre_ast_node_t *node, int subid) +{ + if (node->submatch_id >= 0) { + tre_ast_node_t *n = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1); + if (!n) + return REG_ESPACE; + n = tre_ast_new_catenation(ctx->mem, n, node); + if (!n) + return REG_ESPACE; + n->num_submatches = node->num_submatches; + node = n; + } + node->submatch_id = subid; + node->num_submatches++; + ctx->n = node; + return REG_OK; +} + +/* +BRE grammar: +Regex = Branch | '^' | '$' | '^$' | '^' Branch | Branch '$' | '^' Branch '$' +Branch = Atom | Branch Atom +Atom = char | quoted_char | '.' | Bracket | Atom Dup | '\(' Branch '\)' | back_ref +Dup = '*' | '\{' Count '\}' | '\{' Count ',\}' | '\{' Count ',' Count '\}' + +(leading ^ and trailing $ in a sub expr may be an anchor or literal as well) + +ERE grammar: +Regex = Branch | Regex '|' Branch +Branch = Atom | Branch Atom +Atom = char | quoted_char | '.' | Bracket | Atom Dup | '(' Regex ')' | '^' | '$' +Dup = '*' | '+' | '?' | '{' Count '}' | '{' Count ',}' | '{' Count ',' Count '}' + +(a*+?, ^*, $+, \X, {, (|a) are unspecified) +*/ + +static reg_errcode_t parse_atom(tre_parse_ctx_t *ctx, const char *s) +{ + int len, ere = ctx->cflags & REG_EXTENDED; + const char *p; + tre_ast_node_t *node; + wchar_t wc; + switch (*s) { + case '[': + return parse_bracket(ctx, s+1); + case '\\': + p = tre_expand_macro(s+1); + if (p) { + /* assume \X expansion is a single atom */ + reg_errcode_t err = parse_atom(ctx, p); + ctx->s = s+2; + return err; + } + /* extensions: \b, \B, \<, \>, \xHH \x{HHHH} */ + switch (*++s) { + case 0: + return REG_EESCAPE; + case 'b': + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_WB, -1); + break; + case 'B': + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_WB_NEG, -1); + break; + case '<': + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_BOW, -1); + break; + case '>': + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_EOW, -1); + break; + case 'x': + s++; + int i, v = 0, c; + len = 2; + if (*s == '{') { + len = 8; + s++; + } + for (i=0; imem, v, v, ctx->position++); + s--; + break; + case '{': + case '+': + case '?': + /* extension: treat \+, \? as repetitions in BRE */ + /* reject repetitions after empty expression in BRE */ + if (!ere) + return REG_BADRPT; + case '|': + /* extension: treat \| as alternation in BRE */ + if (!ere) { + node = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1); + s--; + goto end; + } + /* fallthrough */ + default: + if (!ere && (unsigned)*s-'1' < 9) { + /* back reference */ + int val = *s - '0'; + node = tre_ast_new_literal(ctx->mem, BACKREF, val, ctx->position++); + ctx->max_backref = MAX(val, ctx->max_backref); + } else { + /* extension: accept unknown escaped char + as a literal */ + goto parse_literal; + } + } + s++; + break; + case '.': + if (ctx->cflags & REG_NEWLINE) { + tre_ast_node_t *tmp1, *tmp2; + tmp1 = tre_ast_new_literal(ctx->mem, 0, '\n'-1, ctx->position++); + tmp2 = tre_ast_new_literal(ctx->mem, '\n'+1, TRE_CHAR_MAX, ctx->position++); + if (tmp1 && tmp2) + node = tre_ast_new_union(ctx->mem, tmp1, tmp2); + else + node = 0; + } else { + node = tre_ast_new_literal(ctx->mem, 0, TRE_CHAR_MAX, ctx->position++); + } + s++; + break; + case '^': + /* '^' has a special meaning everywhere in EREs, and at beginning of BRE. */ + if (!ere && s != ctx->start) + goto parse_literal; + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_BOL, -1); + s++; + break; + case '$': + /* '$' is special everywhere in EREs, and at the end of a BRE subexpression. */ + if (!ere && s[1] && (s[1]!='\\'|| (s[2]!=')' && s[2]!='|'))) + goto parse_literal; + node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_EOL, -1); + s++; + break; + case '*': + case '{': + case '+': + case '?': + /* reject repetitions after empty expression in ERE */ + if (ere) + return REG_BADRPT; + case '|': + if (!ere) + goto parse_literal; + case 0: + node = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1); + break; + default: +parse_literal: + len = mbtowc(&wc, s, -1); + if (len < 0) + return REG_BADPAT; + if (ctx->cflags & REG_ICASE && (tre_isupper(wc) || tre_islower(wc))) { + tre_ast_node_t *tmp1, *tmp2; + /* multiple opposite case characters are not supported */ + tmp1 = tre_ast_new_literal(ctx->mem, tre_toupper(wc), tre_toupper(wc), ctx->position); + tmp2 = tre_ast_new_literal(ctx->mem, tre_tolower(wc), tre_tolower(wc), ctx->position); + if (tmp1 && tmp2) + node = tre_ast_new_union(ctx->mem, tmp1, tmp2); + else + node = 0; + } else { + node = tre_ast_new_literal(ctx->mem, wc, wc, ctx->position); + } + ctx->position++; + s += len; + break; + } +end: + if (!node) + return REG_ESPACE; + ctx->n = node; + ctx->s = s; + return REG_OK; +} + +#define PUSHPTR(err, s, v) do { \ + if ((err = tre_stack_push_voidptr(s, v)) != REG_OK) \ + return err; \ +} while(0) + +#define PUSHINT(err, s, v) do { \ + if ((err = tre_stack_push_int(s, v)) != REG_OK) \ + return err; \ +} while(0) + +static reg_errcode_t tre_parse(tre_parse_ctx_t *ctx) +{ + tre_ast_node_t *nbranch=0, *nunion=0; + int ere = ctx->cflags & REG_EXTENDED; + const char *s = ctx->start; + int subid = 0; + int depth = 0; + reg_errcode_t err; + tre_stack_t *stack = ctx->stack; + + PUSHINT(err, stack, subid++); + for (;;) { + if ((!ere && *s == '\\' && s[1] == '(') || + (ere && *s == '(')) { + PUSHPTR(err, stack, nunion); + PUSHPTR(err, stack, nbranch); + PUSHINT(err, stack, subid++); + s++; + if (!ere) + s++; + depth++; + nbranch = nunion = 0; + ctx->start = s; + continue; + } + if ((!ere && *s == '\\' && s[1] == ')') || + (ere && *s == ')' && depth)) { + ctx->n = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1); + if (!ctx->n) + return REG_ESPACE; + } else { + err = parse_atom(ctx, s); + if (err != REG_OK) + return err; + s = ctx->s; + } + + parse_iter: + for (;;) { + int min, max; + + if (*s!='\\' && *s!='*') { + if (!ere) + break; + if (*s!='+' && *s!='?' && *s!='{') + break; + } + if (*s=='\\' && ere) + break; + /* extension: treat \+, \? as repetitions in BRE */ + if (*s=='\\' && s[1]!='+' && s[1]!='?' && s[1]!='{') + break; + if (*s=='\\') + s++; + + /* handle ^* at the start of a BRE. */ + if (!ere && s==ctx->start+1 && s[-1]=='^') + break; + + /* extension: multiple consecutive *+?{,} is unspecified, + but (a+)+ has to be supported so accepting a++ makes + sense, note however that the RE_DUP_MAX limit can be + circumvented: (a{255}){255} uses a lot of memory.. */ + if (*s=='{') { + s = parse_dup(s+1, ere, &min, &max); + if (!s) + return REG_BADBR; + } else { + min=0; + max=-1; + if (*s == '+') + min = 1; + if (*s == '?') + max = 1; + s++; + } + if (max == 0) + ctx->n = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1); + else + ctx->n = tre_ast_new_iter(ctx->mem, ctx->n, min, max, 0); + if (!ctx->n) + return REG_ESPACE; + } + + nbranch = tre_ast_new_catenation(ctx->mem, nbranch, ctx->n); + if ((ere && *s == '|') || + (ere && *s == ')' && depth) || + (!ere && *s == '\\' && s[1] == ')') || + /* extension: treat \| as alternation in BRE */ + (!ere && *s == '\\' && s[1] == '|') || + !*s) { + /* extension: empty branch is unspecified (), (|a), (a|) + here they are not rejected but match on empty string */ + int c = *s; + nunion = tre_ast_new_union(ctx->mem, nunion, nbranch); + nbranch = 0; + + if (c == '\\' && s[1] == '|') { + s+=2; + ctx->start = s; + } else if (c == '|') { + s++; + ctx->start = s; + } else { + if (c == '\\') { + if (!depth) return REG_EPAREN; + s+=2; + } else if (c == ')') + s++; + depth--; + err = marksub(ctx, nunion, tre_stack_pop_int(stack)); + if (err != REG_OK) + return err; + if (!c && depth<0) { + ctx->submatch_id = subid; + return REG_OK; + } + if (!c || depth<0) + return REG_EPAREN; + nbranch = tre_stack_pop_voidptr(stack); + nunion = tre_stack_pop_voidptr(stack); + goto parse_iter; + } + } + } +} + + +/*********************************************************************** + from tre-compile.c +***********************************************************************/ + + +/* + TODO: + - Fix tre_ast_to_tnfa() to recurse using a stack instead of recursive + function calls. +*/ + +/* + Algorithms to setup tags so that submatch addressing can be done. +*/ + + +/* Inserts a catenation node to the root of the tree given in `node'. + As the left child a new tag with number `tag_id' to `node' is added, + and the right child is the old root. */ +static reg_errcode_t +tre_add_tag_left(tre_mem_t mem, tre_ast_node_t *node, int tag_id) +{ + tre_catenation_t *c; + + c = tre_mem_alloc(mem, sizeof(*c)); + if (c == NULL) + return REG_ESPACE; + c->left = tre_ast_new_literal(mem, TAG, tag_id, -1); + if (c->left == NULL) + return REG_ESPACE; + c->right = tre_mem_alloc(mem, sizeof(tre_ast_node_t)); + if (c->right == NULL) + return REG_ESPACE; + + c->right->obj = node->obj; + c->right->type = node->type; + c->right->nullable = -1; + c->right->submatch_id = -1; + c->right->firstpos = NULL; + c->right->lastpos = NULL; + c->right->num_tags = 0; + c->right->num_submatches = 0; + node->obj = c; + node->type = CATENATION; + return REG_OK; +} + +/* Inserts a catenation node to the root of the tree given in `node'. + As the right child a new tag with number `tag_id' to `node' is added, + and the left child is the old root. */ +static reg_errcode_t +tre_add_tag_right(tre_mem_t mem, tre_ast_node_t *node, int tag_id) +{ + tre_catenation_t *c; + + c = tre_mem_alloc(mem, sizeof(*c)); + if (c == NULL) + return REG_ESPACE; + c->right = tre_ast_new_literal(mem, TAG, tag_id, -1); + if (c->right == NULL) + return REG_ESPACE; + c->left = tre_mem_alloc(mem, sizeof(tre_ast_node_t)); + if (c->left == NULL) + return REG_ESPACE; + + c->left->obj = node->obj; + c->left->type = node->type; + c->left->nullable = -1; + c->left->submatch_id = -1; + c->left->firstpos = NULL; + c->left->lastpos = NULL; + c->left->num_tags = 0; + c->left->num_submatches = 0; + node->obj = c; + node->type = CATENATION; + return REG_OK; +} + +typedef enum { + ADDTAGS_RECURSE, + ADDTAGS_AFTER_ITERATION, + ADDTAGS_AFTER_UNION_LEFT, + ADDTAGS_AFTER_UNION_RIGHT, + ADDTAGS_AFTER_CAT_LEFT, + ADDTAGS_AFTER_CAT_RIGHT, + ADDTAGS_SET_SUBMATCH_END +} tre_addtags_symbol_t; + + +typedef struct { + int tag; + int next_tag; +} tre_tag_states_t; + + +/* Go through `regset' and set submatch data for submatches that are + using this tag. */ +static void +tre_purge_regset(int *regset, tre_tnfa_t *tnfa, int tag) +{ + int i; + + for (i = 0; regset[i] >= 0; i++) + { + int id = regset[i] / 2; + int start = !(regset[i] % 2); + if (start) + tnfa->submatch_data[id].so_tag = tag; + else + tnfa->submatch_data[id].eo_tag = tag; + } + regset[0] = -1; +} + + +/* Adds tags to appropriate locations in the parse tree in `tree', so that + subexpressions marked for submatch addressing can be traced. */ +static reg_errcode_t +tre_add_tags(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree, + tre_tnfa_t *tnfa) +{ + reg_errcode_t status = REG_OK; + tre_addtags_symbol_t symbol; + tre_ast_node_t *node = tree; /* Tree node we are currently looking at. */ + int bottom = tre_stack_num_objects(stack); + /* True for first pass (counting number of needed tags) */ + int first_pass = (mem == NULL || tnfa == NULL); + int *regset, *orig_regset; + int num_tags = 0; /* Total number of tags. */ + int num_minimals = 0; /* Number of special minimal tags. */ + int tag = 0; /* The tag that is to be added next. */ + int next_tag = 1; /* Next tag to use after this one. */ + int *parents; /* Stack of submatches the current submatch is + contained in. */ + int minimal_tag = -1; /* Tag that marks the beginning of a minimal match. */ + tre_tag_states_t *saved_states; + + tre_tag_direction_t direction = TRE_TAG_MINIMIZE; + if (!first_pass) + { + tnfa->end_tag = 0; + tnfa->minimal_tags[0] = -1; + } + + regset = xmalloc(sizeof(*regset) * ((tnfa->num_submatches + 1) * 2)); + if (regset == NULL) + return REG_ESPACE; + regset[0] = -1; + orig_regset = regset; + + parents = xmalloc(sizeof(*parents) * (tnfa->num_submatches + 1)); + if (parents == NULL) + { + xfree(regset); + return REG_ESPACE; + } + parents[0] = -1; + + saved_states = xmalloc(sizeof(*saved_states) * (tnfa->num_submatches + 1)); + if (saved_states == NULL) + { + xfree(regset); + xfree(parents); + return REG_ESPACE; + } + else + { + unsigned int i; + for (i = 0; i <= tnfa->num_submatches; i++) + saved_states[i].tag = -1; + } + + STACK_PUSH(stack, voidptr, node); + STACK_PUSH(stack, int, ADDTAGS_RECURSE); + + while (tre_stack_num_objects(stack) > bottom) + { + if (status != REG_OK) + break; + + symbol = (tre_addtags_symbol_t)tre_stack_pop_int(stack); + switch (symbol) + { + + case ADDTAGS_SET_SUBMATCH_END: + { + int id = tre_stack_pop_int(stack); + int i; + + /* Add end of this submatch to regset. */ + for (i = 0; regset[i] >= 0; i++); + regset[i] = id * 2 + 1; + regset[i + 1] = -1; + + /* Pop this submatch from the parents stack. */ + for (i = 0; parents[i] >= 0; i++); + parents[i - 1] = -1; + break; + } + + case ADDTAGS_RECURSE: + node = tre_stack_pop_voidptr(stack); + + if (node->submatch_id >= 0) + { + int id = node->submatch_id; + int i; + + + /* Add start of this submatch to regset. */ + for (i = 0; regset[i] >= 0; i++); + regset[i] = id * 2; + regset[i + 1] = -1; + + if (!first_pass) + { + for (i = 0; parents[i] >= 0; i++); + tnfa->submatch_data[id].parents = NULL; + if (i > 0) + { + int *p = xmalloc(sizeof(*p) * (i + 1)); + if (p == NULL) + { + status = REG_ESPACE; + break; + } + assert(tnfa->submatch_data[id].parents == NULL); + tnfa->submatch_data[id].parents = p; + for (i = 0; parents[i] >= 0; i++) + p[i] = parents[i]; + p[i] = -1; + } + } + + /* Add end of this submatch to regset after processing this + node. */ + STACK_PUSHX(stack, int, node->submatch_id); + STACK_PUSHX(stack, int, ADDTAGS_SET_SUBMATCH_END); + } + + switch (node->type) + { + case LITERAL: + { + tre_literal_t *lit = node->obj; + + if (!IS_SPECIAL(lit) || IS_BACKREF(lit)) + { + int i; + if (regset[0] >= 0) + { + /* Regset is not empty, so add a tag before the + literal or backref. */ + if (!first_pass) + { + status = tre_add_tag_left(mem, node, tag); + tnfa->tag_directions[tag] = direction; + if (minimal_tag >= 0) + { + for (i = 0; tnfa->minimal_tags[i] >= 0; i++); + tnfa->minimal_tags[i] = tag; + tnfa->minimal_tags[i + 1] = minimal_tag; + tnfa->minimal_tags[i + 2] = -1; + minimal_tag = -1; + num_minimals++; + } + tre_purge_regset(regset, tnfa, tag); + } + else + { + node->num_tags = 1; + } + + regset[0] = -1; + tag = next_tag; + num_tags++; + next_tag++; + } + } + else + { + assert(!IS_TAG(lit)); + } + break; + } + case CATENATION: + { + tre_catenation_t *cat = node->obj; + tre_ast_node_t *left = cat->left; + tre_ast_node_t *right = cat->right; + int reserved_tag = -1; + + + /* After processing right child. */ + STACK_PUSHX(stack, voidptr, node); + STACK_PUSHX(stack, int, ADDTAGS_AFTER_CAT_RIGHT); + + /* Process right child. */ + STACK_PUSHX(stack, voidptr, right); + STACK_PUSHX(stack, int, ADDTAGS_RECURSE); + + /* After processing left child. */ + STACK_PUSHX(stack, int, next_tag + left->num_tags); + if (left->num_tags > 0 && right->num_tags > 0) + { + /* Reserve the next tag to the right child. */ + reserved_tag = next_tag; + next_tag++; + } + STACK_PUSHX(stack, int, reserved_tag); + STACK_PUSHX(stack, int, ADDTAGS_AFTER_CAT_LEFT); + + /* Process left child. */ + STACK_PUSHX(stack, voidptr, left); + STACK_PUSHX(stack, int, ADDTAGS_RECURSE); + + } + break; + case ITERATION: + { + tre_iteration_t *iter = node->obj; + + if (first_pass) + { + STACK_PUSHX(stack, int, regset[0] >= 0 || iter->minimal); + } + else + { + STACK_PUSHX(stack, int, tag); + STACK_PUSHX(stack, int, iter->minimal); + } + STACK_PUSHX(stack, voidptr, node); + STACK_PUSHX(stack, int, ADDTAGS_AFTER_ITERATION); + + STACK_PUSHX(stack, voidptr, iter->arg); + STACK_PUSHX(stack, int, ADDTAGS_RECURSE); + + /* Regset is not empty, so add a tag here. */ + if (regset[0] >= 0 || iter->minimal) + { + if (!first_pass) + { + int i; + status = tre_add_tag_left(mem, node, tag); + if (iter->minimal) + tnfa->tag_directions[tag] = TRE_TAG_MAXIMIZE; + else + tnfa->tag_directions[tag] = direction; + if (minimal_tag >= 0) + { + for (i = 0; tnfa->minimal_tags[i] >= 0; i++); + tnfa->minimal_tags[i] = tag; + tnfa->minimal_tags[i + 1] = minimal_tag; + tnfa->minimal_tags[i + 2] = -1; + minimal_tag = -1; + num_minimals++; + } + tre_purge_regset(regset, tnfa, tag); + } + + regset[0] = -1; + tag = next_tag; + num_tags++; + next_tag++; + } + direction = TRE_TAG_MINIMIZE; + } + break; + case UNION: + { + tre_union_t *uni = node->obj; + tre_ast_node_t *left = uni->left; + tre_ast_node_t *right = uni->right; + int left_tag; + int right_tag; + + if (regset[0] >= 0) + { + left_tag = next_tag; + right_tag = next_tag + 1; + } + else + { + left_tag = tag; + right_tag = next_tag; + } + + /* After processing right child. */ + STACK_PUSHX(stack, int, right_tag); + STACK_PUSHX(stack, int, left_tag); + STACK_PUSHX(stack, voidptr, regset); + STACK_PUSHX(stack, int, regset[0] >= 0); + STACK_PUSHX(stack, voidptr, node); + STACK_PUSHX(stack, voidptr, right); + STACK_PUSHX(stack, voidptr, left); + STACK_PUSHX(stack, int, ADDTAGS_AFTER_UNION_RIGHT); + + /* Process right child. */ + STACK_PUSHX(stack, voidptr, right); + STACK_PUSHX(stack, int, ADDTAGS_RECURSE); + + /* After processing left child. */ + STACK_PUSHX(stack, int, ADDTAGS_AFTER_UNION_LEFT); + + /* Process left child. */ + STACK_PUSHX(stack, voidptr, left); + STACK_PUSHX(stack, int, ADDTAGS_RECURSE); + + /* Regset is not empty, so add a tag here. */ + if (regset[0] >= 0) + { + if (!first_pass) + { + int i; + status = tre_add_tag_left(mem, node, tag); + tnfa->tag_directions[tag] = direction; + if (minimal_tag >= 0) + { + for (i = 0; tnfa->minimal_tags[i] >= 0; i++); + tnfa->minimal_tags[i] = tag; + tnfa->minimal_tags[i + 1] = minimal_tag; + tnfa->minimal_tags[i + 2] = -1; + minimal_tag = -1; + num_minimals++; + } + tre_purge_regset(regset, tnfa, tag); + } + + regset[0] = -1; + tag = next_tag; + num_tags++; + next_tag++; + } + + if (node->num_submatches > 0) + { + /* The next two tags are reserved for markers. */ + next_tag++; + tag = next_tag; + next_tag++; + } + + break; + } + } + + if (node->submatch_id >= 0) + { + int i; + /* Push this submatch on the parents stack. */ + for (i = 0; parents[i] >= 0; i++); + parents[i] = node->submatch_id; + parents[i + 1] = -1; + } + + break; /* end case: ADDTAGS_RECURSE */ + + case ADDTAGS_AFTER_ITERATION: + { + int minimal = 0; + int enter_tag; + node = tre_stack_pop_voidptr(stack); + if (first_pass) + { + node->num_tags = ((tre_iteration_t *)node->obj)->arg->num_tags + + tre_stack_pop_int(stack); + minimal_tag = -1; + } + else + { + minimal = tre_stack_pop_int(stack); + enter_tag = tre_stack_pop_int(stack); + if (minimal) + minimal_tag = enter_tag; + } + + if (!first_pass) + { + if (minimal) + direction = TRE_TAG_MINIMIZE; + else + direction = TRE_TAG_MAXIMIZE; + } + break; + } + + case ADDTAGS_AFTER_CAT_LEFT: + { + int new_tag = tre_stack_pop_int(stack); + next_tag = tre_stack_pop_int(stack); + if (new_tag >= 0) + { + tag = new_tag; + } + break; + } + + case ADDTAGS_AFTER_CAT_RIGHT: + node = tre_stack_pop_voidptr(stack); + if (first_pass) + node->num_tags = ((tre_catenation_t *)node->obj)->left->num_tags + + ((tre_catenation_t *)node->obj)->right->num_tags; + break; + + case ADDTAGS_AFTER_UNION_LEFT: + /* Lift the bottom of the `regset' array so that when processing + the right operand the items currently in the array are + invisible. The original bottom was saved at ADDTAGS_UNION and + will be restored at ADDTAGS_AFTER_UNION_RIGHT below. */ + while (*regset >= 0) + regset++; + break; + + case ADDTAGS_AFTER_UNION_RIGHT: + { + int added_tags, tag_left, tag_right; + tre_ast_node_t *left = tre_stack_pop_voidptr(stack); + tre_ast_node_t *right = tre_stack_pop_voidptr(stack); + node = tre_stack_pop_voidptr(stack); + added_tags = tre_stack_pop_int(stack); + if (first_pass) + { + node->num_tags = ((tre_union_t *)node->obj)->left->num_tags + + ((tre_union_t *)node->obj)->right->num_tags + added_tags + + ((node->num_submatches > 0) ? 