prototype comparision tools

diff --git a/data/musl.problems b/data/musl.problems
new file mode 100644 (file)
index 0000000..fee26ff
--- /dev/null
+++ b/data/musl.problems
@@ -0,0 +1,144 @@
+header FD_CLR  sys/select.h    sys_select.h
+header FD_ISSET        sys/select.h    sys_select.h
+header FD_SET  sys/select.h    sys_select.h
+header FD_ZERO sys/select.h    sys_select.h
+header accept  sys/socket.h    sys_socket.h
+proto  aio_error       aio.h   int(struct aiocb*)      int(const struct aiocb*)        int aio_error(struct aiocb *);  int aio_error(const struct aiocb *aiocbp);
+proto  aio_suspend     aio.h   int(struct aiocb*const[],int,const struct timespec*)    int(const struct aiocb*const[],int,const struct timespec*)      int aio_suspend(struct aiocb *const [], int, const struct timespec *);  int aio_suspend(const struct aiocb *const list[], int nent, const struct timespec *timeout);
+header bind    sys/socket.h    sys_socket.h
+nodecl cfmakeraw       T
+nodecl cfsetspeed      W
+header chmod   sys/stat.h      sys_stat.h
+nodecl clearerr_unlocked       W
+header connect sys/socket.h    sys_socket.h
+nodecl crypt_r W
+header ctermid unistd.h        stdio.h
+nodecl daemon  T
+nodecl dst_offset      V
+proto  encrypt unistd.h        void(char*,int) void(char[64],int)      void encrypt(char *, int);      void encrypt(char block[64], int edflag);
+nodecl environ V
+proto  erand48 stdlib.h        double(unsigned short[])        double(unsigned short[3])       double erand48 (unsigned short [3]);    double erand48(unsigned short xsubi[3]);
+header fchmod  sys/stat.h      sys_stat.h
+header fchmodat        sys/stat.h      sys_stat.h
+nodecl feof_unlocked   W
+nodecl ferror_unlocked W
+nodecl fflush_unlocked W
+nodecl fgetwc_unlocked W
+nodecl fgetws_unlocked W
+nodecl fileno_unlocked W
+nodecl fputwc_unlocked W
+nodecl fputws_unlocked W
+nodecl fread_unlocked  W
+header fstat   sys/stat.h      sys_stat.h
+header fstatat sys/stat.h      sys_stat.h
+header fstatvfs        sys/statvfs.h   sys_statvfs.h
+header ftok    sys/ipc.h       sys_ipc.h
+header futimens        sys/stat.h      sys_stat.h sys_time.h
+proto  futimens        sys/stat.h      int(int,const struct timespec[])        int(int,const struct timespec[2])       int futimens(int, const struct timespec [2]);   int futimens(int fd, const struct timespec times[2]);
+nodecl fwrite_unlocked W
+proto  getdate time.h  extern struct tm*(const char*)  struct tm*(const char*) extern struct tm *getdate (const char *);       struct tm *getdate(const char *string);
+nodecl getdents        W
+nodecl getdomainname   T
+header getitimer       sys/time.h      sys_time.h
+header getpeername     sys/socket.h    sys_socket.h
+header getpriority     sys/resource.h  sys_resource.h
+header getrlimit       sys/resource.h  sys_resource.h
+header getrusage       sys/resource.h  sys_resource.h
+header getsockname     sys/socket.h    sys_socket.h
+header getsockopt      sys/socket.h    sys_socket.h
+header gettimeofday    sys/time.h      sys_time.h
+nodecl getw    T
+nodecl getwc_unlocked  W
+nodecl getwchar_unlocked       W
+header htonl   netinet/in.h    arpa_inet.h
+header htons   netinet/in.h    arpa_inet.h
+header if_freenameindex        net/if.h        net_if.h
+header if_indextoname  net/if.h        net_if.h
+header if_nameindex    net/if.h        net_if.h
+header if_nametoindex  net/if.h        net_if.h
+header inet_addr       arpa/inet.h     arpa_inet.h
+header inet_ntoa       arpa/inet.h     arpa_inet.h
+header inet_ntop       arpa/inet.h     arpa_inet.h
+header inet_pton       arpa/inet.h     arpa_inet.h
+proto  jrand48 stdlib.h        long(unsigned short[])  long(unsigned short[3]) long jrand48 (unsigned short [3]);      long jrand48(unsigned short xsubi[3]);
+proto  lcong48 stdlib.h        void(unsigned short[])  void(unsigned short[7]) void lcong48 (unsigned short [7]);      void lcong48(unsigned short param[7]);
+header listen  sys/socket.h    sys_socket.h
+header lstat   sys/stat.h      sys_stat.h
+header mkdir   sys/stat.h      sys_stat.h
+header mkdirat sys/stat.h      sys_stat.h
+header mkfifo  sys/stat.h      sys_stat.h
+header mkfifoat        sys/stat.h      sys_stat.h
+header mknod   sys/stat.h      sys_stat.h
+header mknodat sys/stat.h      sys_stat.h
+header mlock   sys/mman.h      sys_mman.h
+header mlockall        sys/mman.h      sys_mman.h
+header mmap    sys/mman.h      sys_mman.h
+header mprotect        sys/mman.h      sys_mman.h
+header msgctl  sys/msg.h       sys_msg.h
+header msgget  sys/msg.h       sys_msg.h
+header msgrcv  sys/msg.h       sys_msg.h
+header msgsnd  sys/msg.h       sys_msg.h
+header msync   sys/mman.h      sys_mman.h