2 : 0); + } + regset = tre_stack_pop_voidptr(stack); + tag_left = tre_stack_pop_int(stack); + tag_right = tre_stack_pop_int(stack); + + /* Add tags after both children, the left child gets a smaller + tag than the right child. This guarantees that we prefer + the left child over the right child. */ + /* XXX - This is not always necessary (if the children have + tags which must be seen for every match of that child). */ + /* XXX - Check if this is the only place where tre_add_tag_right + is used. If so, use tre_add_tag_left (putting the tag before + the child as opposed after the child) and throw away + tre_add_tag_right. */ + if (node->num_submatches > 0) + { + if (!first_pass) + { + status = tre_add_tag_right(mem, left, tag_left); + tnfa->tag_directions[tag_left] = TRE_TAG_MAXIMIZE; + if (status == REG_OK) + status = tre_add_tag_right(mem, right, tag_right); + tnfa->tag_directions[tag_right] = TRE_TAG_MAXIMIZE; + } + num_tags += 2; + } + direction = TRE_TAG_MAXIMIZE; + break; + } + + default: + assert(0); + break; + + } /* end switch(symbol) */ + } /* end while(tre_stack_num_objects(stack) > bottom) */ + + if (!first_pass) + tre_purge_regset(regset, tnfa, tag); + + if (!first_pass && minimal_tag >= 0) + { + int i; + for (i = 0; tnfa->minimal_tags[i] >= 0; i++); + tnfa->minimal_tags[i] = tag; + tnfa->minimal_tags[i + 1] = minimal_tag; + tnfa->minimal_tags[i + 2] = -1; + minimal_tag = -1; + num_minimals++; + } + + assert(tree->num_tags == num_tags); + tnfa->end_tag = num_tags; + tnfa->num_tags = num_tags; + tnfa->num_minimals = num_minimals; + xfree(orig_regset); + xfree(parents); + xfree(saved_states); + return status; +} + + + +/* + AST to TNFA compilation routines. +*/ + +typedef enum { + COPY_RECURSE, + COPY_SET_RESULT_PTR +} tre_copyast_symbol_t; + +/* Flags for tre_copy_ast(). */ +#define COPY_REMOVE_TAGS 1 +#define COPY_MAXIMIZE_FIRST_TAG 2 + +static reg_errcode_t +tre_copy_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast, + int flags, int *pos_add, tre_tag_direction_t *tag_directions, + tre_ast_node_t **copy, int *max_pos) +{ + reg_errcode_t status = REG_OK; + int bottom = tre_stack_num_objects(stack); + int num_copied = 0; + int first_tag = 1; + tre_ast_node_t **result = copy; + tre_copyast_symbol_t symbol; + + STACK_PUSH(stack, voidptr, ast); + STACK_PUSH(stack, int, COPY_RECURSE); + + while (status == REG_OK && tre_stack_num_objects(stack) > bottom) + { + tre_ast_node_t *node; + if (status != REG_OK) + break; + + symbol = (tre_copyast_symbol_t)tre_stack_pop_int(stack); + switch (symbol) + { + case COPY_SET_RESULT_PTR: + result = tre_stack_pop_voidptr(stack); + break; + case COPY_RECURSE: + node = tre_stack_pop_voidptr(stack); + switch (node->type) + { + case LITERAL: + { + tre_literal_t *lit = node->obj; + int pos = lit->position; + int min = lit->code_min; + int max = lit->code_max; + if (!IS_SPECIAL(lit) || IS_BACKREF(lit)) + { + /* XXX - e.g. [ab] has only one position but two + nodes, so we are creating holes in the state space + here. Not fatal, just wastes memory. */ + pos += *pos_add; + num_copied++; + } + else if (IS_TAG(lit) && (flags & COPY_REMOVE_TAGS)) + { + /* Change this tag to empty. */ + min = EMPTY; + max = pos = -1; + } + else if (IS_TAG(lit) && (flags & COPY_MAXIMIZE_FIRST_TAG) + && first_tag) + { + /* Maximize the first tag. */ + tag_directions[max] = TRE_TAG_MAXIMIZE; + first_tag = 0; + } + *result = tre_ast_new_literal(mem, min, max, pos); + if (*result == NULL) + status = REG_ESPACE; + else { + tre_literal_t *p = (*result)->obj; + p->class = lit->class; + p->neg_classes = lit->neg_classes; + } + + if (pos > *max_pos) + *max_pos = pos; + break; + } + case UNION: + { + tre_union_t *uni = node->obj; + tre_union_t *tmp; + *result = tre_ast_new_union(mem, uni->left, uni->right); + if (*result == NULL) + { + status = REG_ESPACE; + break; + } + tmp = (*result)->obj; + result = &tmp->left; + STACK_PUSHX(stack, voidptr, uni->right); + STACK_PUSHX(stack, int, COPY_RECURSE); + STACK_PUSHX(stack, voidptr, &tmp->right); + STACK_PUSHX(stack, int, COPY_SET_RESULT_PTR); + STACK_PUSHX(stack, voidptr, uni->left); + STACK_PUSHX(stack, int, COPY_RECURSE); + break; + } + case CATENATION: + { + tre_catenation_t *cat = node->obj; + tre_catenation_t *tmp; + *result = tre_ast_new_catenation(mem, cat->left, cat->right); + if (*result == NULL) + { + status = REG_ESPACE; + break; + } + tmp = (*result)->obj; + tmp->left = NULL; + tmp->right = NULL; + result = &tmp->left; + + STACK_PUSHX(stack, voidptr, cat->right); + STACK_PUSHX(stack, int, COPY_RECURSE); + STACK_PUSHX(stack, voidptr, &tmp->right); + STACK_PUSHX(stack, int, COPY_SET_RESULT_PTR); + STACK_PUSHX(stack, voidptr, cat->left); + STACK_PUSHX(stack, int, COPY_RECURSE); + break; + } + case ITERATION: + { + tre_iteration_t *iter = node->obj; + STACK_PUSHX(stack, voidptr, iter->arg); + STACK_PUSHX(stack, int, COPY_RECURSE); + *result = tre_ast_new_iter(mem, iter->arg, iter->min, + iter->max, iter->minimal); + if (*result == NULL) + { + status = REG_ESPACE; + break; + } + iter = (*result)->obj; + result = &iter->arg; + break; + } + default: + assert(0); + break; + } + break; + } + } + *pos_add += num_copied; + return status; +} + +typedef enum { + EXPAND_RECURSE, + EXPAND_AFTER_ITER +} tre_expand_ast_symbol_t; + +/* Expands each iteration node that has a finite nonzero minimum or maximum + iteration count to a catenated sequence of copies of the node. */ +static reg_errcode_t +tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast, + int *position, tre_tag_direction_t *tag_directions) +{ + reg_errcode_t status = REG_OK; + int bottom = tre_stack_num_objects(stack); + int pos_add = 0; + int pos_add_total = 0; + int max_pos = 0; + int iter_depth = 0; + + STACK_PUSHR(stack, voidptr, ast); + STACK_PUSHR(stack, int, EXPAND_RECURSE); + while (status == REG_OK && tre_stack_num_objects(stack) > bottom) + { + tre_ast_node_t *node; + tre_expand_ast_symbol_t symbol; + + if (status != REG_OK) + break; + + symbol = (tre_expand_ast_symbol_t)tre_stack_pop_int(stack); + node = tre_stack_pop_voidptr(stack); + switch (symbol) + { + case EXPAND_RECURSE: + switch (node->type) + { + case LITERAL: + { + tre_literal_t *lit= node->obj; + if (!IS_SPECIAL(lit) || IS_BACKREF(lit)) + { + lit->position += pos_add; + if (lit->position > max_pos) + max_pos = lit->position; + } + break; + } + case UNION: + { + tre_union_t *uni = node->obj; + STACK_PUSHX(stack, voidptr, uni->right); + STACK_PUSHX(stack, int, EXPAND_RECURSE); + STACK_PUSHX(stack, voidptr, uni->left); + STACK_PUSHX(stack, int, EXPAND_RECURSE); + break; + } + case CATENATION: + { + tre_catenation_t *cat = node->obj; + STACK_PUSHX(stack, voidptr, cat->right); + STACK_PUSHX(stack, int, EXPAND_RECURSE); + STACK_PUSHX(stack, voidptr, cat->left); + STACK_PUSHX(stack, int, EXPAND_RECURSE); + break; + } + case ITERATION: + { + tre_iteration_t *iter = node->obj; + STACK_PUSHX(stack, int, pos_add); + STACK_PUSHX(stack, voidptr, node); + STACK_PUSHX(stack, int, EXPAND_AFTER_ITER); + STACK_PUSHX(stack, voidptr, iter->arg); + STACK_PUSHX(stack, int, EXPAND_RECURSE); + /* If we are going to expand this node at EXPAND_AFTER_ITER + then don't increase the `pos' fields of the nodes now, it + will get done when expanding. */ + if (iter->min > 1 || iter->max > 1) + pos_add = 0; + iter_depth++; + break; + } + default: + assert(0); + break; + } + break; + case EXPAND_AFTER_ITER: + { + tre_iteration_t *iter = node->obj; + int pos_add_last; + pos_add = tre_stack_pop_int(stack); + pos_add_last = pos_add; + if (iter->min > 1 || iter->max > 1) + { + tre_ast_node_t *seq1 = NULL, *seq2 = NULL; + int j; + int pos_add_save = pos_add; + + /* Create a catenated sequence of copies of the node. */ + for (j = 0; j < iter->min; j++) + { + tre_ast_node_t *copy; + /* Remove tags from all but the last copy. */ + int flags = ((j + 1 < iter->min) + ? COPY_REMOVE_TAGS + : COPY_MAXIMIZE_FIRST_TAG); + pos_add_save = pos_add; + status = tre_copy_ast(mem, stack, iter->arg, flags, + &pos_add, tag_directions, ©, + &max_pos); + if (status != REG_OK) + return status; + if (seq1 != NULL) + seq1 = tre_ast_new_catenation(mem, seq1, copy); + else + seq1 = copy; + if (seq1 == NULL) + return REG_ESPACE; + } + + if (iter->max == -1) + { + /* No upper limit. */ + pos_add_save = pos_add; + status = tre_copy_ast(mem, stack, iter->arg, 0, + &pos_add, NULL, &seq2, &max_pos); + if (status != REG_OK) + return status; + seq2 = tre_ast_new_iter(mem, seq2, 0, -1, 0); + if (seq2 == NULL) + return REG_ESPACE; + } + else + { + for (j = iter->min; j < iter->max; j++) + { + tre_ast_node_t *tmp, *copy; + pos_add_save = pos_add; + status = tre_copy_ast(mem, stack, iter->arg, 0, + &pos_add, NULL, ©, &max_pos); + if (status != REG_OK) + return status; + if (seq2 != NULL) + seq2 = tre_ast_new_catenation(mem, copy, seq2); + else + seq2 = copy; + if (seq2 == NULL) + return REG_ESPACE; + tmp = tre_ast_new_literal(mem, EMPTY, -1, -1); + if (tmp == NULL) + return REG_ESPACE; + seq2 = tre_ast_new_union(mem, tmp, seq2); + if (seq2 == NULL) + return REG_ESPACE; + } + } + + pos_add = pos_add_save; + if (seq1 == NULL) + seq1 = seq2; + else if (seq2 != NULL) + seq1 = tre_ast_new_catenation(mem, seq1, seq2); + if (seq1 == NULL) + return REG_ESPACE; + node->obj = seq1->obj; + node->type = seq1->type; + } + + iter_depth--; + pos_add_total += pos_add - pos_add_last; + if (iter_depth == 0) + pos_add = pos_add_total; + + break; + } + default: + assert(0); + break; + } + } + + *position += pos_add_total; + + /* `max_pos' should never be larger than `*position' if the above + code works, but just an extra safeguard let's make sure + `*position' is set large enough so enough memory will be + allocated for the transition table. */ + if (max_pos > *position) + *position = max_pos; + + return status; +} + +static tre_pos_and_tags_t * +tre_set_empty(tre_mem_t mem) +{ + tre_pos_and_tags_t *new_set; + + new_set = tre_mem_calloc(mem, sizeof(*new_set)); + if (new_set == NULL) + return NULL; + + new_set[0].position = -1; + new_set[0].code_min = -1; + new_set[0].code_max = -1; + + return new_set; +} + +static tre_pos_and_tags_t * +tre_set_one(tre_mem_t mem, int position, int code_min, int code_max, + tre_ctype_t class, tre_ctype_t *neg_classes, int backref) +{ + tre_pos_and_tags_t *new_set; + + new_set = tre_mem_calloc(mem, sizeof(*new_set) * 2); + if (new_set == NULL) + return NULL; + + new_set[0].position = position; + new_set[0].code_min = code_min; + new_set[0].code_max = code_max; + new_set[0].class = class; + new_set[0].neg_classes = neg_classes; + new_set[0].backref = backref; + new_set[1].position = -1; + new_set[1].code_min = -1; + new_set[1].code_max = -1; + + return new_set; +} + +static tre_pos_and_tags_t * +tre_set_union(tre_mem_t mem, tre_pos_and_tags_t *set1, tre_pos_and_tags_t *set2, + int *tags, int assertions) +{ + int s1, s2, i, j; + tre_pos_and_tags_t *new_set; + int *new_tags; + int num_tags; + + for (num_tags = 0; tags != NULL && tags[num_tags] >= 0; num_tags++); + for (s1 = 0; set1[s1].position >= 0; s1++); + for (s2 = 0; set2[s2].position >= 0; s2++); + new_set = tre_mem_calloc(mem, sizeof(*new_set) * (s1 + s2 + 1)); + if (!new_set ) + return NULL; + + for (s1 = 0; set1[s1].position >= 0; s1++) + { + new_set[s1].position = set1[s1].position; + new_set[s1].code_min = set1[s1].code_min; + new_set[s1].code_max = set1[s1].code_max; + new_set[s1].assertions = set1[s1].assertions | assertions; + new_set[s1].class = set1[s1].class; + new_set[s1].neg_classes = set1[s1].neg_classes; + new_set[s1].backref = set1[s1].backref; + if (set1[s1].tags == NULL && tags == NULL) + new_set[s1].tags = NULL; + else + { + for (i = 0; set1[s1].tags != NULL && set1[s1].tags[i] >= 0; i++); + new_tags = tre_mem_alloc(mem, (sizeof(*new_tags) + * (i + num_tags + 1))); + if (new_tags == NULL) + return NULL; + for (j = 0; j < i; j++) + new_tags[j] = set1[s1].tags[j]; + for (i = 0; i < num_tags; i++) + new_tags[j + i] = tags[i]; + new_tags[j + i] = -1; + new_set[s1].tags = new_tags; + } + } + + for (s2 = 0; set2[s2].position >= 0; s2++) + { + new_set[s1 + s2].position = set2[s2].position; + new_set[s1 + s2].code_min = set2[s2].code_min; + new_set[s1 + s2].code_max = set2[s2].code_max; + /* XXX - why not | assertions here as well? */ + new_set[s1 + s2].assertions = set2[s2].assertions; + new_set[s1 + s2].class = set2[s2].class; + new_set[s1 + s2].neg_classes = set2[s2].neg_classes; + new_set[s1 + s2].backref = set2[s2].backref; + if (set2[s2].tags == NULL) + new_set[s1 + s2].tags = NULL; + else + { + for (i = 0; set2[s2].tags[i] >= 0; i++); + new_tags = tre_mem_alloc(mem, sizeof(*new_tags) * (i + 1)); + if (new_tags == NULL) + return NULL; + for (j = 0; j < i; j++) + new_tags[j] = set2[s2].tags[j]; + new_tags[j] = -1; + new_set[s1 + s2].tags = new_tags; + } + } + new_set[s1 + s2].position = -1; + return new_set; +} + +/* Finds the empty path through `node' which is the one that should be + taken according to POSIX.2 rules, and adds the tags on that path to + `tags'. `tags' may be NULL. If `num_tags_seen' is not NULL, it is + set to the number of tags seen on the path. */ +static reg_errcode_t +tre_match_empty(tre_stack_t *stack, tre_ast_node_t *node, int *tags, + int *assertions, int *num_tags_seen) +{ + tre_literal_t *lit; + tre_union_t *uni; + tre_catenation_t *cat; + tre_iteration_t *iter; + int i; + int bottom = tre_stack_num_objects(stack); + reg_errcode_t status = REG_OK; + if (num_tags_seen) + *num_tags_seen = 0; + + status = tre_stack_push_voidptr(stack, node); + + /* Walk through the tree recursively. */ + while (status == REG_OK && tre_stack_num_objects(stack) > bottom) + { + node = tre_stack_pop_voidptr(stack); + + switch (node->type) + { + case LITERAL: + lit = (tre_literal_t *)node->obj; + switch (lit->code_min) + { + case TAG: + if (lit->code_max >= 0) + { + if (tags != NULL) + { + /* Add the tag to `tags'. */ + for (i = 0; tags[i] >= 0; i++) + if (tags[i] == lit->code_max) + break; + if (tags[i] < 0) + { + tags[i] = lit->code_max; + tags[i + 1] = -1; + } + } + if (num_tags_seen) + (*num_tags_seen)++; + } + break; + case ASSERTION: + assert(lit->code_max >= 1 + || lit->code_max <= ASSERT_LAST); + if (assertions != NULL) + *assertions |= lit->code_max; + break; + case EMPTY: + break; + default: + assert(0); + break; + } + break; + + case UNION: + /* Subexpressions starting earlier take priority over ones + starting later, so we prefer the left subexpression over the + right subexpression. */ + uni = (tre_union_t *)node->obj; + if (uni->left->nullable) + STACK_PUSHX(stack, voidptr, uni->left) + else if (uni->right->nullable) + STACK_PUSHX(stack, voidptr, uni->right) + else + assert(0); + break; + + case CATENATION: + /* The path must go through both children. */ + cat = (tre_catenation_t *)node->obj; + assert(cat->left->nullable); + assert(cat->right->nullable); + STACK_PUSHX(stack, voidptr, cat->left); + STACK_PUSHX(stack, voidptr, cat->right); + break; + + case ITERATION: + /* A match with an empty string is preferred over no match at + all, so we go through the argument if possible. */ + iter = (tre_iteration_t *)node->obj; + if (iter->arg->nullable) + STACK_PUSHX(stack, voidptr, iter->arg); + break; + + default: + assert(0); + break; + } + } + + return status; +} + + +typedef enum { + NFL_RECURSE, + NFL_POST_UNION, + NFL_POST_CATENATION, + NFL_POST_ITERATION +} tre_nfl_stack_symbol_t; + + +/* Computes and fills in the fields `nullable', `firstpos', and `lastpos' for + the nodes of the AST `tree'. */ +static reg_errcode_t +tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree) +{ + int bottom = tre_stack_num_objects(stack); + + STACK_PUSHR(stack, voidptr, tree); + STACK_PUSHR(stack, int, NFL_RECURSE); + + while (tre_stack_num_objects(stack) > bottom) + { + tre_nfl_stack_symbol_t symbol; + tre_ast_node_t *node; + + symbol = (tre_nfl_stack_symbol_t)tre_stack_pop_int(stack); + node = tre_stack_pop_voidptr(stack); + switch (symbol) + { + case NFL_RECURSE: + switch (node->type) + { + case LITERAL: + { + tre_literal_t *lit = (tre_literal_t *)node->obj; + if (IS_BACKREF(lit)) + { + /* Back references: nullable = false, firstpos = {i}, + lastpos = {i}. */ + node->nullable = 0; + node->firstpos = tre_set_one(mem, lit->position, 0, + TRE_CHAR_MAX, 0, NULL, -1); + if (!node->firstpos) + return REG_ESPACE; + node->lastpos = tre_set_one(mem, lit->position, 0, + TRE_CHAR_MAX, 0, NULL, + (int)lit->code_max); + if (!node->lastpos) + return REG_ESPACE; + } + else if (lit->code_min < 0) + { + /* Tags, empty strings, params, and zero width assertions: + nullable = true, firstpos = {}, and lastpos = {}. */ + node->nullable = 1; + node->firstpos = tre_set_empty(mem); + if (!node->firstpos) + return REG_ESPACE; + node->lastpos = tre_set_empty(mem); + if (!node->lastpos) + return REG_ESPACE; + } + else + { + /* Literal at position i: nullable = false, firstpos = {i}, + lastpos = {i}. */ + node->nullable = 0; + node->firstpos = + tre_set_one(mem, lit->position, (int)lit->code_min, + (int)lit->code_max, 0, NULL, -1); + if (!node->firstpos) + return REG_ESPACE; + node->lastpos = tre_set_one(mem, lit->position, + (int)lit->code_min, + (int)lit->code_max, + lit->class, lit->neg_classes, + -1); + if (!node->lastpos) + return REG_ESPACE; + } + break; + } + + case UNION: + /* Compute the attributes for the two subtrees, and after that + for this node. */ + STACK_PUSHR(stack, voidptr, node); + STACK_PUSHR(stack, int, NFL_POST_UNION); + STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->right); + STACK_PUSHR(stack, int, NFL_RECURSE); + STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->left); + STACK_PUSHR(stack, int, NFL_RECURSE); + break; + + case CATENATION: + /* Compute the attributes for the two subtrees, and after that + for this node. */ + STACK_PUSHR(stack, voidptr, node); + STACK_PUSHR(stack, int, NFL_POST_CATENATION); + STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->right); + STACK_PUSHR(stack, int, NFL_RECURSE); + STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->left); + STACK_PUSHR(stack, int, NFL_RECURSE); + break; + + case ITERATION: + /* Compute the attributes for the subtree, and after that for + this node. */ + STACK_PUSHR(stack, voidptr, node); + STACK_PUSHR(stack, int, NFL_POST_ITERATION); + STACK_PUSHR(stack, voidptr, ((tre_iteration_t *)node->obj)->arg); + STACK_PUSHR(stack, int, NFL_RECURSE); + break; + } + break; /* end case: NFL_RECURSE */ + + case NFL_POST_UNION: + { + tre_union_t *uni = (tre_union_t *)node->obj; + node->nullable = uni->left->nullable || uni->right->nullable; + node->firstpos = tre_set_union(mem, uni->left->firstpos, + uni->right->firstpos, NULL, 0); + if (!node->firstpos) + return REG_ESPACE; + node->lastpos = tre_set_union(mem, uni->left->lastpos, + uni->right->lastpos, NULL, 0); + if (!node->lastpos) + return REG_ESPACE; + break; + } + + case NFL_POST_ITERATION: + { + tre_iteration_t *iter = (tre_iteration_t *)node->obj; + + if (iter->min == 0 || iter->arg->nullable) + node->nullable = 1; + else + node->nullable = 0; + node->firstpos = iter->arg->firstpos; + node->lastpos = iter->arg->lastpos; + break; + } + + case NFL_POST_CATENATION: + { + int num_tags, *tags, assertions; + reg_errcode_t status; + tre_catenation_t *cat = node->obj; + node->nullable = cat->left->nullable && cat->right->nullable; + + /* Compute firstpos. */ + if (cat->left->nullable) + { + /* The left side matches the empty string. Make a first pass + with tre_match_empty() to get the number of tags and + parameters. */ + status = tre_match_empty(stack, cat->left, + NULL, NULL, &num_tags); + if (status != REG_OK) + return status; + /* Allocate arrays for the tags and parameters. */ + tags = xmalloc(sizeof(*tags) * (num_tags + 1)); + if (!tags) + return REG_ESPACE; + tags[0] = -1; + assertions = 0; + /* Second pass with tre_mach_empty() to get the list of + tags and parameters. */ + status = tre_match_empty(stack, cat->left, tags, + &assertions, NULL); + if (status != REG_OK) + { + xfree(tags); + return status; + } + node->firstpos = + tre_set_union(mem, cat->right->firstpos, cat->left->firstpos, + tags, assertions); + xfree(tags); + if (!node->firstpos) + return REG_ESPACE; + } + else + { + node->firstpos = cat->left->firstpos; + } + + /* Compute lastpos. */ + if (cat->right->nullable) + { + /* The right side matches the empty string. Make a first pass + with tre_match_empty() to get the number of tags and + parameters. */ + status = tre_match_empty(stack, cat->right, + NULL, NULL, &num_tags); + if (status != REG_OK) + return status; + /* Allocate arrays for the tags and parameters. */ + tags = xmalloc(sizeof(int) * (num_tags + 1)); + if (!tags) + return REG_ESPACE; + tags[0] = -1; + assertions = 0; + /* Second pass with tre_mach_empty() to get the list of + tags and parameters. */ + status = tre_match_empty(stack, cat->right, tags, + &assertions, NULL); + if (status != REG_OK) + { + xfree(tags); + return status; + } + node->lastpos = + tre_set_union(mem, cat->left->lastpos, cat->right->lastpos, + tags, assertions); + xfree(tags); + if (!node->lastpos) + return REG_ESPACE; + } + else + { + node->lastpos = cat->right->lastpos; + } + break; + } + + default: + assert(0); + break; + } + } + + return REG_OK; +} + + +/* Adds a transition from each position in `p1' to each position in `p2'. */ +static reg_errcode_t +tre_make_trans(tre_pos_and_tags_t *p1, tre_pos_and_tags_t *p2, + tre_tnfa_transition_t *transitions, + int *counts, int *offs) +{ + tre_pos_and_tags_t *orig_p2 = p2; + tre_tnfa_transition_t *trans; + int i, j, k, l, dup, prev_p2_pos; + + if (transitions != NULL) + while (p1->position >= 0) + { + p2 = orig_p2; + prev_p2_pos = -1; + while (p2->position >= 0) + { + /* Optimization: if this position was already handled, skip it. */ + if (p2->position == prev_p2_pos) + { + p2++; + continue; + } + prev_p2_pos = p2->position; + /* Set `trans' to point to the next unused transition from + position `p1->position'. */ + trans = transitions + offs[p1->position]; + while (trans->state != NULL) + { +#if 0 + /* If we find a previous transition from `p1->position' to + `p2->position', it is overwritten. This can happen only + if there are nested loops in the regexp, like in "((a)*)*". + In POSIX.2 repetition using the outer loop is always + preferred over using the inner loop. Therefore the + transition for the inner loop is useless and can be thrown + away. */ + /* XXX - The same position is used for all nodes in a bracket + expression, so this optimization cannot be used (it will + break bracket expressions) unless I figure out a way to + detect it here. */ + if (trans->state_id == p2->position) + { + break; + } +#endif + trans++; + } + + if (trans->state == NULL) + (trans + 1)->state = NULL; + /* Use the character ranges, assertions, etc. from `p1' for + the transition from `p1' to `p2'. */ + trans->code_min = p1->code_min; + trans->code_max = p1->code_max; + trans->state = transitions + offs[p2->position]; + trans->state_id = p2->position; + trans->assertions = p1->assertions | p2->assertions + | (p1->class ? ASSERT_CHAR_CLASS : 0) + | (p1->neg_classes != NULL ? ASSERT_CHAR_CLASS_NEG : 0); + if (p1->backref >= 0) + { + assert((trans->assertions & ASSERT_CHAR_CLASS) == 0); + assert(p2->backref < 0); + trans->u.backref = p1->backref; + trans->assertions |= ASSERT_BACKREF; + } + else + trans->u.class = p1->class; + if (p1->neg_classes != NULL) + { + for (i = 0; p1->neg_classes[i] != (tre_ctype_t)0; i++); + trans->neg_classes = + xmalloc(sizeof(*trans->neg_classes) * (i + 1)); + if (trans->neg_classes == NULL) + return REG_ESPACE; + for (i = 0; p1->neg_classes[i] != (tre_ctype_t)0; i++) + trans->neg_classes[i] = p1->neg_classes[i]; + trans->neg_classes[i] = (tre_ctype_t)0; + } + else + trans->neg_classes = NULL; + + /* Find out how many tags this transition has. */ + i = 0; + if (p1->tags != NULL) + while(p1->tags[i] >= 0) + i++; + j = 0; + if (p2->tags != NULL) + while(p2->tags[j] >= 0) + j++; + + /* If we are overwriting a transition, free the old tag array. */ + if (trans->tags != NULL) + xfree(trans->tags); + trans->tags = NULL; + + /* If there were any tags, allocate an array and fill it. */ + if (i + j > 0) + { + trans->tags = xmalloc(sizeof(*trans->tags) * (i + j + 1)); + if (!trans->tags) + return REG_ESPACE; + i = 0; + if (p1->tags != NULL) + while(p1->tags[i] >= 0) + { + trans->tags[i] = p1->tags[i]; + i++; + } + l = i; + j = 0; + if (p2->tags != NULL) + while (p2->tags[j] >= 0) + { + /* Don't add duplicates. */ + dup = 0; + for (k = 0; k < i; k++) + if (trans->tags[k] == p2->tags[j]) + { + dup = 1; + break; + } + if (!dup) + trans->tags[l++] = p2->tags[j]; + j++; + } + trans->tags[l] = -1; + } + + p2++; + } + p1++; + } + else + /* Compute a maximum limit for the number of transitions leaving + from each state. */ + while (p1->position >= 0) + { + p2 = orig_p2; + while (p2->position >= 0) + { + counts[p1->position]++; + p2++; + } + p1++; + } + return REG_OK; +} + +/* Converts the syntax tree to a TNFA. All the transitions in the TNFA are + labelled with one character range (there are no transitions on empty + strings). The TNFA takes O(n^2) space in the worst case, `n' is size of + the regexp. */ +static reg_errcode_t +tre_ast_to_tnfa(tre_ast_node_t *node, tre_tnfa_transition_t *transitions, + int *counts, int *offs) +{ + tre_union_t *uni; + tre_catenation_t *cat; + tre_iteration_t *iter; + reg_errcode_t errcode = REG_OK; + + /* XXX - recurse using a stack!. */ + switch (node->type) + { + case LITERAL: + break; + case UNION: + uni = (tre_union_t *)node->obj; + errcode = tre_ast_to_tnfa(uni->left, transitions, counts, offs); + if (errcode != REG_OK) + return errcode; + errcode = tre_ast_to_tnfa(uni->right, transitions, counts, offs); + break; + + case CATENATION: + cat = (tre_catenation_t *)node->obj; + /* Add a transition from each position in cat->left->lastpos + to each position in cat->right->firstpos. */ + errcode = tre_make_trans(cat->left->lastpos, cat->right->firstpos, + transitions, counts, offs); + if (errcode != REG_OK) + return errcode; + errcode = tre_ast_to_tnfa(cat->left, transitions, counts, offs); + if (errcode != REG_OK) + return errcode; + errcode = tre_ast_to_tnfa(cat->right, transitions, counts, offs); + break; + + case ITERATION: + iter = (tre_iteration_t *)node->obj; + assert(iter->max == -1 || iter->max == 1); + + if (iter->max == -1) + { + assert(iter->min == 0 || iter->min == 1); + /* Add a transition from each last position in the iterated + expression to each first position. */ + errcode = tre_make_trans(iter->arg->lastpos, iter->arg->firstpos, + transitions, counts, offs); + if (errcode != REG_OK) + return errcode; + } + errcode = tre_ast_to_tnfa(iter->arg, transitions, counts, offs); + break; + } + return errcode; +} + + +#define ERROR_EXIT(err) \ + do \ + { \ + errcode = err; \ + if (/*CONSTCOND*/1) \ + goto error_exit; \ + } \ + while (/*CONSTCOND*/0) + + +int +regcomp(regex_t *restrict preg, const char *restrict regex, int cflags) +{ + tre_stack_t *stack; + tre_ast_node_t *tree, *tmp_ast_l, *tmp_ast_r; + tre_pos_and_tags_t *p; + int *counts = NULL, *offs = NULL; + int i, add = 0; + tre_tnfa_transition_t *transitions, *initial; + tre_tnfa_t *tnfa = NULL; + tre_submatch_data_t *submatch_data; + tre_tag_direction_t *tag_directions = NULL; + reg_errcode_t errcode; + tre_mem_t mem; + + /* Parse context. */ + tre_parse_ctx_t parse_ctx; + + /* Allocate a stack used throughout the compilation process for various + purposes. */ + stack = tre_stack_new(512, 1024000, 128); + if (!stack) + return REG_ESPACE; + /* Allocate a fast memory allocator. */ + mem = tre_mem_new(); + if (!mem) + { + tre_stack_destroy(stack); + return REG_ESPACE; + } + + /* Parse the regexp. */ + memset(&parse_ctx, 0, sizeof(parse_ctx)); + parse_ctx.mem = mem; + parse_ctx.stack = stack; + parse_ctx.start = regex; + parse_ctx.cflags = cflags; + parse_ctx.max_backref = -1; + errcode = tre_parse(&parse_ctx); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + preg->re_nsub = parse_ctx.submatch_id - 1; + tree = parse_ctx.n; + +#ifdef TRE_DEBUG + tre_ast_print(tree); +#endif /* TRE_DEBUG */ + + /* Referring to nonexistent subexpressions is illegal. */ + if (parse_ctx.max_backref > (int)preg->re_nsub) + ERROR_EXIT(REG_ESUBREG); + + /* Allocate the TNFA struct. */ + tnfa = xcalloc(1, sizeof(tre_tnfa_t)); + if (tnfa == NULL) + ERROR_EXIT(REG_ESPACE); + tnfa->have_backrefs = parse_ctx.max_backref >= 0; + tnfa->have_approx = 0; + tnfa->num_submatches = parse_ctx.submatch_id; + + /* Set up tags for submatch addressing. If REG_NOSUB is set and the + regexp does not have back references, this can be skipped. */ + if (tnfa->have_backrefs || !(cflags & REG_NOSUB)) + { + + /* Figure out how many tags we will need. */ + errcode = tre_add_tags(NULL, stack, tree, tnfa); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + + if (tnfa->num_tags > 0) + { + tag_directions = xmalloc(sizeof(*tag_directions) + * (tnfa->num_tags + 1)); + if (tag_directions == NULL) + ERROR_EXIT(REG_ESPACE); + tnfa->tag_directions = tag_directions; + memset(tag_directions, -1, + sizeof(*tag_directions) * (tnfa->num_tags + 1)); + } + tnfa->minimal_tags = xcalloc((unsigned)tnfa->num_tags * 2 + 1, + sizeof(*tnfa->minimal_tags)); + if (tnfa->minimal_tags == NULL) + ERROR_EXIT(REG_ESPACE); + + submatch_data = xcalloc((unsigned)parse_ctx.submatch_id, + sizeof(*submatch_data)); + if (submatch_data == NULL) + ERROR_EXIT(REG_ESPACE); + tnfa->submatch_data = submatch_data; + + errcode = tre_add_tags(mem, stack, tree, tnfa); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + + } + + /* Expand iteration nodes. */ + errcode = tre_expand_ast(mem, stack, tree, &parse_ctx.position, + tag_directions); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + + /* Add a dummy node for the final state. + XXX - For certain patterns this dummy node can be optimized away, + for example "a*" or "ab*". Figure out a simple way to detect + this possibility. */ + tmp_ast_l = tree; + tmp_ast_r = tre_ast_new_literal(mem, 0, 0, parse_ctx.position++); + if (tmp_ast_r == NULL) + ERROR_EXIT(REG_ESPACE); + + tree = tre_ast_new_catenation(mem, tmp_ast_l, tmp_ast_r); + if (tree == NULL) + ERROR_EXIT(REG_ESPACE); + + errcode = tre_compute_nfl(mem, stack, tree); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + + counts = xmalloc(sizeof(int) * parse_ctx.position); + if (counts == NULL) + ERROR_EXIT(REG_ESPACE); + + offs = xmalloc(sizeof(int) * parse_ctx.position); + if (offs == NULL) + ERROR_EXIT(REG_ESPACE); + + for (i = 0; i < parse_ctx.position; i++) + counts[i] = 0; + tre_ast_to_tnfa(tree, NULL, counts, NULL); + + add = 0; + for (i = 0; i < parse_ctx.position; i++) + { + offs[i] = add; + add += counts[i] + 1; + counts[i] = 0; + } + transitions = xcalloc((unsigned)add + 1, sizeof(*transitions)); + if (transitions == NULL) + ERROR_EXIT(REG_ESPACE); + tnfa->transitions = transitions; + tnfa->num_transitions = add; + + errcode = tre_ast_to_tnfa(tree, transitions, counts, offs); + if (errcode != REG_OK) + ERROR_EXIT(errcode); + + tnfa->firstpos_chars = NULL; + + p = tree->firstpos; + i = 0; + while (p->position >= 0) + { + i++; + p++; + } + + initial = xcalloc((unsigned)i + 1, sizeof(tre_tnfa_transition_t)); + if (initial == NULL) + ERROR_EXIT(REG_ESPACE); + tnfa->initial = initial; + + i = 0; + for (p = tree->firstpos; p->position >= 0; p++) + { + initial[i].state = transitions + offs[p->position]; + initial[i].state_id = p->position; + initial[i].tags = NULL; + /* Copy the arrays p->tags, and p->params, they are allocated + from a tre_mem object. */ + if (p->tags) + { + int j; + for (j = 0; p->tags[j] >= 0; j++); + initial[i].tags = xmalloc(sizeof(*p->tags) * (j + 1)); + if (!initial[i].tags) + ERROR_EXIT(REG_ESPACE); + memcpy(initial[i].tags, p->tags, sizeof(*p->tags) * (j + 1)); + } + initial[i].assertions = p->assertions; + i++; + } + initial[i].state = NULL; + + tnfa->num_transitions = add; + tnfa->final = transitions + offs[tree->lastpos[0].position]; + tnfa->num_states = parse_ctx.position; + tnfa->cflags = cflags; + + tre_mem_destroy(mem); + tre_stack_destroy(stack); + xfree(counts); + xfree(offs); + + preg->TRE_REGEX_T_FIELD = (void *)tnfa; + return REG_OK; + + error_exit: + /* Free everything that was allocated and return the error code. */ + tre_mem_destroy(mem); + if (stack != NULL) + tre_stack_destroy(stack); + if (counts != NULL) + xfree(counts); + if (offs != NULL) + xfree(offs); + preg->TRE_REGEX_T_FIELD = (void *)tnfa; + regfree(preg); + return errcode; +} + + + + +void +regfree(regex_t *preg) +{ + tre_tnfa_t *tnfa; + unsigned int i; + tre_tnfa_transition_t *trans; + + tnfa = (void *)preg->TRE_REGEX_T_FIELD; + if (!tnfa) + return; + + for (i = 0; i < tnfa->num_transitions; i++) + if (tnfa->transitions[i].state) + { + if (tnfa->transitions[i].tags) + xfree(tnfa->transitions[i].tags); + if (tnfa->transitions[i].neg_classes) + xfree(tnfa->transitions[i].neg_classes); + } + if (tnfa->transitions) + xfree(tnfa->transitions); + + if (tnfa->initial) + { + for (trans = tnfa->initial; trans->state; trans++) + { + if (trans->tags) + xfree(trans->tags); + } + xfree(tnfa->initial); + } + + if (tnfa->submatch_data) + { + for (i = 0; i < tnfa->num_submatches; i++) + if (tnfa->submatch_data[i].parents) + xfree(tnfa->submatch_data[i].parents); + xfree(tnfa->submatch_data); + } + + if (tnfa->tag_directions) + xfree(tnfa->tag_directions); + if (tnfa->firstpos_chars) + xfree(tnfa->firstpos_chars); + if (tnfa->minimal_tags) + xfree(tnfa->minimal_tags); + xfree(tnfa); +} diff --git a/kal/posix/musl_src/regex/regexec.c b/kal/posix/musl_src/regex/regexec.c new file mode 100644 index 00000000..253b0e14 --- /dev/null +++ b/kal/posix/musl_src/regex/regexec.c @@ -0,0 +1,1028 @@ +/* + regexec.c - TRE POSIX compatible matching functions (and more). + + Copyright (c) 2001-2009 Ville Laurikari + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + 1. Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + + 2. Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS + ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +*/ + +#include +#include +#include +#include +#include +#include + +#include + +#include "tre.h" + +#include + +static void +tre_fill_pmatch(size_t nmatch, regmatch_t pmatch[], int cflags, + const tre_tnfa_t *tnfa, regoff_t *tags, regoff_t match_eo); + +/*********************************************************************** + from tre-match-utils.h +***********************************************************************/ + +#define GET_NEXT_WCHAR() do { \ + prev_c = next_c; pos += pos_add_next; \ + if ((pos_add_next = mbtowc(&next_c, str_byte, MB_LEN_MAX)) <= 0) { \ + if (pos_add_next < 0) { ret = REG_NOMATCH; goto error_exit; } \ + else pos_add_next++; \ + } \ + str_byte += pos_add_next; \ + } while (0) + +#define IS_WORD_CHAR(c) ((c) == L'_' || tre_isalnum(c)) + +#define CHECK_ASSERTIONS(assertions) \ + (((assertions & ASSERT_AT_BOL) \ + && (pos > 0 || reg_notbol) \ + && (prev_c != L'\n' || !reg_newline)) \ + || ((assertions & ASSERT_AT_EOL) \ + && (next_c != L'\0' || reg_noteol) \ + && (next_c != L'\n' || !reg_newline)) \ + || ((assertions & ASSERT_AT_BOW) \ + && (IS_WORD_CHAR(prev_c) || !IS_WORD_CHAR(next_c))) \ + || ((assertions & ASSERT_AT_EOW) \ + && (!IS_WORD_CHAR(prev_c) || IS_WORD_CHAR(next_c))) \ + || ((assertions & ASSERT_AT_WB) \ + && (pos != 0 && next_c != L'\0' \ + && IS_WORD_CHAR(prev_c) == IS_WORD_CHAR(next_c))) \ + || ((assertions & ASSERT_AT_WB_NEG) \ + && (pos == 0 || next_c == L'\0' \ + || IS_WORD_CHAR(prev_c) != IS_WORD_CHAR(next_c)))) + +#define CHECK_CHAR_CLASSES(trans_i, tnfa, eflags) \ + (((trans_i->assertions & ASSERT_CHAR_CLASS) \ + && !(tnfa->cflags & REG_ICASE) \ + && !tre_isctype((tre_cint_t)prev_c, trans_i->u.class)) \ + || ((trans_i->assertions & ASSERT_CHAR_CLASS) \ + && (tnfa->cflags & REG_ICASE) \ + && !tre_isctype(tre_tolower((tre_cint_t)prev_c),trans_i->u.class) \ + && !tre_isctype(tre_toupper((tre_cint_t)prev_c),trans_i->u.class)) \ + || ((trans_i->assertions & ASSERT_CHAR_CLASS_NEG) \ + && tre_neg_char_classes_match(trans_i->neg_classes,(tre_cint_t)prev_c,\ + tnfa->cflags & REG_ICASE))) + + + + +/* Returns 1 if `t1' wins `t2', 0 otherwise. */ +static int +tre_tag_order(int num_tags, tre_tag_direction_t *tag_directions, + regoff_t *t1, regoff_t *t2) +{ + int i; + for (i = 0; i < num_tags; i++) + { + if (tag_directions[i] == TRE_TAG_MINIMIZE) + { + if (t1[i] < t2[i]) + return 1; + if (t1[i] > t2[i]) + return 0; + } + else + { + if (t1[i] > t2[i]) + return 1; + if (t1[i] < t2[i]) + return 0; + } + } + /* assert(0);*/ + return 0; +} + +static int +tre_neg_char_classes_match(tre_ctype_t *classes, tre_cint_t wc, int icase) +{ + while (*classes != (tre_ctype_t)0) + if ((!icase && tre_isctype(wc, *classes)) + || (icase && (tre_isctype(tre_toupper(wc), *classes) + || tre_isctype(tre_tolower(wc), *classes)))) + return 1; /* Match. */ + else + classes++; + return 0; /* No match. */ +} + + +/*********************************************************************** + from tre-match-parallel.c +***********************************************************************/ + +/* + This algorithm searches for matches basically by reading characters + in the searched string one by one, starting at the beginning. All + matching paths in the TNFA are traversed in parallel. When two or + more paths reach the same state, exactly one is chosen according to + tag ordering rules; if returning submatches is not required it does + not matter which path is chosen. + + The worst case time required for finding the leftmost and longest + match, or determining that there is no match, is always linearly + dependent on the length of the text being searched. + + This algorithm cannot handle TNFAs with back referencing nodes. + See `tre-match-backtrack.c'. +*/ + +typedef struct { + tre_tnfa_transition_t *state; + regoff_t *tags; +} tre_tnfa_reach_t; + +typedef struct { + regoff_t pos; + regoff_t **tags; +} tre_reach_pos_t; + + +static reg_errcode_t +tre_tnfa_run_parallel(const tre_tnfa_t *tnfa, const void *string, + regoff_t *match_tags, int eflags, + regoff_t *match_end_ofs) +{ + /* State variables required by GET_NEXT_WCHAR. */ + tre_char_t prev_c = 0, next_c = 0; + const char *str_byte = string; + regoff_t pos = -1; + regoff_t pos_add_next = 1; +#ifdef TRE_MBSTATE + mbstate_t mbstate; +#endif /* TRE_MBSTATE */ + int reg_notbol = eflags & REG_NOTBOL; + int reg_noteol = eflags & REG_NOTEOL; + int reg_newline = tnfa->cflags & REG_NEWLINE; + reg_errcode_t ret; + + char *buf; + tre_tnfa_transition_t *trans_i; + tre_tnfa_reach_t *reach, *reach_next, *reach_i, *reach_next_i; + tre_reach_pos_t *reach_pos; + int *tag_i; + int num_tags, i; + + regoff_t match_eo = -1; /* end offset of match (-1 if no match found yet) */ + int new_match = 0; + regoff_t *tmp_tags = NULL; + regoff_t *tmp_iptr; + +#ifdef TRE_MBSTATE + memset(&mbstate, '\0', sizeof(mbstate)); +#endif /* TRE_MBSTATE */ + + if (!match_tags) + num_tags = 0; + else + num_tags = tnfa->num_tags; + + /* Allocate memory for temporary data required for matching. This needs to + be done for every matching operation to be thread safe. This allocates + everything in a single large block with calloc(). */ + { + size_t tbytes, rbytes, pbytes, xbytes, total_bytes; + char *tmp_buf; + + /* Ensure that tbytes and xbytes*num_states cannot overflow, and that + * they don't contribute more than 1/8 of SIZE_MAX to total_bytes. */ + if (num_tags > SIZE_MAX/(8 * sizeof(regoff_t) * tnfa->num_states)) + return REG_ESPACE; + + /* Likewise check rbytes. */ + if (tnfa->num_states+1 > SIZE_MAX/(8 * sizeof(*reach_next))) + return REG_ESPACE; + + /* Likewise check pbytes. */ + if (tnfa->num_states > SIZE_MAX/(8 * sizeof(*reach_pos))) + return REG_ESPACE; + + /* Compute the length of the block we need. */ + tbytes = sizeof(*tmp_tags) * num_tags; + rbytes = sizeof(*reach_next) * (tnfa->num_states + 1); + pbytes = sizeof(*reach_pos) * tnfa->num_states; + xbytes = sizeof(regoff_t) * num_tags; + total_bytes = + (sizeof(long) - 1) * 4 /* for alignment paddings */ + + (rbytes + xbytes * tnfa->num_states) * 2 + tbytes + pbytes; + + /* Allocate the memory. */ + buf = calloc(total_bytes, 1); + if (buf == NULL) + return REG_ESPACE; + + /* Get the various pointers within tmp_buf (properly aligned). */ + tmp_tags = (void *)buf; + tmp_buf = buf + tbytes; + tmp_buf += ALIGN(tmp_buf, long); + reach_next = (void *)tmp_buf; + tmp_buf += rbytes; + tmp_buf += ALIGN(tmp_buf, long); + reach = (void *)tmp_buf; + tmp_buf += rbytes; + tmp_buf += ALIGN(tmp_buf, long); + reach_pos = (void *)tmp_buf; + tmp_buf += pbytes; + tmp_buf += ALIGN(tmp_buf, long); + for (i = 0; i < tnfa->num_states; i++) + { + reach[i].tags = (void *)tmp_buf; + tmp_buf += xbytes; + reach_next[i].tags = (void *)tmp_buf; + tmp_buf += xbytes; + } + } + + for (i = 0; i < tnfa->num_states; i++) + reach_pos[i].pos = -1; + + GET_NEXT_WCHAR(); + pos = 0; + + reach_next_i = reach_next; + while (1) + { + /* If no match found yet, add the initial states to `reach_next'. */ + if (match_eo < 0) + { + trans_i = tnfa->initial; + while (trans_i->state != NULL) + { + if (reach_pos[trans_i->state_id].pos < pos) + { + if (trans_i->assertions + && CHECK_ASSERTIONS(trans_i->assertions)) + { + trans_i++; + continue; + } + + reach_next_i->state = trans_i->state; + for (i = 0; i < num_tags; i++) + reach_next_i->tags[i] = -1; + tag_i = trans_i->tags; + if (tag_i) + while (*tag_i >= 0) + { + if (*tag_i < num_tags) + reach_next_i->tags[*tag_i] = pos; + tag_i++; + } + if (reach_next_i->state == tnfa->final) + { + match_eo = pos; + new_match = 1; + for (i = 0; i < num_tags; i++) + match_tags[i] = reach_next_i->tags[i]; + } + reach_pos[trans_i->state_id].pos = pos; + reach_pos[trans_i->state_id].tags = &reach_next_i->tags; + reach_next_i++; + } + trans_i++; + } + reach_next_i->state = NULL; + } + else + { + if (num_tags == 0 || reach_next_i == reach_next) + /* We have found a match. */ + break; + } + + /* Check for end of string. */ + if (!next_c) break; + + GET_NEXT_WCHAR(); + + /* Swap `reach' and `reach_next'. */ + reach_i = reach; + reach = reach_next; + reach_next = reach_i; + + /* For each state in `reach', weed out states that don't fulfill the + minimal matching conditions. */ + if (tnfa->num_minimals && new_match) + { + new_match = 0; + reach_next_i = reach_next; + for (reach_i = reach; reach_i->state; reach_i++) + { + int skip = 0; + for (i = 0; tnfa->minimal_tags[i] >= 0; i += 2) + { + int end = tnfa->minimal_tags[i]; + int start = tnfa->minimal_tags[i + 1]; + if (end >= num_tags) + { + skip = 1; + break; + } + else if (reach_i->tags[start] == match_tags[start] + && reach_i->tags[end] < match_tags[end]) + { + skip = 1; + break; + } + } + if (!skip) + { + reach_next_i->state = reach_i->state; + tmp_iptr = reach_next_i->tags; + reach_next_i->tags = reach_i->tags; + reach_i->tags = tmp_iptr; + reach_next_i++; + } + } + reach_next_i->state = NULL; + + /* Swap `reach' and `reach_next'. */ + reach_i = reach; + reach = reach_next; + reach_next = reach_i; + } + + /* For each state in `reach' see if there is a transition leaving with + the current input symbol to a state not yet in `reach_next', and + add the destination states to `reach_next'. */ + reach_next_i = reach_next; + for (reach_i = reach; reach_i->state; reach_i++) + { + for (trans_i = reach_i->state; trans_i->state; trans_i++) + { + /* Does this transition match the input symbol? */ + if (trans_i->code_min <= (tre_cint_t)prev_c && + trans_i->code_max >= (tre_cint_t)prev_c) + { + if (trans_i->assertions + && (CHECK_ASSERTIONS(trans_i->assertions) + || CHECK_CHAR_CLASSES(trans_i, tnfa, eflags))) + { + continue; + } + + /* Compute the tags after this transition. */ + for (i = 0; i < num_tags; i++) + tmp_tags[i] = reach_i->tags[i]; + tag_i = trans_i->tags; + if (tag_i != NULL) + while (*tag_i >= 0) + { + if (*tag_i < num_tags) + tmp_tags[*tag_i] = pos; + tag_i++; + } + + if (reach_pos[trans_i->state_id].pos < pos) + { + /* Found an unvisited node. */ + reach_next_i->state = trans_i->state; + tmp_iptr = reach_next_i->tags; + reach_next_i->tags = tmp_tags; + tmp_tags = tmp_iptr; + reach_pos[trans_i->state_id].pos = pos; + reach_pos[trans_i->state_id].tags = &reach_next_i->tags; + + if (reach_next_i->state == tnfa->final + && (match_eo == -1 + || (num_tags > 0 + && reach_next_i->tags[0] <= match_tags[0]))) + { + match_eo = pos; + new_match = 1; + for (i = 0; i < num_tags; i++) + match_tags[i] = reach_next_i->tags[i]; + } + reach_next_i++; + + } + else + { + assert(reach_pos[trans_i->state_id].pos == pos); + /* Another path has also reached this state. We choose + the winner by examining the tag values for both + paths. */ + if (tre_tag_order(num_tags, tnfa->tag_directions, + tmp_tags, + *reach_pos[trans_i->state_id].tags)) + { + /* The new path wins. */ + tmp_iptr = *reach_pos[trans_i->state_id].tags; + *reach_pos[trans_i->state_id].tags = tmp_tags; + if (trans_i->state == tnfa->final) + { + match_eo = pos; + new_match = 1; + for (i = 0; i < num_tags; i++) + match_tags[i] = tmp_tags[i]; + } + tmp_tags = tmp_iptr; + } + } + } + } + } + reach_next_i->state = NULL; + } + + *match_end_ofs = match_eo; + ret = match_eo >= 0 ? REG_OK : REG_NOMATCH; +error_exit: + xfree(buf); + return ret; +} + + + +/*********************************************************************** + from tre-match-backtrack.c +***********************************************************************/ + +/* + This matcher is for regexps that use back referencing. Regexp matching + with back referencing is an NP-complete problem on the number of back + references. The easiest way to match them is to use a backtracking + routine which basically goes through all possible paths in the TNFA + and chooses the one which results in the best (leftmost and longest) + match. This can be spectacularly expensive and may run out of stack + space, but there really is no better known generic algorithm. Quoting + Henry Spencer from comp.compilers: + + + POSIX.2 REs require longest match, which is really exciting to + implement since the obsolete ("basic") variant also includes + \. I haven't found a better way of tackling this than doing + a preliminary match using a DFA (or simulation) on a modified RE + that just replicates subREs for \, and then doing a + backtracking match to determine whether the subRE matches were + right. This can be rather slow, but I console myself with the + thought that people who use \ deserve very slow execution. + (Pun unintentional but very appropriate.) + +*/ + +typedef struct { + regoff_t pos; + const char *str_byte; + tre_tnfa_transition_t *state; + int state_id; + int next_c; + regoff_t *tags; +#ifdef TRE_MBSTATE + mbstate_t mbstate; +#endif /* TRE_MBSTATE */ +} tre_backtrack_item_t; + +typedef struct tre_backtrack_struct { + tre_backtrack_item_t item; + struct tre_backtrack_struct *prev; + struct tre_backtrack_struct *next; +} *tre_backtrack_t; + +#ifdef TRE_MBSTATE +#define BT_STACK_MBSTATE_IN stack->item.mbstate = (mbstate) +#define BT_STACK_MBSTATE_OUT (mbstate) = stack->item.mbstate +#else /* !TRE_MBSTATE */ +#define BT_STACK_MBSTATE_IN +#define BT_STACK_MBSTATE_OUT +#endif /* !TRE_MBSTATE */ + +#define tre_bt_mem_new tre_mem_new +#define tre_bt_mem_alloc tre_mem_alloc +#define tre_bt_mem_destroy tre_mem_destroy + + +#define BT_STACK_PUSH(_pos, _str_byte, _str_wide, _state, _state_id, _next_c, _tags, _mbstate) \ + do \ + { \ + int i; \ + if (!stack->next) \ + { \ + tre_backtrack_t s; \ + s = tre_bt_mem_alloc(mem, sizeof(*s)); \ + if (!s) \ + { \ + tre_bt_mem_destroy(mem); \ + if (tags) \ + xfree(tags); \ + if (pmatch) \ + xfree(pmatch); \ + if (states_seen) \ + xfree(states_seen); \ + return REG_ESPACE; \ + } \ + s->prev = stack; \ + s->next = NULL; \ + s->item.tags = tre_bt_mem_alloc(mem, \ + sizeof(*tags) * tnfa->num_tags); \ + if (!s->item.tags) \ + { \ + tre_bt_mem_destroy(mem); \ + if (tags) \ + xfree(tags); \ + if (pmatch) \ + xfree(pmatch); \ + if (states_seen) \ + xfree(states_seen); \ + return REG_ESPACE; \ + } \ + stack->next = s; \ + stack = s; \ + } \ + else \ + stack = stack->next; \ + stack->item.pos = (_pos); \ + stack->item.str_byte = (_str_byte); \ + stack->item.state = (_state); \ + stack->item.state_id = (_state_id); \ + stack->item.next_c = (_next_c); \ + for (i = 0; i < tnfa->num_tags; i++) \ + stack->item.tags[i] = (_tags)[i]; \ + BT_STACK_MBSTATE_IN; \ + } \ + while (0) + +#define BT_STACK_POP() \ + do \ + { \ + int i; \ + assert(stack->prev); \ + pos = stack->item.pos; \ + str_byte = stack->item.str_byte; \ + state = stack->item.state; \ + next_c = stack->item.next_c; \ + for (i = 0; i < tnfa->num_tags; i++) \ + tags[i] = stack->item.tags[i]; \ + BT_STACK_MBSTATE_OUT; \ + stack = stack->prev; \ + } \ + while (0) + +#undef MIN +#define MIN(a, b) ((a) <= (b) ? (a) : (b)) + +static reg_errcode_t +tre_tnfa_run_backtrack(const tre_tnfa_t *tnfa, const void *string, + regoff_t *match_tags, int eflags, regoff_t *match_end_ofs) +{ + /* State variables required by GET_NEXT_WCHAR. */ + tre_char_t prev_c = 0, next_c = 0; + const char *str_byte = string; + regoff_t pos = 0; + regoff_t pos_add_next = 1; +#ifdef TRE_MBSTATE + mbstate_t mbstate; +#endif /* TRE_MBSTATE */ + int reg_notbol = eflags & REG_NOTBOL; + int reg_noteol = eflags & REG_NOTEOL; + int reg_newline = tnfa->cflags & REG_NEWLINE; + + /* These are used to remember the necessary values of the above + variables to return to the position where the current search + started from. */ + int next_c_start; + const char *str_byte_start; + regoff_t pos_start = -1; +#ifdef TRE_MBSTATE + mbstate_t mbstate_start; +#endif /* TRE_MBSTATE */ + + /* End offset of best match so far, or -1 if no match found yet. */ + regoff_t match_eo = -1; + /* Tag arrays. */ + int *next_tags; + regoff_t *tags = NULL; + /* Current TNFA state. */ + tre_tnfa_transition_t *state; + int *states_seen = NULL; + + /* Memory allocator to for allocating the backtracking stack. */ + tre_mem_t mem = tre_bt_mem_new(); + + /* The backtracking stack. */ + tre_backtrack_t stack; + + tre_tnfa_transition_t *trans_i; + regmatch_t *pmatch = NULL; + int ret; + +#ifdef TRE_MBSTATE + memset(&mbstate, '\0', sizeof(mbstate)); +#endif /* TRE_MBSTATE */ + + if (!mem) + return REG_ESPACE; + stack = tre_bt_mem_alloc(mem, sizeof(*stack)); + if (!stack) + { + ret = REG_ESPACE; + goto error_exit; + } + stack->prev = NULL; + stack->next = NULL; + + if (tnfa->num_tags) + { + tags = xmalloc(sizeof(*tags) * tnfa->num_tags); + if (!tags) + { + ret = REG_ESPACE; + goto error_exit; + } + } + if (tnfa->num_submatches) + { + pmatch = xmalloc(sizeof(*pmatch) * tnfa->num_submatches); + if (!pmatch) + { + ret = REG_ESPACE; + goto error_exit; + } + } + if (tnfa->num_states) + { + states_seen = xmalloc(sizeof(*states_seen) * tnfa->num_states); + if (!states_seen) + { + ret = REG_ESPACE; + goto error_exit; + } + } + + retry: + { + int i; + for (i = 0; i < tnfa->num_tags; i++) + { + tags[i] = -1; + if (match_tags) + match_tags[i] = -1; + } + for (i = 0; i < tnfa->num_states; i++) + states_seen[i] = 0; + } + + state = NULL; + pos = pos_start; + GET_NEXT_WCHAR(); + pos_start = pos; + next_c_start = next_c; + str_byte_start = str_byte; +#ifdef TRE_MBSTATE + mbstate_start = mbstate; +#endif /* TRE_MBSTATE */ + + /* Handle initial states. */ + next_tags = NULL; + for (trans_i = tnfa->initial; trans_i->state; trans_i++) + { + if (trans_i->assertions && CHECK_ASSERTIONS(trans_i->assertions)) + { + continue; + } + if (state == NULL) + { + /* Start from this state. */ + state = trans_i->state; + next_tags = trans_i->tags; + } + else + { + /* Backtrack to this state. */ + BT_STACK_PUSH(pos, str_byte, 0, trans_i->state, + trans_i->state_id, next_c, tags, mbstate); + { + int *tmp = trans_i->tags; + if (tmp) + while (*tmp >= 0) + stack->item.tags[*tmp++] = pos; + } + } + } + + if (next_tags) + for (; *next_tags >= 0; next_tags++) + tags[*next_tags] = pos; + + + if (state == NULL) + goto backtrack; + + while (1) + { + tre_tnfa_transition_t *next_state; + int empty_br_match; + + if (state == tnfa->final) + { + if (match_eo < pos + || (match_eo == pos + && match_tags + && tre_tag_order(tnfa->num_tags, tnfa->tag_directions, + tags, match_tags))) + { + int i; + /* This match wins the previous match. */ + match_eo = pos; + if (match_tags) + for (i = 0; i < tnfa->num_tags; i++) + match_tags[i] = tags[i]; + } + /* Our TNFAs never have transitions leaving from the final state, + so we jump right to backtracking. */ + goto backtrack; + } + + /* Go to the next character in the input string. */ + empty_br_match = 0; + trans_i = state; + if (trans_i->state && trans_i->assertions & ASSERT_BACKREF) + { + /* This is a back reference state. All transitions leaving from + this state have the same back reference "assertion". Instead + of reading the next character, we match the back reference. */ + regoff_t so, eo; + int bt = trans_i->u.backref; + regoff_t bt_len; + int result; + + /* Get the substring we need to match against. Remember to + turn off REG_NOSUB temporarily. */ + tre_fill_pmatch(bt + 1, pmatch, tnfa->cflags & ~REG_NOSUB, + tnfa, tags, pos); + so = pmatch[bt].rm_so; + eo = pmatch[bt].rm_eo; + bt_len = eo - so; + + result = strncmp((const char*)string + so, str_byte - 1, + (size_t)bt_len); + + if (result == 0) + { + /* Back reference matched. Check for infinite loop. */ + if (bt_len == 0) + empty_br_match = 1; + if (empty_br_match && states_seen[trans_i->state_id]) + { + goto backtrack; + } + + states_seen[trans_i->state_id] = empty_br_match; + + /* Advance in input string and resync `prev_c', `next_c' + and pos. */ + str_byte += bt_len - 1; + pos += bt_len - 1; + GET_NEXT_WCHAR(); + } + else + { + goto backtrack; + } + } + else + { + /* Check for end of string. */ + if (next_c == L'\0') + goto backtrack; + + /* Read the next character. */ + GET_NEXT_WCHAR(); + } + + next_state = NULL; + for (trans_i = state; trans_i->state; trans_i++) + { + if (trans_i->code_min <= (tre_cint_t)prev_c + && trans_i->code_max >= (tre_cint_t)prev_c) + { + if (trans_i->assertions + && (CHECK_ASSERTIONS(trans_i->assertions) + || CHECK_CHAR_CLASSES(trans_i, tnfa, eflags))) + { + continue; + } + + if (next_state == NULL) + { + /* First matching transition. */ + next_state = trans_i->state; + next_tags = trans_i->tags; + } + else + { + /* Second matching transition. We may need to backtrack here + to take this transition instead of the first one, so we + push this transition in the backtracking stack so we can + jump back here if needed. */ + BT_STACK_PUSH(pos, str_byte, 0, trans_i->state, + trans_i->state_id, next_c, tags, mbstate); + { + int *tmp; + for (tmp = trans_i->tags; tmp && *tmp >= 0; tmp++) + stack->item.tags[*tmp] = pos; + } +#if 0 /* XXX - it's important not to look at all transitions here to keep + the stack small! */ + break; +#endif + } + } + } + + if (next_state != NULL) + { + /* Matching transitions were found. Take the first one. */ + state = next_state; + + /* Update the tag values. */ + if (next_tags) + while (*next_tags >= 0) + tags[*next_tags++] = pos; + } + else + { + backtrack: + /* A matching transition was not found. Try to backtrack. */ + if (stack->prev) + { + if (stack->item.state->assertions & ASSERT_BACKREF) + { + states_seen[stack->item.state_id] = 0; + } + + BT_STACK_POP(); + } + else if (match_eo < 0) + { + /* Try starting from a later position in the input string. */ + /* Check for end of string. */ + if (next_c == L'\0') + { + break; + } + next_c = next_c_start; +#ifdef TRE_MBSTATE + mbstate = mbstate_start; +#endif /* TRE_MBSTATE */ + str_byte = str_byte_start; + goto retry; + } + else + { + break; + } + } + } + + ret = match_eo >= 0 ? REG_OK : REG_NOMATCH; + *match_end_ofs = match_eo; + + error_exit: + tre_bt_mem_destroy(mem); +#ifndef TRE_USE_ALLOCA + if (tags) + xfree(tags); + if (pmatch) + xfree(pmatch); + if (states_seen) + xfree(states_seen); +#endif /* !TRE_USE_ALLOCA */ + + return ret; +} + +/*********************************************************************** + from regexec.c +***********************************************************************/ + +/* Fills the POSIX.2 regmatch_t array according to the TNFA tag and match + endpoint values. */ +static void +tre_fill_pmatch(size_t nmatch, regmatch_t pmatch[], int cflags, + const tre_tnfa_t *tnfa, regoff_t *tags, regoff_t match_eo) +{ + tre_submatch_data_t *submatch_data; + unsigned int i, j; + int *parents; + + i = 0; + if (match_eo >= 0 && !(cflags & REG_NOSUB)) + { + /* Construct submatch offsets from the tags. */ + submatch_data = tnfa->submatch_data; + while (i < tnfa->num_submatches && i < nmatch) + { + if (submatch_data[i].so_tag == tnfa->end_tag) + pmatch[i].rm_so = match_eo; + else + pmatch[i].rm_so = tags[submatch_data[i].so_tag]; + + if (submatch_data[i].eo_tag == tnfa->end_tag) + pmatch[i].rm_eo = match_eo; + else + pmatch[i].rm_eo = tags[submatch_data[i].eo_tag]; + + /* If either of the endpoints were not used, this submatch + was not part of the match. */ + if (pmatch[i].rm_so == -1 || pmatch[i].rm_eo == -1) + pmatch[i].rm_so = pmatch[i].rm_eo = -1; + + i++; + } + /* Reset all submatches that are not within all of their parent + submatches. */ + i = 0; + while (i < tnfa->num_submatches && i < nmatch) + { + if (pmatch[i].rm_eo == -1) + assert(pmatch[i].rm_so == -1); + assert(pmatch[i].rm_so <= pmatch[i].rm_eo); + + parents = submatch_data[i].parents; + if (parents != NULL) + for (j = 0; parents[j] >= 0; j++) + { + if (pmatch[i].rm_so < pmatch[parents[j]].rm_so + || pmatch[i].rm_eo > pmatch[parents[j]].rm_eo) + pmatch[i].rm_so = pmatch[i].rm_eo = -1; + } + i++; + } + } + + while (i < nmatch) + { + pmatch[i].rm_so = -1; + pmatch[i].rm_eo = -1; + i++; + } +} + + +/* + Wrapper functions for POSIX compatible regexp matching. +*/ + +int +regexec(const regex_t *restrict preg, const char *restrict string, + size_t nmatch, regmatch_t pmatch[restrict], int eflags) +{ + tre_tnfa_t *tnfa = (void *)preg->TRE_REGEX_T_FIELD; + reg_errcode_t status; + regoff_t *tags = NULL, eo; + if (tnfa->cflags & REG_NOSUB) nmatch = 0; + if (tnfa->num_tags > 0 && nmatch > 0) + { + tags = xmalloc(sizeof(*tags) * tnfa->num_tags); + if (tags == NULL) + return REG_ESPACE; + } + + /* Dispatch to the appropriate matcher. */ + if (tnfa->have_backrefs) + { + /* The regex has back references, use the backtracking matcher. */ + status = tre_tnfa_run_backtrack(tnfa, string, tags, eflags, &eo); + } + else + { + /* Exact matching, no back references, use the parallel matcher. */ + status = tre_tnfa_run_parallel(tnfa, string, tags, eflags, &eo); + } + + if (status == REG_OK) + /* A match was found, so fill the submatch registers. */ + tre_fill_pmatch(nmatch, pmatch, tnfa->cflags, tnfa, tags, eo); + if (tags) + xfree(tags); + return status; +} diff --git a/kal/posix/musl_src/regex/tre-mem.c b/kal/posix/musl_src/regex/tre-mem.c new file mode 100644 index 00000000..86f809d4 --- /dev/null +++ b/kal/posix/musl_src/regex/tre-mem.c @@ -0,0 +1,158 @@ +/* + tre-mem.c - TRE memory allocator + + Copyright (c) 2001-2009 Ville Laurikari + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + 1. Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + + 2. Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS + ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +*/ + +/* + This memory allocator is for allocating small memory blocks efficiently + in terms of memory overhead and execution speed. The allocated blocks + cannot be freed individually, only all at once. There can be multiple + allocators, though. +*/ + +#include +#include + +#include "tre.h" + +/* + This memory allocator is for allocating small memory blocks efficiently + in terms of memory overhead and execution speed. The allocated blocks + cannot be freed individually, only all at once. There can be multiple + allocators, though. +*/ + +/* Returns a new memory allocator or NULL if out of memory. */ +tre_mem_t +tre_mem_new_impl(int provided, void *provided_block) +{ + tre_mem_t mem; + if (provided) + { + mem = provided_block; + memset(mem, 0, sizeof(*mem)); + } + else + mem = xcalloc(1, sizeof(*mem)); + if (mem == NULL) + return NULL; + return mem; +} + + +/* Frees the memory allocator and all memory allocated with it. */ +void +tre_mem_destroy(tre_mem_t mem) +{ + tre_list_t *tmp, *l = mem->blocks; + + while (l != NULL) + { + xfree(l->data); + tmp = l->next; + xfree(l); + l = tmp; + } + xfree(mem); +} + + +/* Allocates a block of `size' bytes from `mem'. Returns a pointer to the + allocated block or NULL if an underlying malloc() failed. */ +void * +tre_mem_alloc_impl(tre_mem_t mem, int provided, void *provided_block, + int zero, size_t size) +{ + void *ptr; + + if (mem->failed) + { + return NULL; + } + + if (mem->n < size) + { + /* We need more memory than is available in the current block. + Allocate a new block. */ + tre_list_t *l; + if (provided) + { + if (provided_block == NULL) + { + mem->failed = 1; + return NULL; + } + mem->ptr = provided_block; + mem->n = TRE_MEM_BLOCK_SIZE; + } + else + { + int block_size; + if (size * 8 > TRE_MEM_BLOCK_SIZE) + block_size = size * 8; + else + block_size = TRE_MEM_BLOCK_SIZE; + l = xmalloc(sizeof(*l)); + if (l == NULL) + { + mem->failed = 1; + return NULL; + } + l->data = xmalloc(block_size); + if (l->data == NULL) + { + xfree(l); + mem->failed = 1; + return NULL; + } + l->next = NULL; + if (mem->current != NULL) + mem->current->next = l; + if (mem->blocks == NULL) + mem->blocks = l; + mem->current = l; + mem->ptr = l->data; + mem->n = block_size; + } + } + + /* Make sure the next pointer will be aligned. */ + size += ALIGN(mem->ptr + size, long); + + /* Allocate from current block. */ + ptr = mem->ptr; + mem->ptr += size; + mem->n -= size; + + /* Set to zero if needed. */ + if (zero) + memset(ptr, 0, size); + + return ptr; +} diff --git a/kal/posix/musl_src/regex/tre.h b/kal/posix/musl_src/regex/tre.h new file mode 100644 index 00000000..6e099375 --- /dev/null +++ b/kal/posix/musl_src/regex/tre.h @@ -0,0 +1,232 @@ +/* + tre-internal.h - TRE internal definitions + + Copyright (c) 2001-2009 Ville Laurikari + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + 1. Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + + 2. Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS + ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +*/ + +#include +#include +#include +#include "../include/features.h" + +#undef TRE_MBSTATE + +#define NDEBUG + +#define TRE_REGEX_T_FIELD __opaque +typedef int reg_errcode_t; + +typedef wchar_t tre_char_t; + +#define DPRINT(msg) do { } while(0) + +#define elementsof(x) ( sizeof(x) / sizeof(x[0]) ) + +#define tre_mbrtowc(pwc, s, n, ps) (mbtowc((pwc), (s), (n))) + +/* Wide characters. */ +typedef wint_t tre_cint_t; +#define TRE_CHAR_MAX 0x10ffff + +#define tre_isalnum iswalnum +#define tre_isalpha iswalpha +#define tre_isblank iswblank +#define tre_iscntrl iswcntrl +#define tre_isdigit iswdigit +#define tre_isgraph iswgraph +#define tre_islower iswlower +#define tre_isprint iswprint +#define tre_ispunct iswpunct +#define tre_isspace iswspace +#define tre_isupper iswupper +#define tre_isxdigit iswxdigit + +#define tre_tolower towlower +#define tre_toupper towupper +#define tre_strlen wcslen + +/* Use system provided iswctype() and wctype(). */ +typedef wctype_t tre_ctype_t; +#define tre_isctype iswctype +#define tre_ctype wctype + +/* Returns number of bytes to add to (char *)ptr to make it + properly aligned for the type. */ +#define ALIGN(ptr, type) \ + ((((long)ptr) % sizeof(type)) \ + ? (sizeof(type) - (((long)ptr) % sizeof(type))) \ + : 0) + +#undef MAX +#undef MIN +#define MAX(a, b) (((a) >= (b)) ? (a) : (b)) +#define MIN(a, b) (((a) <= (b)) ? (a) : (b)) + +/* TNFA transition type. A TNFA state is an array of transitions, + the terminator is a transition with NULL `state'. */ +typedef struct tnfa_transition tre_tnfa_transition_t; + +struct tnfa_transition { + /* Range of accepted characters. */ + tre_cint_t code_min; + tre_cint_t code_max; + /* Pointer to the destination state. */ + tre_tnfa_transition_t *state; + /* ID number of the destination state. */ + int state_id; + /* -1 terminated array of tags (or NULL). */ + int *tags; + /* Assertion bitmap. */ + int assertions; + /* Assertion parameters. */ + union { + /* Character class assertion. */ + tre_ctype_t class; + /* Back reference assertion. */ + int backref; + } u; + /* Negative character class assertions. */ + tre_ctype_t *neg_classes; +}; + + +/* Assertions. */ +#define ASSERT_AT_BOL 1 /* Beginning of line. */ +#define ASSERT_AT_EOL 2 /* End of line. */ +#define ASSERT_CHAR_CLASS 4 /* Character class in `class'. */ +#define ASSERT_CHAR_CLASS_NEG 8 /* Character classes in `neg_classes'. */ +#define ASSERT_AT_BOW 16 /* Beginning of word. */ +#define ASSERT_AT_EOW 32 /* End of word. */ +#define ASSERT_AT_WB 64 /* Word boundary. */ +#define ASSERT_AT_WB_NEG 128 /* Not a word boundary. */ +#define ASSERT_BACKREF 256 /* A back reference in `backref'. */ +#define ASSERT_LAST 256 + +/* Tag directions. */ +typedef enum { + TRE_TAG_MINIMIZE = 0, + TRE_TAG_MAXIMIZE = 1 +} tre_tag_direction_t; + +/* Instructions to compute submatch register values from tag values + after a successful match. */ +struct tre_submatch_data { + /* Tag that gives the value for rm_so (submatch start offset). */ + int so_tag; + /* Tag that gives the value for rm_eo (submatch end offset). */ + int eo_tag; + /* List of submatches this submatch is contained in. */ + int *parents; +}; + +typedef struct tre_submatch_data tre_submatch_data_t; + + +/* TNFA definition. */ +typedef struct tnfa tre_tnfa_t; + +struct tnfa { + tre_tnfa_transition_t *transitions; + unsigned int num_transitions; + tre_tnfa_transition_t *initial; + tre_tnfa_transition_t *final; + tre_submatch_data_t *submatch_data; + char *firstpos_chars; + int first_char; + unsigned int num_submatches; + tre_tag_direction_t *tag_directions; + int *minimal_tags; + int num_tags; + int num_minimals; + int end_tag; + int num_states; + int cflags; + int have_backrefs; + int have_approx; +}; + +/* from tre-mem.h: */ + +#define TRE_MEM_BLOCK_SIZE 1024 + +typedef struct tre_list { + void *data; + struct tre_list *next; +} tre_list_t; + +typedef struct tre_mem_struct { + tre_list_t *blocks; + tre_list_t *current; + char *ptr; + size_t n; + int failed; + void **provided; +} *tre_mem_t; + +#define tre_mem_new_impl __tre_mem_new_impl +#define tre_mem_alloc_impl __tre_mem_alloc_impl +#define tre_mem_destroy __tre_mem_destroy + +hidden tre_mem_t tre_mem_new_impl(int provided, void *provided_block); +hidden void *tre_mem_alloc_impl(tre_mem_t mem, int provided, void *provided_block, + int zero, size_t size); + +/* Returns a new memory allocator or NULL if out of memory. */ +#define tre_mem_new() tre_mem_new_impl(0, NULL) + +/* Allocates a block of `size' bytes from `mem'. Returns a pointer to the + allocated block or NULL if an underlying malloc() failed. */ +#define tre_mem_alloc(mem, size) tre_mem_alloc_impl(mem, 0, NULL, 0, size) + +/* Allocates a block of `size' bytes from `mem'. Returns a pointer to the + allocated block or NULL if an underlying malloc() failed. The memory + is set to zero. */ +#define tre_mem_calloc(mem, size) tre_mem_alloc_impl(mem, 0, NULL, 1, size) + +#ifdef TRE_USE_ALLOCA +/* alloca() versions. Like above, but memory is allocated with alloca() + instead of malloc(). */ + +#define tre_mem_newa() \ + tre_mem_new_impl(1, alloca(sizeof(struct tre_mem_struct))) + +#define tre_mem_alloca(mem, size) \ + ((mem)->n >= (size) \ + ? tre_mem_alloc_impl((mem), 1, NULL, 0, (size)) \ + : tre_mem_alloc_impl((mem), 1, alloca(TRE_MEM_BLOCK_SIZE), 0, (size))) +#endif /* TRE_USE_ALLOCA */ + + +/* Frees the memory allocator and all memory allocated with it. */ +hidden void tre_mem_destroy(tre_mem_t mem); + +#define xmalloc malloc +#define xcalloc calloc +#define xfree free +#define xrealloc realloc + diff --git a/kal/posix/musl_src/stdio/__fdopen.c b/kal/posix/musl_src/stdio/__fdopen.c new file mode 100644 index 00000000..e89f930d --- /dev/null +++ b/kal/posix/musl_src/stdio/__fdopen.c @@ -0,0 +1,62 @@ +#include "stdio_impl.h" +#include +#include +#include +#include +#include +#include + +FILE *__fdopen(int fd, const char *mode) +{ + FILE *f; + struct winsize wsz; + pthread_mutex_t filelockinit = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP; + + /* Check for valid initial mode character */ + if (!strchr("rwa", *mode)) { + errno = EINVAL; + return 0; + } + + /* Allocate FILE+buffer or fail */ + if (!(f=malloc(sizeof *f + UNGET + BUFSIZ + sizeof(pthread_mutex_t)))) return 0; + + /* Zero-fill only the struct, not the buffer */ + memset(f, 0, sizeof *f); + + /* Impose mode restrictions */ + if (!strchr(mode, '+')) f->flags = (*mode == 'r') ? F_NOWR : F_NORD; + + /* Apply close-on-exec flag */ + if (strchr(mode, 'e')) fcntl(fd, F_SETFD, FD_CLOEXEC); + + /* Set append mode on fd if opened for append */ + if (*mode == 'a') { + int flags = fcntl(fd, F_GETFL); + if (!(flags & O_APPEND)) + fcntl(fd, F_SETFL, flags | O_APPEND); + f->flags |= F_APP; + } + + f->fd = fd; + f->buf = (unsigned char *)f + sizeof *f + UNGET; + f->buf_size = BUFSIZ; + f->lock = (pthread_mutex_t *)((unsigned char *)f + sizeof *f + UNGET + BUFSIZ); + memcpy(f->lock, &filelockinit, sizeof(pthread_mutex_t)); + + /* Activate line buffered mode for terminals */ + f->lbf = EOF; + if (!(f->flags & F_NOWR) && !ioctl(fd, TIOCGWINSZ, &wsz)) + f->lbf = '\n'; + + /* Initialize op ptrs. No problem if some are unneeded. */ + f->read = __stdio_read; + f->write = __stdio_write; + f->seek = __stdio_seek; + f->close = __stdio_close; + + /* Add new FILE to open file list */ + return __ofl_add(f); +} + +weak_alias(__fdopen, fdopen); diff --git a/kal/posix/musl_src/stdio/__fmodeflags.c b/kal/posix/musl_src/stdio/__fmodeflags.c new file mode 100644 index 00000000..3c2370a5 --- /dev/null +++ b/kal/posix/musl_src/stdio/__fmodeflags.c @@ -0,0 +1,16 @@ +#include +#include + +int __fmodeflags(const char *mode) +{ + int flags; + if (strchr(mode, '+')) flags = O_RDWR; + else if (*mode == 'r') flags = O_RDONLY; + else flags = O_WRONLY; + if (strchr(mode, 'x')) flags |= O_EXCL; + if (strchr(mode, 'e')) flags |= O_CLOEXEC; + if (*mode != 'r') flags |= O_CREAT; + if (*mode == 'w') flags |= O_TRUNC; + if (*mode == 'a') flags |= O_APPEND; + return flags; +} diff --git a/kal/posix/musl_src/stdio/__lockfile.c b/kal/posix/musl_src/stdio/__lockfile.c new file mode 100644 index 00000000..1399f969 --- /dev/null +++ b/kal/posix/musl_src/stdio/__lockfile.c @@ -0,0 +1,12 @@ +#include "stdio_impl.h" +#include "pthread.h" + +int __lockfile(FILE *f) +{ + return pthread_mutex_lock((pthread_mutex_t *)f->lock); +} + +int __unlockfile(FILE *f) +{ + return pthread_mutex_unlock((pthread_mutex_t *)f->lock); +} diff --git a/kal/posix/musl_src/stdio/__stdio_close.c b/kal/posix/musl_src/stdio/__stdio_close.c new file mode 100644 index 00000000..c97688b2 --- /dev/null +++ b/kal/posix/musl_src/stdio/__stdio_close.c @@ -0,0 +1,7 @@ +#include "stdio_impl.h" +#include + +int __stdio_close(FILE *f) +{ + return close(f->fd); +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/__stdio_read.c b/kal/posix/musl_src/stdio/__stdio_read.c new file mode 100644 index 00000000..fcacffee --- /dev/null +++ b/kal/posix/musl_src/stdio/__stdio_read.c @@ -0,0 +1,25 @@ +#include "stdio_impl.h" +#include +#include + +size_t __stdio_read(FILE *f, unsigned char *buf, size_t len) +{ + struct iovec iov[2] = { + { .iov_base = buf, .iov_len = len - !!f->buf_size }, + { .iov_base = f->buf, .iov_len = f->buf_size } + }; + ssize_t cnt; + + cnt = read(f->fd, iov[0].iov_base, iov[0].iov_len); + if (iov[1].iov_len - iov[0].iov_len > 0) cnt += read(f->fd, iov[1].iov_base, iov[1].iov_len - iov[0].iov_len); + if (cnt <= 0) { + f->flags |= cnt ? F_ERR : F_EOF; + return 0; + } + if (cnt <= iov[0].iov_len) return cnt; + cnt -= iov[0].iov_len; + f->rpos = f->buf; + f->rend = f->buf + cnt; + if (f->buf_size) buf[len-1] = *f->rpos++; + return len; +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/__stdio_seek.c b/kal/posix/musl_src/stdio/__stdio_seek.c new file mode 100644 index 00000000..57060258 --- /dev/null +++ b/kal/posix/musl_src/stdio/__stdio_seek.c @@ -0,0 +1,53 @@ +#include "stdio_impl.h" +#include + +#if 0 +static off64_t __stdio_lseek64(int fd, int offsetHigh, int offsetLow, off64_t *result, int whence) +{ + off64_t ret; + struct file *filep = NULL; + off64_t offset = ((off64_t)offsetHigh << 32) + (uint)offsetLow; /* 32: offsetHigh is high 32 bits */ + + /* Get the file structure corresponding to the file descriptor. */ + ret = fs_getfilep(fd, &filep); + if (ret < 0) { + /* The errno value has already been set */ + return (off64_t)-get_errno(); + } + + /* libc seekdir function should set the whence to SEEK_SET, so we can discard + * the whence argument here */ + if (filep->f_oflags & O_DIRECTORY) { + /* defensive coding */ + if (filep->f_dir == NULL) { + return (off64_t)-EINVAL; + } + if (offsetLow == 0) { + rewinddir(filep->f_dir); + } else { + seekdir(filep->f_dir, offsetLow); + } + ret = telldir(filep->f_dir); + if (ret < 0) { + return (off64_t)-get_errno(); + } + goto out; + } + + /* Then let file_seek do the real work */ + ret = file_seek64(filep, offset, whence); + if (ret < 0) { + return (off64_t)-get_errno(); + } + +out: + *result = ret; + + return 0; +} +#endif +off_t __stdio_seek(FILE *f, off_t off, int whence) +{ + off_t result = 0; + return lseek(f->fd, (unsigned int)off, whence) ? -1 : result; +} diff --git a/kal/posix/musl_src/stdio/__stdio_write.c b/kal/posix/musl_src/stdio/__stdio_write.c new file mode 100644 index 00000000..cf9c1d31 --- /dev/null +++ b/kal/posix/musl_src/stdio/__stdio_write.c @@ -0,0 +1,35 @@ +#include "stdio_impl.h" +#include + +size_t __stdio_write(FILE *f, const unsigned char *buf, size_t len) +{ + struct iovec iovs[2] = { + { .iov_base = f->wbase, .iov_len = f->wpos-f->wbase }, + { .iov_base = (void *)buf, .iov_len = len } + }; + struct iovec *iov = iovs; + size_t rem = iov[0].iov_len + iov[1].iov_len; + int iovcnt = 2; + ssize_t cnt; + for (;;) { + cnt = write(f->fd, iov[0].iov_base, iov[0].iov_len); + if (0 != iov[1].iov_len) cnt += write(f->fd, iov[1].iov_base, iov[1].iov_len); + if (cnt == rem) { + f->wend = f->buf + f->buf_size; + f->wpos = f->wbase = f->buf; + return len; + } + if (cnt < 0) { + f->wpos = f->wbase = f->wend = 0; + f->flags |= F_ERR; + return iovcnt == 2 ? 0 : len-iov[0].iov_len; + } + rem -= cnt; + if (cnt > iov[0].iov_len) { + cnt -= iov[0].iov_len; + iov++; iovcnt--; + } + iov[0].iov_base = (char *)iov[0].iov_base + cnt; + iov[0].iov_len -= cnt; + } +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/__stdout_write.c b/kal/posix/musl_src/stdio/__stdout_write.c new file mode 100644 index 00000000..8205b7b7 --- /dev/null +++ b/kal/posix/musl_src/stdio/__stdout_write.c @@ -0,0 +1,11 @@ +#include "stdio_impl.h" +#include + +size_t __stdout_write(FILE *f, const unsigned char *buf, size_t len) +{ + struct winsize wsz; + if (!(f->flags & F_SVB) && ioctl(f->fd, TIOCGWINSZ, &wsz)) { + f->lbf = EOF; + } + return __stdio_write(f, buf, len); +} diff --git a/kal/posix/musl_src/stdio/__toread.c b/kal/posix/musl_src/stdio/__toread.c new file mode 100644 index 00000000..d2a377e0 --- /dev/null +++ b/kal/posix/musl_src/stdio/__toread.c @@ -0,0 +1,14 @@ +#include + +int __toread(FILE *f) +{ + f->mode |= f->mode-1; + if (f->wpos != f->wbase) f->write(f, 0, 0); + f->wpos = f->wbase = f->wend = 0; + if (f->flags & F_NORD) { + f->flags |= F_ERR; + return EOF; + } + f->rpos = f->rend = f->buf + f->buf_size; + return (f->flags & F_EOF) ? EOF : 0; +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/__towrite.c b/kal/posix/musl_src/stdio/__towrite.c new file mode 100644 index 00000000..433a9e67 --- /dev/null +++ b/kal/posix/musl_src/stdio/__towrite.c @@ -0,0 +1,18 @@ +#include "stdio_impl.h" + +int __towrite(FILE *f) +{ + f->mode |= f->mode-1; + if (f->flags & F_NOWR) { + f->flags |= F_ERR; + return EOF; + } + /* Clear read buffer (easier than summoning nasal demons) */ + f->rpos = f->rend = 0; + + /* Activate write through the buffer. */ + f->wpos = f->wbase = f->buf; + f->wend = f->buf + f->buf_size; + + return 0; +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/__uflow.c b/kal/posix/musl_src/stdio/__uflow.c new file mode 100644 index 00000000..2a88bca6 --- /dev/null +++ b/kal/posix/musl_src/stdio/__uflow.c @@ -0,0 +1,11 @@ +#include "stdio_impl.h" + +/* This function assumes it will never be called if there is already + * data buffered for reading. */ + +int __uflow(FILE *f) +{ + unsigned char c; + if (!__toread(f) && f->read(f, &c, 1)==1) return c; + return EOF; +} diff --git a/kal/posix/musl_src/stdio/clearerr.c b/kal/posix/musl_src/stdio/clearerr.c new file mode 100644 index 00000000..3bf94d30 --- /dev/null +++ b/kal/posix/musl_src/stdio/clearerr.c @@ -0,0 +1,10 @@ +#include "stdio_impl.h" + +void clearerr(FILE *f) +{ + FLOCK(f); + f->flags &= ~(F_EOF|F_ERR); + FUNLOCK(f); +} + +weak_alias(clearerr, clearerr_unlocked); diff --git a/kal/posix/musl_src/stdio/fclose.c b/kal/posix/musl_src/stdio/fclose.c new file mode 100644 index 00000000..bd6a890d --- /dev/null +++ b/kal/posix/musl_src/stdio/fclose.c @@ -0,0 +1,36 @@ +#include "stdio_impl.h" +#include + +static void dummy(FILE *f) { } +weak_alias(dummy, __unlist_locked_file); + +int fclose(FILE *f) +{ + int r; + + FLOCK(f); + r = fflush(f); + r |= f->close(f); + FUNLOCK(f); + + /* Past this point, f is closed and any further explict access + * to it is undefined. However, it still exists as an entry in + * the open file list and possibly in the thread's locked files + * list, if it was closed while explicitly locked. Functions + * which process these lists must tolerate dead FILE objects + * (which necessarily have inactive buffer pointers) without + * producing any side effects. */ + + if (f->flags & F_PERM) return r; + + FILE **head = __ofl_lock(); + if (f->prev) f->prev->next = f->next; + if (f->next) f->next->prev = f->prev; + if (*head == f) *head = f->next; + __ofl_unlock(); + + free(f->getln_buf); + free(f); + + return r; +} diff --git a/kal/posix/musl_src/stdio/feof.c b/kal/posix/musl_src/stdio/feof.c new file mode 100644 index 00000000..56da6b91 --- /dev/null +++ b/kal/posix/musl_src/stdio/feof.c @@ -0,0 +1,14 @@ +#include "stdio_impl.h" + +#undef feof + +int feof(FILE *f) +{ + FLOCK(f); + int ret = !!(f->flags & F_EOF); + FUNLOCK(f); + return ret; +} + +weak_alias(feof, feof_unlocked); +weak_alias(feof, _IO_feof_unlocked); diff --git a/kal/posix/musl_src/stdio/fflush.c b/kal/posix/musl_src/stdio/fflush.c new file mode 100644 index 00000000..71606ac9 --- /dev/null +++ b/kal/posix/musl_src/stdio/fflush.c @@ -0,0 +1,47 @@ +#include "stdio_impl.h" + +/* stdout.c will override this if linked */ +static FILE *volatile dummy = 0; +weak_alias(dummy, __stdout_used); +weak_alias(dummy, __stderr_used); + +int fflush(FILE *f) +{ + if (!f) { + int r = 0; + if (__stdout_used) r |= fflush(__stdout_used); + if (__stderr_used) r |= fflush(__stderr_used); + + for (f=*__ofl_lock(); f; f=f->next) { + FLOCK(f); + if (f->wpos != f->wbase) r |= fflush(f); + FUNLOCK(f); + } + __ofl_unlock(); + + return r; + } + + FLOCK(f); + + /* If writing, flush output */ + if (f->wpos != f->wbase) { + f->write(f, 0, 0); + if (!f->wpos) { + FUNLOCK(f); + return EOF; + } + } + + /* If reading, sync position, per POSIX */ + if (f->rpos != f->rend) lseek(f->fd, f->rpos-f->rend, SEEK_CUR); + + /* Clear read and write modes */ + f->wpos = f->wbase = f->wend = 0; + f->rpos = f->rend = 0; + + FUNLOCK(f); + return 0; +} + +weak_alias(fflush, fflush_unlocked); diff --git a/kal/posix/musl_src/stdio/fgets.c b/kal/posix/musl_src/stdio/fgets.c new file mode 100644 index 00000000..d1109e75 --- /dev/null +++ b/kal/posix/musl_src/stdio/fgets.c @@ -0,0 +1,46 @@ +#include "stdio_impl.h" +#include + +#define MIN(a,b) ((a)<(b) ? (a) : (b)) + +char *fgets(char *restrict s, int n, FILE *restrict f) +{ + char *p = s; + unsigned char *z; + size_t k; + int c; + + FLOCK(f); + + if (n--<=1) { + f->mode |= f->mode-1; + FUNLOCK(f); + if (n) return 0; + *s = 0; + return s; + } + + while (n) { + if (f->rpos != f->rend) { + z = memchr(f->rpos, '\n', f->rend - f->rpos); + k = z ? z - f->rpos + 1 : f->rend - f->rpos; + k = MIN(k, (size_t)n); + memcpy(p, f->rpos, k); + f->rpos += k; + p += k; + n -= k; + if (z || !n) break; + } + if ((c = getc_unlocked(f)) < 0) { + if (p==s || !feof(f)) s = 0; + break; + } + n--; + if ((*p++ = c) == '\n') break; + } + if (s) *p = 0; + + FUNLOCK(f); + + return s; +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/fileno.c b/kal/posix/musl_src/stdio/fileno.c new file mode 100644 index 00000000..e6e66908 --- /dev/null +++ b/kal/posix/musl_src/stdio/fileno.c @@ -0,0 +1,14 @@ +#include "stdio_impl.h" +#include + +int fileno(FILE *f) +{ + FLOCK(f); + int fd = f->fd; + FUNLOCK(f); + if (fd < 0) { + errno = EBADF; + return -1; + } + return fd; +} \ No newline at end of file diff --git a/kal/posix/musl_src/stdio/fopen.c b/kal/posix/musl_src/stdio/fopen.c new file mode 100644 index 00000000..7dddf930 --- /dev/null +++ b/kal/posix/musl_src/stdio/fopen.c @@ -0,0 +1,42 @@ +#include "stdio_impl.h" +#include +#include +#include +#include +#include + +FILE *fopen(const char *restrict filename, const char *restrict mode) +{ + FILE *f; + int fd; + int flags; + + /* Check for valid initial mode character */ + if (!strchr("rwa", *mode)) { + errno = EINVAL; + return 0; + } + + /* Compute the flags to pass to open() */ + flags = __fmodeflags(mode); + + fd = open(filename, flags, 0666); + if (fd < 0) return 0; + if (flags & O_CLOEXEC) + fcntl(fd, F_SETFD, FD_CLOEXEC); + +#if !defined(__LP64__) + if (fd > SHRT_MAX) { + errno = EMFILE; + return 0; + } +#endif + + f = __fdopen(fd, mode); + if (f) return f; + + close(fd); + return 0; +} + +weak_alias(fopen, fopen64); diff --git a/kal/posix/musl_src/stdio/fputs.c b/kal/posix/musl_src/stdio/fputs.c new file mode 100644 index 00000000..1cf344f2 --- /dev/null +++ b/kal/posix/musl_src/stdio/fputs.c @@ -0,0 +1,10 @@ +#include "stdio_impl.h" +#include + +int fputs(const char *restrict s, FILE *restrict f) +{ + size_t l = strlen(s); + return (fwrite(s, 1, l, f)==l) - 1; +} + +weak_alias(fputs, fputs_unlocked); diff --git a/kal/posix/musl_src/stdio/fread.c b/kal/posix/musl_src/stdio/fread.c new file mode 100644 index 00000000..a2116da6 --- /dev/null +++ b/kal/posix/musl_src/stdio/fread.c @@ -0,0 +1,38 @@ +#include "stdio_impl.h" +#include + +#define MIN(a,b) ((a)<(b) ? (a) : (b)) + +size_t fread(void *restrict destv, size_t size, size_t nmemb, FILE *restrict f) +{ + unsigned char *dest = destv; + size_t len = size*nmemb, l = len, k; + if (!size) nmemb = 0; + + FLOCK(f); + + f->mode |= f->mode-1; + + if (f->rpos != f->rend) { + /* First exhaust the buffer. */ + k = MIN(f->rend - f->rpos, l); + memcpy(dest, f->rpos, k); + f->rpos += k; + dest += k; + l -= k; + } + + /* Read the remainder directly */ + for (; l; l-=k, dest+=k) { + k = __toread(f) ? 0 : f->read(f, dest, l); + if (!k) { + FUNLOCK(f); + return (len-l)/size; + } + } + + FUNLOCK(f); + return nmemb; +} + +weak_alias(fread, fread_unlocked); diff --git a/kal/posix/musl_src/stdio/fseek.c b/kal/posix/musl_src/stdio/fseek.c new file mode 100644 index 00000000..2fc16d84 --- /dev/null +++ b/kal/posix/musl_src/stdio/fseek.c @@ -0,0 +1,41 @@ +#include "stdio_impl.h" + +int __fseeko_unlocked(FILE *f, off_t off, int whence) +{ + /* Adjust relative offset for unread data in buffer, if any. */ + if (whence == SEEK_CUR && f->rend) off -= f->rend - f->rpos; + + /* Flush write buffer, and report error on failure. */ + if (f->wpos != f->wbase) { + f->write(f, 0, 0); + if (!f->wpos) return -1; + } + + /* Leave writing mode */ + f->wpos = f->wbase = f->wend = 0; + + /* Perform the underlying seek. */ + if (lseek(f->fd, (unsigned int)off, whence) < 0) return -1; + + /* If seek succeeded, file is seekable and we discard read buffer. */ + f->rpos = f->rend = 0; + f->flags &= ~F_EOF; + + return 0; +} + +int __fseeko(FILE *f, off_t off, int whence) +{ + int result; + FLOCK(f); + result = __fseeko_unlocked(f, off, whence); + FUNLOCK(f); + return result; +} + +int fseek(FILE *f, long off, int whence) +{ + return __fseeko(f, off, whence); +} + +weak_alias(__fseeko, fseeko); diff --git a/kal/posix/musl_src/stdio/ftell.c b/kal/posix/musl_src/stdio/ftell.c new file mode 100644 index 00000000..d15cc665 --- /dev/null +++ b/kal/posix/musl_src/stdio/ftell.c @@ -0,0 +1,39 @@ +#include "stdio_impl.h" +#include +#include + +off_t __ftello_unlocked(FILE *f) +{ + off_t pos = lseek(f->fd, 0, + (f->flags & F_APP) && f->wpos != f->wbase + ? SEEK_END : SEEK_CUR); + if (pos < 0) return pos; + + /* Adjust for data in buffer. */ + if (f->rend) + pos += f->rpos - f->rend; + else if (f->wbase) + pos += f->wpos - f->wbase; + return pos; +} + +off_t __ftello(FILE *f) +{ + off_t pos; + FLOCK(f); + pos = __ftello_unlocked(f); + FUNLOCK(f); + return pos; +} + +long ftell(FILE *f) +{ + off_t pos = __ftello(f); + if (pos > LONG_MAX) { + errno = EOVERFLOW; + return -1; + } + return pos; +} + +weak_alias(__ftello, ftello); diff --git a/kal/posix/musl_src/stdio/fwrite.c b/kal/posix/musl_src/stdio/fwrite.c new file mode 100644 index 00000000..7a567b2c --- /dev/null +++ b/kal/posix/musl_src/stdio/fwrite.c @@ -0,0 +1,38 @@ +#include "stdio_impl.h" +#include + +size_t __fwritex(const unsigned char *restrict s, size_t l, FILE *restrict f) +{ + size_t i=0; + + if (!f->wend && __towrite(f)) return 0; + + if (l > f->wend - f->wpos) return f->write(f, s, l); + + if (f->lbf >= 0) { + /* Match /^(.*\n|)/ */ + for (i=l; i && s[i-1] != '\n'; i--); + if (i) { + size_t n = f->write(f, s, i); + if (n < i) return n; + s += i; + l -= i; + } + } + + memcpy(f->wpos, s, l); + f->wpos += l; + return l+i; +} + +size_t fwrite(const void *restrict src, size_t size, size_t nmemb, FILE *restrict f) +{ + size_t k, l = size*nmemb; + if (!size) nmemb = 0; + FLOCK(f); + k = __fwritex(src, l, f); + FUNLOCK(f); + return k==l ? nmemb : k/size; +} + +weak_alias(fwrite, fwrite_unlocked); diff --git a/kal/posix/musl_src/stdio/ofl.c b/kal/posix/musl_src/stdio/ofl.c new file mode 100644 index 00000000..4dc74ce9 --- /dev/null +++ b/kal/posix/musl_src/stdio/ofl.c @@ -0,0 +1,28 @@ +#include "stdio_impl.h" +#include + +static FILE *ofl_head; + +static pthread_mutex_t locallock = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP; + +static int LOCK(void) +{ + return pthread_mutex_lock(&locallock); +} + +static void UNLOCK(void) +{ + (void)pthread_mutex_unlock(&locallock); +} + + +FILE **__ofl_lock() +{ + LOCK(); + return &ofl_head; +} + +void __ofl_unlock() +{ + UNLOCK(); +} diff --git a/kal/posix/musl_src/stdio/ofl_add.c b/kal/posix/musl_src/stdio/ofl_add.c new file mode 100644 index 00000000..d7de9f15 --- /dev/null +++ b/kal/posix/musl_src/stdio/ofl_add.c @@ -0,0 +1,11 @@ +#include "stdio_impl.h" + +FILE *__ofl_add(FILE *f) +{ + FILE **head = __ofl_lock(); + f->next = *head; + if (*head) (*head)->prev = f; + *head = f; + __ofl_unlock(); + return f; +} diff --git a/kal/posix/musl_src/stdio/perror.c b/kal/posix/musl_src/stdio/perror.c new file mode 100644 index 00000000..03ffeab8 --- /dev/null +++ b/kal/posix/musl_src/stdio/perror.c @@ -0,0 +1,30 @@ +#include +#include +#include +#include "stdio_impl.h" + +void perror(const char *msg) +{ + FILE *f = stderr; + char *errstr = strerror(errno); + + FLOCK(f); + + /* Save stderr's orientation and encoding rule, since perror is not + * permitted to change them. */ + void *old_locale = f->locale; + int old_mode = f->mode; + + if (msg && *msg) { + fwrite(msg, strlen(msg), 1, f); + fputc(':', f); + fputc(' ', f); + } + fwrite(errstr, strlen(errstr), 1, f); + fputc('\n', f); + + f->mode = old_mode; + f->locale = old_locale; + + FUNLOCK(f); +} diff --git a/kal/posix/musl_src/stdio/rewind.c b/kal/posix/musl_src/stdio/rewind.c new file mode 100644 index 00000000..6387b995 --- /dev/null +++ b/kal/posix/musl_src/stdio/rewind.c @@ -0,0 +1,9 @@ +#include "stdio_impl.h" + +void rewind(FILE *f) +{ + FLOCK(f); + __fseeko_unlocked(f, 0, SEEK_SET); + f->flags &= ~F_ERR; + FUNLOCK(f); +} diff --git a/kal/posix/musl_src/stdio/stderr.c b/kal/posix/musl_src/stdio/stderr.c new file mode 100644 index 00000000..008ab010 --- /dev/null +++ b/kal/posix/musl_src/stdio/stderr.c @@ -0,0 +1,20 @@ +#include "stdio_impl.h" +#include + +#undef stderr + +static pthread_mutex_t locallock = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP; +static unsigned char buf[UNGET]; +hidden FILE __stderr_FILE = { + .buf = buf+UNGET, + .buf_size = 0, + .fd = 2, + .flags = F_PERM | F_NORD, + .lbf = -1, + .write = __stdio_write, + .seek = __stdio_seek, + .close = __stdio_close, + .lock = &locallock, +}; +FILE *const stderr = &__stderr_FILE; +FILE *volatile __stderr_used = &__stderr_FILE; diff --git a/kal/posix/musl_src/stdio/stdout.c b/kal/posix/musl_src/stdio/stdout.c new file mode 100644 index 00000000..226fffce --- /dev/null +++ b/kal/posix/musl_src/stdio/stdout.c @@ -0,0 +1,20 @@ +#include "stdio_impl.h" +#include + +#undef stdout + +static pthread_mutex_t locallock = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP; +static unsigned char buf[BUFSIZ+UNGET]; +hidden FILE __stdout_FILE = { + .buf = buf+UNGET, + .buf_size = sizeof buf-UNGET, + .fd = 1, + .flags = F_PERM | F_NORD, + .lbf = '\n', + .write = __stdout_write, + .seek = __stdio_seek, + .close = __stdio_close, + .lock = &locallock, +}; +FILE *const stdout = &__stdout_FILE; +FILE *volatile __stdout_used = &__stdout_FILE; diff --git a/kal/posix/musl_src/stdlib/atoi.c b/kal/posix/musl_src/stdlib/atoi.c new file mode 100644 index 00000000..9baca7b8 --- /dev/null +++ b/kal/posix/musl_src/stdlib/atoi.c @@ -0,0 +1,16 @@ +#include +#include + +int atoi(const char *s) +{ + int n=0, neg=0; + while (isspace(*s)) s++; + switch (*s) { + case '-': neg=1; + case '+': s++; + } + /* Compute n as a negative number to avoid overflow on INT_MIN */ + while (isdigit(*s)) + n = 10*n - (*s++ - '0'); + return neg ? n : -n; +} diff --git a/kal/posix/musl_src/stdlib/atol.c b/kal/posix/musl_src/stdlib/atol.c new file mode 100644 index 00000000..140ea3ea --- /dev/null +++ b/kal/posix/musl_src/stdlib/atol.c @@ -0,0 +1,17 @@ +#include +#include + +long atol(const char *s) +{ + long n=0; + int neg=0; + while (isspace(*s)) s++; + switch (*s) { + case '-': neg=1; + case '+': s++; + } + /* Compute n as a negative number to avoid overflow on LONG_MIN */ + while (isdigit(*s)) + n = 10*n - (*s++ - '0'); + return neg ? n : -n; +} diff --git a/kal/posix/musl_src/stdlib/atoll.c b/kal/posix/musl_src/stdlib/atoll.c new file mode 100644 index 00000000..b6930489 --- /dev/null +++ b/kal/posix/musl_src/stdlib/atoll.c @@ -0,0 +1,17 @@ +#include +#include + +long long atoll(const char *s) +{ + long long n=0; + int neg=0; + while (isspace(*s)) s++; + switch (*s) { + case '-': neg=1; + case '+': s++; + } + /* Compute n as a negative number to avoid overflow on LLONG_MIN */ + while (isdigit(*s)) + n = 10*n - (*s++ - '0'); + return neg ? n : -n; +} diff --git a/kal/posix/musl_src/stdlib/strtol.c b/kal/posix/musl_src/stdlib/strtol.c new file mode 100644 index 00000000..4a3f7942 --- /dev/null +++ b/kal/posix/musl_src/stdlib/strtol.c @@ -0,0 +1,38 @@ +#include "stdio_impl.h" +#include "intscan.h" +#include "shgetc.h" +#include +#include +#include + +static unsigned long long strtox(const char *s, char **p, int base, unsigned long long lim) +{ + FILE f; + sh_fromstring(&f, s); + shlim(&f, 0); + unsigned long long y = __intscan(&f, base, 1, lim); + if (p) { + size_t cnt = shcnt(&f); + *p = (char *)s + cnt; + } + return y; +} + +unsigned long long strtoull(const char *restrict s, char **restrict p, int base) +{ + return strtox(s, p, base, ULLONG_MAX); +} + +unsigned long strtoul(const char *restrict s, char **restrict p, int base) +{ + return strtox(s, p, base, ULONG_MAX); +} + +long strtol(const char *restrict s, char **restrict p, int base) +{ + return strtox(s, p, base, 0UL+LONG_MIN); +} + +weak_alias(strtol, __strtol_internal); +weak_alias(strtoul, __strtoul_internal); +weak_alias(strtoull, __strtoull_internal); diff --git a/kal/posix/musl_src/string/memchr.c b/kal/posix/musl_src/string/memchr.c new file mode 100644 index 00000000..65f0d789 --- /dev/null +++ b/kal/posix/musl_src/string/memchr.c @@ -0,0 +1,27 @@ +#include +#include +#include + +#define SS (sizeof(size_t)) +#define ALIGN (sizeof(size_t)-1) +#define ONES ((size_t)-1/UCHAR_MAX) +#define HIGHS (ONES * (UCHAR_MAX/2+1)) +#define HASZERO(x) ((x)-ONES & ~(x) & HIGHS) + +void *memchr(const void *src, int c, size_t n) +{ + const unsigned char *s = src; + c = (unsigned char)c; +#ifdef __GNUC__ + for (; ((uintptr_t)s & ALIGN) && n && *s != c; s++, n--); + if (n && *s != c) { + typedef size_t __attribute__((__may_alias__)) word; + const word *w; + size_t k = ONES * c; + for (w = (const void *)s; n>=SS && !HASZERO(*w^k); w++, n-=SS); + s = (const void *)w; + } +#endif + for (; n && *s != c; s++, n--); + return n ? (void *)s : 0; +} diff --git a/kal/posix/musl_src/string/memcmp.c b/kal/posix/musl_src/string/memcmp.c new file mode 100644 index 00000000..bdbce9f0 --- /dev/null +++ b/kal/posix/musl_src/string/memcmp.c @@ -0,0 +1,8 @@ +#include + +int memcmp(const void *vl, const void *vr, size_t n) +{ + const unsigned char *l=vl, *r=vr; + for (; n && *l == *r; n--, l++, r++); + return n ? *l-*r : 0; +} diff --git a/kal/posix/musl_src/string/memcpy.c b/kal/posix/musl_src/string/memcpy.c new file mode 100644 index 00000000..06e88742 --- /dev/null +++ b/kal/posix/musl_src/string/memcpy.c @@ -0,0 +1,124 @@ +#include +#include +#include + +void *memcpy(void *restrict dest, const void *restrict src, size_t n) +{ + unsigned char *d = dest; + const unsigned char *s = src; + +#ifdef __GNUC__ + +#if __BYTE_ORDER == __LITTLE_ENDIAN +#define LS >> +#define RS << +#else +#define LS << +#define RS >> +#endif + + typedef uint32_t __attribute__((__may_alias__)) u32; + uint32_t w, x; + + for (; (uintptr_t)s % 4 && n; n--) *d++ = *s++; + + if ((uintptr_t)d % 4 == 0) { + for (; n>=16; s+=16, d+=16, n-=16) { + *(u32 *)(d+0) = *(u32 *)(s+0); + *(u32 *)(d+4) = *(u32 *)(s+4); + *(u32 *)(d+8) = *(u32 *)(s+8); + *(u32 *)(d+12) = *(u32 *)(s+12); + } + if (n&8) { + *(u32 *)(d+0) = *(u32 *)(s+0); + *(u32 *)(d+4) = *(u32 *)(s+4); + d += 8; s += 8; + } + if (n&4) { + *(u32 *)(d+0) = *(u32 *)(s+0); + d += 4; s += 4; + } + if (n&2) { + *d++ = *s++; *d++ = *s++; + } + if (n&1) { + *d = *s; + } + return dest; + } + + if (n >= 32) switch ((uintptr_t)d % 4) { + case 1: + w = *(u32 *)s; + *d++ = *s++; + *d++ = *s++; + *d++ = *s++; + n -= 3; + for (; n>=17; s+=16, d+=16, n-=16) { + x = *(u32 *)(s+1); + *(u32 *)(d+0) = (w LS 24) | (x RS 8); + w = *(u32 *)(s+5); + *(u32 *)(d+4) = (x LS 24) | (w RS 8); + x = *(u32 *)(s+9); + *(u32 *)(d+8) = (w LS 24) | (x RS 8); + w = *(u32 *)(s+13); + *(u32 *)(d+12) = (x LS 24) | (w RS 8); + } + break; + case 2: + w = *(u32 *)s; + *d++ = *s++; + *d++ = *s++; + n -= 2; + for (; n>=18; s+=16, d+=16, n-=16) { + x = *(u32 *)(s+2); + *(u32 *)(d+0) = (w LS 16) | (x RS 16); + w = *(u32 *)(s+6); + *(u32 *)(d+4) = (x LS 16) | (w RS 16); + x = *(u32 *)(s+10); + *(u32 *)(d+8) = (w LS 16) | (x RS 16); + w = *(u32 *)(s+14); + *(u32 *)(d+12) = (x LS 16) | (w RS 16); + } + break; + case 3: + w = *(u32 *)s; + *d++ = *s++; + n -= 1; + for (; n>=19; s+=16, d+=16, n-=16) { + x = *(u32 *)(s+3); + *(u32 *)(d+0) = (w LS 8) | (x RS 24); + w = *(u32 *)(s+7); + *(u32 *)(d+4) = (x LS 8) | (w RS 24); + x = *(u32 *)(s+11); + *(u32 *)(d+8) = (w LS 8) | (x RS 24); + w = *(u32 *)(s+15); + *(u32 *)(d+12) = (x LS 8) | (w RS 24); + } + break; + } + if (n&16) { + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + } + if (n&8) { + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + } + if (n&4) { + *d++ = *s++; *d++ = *s++; *d++ = *s++; *d++ = *s++; + } + if (n&2) { + *d++ = *s++; *d++ = *s++; + } + if (n&1) { + *d = *s; + } + return dest; +#endif + + for (; n; n--) *d++ = *s++; + return dest; +} diff --git a/kal/posix/musl_src/string/memrchr.c b/kal/posix/musl_src/string/memrchr.c new file mode 100644 index 00000000..d62bb625 --- /dev/null +++ b/kal/posix/musl_src/string/memrchr.c @@ -0,0 +1,10 @@ +#include + +void *__memrchr(const void *m, int c, size_t n) +{ + const unsigned char *s = m; + c = (unsigned char)c; + while (n--) if (s[n]==c) return (void *)(s+n); + return 0; +} + diff --git a/kal/posix/musl_src/string/memset.c b/kal/posix/musl_src/string/memset.c new file mode 100644 index 00000000..5613a148 --- /dev/null +++ b/kal/posix/musl_src/string/memset.c @@ -0,0 +1,90 @@ +#include +#include + +void *memset(void *dest, int c, size_t n) +{ + unsigned char *s = dest; + size_t k; + + /* Fill head and tail with minimal branching. Each + * conditional ensures that all the subsequently used + * offsets are well-defined and in the dest region. */ + + if (!n) return dest; + s[0] = c; + s[n-1] = c; + if (n <= 2) return dest; + s[1] = c; + s[2] = c; + s[n-2] = c; + s[n-3] = c; + if (n <= 6) return dest; + s[3] = c; + s[n-4] = c; + if (n <= 8) return dest; + + /* Advance pointer to align it at a 4-byte boundary, + * and truncate n to a multiple of 4. The previous code + * already took care of any head/tail that get cut off + * by the alignment. */ + + k = -(uintptr_t)s & 3; + s += k; + n -= k; + n &= -4; + +#ifdef __GNUC__ + typedef uint32_t __attribute__((__may_alias__)) u32; + typedef uint64_t __attribute__((__may_alias__)) u64; + + u32 c32 = ((u32)-1)/255 * (unsigned char)c; + + /* In preparation to copy 32 bytes at a time, aligned on + * an 8-byte bounary, fill head/tail up to 28 bytes each. + * As in the initial byte-based head/tail fill, each + * conditional below ensures that the subsequent offsets + * are valid (e.g. !(n<=24) implies n>=28). */ + + *(u32 *)(s+0) = c32; + *(u32 *)(s+n-4) = c32; + if (n <= 8) return dest; + *(u32 *)(s+4) = c32; + *(u32 *)(s+8) = c32; + *(u32 *)(s+n-12) = c32; + *(u32 *)(s+n-8) = c32; + if (n <= 24) return dest; + *(u32 *)(s+12) = c32; + *(u32 *)(s+16) = c32; + *(u32 *)(s+20) = c32; + *(u32 *)(s+24) = c32; + *(u32 *)(s+n-28) = c32; + *(u32 *)(s+n-24) = c32; + *(u32 *)(s+n-20) = c32; + *(u32 *)(s+n-16) = c32; + + /* Align to a multiple of 8 so we can fill 64 bits at a time, + * and avoid writing the same bytes twice as much as is + * practical without introducing additional branching. */ + + k = 24 + ((uintptr_t)s & 4); + s += k; + n -= k; + + /* If this loop is reached, 28 tail bytes have already been + * filled, so any remainder when n drops below 32 can be + * safely ignored. */ + + u64 c64 = c32 | ((u64)c32 << 32); + for (; n >= 32; n-=32, s+=32) { + *(u64 *)(s+0) = c64; + *(u64 *)(s+8) = c64; + *(u64 *)(s+16) = c64; + *(u64 *)(s+24) = c64; + } +#else + /* Pure C fallback with no aliasing violations. */ + for (; n; n--, s++) *s = c; +#endif + + return dest; +} diff --git a/kal/posix/musl_src/string/strcasecmp.c b/kal/posix/musl_src/string/strcasecmp.c new file mode 100644 index 00000000..fe34ad5c --- /dev/null +++ b/kal/posix/musl_src/string/strcasecmp.c @@ -0,0 +1,16 @@ +#include +#include + +int strcasecmp(const char *_l, const char *_r) +{ + const unsigned char *l=(void *)_l, *r=(void *)_r; + for (; *l && *r && (*l == *r || tolower(*l) == tolower(*r)); l++, r++); + return tolower(*l) - tolower(*r); +} + +int __strcasecmp_l(const char *l, const char *r, locale_t loc) +{ + return strcasecmp(l, r); +} + +weak_alias(__strcasecmp_l, strcasecmp_l); \ No newline at end of file diff --git a/kal/posix/musl_src/string/strchr.c b/kal/posix/musl_src/string/strchr.c new file mode 100644 index 00000000..3cbc828b --- /dev/null +++ b/kal/posix/musl_src/string/strchr.c @@ -0,0 +1,7 @@ +#include + +char *strchr(const char *s, int c) +{ + char *r = __strchrnul(s, c); + return *(unsigned char *)r == (unsigned char)c ? r : 0; +} diff --git a/kal/posix/musl_src/string/strchrnul.c b/kal/posix/musl_src/string/strchrnul.c new file mode 100644 index 00000000..39e2635b --- /dev/null +++ b/kal/posix/musl_src/string/strchrnul.c @@ -0,0 +1,28 @@ +#include +#include +#include + +#define ALIGN (sizeof(size_t)) +#define ONES ((size_t)-1/UCHAR_MAX) +#define HIGHS (ONES * (UCHAR_MAX/2+1)) +#define HASZERO(x) ((x)-ONES & ~(x) & HIGHS) + +char *__strchrnul(const char *s, int c) +{ + c = (unsigned char)c; + if (!c) return (char *)s + strlen(s); + +#ifdef __GNUC__ + typedef size_t __attribute__((__may_alias__)) word; + const word *w; + for (; (uintptr_t)s % ALIGN; s++) + if (!*s || *(unsigned char *)s == c) return (char *)s; + size_t k = ONES * c; + for (w = (void *)s; !HASZERO(*w) && !HASZERO(*w^k); w++); + s = (void *)w; +#endif + for (; *s && *(unsigned char *)s != c; s++); + return (char *)s; +} + +weak_alias(__strchrnul, strchrnul); diff --git a/kal/posix/musl_src/string/strcmp.c b/kal/posix/musl_src/string/strcmp.c new file mode 100644 index 00000000..808bd837 --- /dev/null +++ b/kal/posix/musl_src/string/strcmp.c @@ -0,0 +1,7 @@ +#include + +int strcmp(const char *l, const char *r) +{ + for (; *l==*r && *l; l++, r++); + return *(unsigned char *)l - *(unsigned char *)r; +} diff --git a/kal/posix/musl_src/string/strcspn.c b/kal/posix/musl_src/string/strcspn.c new file mode 100644 index 00000000..75e262b2 --- /dev/null +++ b/kal/posix/musl_src/string/strcspn.c @@ -0,0 +1,17 @@ +#include "../include/string.h" + +#define BITOP(a,b,op) \ + ((a)[(size_t)(b)/(8*sizeof *(a))] op (size_t)1<<((size_t)(b)%(8*sizeof *(a)))) + +size_t strcspn(const char *s, const char *c) +{ + const char *a = s; + size_t byteset[32/sizeof(size_t)]; + + if (!c[0] || !c[1]) return __strchrnul(s, *c)-a; + + memset(byteset, 0, sizeof byteset); + for (; *c && BITOP(byteset, *(unsigned char *)c, |=); c++); + for (; *s && !BITOP(byteset, *(unsigned char *)s, &); s++); + return s-a; +} diff --git a/kal/posix/musl_src/string/strdup.c b/kal/posix/musl_src/string/strdup.c new file mode 100644 index 00000000..d4c27449 --- /dev/null +++ b/kal/posix/musl_src/string/strdup.c @@ -0,0 +1,10 @@ +#include +#include + +char *strdup(const char *s) +{ + size_t l = strlen(s); + char *d = malloc(l+1); + if (!d) return NULL; + return memcpy(d, s, l+1); +} diff --git a/kal/posix/musl_src/string/strlen.c b/kal/posix/musl_src/string/strlen.c new file mode 100644 index 00000000..309990f0 --- /dev/null +++ b/kal/posix/musl_src/string/strlen.c @@ -0,0 +1,22 @@ +#include +#include +#include + +#define ALIGN (sizeof(size_t)) +#define ONES ((size_t)-1/UCHAR_MAX) +#define HIGHS (ONES * (UCHAR_MAX/2+1)) +#define HASZERO(x) ((x)-ONES & ~(x) & HIGHS) + +size_t strlen(const char *s) +{ + const char *a = s; +#ifdef __GNUC__ + typedef size_t __attribute__((__may_alias__)) word; + const word *w; + for (; (uintptr_t)s % ALIGN; s++) if (!*s) return s-a; + for (w = (const void *)s; !HASZERO(*w); w++); + s = (const void *)w; +#endif + for (; *s; s++); + return s-a; +} diff --git a/kal/posix/musl_src/string/strncasecmp.c b/kal/posix/musl_src/string/strncasecmp.c new file mode 100644 index 00000000..24659721 --- /dev/null +++ b/kal/posix/musl_src/string/strncasecmp.c @@ -0,0 +1,10 @@ +#include +#include + +int strncasecmp(const char *_l, const char *_r, size_t n) +{ + const unsigned char *l=(void *)_l, *r=(void *)_r; + if (!n--) return 0; + for (; *l && *r && n && (*l == *r || tolower(*l) == tolower(*r)); l++, r++, n--); + return tolower(*l) - tolower(*r); +} diff --git a/kal/posix/musl_src/string/strncmp.c b/kal/posix/musl_src/string/strncmp.c new file mode 100644 index 00000000..e228843f --- /dev/null +++ b/kal/posix/musl_src/string/strncmp.c @@ -0,0 +1,9 @@ +#include + +int strncmp(const char *_l, const char *_r, size_t n) +{ + const unsigned char *l=(void *)_l, *r=(void *)_r; + if (!n--) return 0; + for (; *l && *r && n && *l == *r ; l++, r++, n--); + return *l - *r; +} diff --git a/kal/posix/musl_src/string/strrchr.c b/kal/posix/musl_src/string/strrchr.c new file mode 100644 index 00000000..98ad1b04 --- /dev/null +++ b/kal/posix/musl_src/string/strrchr.c @@ -0,0 +1,6 @@ +#include + +char *strrchr(const char *s, int c) +{ + return __memrchr(s, c, strlen(s) + 1); +} diff --git a/kal/posix/musl_src/string/strstr.c b/kal/posix/musl_src/string/strstr.c new file mode 100644 index 00000000..55ba1c7b --- /dev/null +++ b/kal/posix/musl_src/string/strstr.c @@ -0,0 +1,154 @@ +#include +#include + +static char *twobyte_strstr(const unsigned char *h, const unsigned char *n) +{ + uint16_t nw = n[0]<<8 | n[1], hw = h[0]<<8 | h[1]; + for (h++; *h && hw != nw; hw = hw<<8 | *++h); + return *h ? (char *)h-1 : 0; +} + +static char *threebyte_strstr(const unsigned char *h, const unsigned char *n) +{ + uint32_t nw = n[0]<<24 | n[1]<<16 | n[2]<<8; + uint32_t hw = h[0]<<24 | h[1]<<16 | h[2]<<8; + for (h+=2; *h && hw != nw; hw = (hw|*++h)<<8); + return *h ? (char *)h-2 : 0; +} + +static char *fourbyte_strstr(const unsigned char *h, const unsigned char *n) +{ + uint32_t nw = n[0]<<24 | n[1]<<16 | n[2]<<8 | n[3]; + uint32_t hw = h[0]<<24 | h[1]<<16 | h[2]<<8 | h[3]; + for (h+=3; *h && hw != nw; hw = hw<<8 | *++h); + return *h ? (char *)h-3 : 0; +} + +#define MAX(a,b) ((a)>(b)?(a):(b)) +#define MIN(a,b) ((a)<(b)?(a):(b)) + +#define BITOP(a,b,op) \ + ((a)[(size_t)(b)/(8*sizeof *(a))] op (size_t)1<<((size_t)(b)%(8*sizeof *(a)))) + +static char *twoway_strstr(const unsigned char *h, const unsigned char *n) +{ + const unsigned char *z; + size_t l, ip, jp, k, p, ms, p0, mem, mem0; + size_t byteset[32 / sizeof(size_t)] = { 0 }; + size_t shift[256]; + + /* Computing length of needle and fill shift table */ + for (l=0; n[l] && h[l]; l++) + BITOP(byteset, n[l], |=), shift[n[l]] = l+1; + if (n[l]) return 0; /* hit the end of h */ + + /* Compute maximal suffix */ + ip = -1; jp = 0; k = p = 1; + while (jp+k n[jp+k]) { + jp += k; + k = 1; + p = jp - ip; + } else { + ip = jp++; + k = p = 1; + } + } + ms = ip; + p0 = p; + + /* And with the opposite comparison */ + ip = -1; jp = 0; k = p = 1; + while (jp+k ms+1) ms = ip; + else p = p0; + + /* Periodic needle? */ + if (memcmp(n, n+p, ms+1)) { + mem0 = 0; + p = MAX(ms, l-ms-1) + 1; + } else mem0 = l-p; + mem = 0; + + /* Initialize incremental end-of-haystack pointer */ + z = h; + + /* Search loop */ + for (;;) { + /* Update incremental end-of-haystack pointer */ + if (z-h < l) { + /* Fast estimate for MIN(l,63) */ + size_t grow = l | 63; + const unsigned char *z2 = memchr(z, 0, grow); + if (z2) { + z = z2; + if (z-h < l) return 0; + } else z += grow; + } + + /* Check last byte first; advance by shift on mismatch */ + if (BITOP(byteset, h[l-1], &)) { + k = l-shift[h[l-1]]; + if (k) { + if (k < mem) k = mem; + h += k; + mem = 0; + continue; + } + } else { + h += l; + mem = 0; + continue; + } + + /* Compare right half */ + for (k=MAX(ms+1,mem); n[k] && n[k] == h[k]; k++); + if (n[k]) { + h += k-ms; + mem = 0; + continue; + } + /* Compare left half */ + for (k=ms+1; k>mem && n[k-1] == h[k-1]; k--); + if (k <= mem) return (char *)h; + h += p; + mem = mem0; + } +} + +char *strstr(const char *h, const char *n) +{ + /* Return immediately on empty needle */ + if (!n[0]) return (char *)h; + + /* Use faster algorithms for short needles */ + h = strchr(h, *n); + if (!h || !n[1]) return (char *)h; + if (!h[1]) return 0; + if (!n[2]) return twobyte_strstr((void *)h, (void *)n); + if (!h[2]) return 0; + if (!n[3]) return threebyte_strstr((void *)h, (void *)n); + if (!h[3]) return 0; + if (!n[4]) return fourbyte_strstr((void *)h, (void *)n); + + return twoway_strstr((void *)h, (void *)n); +} diff --git a/kal/posix/musl_src/string/wcschr.c b/kal/posix/musl_src/string/wcschr.c new file mode 100644 index 00000000..8dfc2f31 --- /dev/null +++ b/kal/posix/musl_src/string/wcschr.c @@ -0,0 +1,8 @@ +#include + +wchar_t *wcschr(const wchar_t *s, wchar_t c) +{ + if (!c) return (wchar_t *)s + wcslen(s); + for (; *s && *s != c; s++); + return *s ? (wchar_t *)s : 0; +} diff --git a/kal/posix/musl_src/string/wcslen.c b/kal/posix/musl_src/string/wcslen.c new file mode 100644 index 00000000..1b7b6655 --- /dev/null +++ b/kal/posix/musl_src/string/wcslen.c @@ -0,0 +1,8 @@ +#include + +size_t wcslen(const wchar_t *s) +{ + const wchar_t *a; + for (a=s; *s; s++); + return s-a; +} diff --git a/kal/posix/musl_src/time/strftime.c b/kal/posix/musl_src/time/strftime.c new file mode 100644 index 00000000..c8fd32d2 --- /dev/null +++ b/kal/posix/musl_src/time/strftime.c @@ -0,0 +1,281 @@ +#include +#include +#include +#include +#include +#include +#include +#include "locale_impl.h" +#include "time_impl.h" + +#define CURRENT_LOCALE NULL + +static int is_leap(int y) +{ + /* Avoid overflow */ + if (y>INT_MAX-1900) y -= 2000; + y += 1900; + return !(y%4) && ((y%100) || !(y%400)); +} + +static int week_num(const struct tm *tm) +{ + int val = (tm->tm_yday + 7U - (tm->tm_wday+6U)%7) / 7; + /* If 1 Jan is just 1-3 days past Monday, + * the previous week is also in this year. */ + if ((tm->tm_wday + 371U - tm->tm_yday - 2) % 7 <= 2) + val++; + if (!val) { + val = 52; + /* If 31 December of prev year a Thursday, + * or Friday of a leap year, then the + * prev year has 53 weeks. */ + int dec31 = (tm->tm_wday + 7U - tm->tm_yday - 1) % 7; + if (dec31 == 4 || (dec31 == 5 && is_leap(tm->tm_year%400-1))) + val++; + } else if (val == 53) { + /* If 1 January is not a Thursday, and not + * a Wednesday of a leap year, then this + * year has only 52 weeks. */ + int jan1 = (tm->tm_wday + 371U - tm->tm_yday) % 7; + if (jan1 != 4 && (jan1 != 3 || !is_leap(tm->tm_year))) + val = 1; + } + return val; +} + +const char *__strftime_fmt_1(char (*s)[100], size_t *l, int f, const struct tm *tm, locale_t loc, int pad) +{ + nl_item item; + long long val; + const char *fmt = "-"; + int width = 2, def_pad = '0'; + + switch (f) { + case 'a': + if (tm->tm_wday > 6U) goto string; + item = ABDAY_1 + tm->tm_wday; + goto nl_strcat; + case 'A': + if (tm->tm_wday > 6U) goto string; + item = DAY_1 + tm->tm_wday; + goto nl_strcat; + case 'h': + case 'b': + if (tm->tm_mon > 11U) goto string; + item = ABMON_1 + tm->tm_mon; + goto nl_strcat; + case 'B': + if (tm->tm_mon > 11U) goto string; + item = MON_1 + tm->tm_mon; + goto nl_strcat; + case 'c': + item = D_T_FMT; + goto nl_strftime; + case 'C': + val = (1900LL+tm->tm_year) / 100; + goto number; + case 'e': + def_pad = '_'; + case 'd': + val = tm->tm_mday; + goto number; + case 'D': + fmt = "%m/%d/%y"; + goto recu_strftime; + case 'F': + fmt = "%Y-%m-%d"; + goto recu_strftime; + case 'g': + case 'G': + val = tm->tm_year + 1900LL; + if (tm->tm_yday < 3 && week_num(tm) != 1) val--; + else if (tm->tm_yday > 360 && week_num(tm) == 1) val++; + if (f=='g') val %= 100; + else width = 4; + goto number; + case 'H': + val = tm->tm_hour; + goto number; + case 'I': + val = tm->tm_hour; + if (!val) val = 12; + else if (val > 12) val -= 12; + goto number; + case 'j': + val = tm->tm_yday+1; + width = 3; + goto number; + case 'm': + val = tm->tm_mon+1; + goto number; + case 'M': + val = tm->tm_min; + goto number; + case 'n': + *l = 1; + return "\n"; + case 'p': + item = tm->tm_hour >= 12 ? PM_STR : AM_STR; + goto nl_strcat; + case 'r': + item = T_FMT_AMPM; + goto nl_strftime; + case 'R': + fmt = "%H:%M"; + goto recu_strftime; + case 's': + val = __tm_to_secs(tm) - tm->__tm_gmtoff; + width = 1; + goto number; + case 'S': + val = tm->tm_sec; + goto number; + case 't': + *l = 1; + return "\t"; + case 'T': + fmt = "%H:%M:%S"; + goto recu_strftime; + case 'u': + val = tm->tm_wday ? tm->tm_wday : 7; + width = 1; + goto number; + case 'U': + val = (tm->tm_yday + 7U - tm->tm_wday) / 7; + goto number; + case 'W': + val = (tm->tm_yday + 7U - (tm->tm_wday+6U)%7) / 7; + goto number; + case 'V': + val = week_num(tm); + goto number; + case 'w': + val = tm->tm_wday; + width = 1; + goto number; + case 'x': + item = D_FMT; + goto nl_strftime; + case 'X': + item = T_FMT; + goto nl_strftime; + case 'y': + val = (tm->tm_year + 1900LL) % 100; + if (val < 0) val = -val; + goto number; + case 'Y': + val = tm->tm_year + 1900LL; + if (val >= 10000) { + *l = snprintf(*s, sizeof *s, "+%lld", val); + return *s; + } + width = 4; + goto number; + case 'z': + if (tm->tm_isdst < 0) { + *l = 0; + return ""; + } + *l = snprintf(*s, sizeof *s, "%+.4ld", + tm->__tm_gmtoff/3600*100 + tm->__tm_gmtoff%3600/60); + return *s; + case 'Z': + // unsupport tz + return 0; + case '%': + *l = 1; + return "%"; + default: + return 0; + } +number: + switch (pad ? pad : def_pad) { + case '-': *l = snprintf(*s, sizeof *s, "%lld", val); break; + case '_': *l = snprintf(*s, sizeof *s, "%*lld", width, val); break; + case '0': + default: *l = snprintf(*s, sizeof *s, "%0*lld", width, val); break; + } + return *s; +nl_strcat: + // unsuport loc + return 0; +string: + *l = strlen(fmt); + return fmt; +nl_strftime: + // unsuport loc + return 0; +recu_strftime: + *l = __strftime_l(*s, sizeof *s, fmt, tm, loc); + if (!*l) return 0; + return *s; +} + +size_t __strftime_l(char *restrict s, size_t n, const char *restrict f, const struct tm *restrict tm, locale_t loc) +{ + size_t l, k; + char buf[100]; + char *p; + const char *t; + int pad, plus; + unsigned long width; + for (l=0; ltm_year < -1900) { + s[l++] = '-'; + width--; + } else if (plus && d+(width-k) >= (*p=='C'?3:5)) { + s[l++] = '+'; + width--; + } + for (; width > k && l < n; width--) + s[l++] = '0'; + } + if (k > n-l) k = n-l; + memcpy(s+l, t, k); + l += k; + } + if (n) { + if (l==n) l=n-1; + s[l] = 0; + } + return 0; +} + +size_t strftime(char *restrict s, size_t n, const char *restrict f, const struct tm *restrict tm) +{ + return __strftime_l(s, n, f, tm, NULL); +} + +weak_alias(__strftime_l, strftime_l); diff --git a/kal/posix/musl_src/time/strptime.c b/kal/posix/musl_src/time/strptime.c new file mode 100644 index 00000000..d4b8f253 --- /dev/null +++ b/kal/posix/musl_src/time/strptime.c @@ -0,0 +1,206 @@ +#include +#include +#include +#include +#include +#include +#include + +char *strptime(const char *restrict s, const char *restrict f, struct tm *restrict tm) +{ + int i, w, neg, adj, min, range, *dest, dummy; + const char *ex; + size_t len; + int want_century = 0, century = 0, relyear = 0; + while (*f) { + if (*f != '%') { + if (isspace(*f)) for (; *s && isspace(*s); s++); + else if (*s != *f) return 0; + else s++; + f++; + continue; + } + f++; + if (*f == '+') f++; + if (isdigit(*f)) { + char *new_f; + w=strtoul(f, &new_f, 10); + f = new_f; + } else { + w=-1; + } + adj=0; + switch (*f++) { + case 'a': case 'A': + dest = &tm->tm_wday; + min = ABDAY_1; + range = 7; + goto symbolic_range; + case 'b': case 'B': case 'h': + dest = &tm->tm_mon; + min = ABMON_1; + range = 12; + goto symbolic_range; + case 'c': + s = strptime(s, nl_langinfo(D_T_FMT), tm); + if (!s) return 0; + break; + case 'C': + dest = ¢ury; + if (w<0) w=2; + want_century |= 2; + goto numeric_digits; + case 'd': case 'e': + dest = &tm->tm_mday; + min = 1; + range = 31; + goto numeric_range; + case 'D': + s = strptime(s, "%m/%d/%y", tm); + if (!s) return 0; + break; + case 'H': + dest = &tm->tm_hour; + min = 0; + range = 24; + goto numeric_range; + case 'I': + dest = &tm->tm_hour; + min = 1; + range = 12; + goto numeric_range; + case 'j': + dest = &tm->tm_yday; + min = 1; + range = 366; + adj = 1; + goto numeric_range; + case 'm': + dest = &tm->tm_mon; + min = 1; + range = 12; + adj = 1; + goto numeric_range; + case 'M': + dest = &tm->tm_min; + min = 0; + range = 60; + goto numeric_range; + case 'n': case 't': + for (; *s && isspace(*s); s++); + break; + case 'p': + ex = nl_langinfo(AM_STR); + len = strlen(ex); + if (!strncasecmp(s, ex, len)) { + tm->tm_hour %= 12; + s += len; + break; + } + ex = nl_langinfo(PM_STR); + len = strlen(ex); + if (!strncasecmp(s, ex, len)) { + tm->tm_hour %= 12; + tm->tm_hour += 12; + s += len; + break; + } + return 0; + case 'r': + s = strptime(s, nl_langinfo(T_FMT_AMPM), tm); + if (!s) return 0; + break; + case 'R': + s = strptime(s, "%H:%M", tm); + if (!s) return 0; + break; + case 'S': + dest = &tm->tm_sec; + min = 0; + range = 61; + goto numeric_range; + case 'T': + s = strptime(s, "%H:%M:%S", tm); + if (!s) return 0; + break; + case 'U': + case 'W': + /* Throw away result, for now. (FIXME?) */ + dest = &dummy; + min = 0; + range = 54; + goto numeric_range; + case 'w': + dest = &tm->tm_wday; + min = 0; + range = 7; + goto numeric_range; + case 'x': + s = strptime(s, nl_langinfo(D_FMT), tm); + if (!s) return 0; + break; + case 'X': + s = strptime(s, nl_langinfo(T_FMT), tm); + if (!s) return 0; + break; + case 'y': + dest = &relyear; + w = 2; + want_century |= 1; + goto numeric_digits; + case 'Y': + dest = &tm->tm_year; + if (w<0) w=4; + adj = 1900; + want_century = 0; + goto numeric_digits; + case '%': + if (*s++ != '%') return 0; + break; + default: + return 0; + numeric_range: + if (!isdigit(*s)) return 0; + *dest = 0; + for (i=1; i<=min+range && isdigit(*s); i*=10) + *dest = *dest * 10 + *s++ - '0'; + if (*dest - min >= (int)range) return 0; + *dest -= adj; + switch((char *)dest - (char *)tm) { + case offsetof(struct tm, tm_yday): + ; + } + goto update; + numeric_digits: + neg = 0; + if (*s == '+') s++; + else if (*s == '-') neg=1, s++; + if (!isdigit(*s)) return 0; + for (*dest=i=0; i=0; i--) { + ex = nl_langinfo(min+i); + len = strlen(ex); + if (strncasecmp(s, ex, len)) continue; + s += len; + *dest = i % range; + break; + } + if (i<0) return 0; + goto update; + update: + //FIXME + ; + } + } + if (want_century) { + tm->tm_year = relyear; + if (want_century & 2) tm->tm_year += century * 100 - 1900; + else if (tm->tm_year <= 68) tm->tm_year += 100; + } + return (char *)s; +} diff --git a/kal/posix/musl_src/time/time_impl.h b/kal/posix/musl_src/time/time_impl.h new file mode 100644 index 00000000..c3c136d8 --- /dev/null +++ b/kal/posix/musl_src/time/time_impl.h @@ -0,0 +1,14 @@ +#include +#include "features.h" + +hidden int __days_in_month(int, int); +hidden int __month_to_secs(int, int); +hidden long long __year_to_secs(long long, int *); +hidden long long __tm_to_secs(const struct tm *); +hidden const char *__tm_to_tzname(const struct tm *); +hidden int __secs_to_tm(long long, struct tm *); +hidden void __secs_to_zone(long long, int, int *, long *, long *, const char **); +hidden const char *__strftime_fmt_1(char (*)[100], size_t *, int, const struct tm *, locale_t, int); +extern hidden const char __utc[]; + +hidden size_t __strftime_l(char *restrict, size_t, const char *restrict, const struct tm *restrict, locale_t); diff --git a/kal/posix/src/mqueue.c b/kal/posix/src/mqueue.c new file mode 100644 index 00000000..48b6667f --- /dev/null +++ b/kal/posix/src/mqueue.c @@ -0,0 +1,606 @@ +/* + * Copyright (c) 2013-2019 Huawei Technologies Co., Ltd. All rights reserved. + * Copyright (c) 2020-2021 Huawei Device Co., Ltd. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without modification, + * are permitted provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright notice, this list + * of conditions and the following disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 3. Neither the name of the copyright holder nor the names of its contributors may be used + * to endorse or promote products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; + * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, + * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR + * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF + * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "mqueue_impl.h" +#include "time_internal.h" + +/* GLOBALS */ +STATIC struct mqarray g_queueTable[LOSCFG_BASE_IPC_QUEUE_LIMIT]; +STATIC pthread_mutex_t g_mqueueMutex = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP; + +int MapMqErrno(UINT32 err) +{ + if (err == LOS_OK) { + return ENOERR; + } + switch (err) { + case LOS_ERRNO_QUEUE_INVALID: + case LOS_ERRNO_QUEUE_WRITE_PTR_NULL: + case LOS_ERRNO_QUEUE_WRITESIZE_ISZERO: + case LOS_ERRNO_QUEUE_SIZE_TOO_BIG: + case LOS_ERRNO_QUEUE_CREAT_PTR_NULL: + case LOS_ERRNO_QUEUE_PARA_ISZERO: + case LOS_ERRNO_QUEUE_WRITE_SIZE_TOO_BIG: + errno = EINVAL; + break; + case LOS_ERRNO_QUEUE_ISFULL: + case LOS_ERRNO_QUEUE_ISEMPTY: + errno = EAGAIN; + break; + case LOS_ERRNO_QUEUE_CREATE_NO_MEMORY: + errno = ENOSPC; + break; + case LOS_ERRNO_QUEUE_TIMEOUT: + errno = ETIMEDOUT; + break; + case LOS_ERRNO_QUEUE_CB_UNAVAILABLE: + errno = ENFILE; + break; + case LOS_ERRNO_QUEUE_READ_IN_INTERRUPT: + case LOS_ERRNO_QUEUE_WRITE_IN_INTERRUPT: + errno = EINTR; + break; + default: + errno = EINVAL; + break; + } + return errno; +} + +STATIC INLINE INT32 MqNameCheck(const CHAR *mqName) +{ + if (mqName == NULL) { + errno = EINVAL; + return -1; + } + + if (strlen(mqName) == 0) { + errno = EINVAL; + return -1; + } + + if (strlen(mqName) > (PATH_MAX - 1)) { + errno = ENAMETOOLONG; + return -1; + } + return 0; +} + +STATIC INLINE UINT32 GetMqueueCBByID(UINT32 queueID, LosQueueCB **queueCB) +{ + LosQueueCB *tmpQueueCB = NULL; + if (queueCB == NULL) { + errno = EINVAL; + return LOS_ERRNO_QUEUE_READ_PTR_NULL; + } + tmpQueueCB = GET_QUEUE_HANDLE(queueID); + if ((GET_QUEUE_INDEX(queueID) >= LOSCFG_BASE_IPC_QUEUE_LIMIT) || (tmpQueueCB->queueID != queueID)) { + return LOS_ERRNO_QUEUE_INVALID; + } + *queueCB = tmpQueueCB; + + return LOS_OK; +} + +STATIC INLINE struct mqarray *GetMqueueCBByName(const CHAR *name) +{ + UINT32 index; + UINT32 mylen = strlen(name); + + for (index = 0; index < LOSCFG_BASE_IPC_QUEUE_LIMIT; index++) { + if ((g_queueTable[index].mq_name == NULL) || (strlen(g_queueTable[index].mq_name) != mylen)) { + continue; + } + + if (strncmp(name, (const CHAR *)(g_queueTable[index].mq_name), mylen) == 0) { + return &(g_queueTable[index]); + } + } + + return NULL; +} + +STATIC INT32 DoMqueueDelete(struct mqarray *mqueueCB) +{ + UINT32 ret; + + if (mqueueCB->mq_name != NULL) { + LOS_MemFree(OS_SYS_MEM_ADDR, mqueueCB->mq_name); + mqueueCB->mq_name = NULL; + } + + mqueueCB->mqcb = NULL; + + ret = LOS_QueueDelete(mqueueCB->mq_id); + switch (ret) { + case LOS_OK: + return 0; + case LOS_ERRNO_QUEUE_NOT_FOUND: + case LOS_ERRNO_QUEUE_NOT_CREATE: + case LOS_ERRNO_QUEUE_IN_TSKUSE: + case LOS_ERRNO_QUEUE_IN_TSKWRITE: + errno = EAGAIN; + return -1; + default: + errno = EINVAL; + return -1; + } +} + +STATIC int SaveMqueueName(const CHAR *mqName, struct mqarray *mqueueCB) +{ + size_t nameLen; + + nameLen = strlen(mqName); /* sys_mq_open has checked name and name length */ + mqueueCB->mq_name = (char *)LOS_MemAlloc(OS_SYS_MEM_ADDR, nameLen + 1); + if (mqueueCB->mq_name == NULL) { + errno = ENOMEM; + return LOS_NOK; + } + + if (strncpy_s(mqueueCB->mq_name, (nameLen + 1), mqName, nameLen) != EOK) { + LOS_MemFree(OS_SYS_MEM_ADDR, mqueueCB->mq_name); + mqueueCB->mq_name = NULL; + errno = EINVAL; + return LOS_NOK; + } + mqueueCB->mq_name[nameLen] = '\0'; + return LOS_OK; +} + +STATIC struct mqpersonal *DoMqueueCreate(const struct mq_attr *attr, const CHAR *mqName, INT32 openFlag) +{ + struct mqarray *mqueueCB = NULL; + UINT32 mqueueID; + + UINT32 err = LOS_QueueCreate(NULL, attr->mq_maxmsg, &mqueueID, 0, attr->mq_msgsize); + if (MapMqErrno(err) != ENOERR) { + goto ERROUT; + } + + if (g_queueTable[GET_QUEUE_INDEX(mqueueID)].mqcb == NULL) { + mqueueCB = &(g_queueTable[GET_QUEUE_INDEX(mqueueID)]); + mqueueCB->mq_id = mqueueID; + } + + if (mqueueCB == NULL) { + errno = EINVAL; + goto ERROUT; + } + + if (SaveMqueueName(mqName, mqueueCB) != LOS_OK) { + goto ERROUT; + } + + if (GetMqueueCBByID(mqueueCB->mq_id, &(mqueueCB->mqcb)) != LOS_OK) { + errno = ENOSPC; + goto ERROUT; + } + + mqueueCB->mq_personal = (struct mqpersonal *)LOS_MemAlloc(OS_SYS_MEM_ADDR, sizeof(struct mqpersonal)); + if (mqueueCB->mq_personal == NULL) { + (VOID)LOS_QueueDelete(mqueueCB->mq_id); + mqueueCB->mqcb->queue = NULL; + mqueueCB->mqcb = NULL; + errno = ENOSPC; + goto ERROUT; + } + + mqueueCB->unlinkflag = FALSE; + mqueueCB->mq_personal->mq_status = MQ_USE_MAGIC; + mqueueCB->mq_personal->mq_next = NULL; + mqueueCB->mq_personal->mq_posixdes = mqueueCB; + mqueueCB->mq_personal->mq_flags = (INT32)((UINT32)openFlag | ((UINT32)attr->mq_flags & (UINT32)FNONBLOCK)); + + return mqueueCB->mq_personal; +ERROUT: + + if ((mqueueCB != NULL) && (mqueueCB->mq_name != NULL)) { + LOS_MemFree(OS_SYS_MEM_ADDR, mqueueCB->mq_name); + mqueueCB->mq_name = NULL; + } + return (struct mqpersonal *)-1; +} + +STATIC struct mqpersonal *DoMqueueOpen(struct mqarray *mqueueCB, INT32 openFlag) +{ + struct mqpersonal *privateMqPersonal = NULL; + + /* already have the same name of g_squeuetable */ + if (mqueueCB->unlinkflag == TRUE) { + errno = EINVAL; + goto ERROUT; + } + /* alloc mqprivate and add to mqarray */ + privateMqPersonal = (struct mqpersonal *)LOS_MemAlloc(OS_SYS_MEM_ADDR, sizeof(struct mqpersonal)); + if (privateMqPersonal == NULL) { + errno = ENOSPC; + goto ERROUT; + } + + privateMqPersonal->mq_next = mqueueCB->mq_personal; + mqueueCB->mq_personal = privateMqPersonal; + + privateMqPersonal->mq_posixdes = mqueueCB; + privateMqPersonal->mq_flags = openFlag; + privateMqPersonal->mq_status = MQ_USE_MAGIC; + + return privateMqPersonal; + +ERROUT: + return (struct mqpersonal *)-1; +} + +mqd_t mq_open(const char *mqName, int openFlag, ...) +{ + struct mqarray *mqueueCB = NULL; + struct mqpersonal *privateMqPersonal = (struct mqpersonal *)-1; + struct mq_attr *attr = NULL; + struct mq_attr defaultAttr = { 0, MQ_MAX_MSG_NUM, MQ_MAX_MSG_LEN, 0, {0} }; + + va_list ap; + + if (MqNameCheck(mqName) == -1) { + return (mqd_t)-1; + } + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + mqueueCB = GetMqueueCBByName(mqName); + if ((UINT32)openFlag & (UINT32)O_CREAT) { + if (mqueueCB != NULL) { + if (((UINT32)openFlag & (UINT32)O_EXCL)) { + errno = EEXIST; + goto OUT; + } + privateMqPersonal = DoMqueueOpen(mqueueCB, openFlag); + } else { + va_start(ap, openFlag); + (VOID)va_arg(ap, int); + attr = va_arg(ap, struct mq_attr *); + va_end(ap); + + if (attr != NULL) { + (VOID)memcpy_s(&defaultAttr, sizeof(struct mq_attr), attr, sizeof(struct mq_attr)); + if ((defaultAttr.mq_maxmsg < 0) || (defaultAttr.mq_maxmsg > (long int)USHRT_MAX) || + (defaultAttr.mq_msgsize < 0) || (defaultAttr.mq_msgsize > (long int)(USHRT_MAX - sizeof(UINT32)))) { + errno = EINVAL; + goto OUT; + } + } + privateMqPersonal = DoMqueueCreate(&defaultAttr, mqName, openFlag); + } + } else { + if (mqueueCB == NULL) { + errno = ENOENT; + goto OUT; + } + privateMqPersonal = DoMqueueOpen(mqueueCB, openFlag); + } + +OUT: + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return (mqd_t)privateMqPersonal; +} + +int mq_close(mqd_t personal) +{ + INT32 ret = 0; + struct mqarray *mqueueCB = NULL; + struct mqpersonal *privateMqPersonal = NULL; + struct mqpersonal *tmp = NULL; + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + privateMqPersonal = (struct mqpersonal *)personal; + if (privateMqPersonal->mq_status != MQ_USE_MAGIC) { + errno = EBADF; + goto OUT_UNLOCK; + } + + mqueueCB = privateMqPersonal->mq_posixdes; + if (mqueueCB->mq_personal == NULL) { + errno = EBADF; + goto OUT_UNLOCK; + } + + /* find the personal and remove */ + if (mqueueCB->mq_personal == privateMqPersonal) { + mqueueCB->mq_personal = privateMqPersonal->mq_next; + } else { + for (tmp = mqueueCB->mq_personal; tmp->mq_next != NULL; tmp = tmp->mq_next) { + if (tmp->mq_next == privateMqPersonal) { + break; + } + } + if (tmp->mq_next == NULL) { + errno = EBADF; + goto OUT_UNLOCK; + } + tmp->mq_next = privateMqPersonal->mq_next; + } + /* flag no use */ + privateMqPersonal->mq_status = 0; + + /* free the personal */ + ret = LOS_MemFree(OS_SYS_MEM_ADDR, privateMqPersonal); + if (ret != LOS_OK) { + errno = EFAULT; + ret = -1; + goto OUT_UNLOCK; + } + + if ((mqueueCB->unlinkflag == TRUE) && (mqueueCB->mq_personal == NULL)) { + ret = DoMqueueDelete(mqueueCB); + } +OUT_UNLOCK: + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return ret; +} + +int OsMqGetAttr(mqd_t personal, struct mq_attr *mqAttr) +{ + struct mqarray *mqueueCB = NULL; + struct mqpersonal *privateMqPersonal = NULL; + + if (mqAttr == NULL) { + errno = EINVAL; + return -1; + } + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + privateMqPersonal = (struct mqpersonal *)personal; + if (privateMqPersonal->mq_status != MQ_USE_MAGIC) { + errno = EBADF; + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return -1; + } + + mqueueCB = privateMqPersonal->mq_posixdes; + mqAttr->mq_maxmsg = mqueueCB->mqcb->queueLen; + mqAttr->mq_msgsize = mqueueCB->mqcb->queueSize - sizeof(UINT32); + mqAttr->mq_curmsgs = mqueueCB->mqcb->readWriteableCnt[OS_QUEUE_READ]; + mqAttr->mq_flags = privateMqPersonal->mq_flags; + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return 0; +} + +int OsMqSetAttr(mqd_t personal, const struct mq_attr *mqSetAttr, struct mq_attr *mqOldAttr) +{ + struct mqpersonal *privateMqPersonal = NULL; + + if (mqSetAttr == NULL) { + errno = EINVAL; + return -1; + } + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + privateMqPersonal = (struct mqpersonal *)personal; + if (privateMqPersonal->mq_status != MQ_USE_MAGIC) { + errno = EBADF; + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return -1; + } + + if (mqOldAttr != NULL) { + (VOID)OsMqGetAttr((mqd_t)privateMqPersonal, mqOldAttr); + } + + privateMqPersonal->mq_flags = (INT32)((UINT32)privateMqPersonal->mq_flags & (UINT32)(~FNONBLOCK)); /* clear */ + if (((UINT32)mqSetAttr->mq_flags & (UINT32)FNONBLOCK) == (UINT32)FNONBLOCK) { + privateMqPersonal->mq_flags = (INT32)((UINT32)privateMqPersonal->mq_flags | (UINT32)FNONBLOCK); + } + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return 0; +} + +int mq_getsetattr(mqd_t mqd, const struct mq_attr *new, struct mq_attr *old) +{ + if (new == NULL) { + return OsMqGetAttr(mqd, old); + } + return OsMqSetAttr(mqd, new, old); +} + +int mq_unlink(const char *mqName) +{ + INT32 ret = 0; + struct mqarray *mqueueCB = NULL; + + if (MqNameCheck(mqName) == -1) { + return -1; + } + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + mqueueCB = GetMqueueCBByName(mqName); + if (mqueueCB == NULL) { + errno = ENOENT; + goto ERROUT_UNLOCK; + } + + if (mqueueCB->mq_personal != NULL) { + mqueueCB->unlinkflag = TRUE; + } else { + ret = DoMqueueDelete(mqueueCB); + } + + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return ret; + +ERROUT_UNLOCK: + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + return -1; +} + +STATIC INT32 ConvertTimeout(long flags, const struct timespec *absTimeout, UINT64 *ticks) +{ + if ((UINT32)flags & (UINT32)FNONBLOCK) { + *ticks = LOS_NO_WAIT; + return 0; + } + + if (absTimeout == NULL) { + *ticks = LOS_WAIT_FOREVER; + return 0; + } + + if (!ValidTimeSpec(absTimeout)) { + errno = EINVAL; + return -1; + } + + *ticks = OsTimeSpec2Tick(absTimeout); + return 0; +} + +STATIC INLINE BOOL MqParamCheck(mqd_t personal, const char *msg, size_t msgLen) +{ + (void)personal; + if ((msg == NULL) || (msgLen == 0)) { + errno = EINVAL; + return FALSE; + } + return TRUE; +} + +#define OS_MQ_GOTO_ERROUT_UNLOCK_IF(expr, errcode) \ + if (expr) { \ + errno = errcode; \ + goto ERROUT_UNLOCK; \ + } +#define OS_MQ_GOTO_ERROUT_IF(expr, errcode) \ + if (expr) { \ + errno = errcode; \ + goto ERROUT; \ + } +int mq_timedsend(mqd_t personal, const char *msg, size_t msgLen, unsigned int msgPrio, + const struct timespec *absTimeout) +{ + UINT32 mqueueID, err; + UINT64 absTicks; + struct mqarray *mqueueCB = NULL; + struct mqpersonal *privateMqPersonal = NULL; + + OS_MQ_GOTO_ERROUT_IF(!MqParamCheck(personal, msg, msgLen), errno); + + OS_MQ_GOTO_ERROUT_IF(msgPrio > (MQ_PRIO_MAX - 1), EINVAL); + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + privateMqPersonal = (struct mqpersonal *)personal; + OS_MQ_GOTO_ERROUT_UNLOCK_IF(privateMqPersonal->mq_status != MQ_USE_MAGIC, EBADF); + + mqueueCB = privateMqPersonal->mq_posixdes; + OS_MQ_GOTO_ERROUT_UNLOCK_IF(msgLen > (size_t)(mqueueCB->mqcb->queueSize - sizeof(UINT32)), EMSGSIZE); + + OS_MQ_GOTO_ERROUT_UNLOCK_IF((((UINT32)privateMqPersonal->mq_flags & (UINT32)O_WRONLY) != (UINT32)O_WRONLY) && + (((UINT32)privateMqPersonal->mq_flags & (UINT32)O_RDWR) != (UINT32)O_RDWR), + EBADF); + + OS_MQ_GOTO_ERROUT_UNLOCK_IF(ConvertTimeout(privateMqPersonal->mq_flags, absTimeout, &absTicks) == -1, errno); + mqueueID = mqueueCB->mq_id; + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + + err = LOS_QueueWriteCopy(mqueueID, (VOID *)msg, (UINT32)msgLen, (UINT32)absTicks); + if (MapMqErrno(err) != ENOERR) { + goto ERROUT; + } + return 0; +ERROUT_UNLOCK: + (VOID)pthread_mutex_unlock(&g_mqueueMutex); +ERROUT: + return -1; +} + +ssize_t mq_timedreceive(mqd_t personal, char *msg, size_t msgLen, unsigned int *msgPrio, + const struct timespec *absTimeout) +{ + UINT32 mqueueID, err; + UINT32 receiveLen; + UINT64 absTicks; + struct mqarray *mqueueCB = NULL; + struct mqpersonal *privateMqPersonal = NULL; + + if (!MqParamCheck(personal, msg, msgLen)) { + goto ERROUT; + } + + if (msgPrio != NULL) { + *msgPrio = 0; + } + + (VOID)pthread_mutex_lock(&g_mqueueMutex); + privateMqPersonal = (struct mqpersonal *)personal; + if (privateMqPersonal->mq_status != MQ_USE_MAGIC) { + errno = EBADF; + goto ERROUT_UNLOCK; + } + + mqueueCB = privateMqPersonal->mq_posixdes; + if (msgLen < (size_t)(mqueueCB->mqcb->queueSize - sizeof(UINT32))) { + errno = EMSGSIZE; + goto ERROUT_UNLOCK; + } + + if (((UINT32)privateMqPersonal->mq_flags & (UINT32)O_WRONLY) == (UINT32)O_WRONLY) { + errno = EBADF; + goto ERROUT_UNLOCK; + } + + if (ConvertTimeout(privateMqPersonal->mq_flags, absTimeout, &absTicks) == -1) { + goto ERROUT_UNLOCK; + } + + receiveLen = msgLen; + mqueueID = mqueueCB->mq_id; + (VOID)pthread_mutex_unlock(&g_mqueueMutex); + + err = LOS_QueueReadCopy(mqueueID, (VOID *)msg, &receiveLen, (UINT32)absTicks); + if (MapMqErrno(err) == ENOERR) { + return (ssize_t)receiveLen; + } else { + goto ERROUT; + } + +ERROUT_UNLOCK: + (VOID)pthread_mutex_unlock(&g_mqueueMutex); +ERROUT: + return -1; +} + +/* not support the prio */ +int mq_send(mqd_t personal, const char *msg_ptr, size_t msg_len, unsigned int msg_prio) +{ + return mq_timedsend(personal, msg_ptr, msg_len, msg_prio, NULL); +} + +ssize_t mq_receive(mqd_t personal, char *msg_ptr, size_t msg_len, unsigned int *msg_prio) +{ + return mq_timedreceive(personal, msg_ptr, msg_len, msg_prio, NULL); +} + diff --git a/kal/posix/src/mqueue_impl.h b/kal/posix/src/mqueue_impl.h new file mode 100644 index 00000000..3f824fde --- /dev/null +++ b/kal/posix/src/mqueue_impl.h @@ -0,0 +1,145 @@ +/* + * Copyright (c) 2013-2019 Huawei Technologies Co., Ltd. All rights reserved. + * Copyright (c) 2020-2021 Huawei Device Co., Ltd. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without modification, + * are permitted provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright notice, this list + * of conditions and the following disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 3. Neither the name of the copyright holder nor the names of its contributors may be used + * to endorse or promote products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; + * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, + * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR + * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF + * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _MQUEUE_IMPL_H +#define _MQUEUE_IMPL_H + +#define _GNU_SOURCE + +#include +#include +#include +#include +#include +#include +#include +#include +#include "los_queue.h" +#include "los_memory.h" +#include "los_task.h" +#include "los_compiler.h" + +#define OS_QUEUE_OPERATE_TYPE(ReadOrWrite, HeadOrTail, PointOrNot) \ + (((UINT32)(PointOrNot) << 2) | ((UINT32)(HeadOrTail) << 1) | (ReadOrWrite)) +#define OS_QUEUE_READ_WRITE_GET(type) ((type) & (0x01)) +#define OS_QUEUE_READ_HEAD (OS_QUEUE_READ | (OS_QUEUE_HEAD << 1)) +#define OS_QUEUE_READ_TAIL (OS_QUEUE_READ | (OS_QUEUE_TAIL << 1)) +#define OS_QUEUE_WRITE_HEAD (OS_QUEUE_WRITE | (OS_QUEUE_HEAD << 1)) +#define OS_QUEUE_WRITE_TAIL (OS_QUEUE_WRITE | (OS_QUEUE_TAIL << 1)) +#define OS_QUEUE_OPERATE_GET(type) ((type) & (0x03)) +#define OS_QUEUE_IS_POINT(type) ((type) & (0x04)) +#define OS_QUEUE_IS_READ(type) (OS_QUEUE_READ_WRITE_GET(type) == OS_QUEUE_READ) +#define OS_QUEUE_IS_WRITE(type) (OS_QUEUE_READ_WRITE_GET(type) == OS_QUEUE_WRITE) +#define OS_READWRITE_LEN 2 + +/** + * @ingroup mqueue + * Maximum number of messages in a message queue + */ +#define MQ_MAX_MSG_NUM 16 + +/** + * @ingroup mqueue + * Maximum size of a single message in a message queue + */ +#define MQ_MAX_MSG_LEN 64 + +#define ENOERR 0 + +/* CONSTANTS */ + +#define MQ_USE_MAGIC 0x89abcdef +#define MQ_PRIO_MAX 1 + + +#define FNONBLOCK O_NONBLOCK + +#define QUEUE_SPLIT_BIT 16 + +#define SET_QUEUE_ID(count, queueID) (((count) << QUEUE_SPLIT_BIT) | (queueID)) + +/** + * @ingroup los_queue + * get the queue index + */ +#define GET_QUEUE_INDEX(queueID) ((queueID) & ((1U << QUEUE_SPLIT_BIT) - 1)) + +/** + * @ingroup los_queue + * get the queue count + */ +#define GET_QUEUE_COUNT(queueID) ((queueID) >> QUEUE_SPLIT_BIT) + + +/** + * @ingroup los_queue + * Obtain the head node in a queue doubly linked list. + */ +#define GET_QUEUE_LIST(ptr) LOS_DL_LIST_ENTRY(ptr, LosQueueCB, readWriteList[OS_QUEUE_WRITE]) + +#define STATIC static +#define INLNIE inline + +typedef union send_receive_t { + unsigned oth : 3; + unsigned grp : 6; + unsigned usr : 9; + short data; +} mode_s; + +/* TYPE DEFINITIONS */ +struct mqarray { + UINT32 mq_id : 31; + UINT32 unlinkflag : 1; + char *mq_name; + UINT32 unlink_ref; + mode_s mode_data; /* mode data of mqueue */ + uid_t euid; /* euid of mqueue */ + gid_t egid; /* egid of mqueue */ + fd_set mq_fdset; /* mqueue sysFd bit map */ + LosQueueCB *mqcb; + struct mqpersonal *mq_personal; +}; + +struct mqpersonal { + struct mqarray *mq_posixdes; + struct mqpersonal *mq_next; + int mq_flags; + int mq_mode; /* Mode of mqueue */ + UINT32 mq_status; + UINT32 mq_refcount; +}; +#ifdef LOSCFG_BASE_IPC_QUEUE_LIMIT +#undef LOSCFG_BASE_IPC_QUEUE_LIMIT +#define LOSCFG_BASE_IPC_QUEUE_LIMIT 20 +#endif + +#endif diff --git a/kal/posix/src/pthread_cond.c b/kal/posix/src/pthread_cond.c new file mode 100644 index 00000000..9c89a682 --- /dev/null +++ b/kal/posix/src/pthread_cond.c @@ -0,0 +1,288 @@ +/* + * Copyright (c) 2013-2019 Huawei Technologies Co., Ltd. All rights reserved. + * Copyright (c) 2020-2021 Huawei Device Co., Ltd. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without modification, + * are permitted provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright notice, this list + * of conditions and the following disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 3. Neither the name of the copyright holder nor the names of its contributors may be used + * to endorse or promote products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; + * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, + * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR + * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF + * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "time_internal.h" +#include +#include +#include +#include +#include +#include "los_config.h" +#include "los_task.h" +#include +#include "los_atomic.h" +#include "los_event.h" + +typedef struct { + volatile INT32 *realValue; + INT32 value; + UINT32 clearEvent; +} EventCond; + +#define INLNIE inline + +#define BROADCAST_EVENT 1 +#define COND_COUNTER_STEP 0x0004U +#define COND_FLAGS_MASK 0x0003U +#define COND_COUNTER_MASK (~COND_FLAGS_MASK) + +int pthread_condattr_destroy(pthread_condattr_t *attr) +{ + if (attr == NULL) { + return EINVAL; + } + + return 0; +} + +int pthread_condattr_init(pthread_condattr_t *attr) +{ + if (attr == NULL) { + return EINVAL; + } + + return 0; +} + +STATIC INLINE INT32 CondInitCheck(const pthread_cond_t *cond) +{ + if ((cond->event.stEventList.pstPrev == NULL) && + (cond->event.stEventList.pstNext == NULL)) { + return 1; + } + return 0; +} + +int pthread_cond_init(pthread_cond_t *cond, const pthread_condattr_t *attr) +{ + int ret = ENOERR; + + if (cond == NULL) { + return EINVAL; + } + (VOID)attr; + (VOID)LOS_EventInit(&(cond->event)); + + cond->mutex = (pthread_mutex_t *)malloc(sizeof(pthread_mutex_t)); + if (cond->mutex == NULL) { + return ENOMEM; + } + + (VOID)pthread_mutex_init(cond->mutex, NULL); + + cond->value = 0; + (VOID)pthread_mutex_lock(cond->mutex); + cond->count = 0; + (VOID)pthread_mutex_unlock(cond->mutex); + + return ret; +} + +int pthread_cond_destroy(pthread_cond_t *cond) +{ + if (cond == NULL) { + return EINVAL; + } + + if (CondInitCheck(cond)) { + return ENOERR; + } + + if (LOS_EventDestroy(&cond->event) != LOS_OK) { + return EBUSY; + } + if (pthread_mutex_destroy(cond->mutex) != ENOERR) { + PRINT_ERR("%s mutex destroy fail!\n", __FUNCTION__); + return EINVAL; + } + free(cond->mutex); + cond->mutex = NULL; + return ENOERR; +} + +STATIC VOID PthreadCountSub(pthread_cond_t *cond) +{ + (VOID)pthread_mutex_lock(cond->mutex); + if (cond->count > 0) { + cond->count--; + } + (VOID)pthread_mutex_unlock(cond->mutex); +} + + +int pthread_cond_broadcast(pthread_cond_t *cond) +{ + int ret = ENOERR; + + if (cond == NULL) { + return EINVAL; + } + + (VOID)pthread_mutex_lock(cond->mutex); + if (cond->count > 0) { + cond->count = 0; + (VOID)pthread_mutex_unlock(cond->mutex); + (VOID)LOS_EventWrite(&(cond->event), BROADCAST_EVENT); + return ret; + } + (VOID)pthread_mutex_unlock(cond->mutex); + + return ret; +} + +int pthread_cond_signal(pthread_cond_t *cond) +{ + int ret = ENOERR; + + if (cond == NULL) { + return EINVAL; + } + + (VOID)pthread_mutex_lock(cond->mutex); + if (cond->count > 0) { + cond->count--; + (VOID)pthread_mutex_unlock(cond->mutex); + // This should modify to once. + (VOID)LOS_EventWrite(&(cond->event), BROADCAST_EVENT); + + return ret; + } + (VOID)pthread_mutex_unlock(cond->mutex); + + return ret; +} + +STATIC INT32 ProcessReturnVal(pthread_cond_t *cond, INT32 val) +{ + INT32 ret; + switch (val) { + /* 0: event does not occur */ + case 0: + case BROADCAST_EVENT: + ret = ENOERR; + break; + case LOS_ERRNO_EVENT_READ_TIMEOUT: + PthreadCountSub(cond); + ret = ETIMEDOUT; + break; + default: + PthreadCountSub(cond); + ret = EINVAL; + break; + } + return ret; +} + +int pthread_cond_timedwait(pthread_cond_t *cond, pthread_mutex_t *mutex, + const struct timespec *absTime) +{ + UINT32 absTicks; + INT32 ret; + + if ((cond == NULL) || (mutex == NULL) || (absTime == NULL)) { + return EINVAL; + } + + if (CondInitCheck(cond)) { + ret = pthread_cond_init(cond, NULL); + if (ret != ENOERR) { + return ret; + } + } + + (VOID)pthread_mutex_lock(cond->mutex); + cond->count++; + (VOID)pthread_mutex_unlock(cond->mutex); + + if ((absTime->tv_sec == 0) && (absTime->tv_nsec == 0)) { + return ETIMEDOUT; + } + + if (!ValidTimeSpec(absTime)) { + return EINVAL; + } + + absTicks = OsTimeSpec2Tick(absTime); + if (pthread_mutex_unlock(mutex) != ENOERR) { + PRINT_ERR("%s: %d failed\n", __FUNCTION__, __LINE__); + } + + ret = (INT32)LOS_EventRead(&(cond->event), 0x0f, LOS_WAITMODE_OR | LOS_WAITMODE_CLR, absTicks); + + if (pthread_mutex_lock(mutex) != ENOERR) { + PRINT_ERR("%s: %d failed\n", __FUNCTION__, __LINE__); + } + + ret = ProcessReturnVal(cond, ret); + return ret; +} + +int pthread_cond_wait(pthread_cond_t *cond, pthread_mutex_t *mutex) +{ + int ret; + + if ((cond == NULL) || (mutex == NULL)) { + return EINVAL; + } + + if (CondInitCheck(cond)) { + ret = pthread_cond_init(cond, NULL); + if (ret != ENOERR) { + return ret; + } + } + + (VOID)pthread_mutex_lock(cond->mutex); + cond->count++; + (VOID)pthread_mutex_unlock(cond->mutex); + + if (pthread_mutex_unlock(mutex) != ENOERR) { + PRINT_ERR("%s: %d failed\n", __FUNCTION__, __LINE__); + } + ret = (INT32)LOS_EventRead(&(cond->event), 0x0f, LOS_WAITMODE_OR | LOS_WAITMODE_CLR, LOS_WAIT_FOREVER); + if (pthread_mutex_lock(mutex) != ENOERR) { + PRINT_ERR("%s: %d failed\n", __FUNCTION__, __LINE__); + } + + switch (ret) { + /* 0: event does not occur */ + case 0: + case BROADCAST_EVENT: + ret = ENOERR; + break; + default: + PthreadCountSub(cond); + ret = EINVAL; + break; + } + + return ret; +} diff --git a/kal/posix/src/semaphore.c b/kal/posix/src/semaphore.c index 395ed30b..2c9535fc 100644 --- a/kal/posix/src/semaphore.c +++ b/kal/posix/src/semaphore.c @@ -32,6 +32,7 @@ #include #include #include "los_sem.h" +#include "time_internal.h" #define _SEM_MAGIC 0xEBCFDEA1 @@ -137,3 +138,52 @@ int sem_post(sem_t *sem) return 0; } + +static long long GetTickTimeFromNow(const struct timespec *absTimeSpec) +{ + struct timespec tsNow = { 0 }; + long long ns; + long long tick; + + clock_gettime(CLOCK_REALTIME, &tsNow); + ns = (absTimeSpec->tv_sec - tsNow.tv_sec) * OS_SYS_NS_PER_SECOND + (absTimeSpec->tv_nsec - tsNow.tv_nsec); + + /* Round up for ticks */ + tick = (ns * LOSCFG_BASE_CORE_TICK_PER_SECOND + (OS_SYS_NS_PER_SECOND - 1)) / OS_SYS_NS_PER_SECOND; + return tick; +} + +int sem_timedwait(sem_t *sem, const struct timespec *timeout) +{ + UINT32 ret; + long long tickCnt; + struct timespec tsNow = { 0 }; + + if ((sem == NULL) || (sem->s_magic != _SEM_MAGIC)) { + errno = EINVAL; + return -1; + } + + if (!ValidTimeSpec(timeout)) { + errno = EINVAL; + return -1; + } + + tickCnt = GetTickTimeFromNow(timeout); + if (tickCnt < 0) { + errno = ETIMEDOUT; + return -1; + } + + if (tickCnt > LOS_WAIT_FOREVER) { + tickCnt = LOS_WAIT_FOREVER; + } + + ret = LOS_SemPend((UINT32)sem->s_handle, (UINT32)tickCnt); + if (ret != LOS_OK) { + errno = MapError(ret); + return -1; + } + + return 0; +} diff --git a/kal/posix/src/time.c b/kal/posix/src/time.c index e9fdb3af..e11c01f7 100755 --- a/kal/posix/src/time.c +++ b/kal/posix/src/time.c @@ -36,35 +36,13 @@ #include #include #include +#include "time_internal.h" #include "los_debug.h" #include "los_task.h" #include "los_swtmr.h" #include "los_timer.h" #include "los_context.h" -#ifndef STATIC -#define STATIC static -#endif - -#define OS_SYS_NS_PER_US 1000 -#define OS_SYS_NS_PER_SECOND 1000000000 -#define OS_SYS_US_PER_SECOND 1000000 -#define OS_SYS_MS_PER_SECOND 1000 - -#define TM_YEAR_BASE 1900 -#define EPOCH_YEAR 1970 -#define SECS_PER_MIN 60 -#define MINS_PER_HOUR 60 -#define SECS_PER_HOUR 3600 /* 60 * 60 */ -#define SECS_PER_DAY 86400 /* 60 * 60 * 24 */ -#define SECS_PER_NORMAL_YEAR 31536000 /* 60 * 60 * 24 * 365 */ -#define DAYS_PER_WEEK 7 -#define DAYS_PER_NORMAL_YEAR 365 -#define DAYS_PER_LEAP_YEAR 366 -#define BEGIN_WEEKDAY 4 -#define TIME_ZONE_MAX 720 /* 12 * 60 */ -#define TIME_ZONE_MIN (-840) /* -14 * 60 */ - /* accumulative time delta from discontinuous modify */ STATIC struct timespec g_accDeltaFromSet; @@ -77,26 +55,6 @@ STATIC const UINT16 g_daysInMonth[2][13] = { STATIC const UINT8 g_montbl[12] = { 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31 }; -/* - * Nonzero if YEAR is a leap year (every 4 years, - * except every 100th isn't, and every 400th is). - */ -#ifndef IS_LEAP_YEAR -#define IS_LEAP_YEAR(year) \ - (((year) % 4 == 0) && (((year) % 100 != 0) || ((year) % 400 == 0))) -#endif -/* The lowest two bytes indicate minutes of the time zone */ -#ifndef OFFSET_TO_MINUTE -#define OFFSET_TO_MINUTE(time) (((time) < 0) ? (-(time)) : (time)) -#endif -/* The highest 31 bytes, 1 indicates eastern time zone，0 indicates western time zone */ -#ifndef TIME_ZONE_SIGN -#define TIME_ZONE_SIGN(time) ((time) >> 31) -#endif - -#define DIV(a, b) (((a) / (b)) - ((a) % (b) < 0)) -#define LEAPS_THRU_END_OF(y) (DIV (y, 4) - DIV (y, 100) + DIV (y, 400)) - static UINT64 g_rtcTimeBase = 0; static UINT64 g_systickBase = 0; @@ -108,42 +66,6 @@ static UINT64 g_systickBase = 0; static INT32 g_rtcTimeZone = -480; static struct tm g_tm = {0}; -/* internal functions */ -STATIC INLINE BOOL ValidTimeSpec(const struct timespec *tp) -{ - /* Fail a NULL pointer */ - if (tp == NULL) { - return FALSE; - } - - /* Fail illegal nanosecond values */ - if ((tp->tv_nsec < 0) || (tp->tv_nsec >= OS_SYS_NS_PER_SECOND) || (tp->tv_sec < 0)) { - return FALSE; - } - - return TRUE; -} - -STATIC INLINE UINT32 OsTimeSpec2Tick(const struct timespec *tp) -{ - UINT64 tick, ns; - - ns = (UINT64)tp->tv_sec * OS_SYS_NS_PER_SECOND + tp->tv_nsec; - /* Round up for ticks */ - tick = (ns * LOSCFG_BASE_CORE_TICK_PER_SECOND + (OS_SYS_NS_PER_SECOND - 1)) / OS_SYS_NS_PER_SECOND; - if (tick > LOS_WAIT_FOREVER) { - tick = LOS_WAIT_FOREVER; - } - return (UINT32)tick; -} - -STATIC INLINE VOID OsTick2TimeSpec(struct timespec *tp, UINT32 tick) -{ - UINT64 ns = ((UINT64)tick * OS_SYS_NS_PER_SECOND) / LOSCFG_BASE_CORE_TICK_PER_SECOND; - tp->tv_sec = (time_t)(ns / OS_SYS_NS_PER_SECOND); - tp->tv_nsec = (long)(ns % OS_SYS_NS_PER_SECOND); -} - int nanosleep(const struct timespec *rqtp, struct timespec *rmtp) { UINT64 nseconds; @@ -721,3 +643,13 @@ int settimeofday(const struct timeval *tv, const struct timezone *tz) HalSetRtcTimeZone(g_rtcTimeZone); return 0; } + +int usleep(unsigned useconds) +{ + struct timespec specTime = { 0 }; + UINT64 nanoseconds = useconds * OS_SYS_NS_PER_US; + + specTime.tv_sec = (time_t)(nanoseconds / OS_SYS_NS_PER_SECOND); + specTime.tv_nsec = (long)(nanoseconds % OS_SYS_NS_PER_SECOND); + return nanosleep(&specTime, NULL); +} \ No newline at end of file diff --git a/kal/posix/src/time_internal.h b/kal/posix/src/time_internal.h new file mode 100644 index 00000000..e7181d19 --- /dev/null +++ b/kal/posix/src/time_internal.h @@ -0,0 +1,123 @@ +/* + * Copyright (c) 2013-2019 Huawei Technologies Co., Ltd. All rights reserved. + * Copyright (c) 2020-2021 Huawei Device Co., Ltd. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without modification, + * are permitted provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright notice, this list + * of conditions and the following disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 3. Neither the name of the copyright holder nor the names of its contributors may be used + * to endorse or promote products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, + * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, + * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; + * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, + * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR + * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF + * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _TIME_IMPL_H +#define _TIME_IMPL_H + +#include +#include +#include +#include +#include +#include +#include "los_debug.h" +#include "los_task.h" +#include "los_swtmr.h" +#include "los_timer.h" +#include "los_context.h" +#include "los_compiler.h" + +#define OS_SYS_NS_PER_US 1000 +#define OS_SYS_NS_PER_SECOND 1000000000 +#define OS_SYS_US_PER_SECOND 1000000 +#define OS_SYS_MS_PER_SECOND 1000 + +#define TM_YEAR_BASE 1900 +#define EPOCH_YEAR 1970 +#define SECS_PER_MIN 60 +#define MINS_PER_HOUR 60 +#define SECS_PER_HOUR 3600 /* 60 * 60 */ +#define SECS_PER_DAY 86400 /* 60 * 60 * 24 */ +#define SECS_PER_NORMAL_YEAR 31536000 /* 60 * 60 * 24 * 365 */ +#define DAYS_PER_WEEK 7 +#define DAYS_PER_NORMAL_YEAR 365 +#define DAYS_PER_LEAP_YEAR 366 +#define BEGIN_WEEKDAY 4 +#define TIME_ZONE_MAX 720 /* 12 * 60 */ +#define TIME_ZONE_MIN (-840) /* -14 * 60 */ + +/* + * Nonzero if YEAR is a leap year (every 4 years, + * except every 100th isn't, and every 400th is). + */ +#ifndef IS_LEAP_YEAR +#define IS_LEAP_YEAR(year) \ + (((year) % 4 == 0) && (((year) % 100 != 0) || ((year) % 400 == 0))) +#endif +/* The lowest two bytes indicate minutes of the time zone */ +#ifndef OFFSET_TO_MINUTE +#define OFFSET_TO_MINUTE(time) (((time) < 0) ? (-(time)) : (time)) +#endif +/* The highest 31 bytes, 1 indicates eastern time zoneï¼? indicates western time zone */ +#ifndef TIME_ZONE_SIGN +#define TIME_ZONE_SIGN(time) ((time) >> 31) +#endif + +#define DIV(a, b) (((a) / (b)) - ((a) % (b) < 0)) +#define LEAPS_THRU_END_OF(y) (DIV (y, 4) - DIV (y, 100) + DIV (y, 400)) + +/* internal functions */ +STATIC INLINE BOOL ValidTimeSpec(const struct timespec *tp) +{ + /* Fail a NULL pointer */ + if (tp == NULL) { + return FALSE; + } + + /* Fail illegal nanosecond values */ + if ((tp->tv_nsec < 0) || (tp->tv_nsec >= OS_SYS_NS_PER_SECOND) || (tp->tv_sec < 0)) { + return FALSE; + } + + return TRUE; +} + +STATIC INLINE UINT32 OsTimeSpec2Tick(const struct timespec *tp) +{ + UINT64 tick, ns; + + ns = (UINT64)tp->tv_sec * OS_SYS_NS_PER_SECOND + tp->tv_nsec; + /* Round up for ticks */ + tick = (ns * LOSCFG_BASE_CORE_TICK_PER_SECOND + (OS_SYS_NS_PER_SECOND - 1)) / OS_SYS_NS_PER_SECOND; + if (tick > LOS_WAIT_FOREVER) { + tick = LOS_WAIT_FOREVER; + } + return (UINT32)tick; +} + +STATIC INLINE VOID OsTick2TimeSpec(struct timespec *tp, UINT32 tick) +{ + UINT64 ns = ((UINT64)tick * OS_SYS_NS_PER_SECOND) / LOSCFG_BASE_CORE_TICK_PER_SECOND; + tp->tv_sec = (time_t)(ns / OS_SYS_NS_PER_SECOND); + tp->tv_nsec = (long)(ns % OS_SYS_NS_PER_SECOND); +} +#endif +