+header munlock sys/mman.h      sys_mman.h
+header munlockall      sys/mman.h      sys_mman.h
+header munmap  sys/mman.h      sys_mman.h
+proto  nrand48 stdlib.h        long(unsigned short[])  long(unsigned short[3]) long int nrand48 (unsigned short [3]);  long nrand48(unsigned short xsubi[3]);
+header ntohl   netinet/in.h    arpa_inet.h
+header ntohs   netinet/in.h    arpa_inet.h
+proto  pipe    unistd.h        int(int[])      int(int[2])     int pipe(int [2]);      int pipe(int fildes[2]);
+nodecl pivot_root      T
+proto  poll    poll.h  int(struct pollfd*,nfds_t,int)  int(struct pollfd[],nfds_t,int) int poll (struct pollfd *, nfds_t, int);        int poll(struct pollfd fds[], nfds_t nfds, int timeout);
+header posix_madvise   sys/mman.h      sys_mman.h
+proto  posix_spawn     spawn.h int(pid_t*,const char*,const posix_spawn_file_actions_t*,const posix_spawnattr_t*,char**,char**)        int(pid_t*,const char*,const posix_spawn_file_actions_t*,const posix_spawnattr_t*,char*const[],char*const[])    int posix_spawn(pid_t *, const char *, const posix_spawn_file_actions_t *, const posix_spawnattr_t *, char **, char **);        int posix_spawn(pid_t *restrict pid, const char *restrict path, const posix_spawn_file_actions_t *file_actions, const posix_spawnattr_t *restrict attrp, char *const argv[restrict], char *const envp[restrict]);
+proto  posix_spawnp    spawn.h int(pid_t*,const char*,const posix_spawn_file_actions_t*,const posix_spawnattr_t*,char**,char**)        int(pid_t*,const char*,const posix_spawn_file_actions_t*,const posix_spawnattr_t*,char*const[],char*const[])    int posix_spawnp(pid_t *, const char *, const posix_spawn_file_actions_t *, const posix_spawnattr_t *, char **, char **);       int posix_spawnp(pid_t *restrict pid, const char *restrict file, const posix_spawn_file_actions_t *file_actions, const posix_spawnattr_t *restrict attrp, char *const argv[restrict], char *const envp[restrict]);
+header pselect sys/select.h    sys_select.h
+nodecl putw    T
+nodecl putwc_unlocked  W
+nodecl putwchar_unlocked       W
+proto  readlink        unistd.h        int(const char*,char*,size_t)   ssize_t(const char*,char*,size_t)       int readlink(const char *, char *, size_t);     ssize_t readlink(const char *restrict path, char *restrict buf, size_t bufsize);
+proto  readlinkat      unistd.h        int(int,const char*,char*,size_t)       ssize_t(int,const char*,char*,size_t)   int readlinkat(int, const char *, char *, size_t);      ssize_t readlinkat(int fd, const char *restrict path, char *restrict buf, size_t bufsize);
+header readv   sys/uio.h       sys_uio.h
+header recv    sys/socket.h    sys_socket.h
+header recvfrom        sys/socket.h    sys_socket.h
+header recvmsg sys/socket.h    sys_socket.h
+nodecl scalbf  T
+proto  seed48  stdlib.h        unsigned short*(unsigned short[])       unsigned short*(unsigned short[3])      unsigned short *seed48 (unsigned short [3]);    unsigned short *seed48(unsigned short seed16v[3]);
+header select  sys/select.h    sys_select.h
+header semget  sys/sem.h       sys_sem.h
+header semop   sys/sem.h       sys_sem.h
+header send    sys/socket.h    sys_socket.h
+header sendmsg sys/socket.h    sys_socket.h
+header sendto  sys/socket.h    sys_socket.h
+header setitimer       sys/time.h      sys_time.h
+proto  setpgrp unistd.h        int(void)       pid_t(void)     int setpgrp(void);      pid_t setpgrp(void);
+header setpriority     sys/resource.h  sys_resource.h
+header setrlimit       sys/resource.h  sys_resource.h
+header setsockopt      sys/socket.h    sys_socket.h
+header shm_open        sys/mman.h      sys_mman.h
+header shm_unlink      sys/mman.h      sys_mman.h
+header shmat   sys/shm.h       sys_shm.h
+header shmctl  sys/shm.h       sys_shm.h
+header shmdt   sys/shm.h       sys_shm.h
+header shmget  sys/shm.h       sys_shm.h
+header shutdown        sys/socket.h    sys_socket.h
+nodecl sincos  T
+header sockatmark      sys/socket.h    sys_socket.h
+header socket  sys/socket.h    sys_socket.h
+header socketpair      sys/socket.h    sys_socket.h
+proto  socketpair      sys/socket.h    int(int,int,int,int[])  int(int,int,int,int[2]) int socketpair (int, int, int, int [2]);        int socketpair(int domain, int type, int protocol, int socket_vector[2]);
+nodecl start   T
+header stat    sys/stat.h      sys_stat.h
+header statvfs sys/statvfs.h   sys_statvfs.h
+nodecl stime   T
+header times   sys/times.h     sys_times.h
+header umask   sys/stat.h      sys_stat.h
+header uname   sys/utsname.h   sys_utsname.h
+header utimensat       sys/stat.h      sys_stat.h sys_time.h
+proto  utimensat       sys/stat.h      int(int,const char*,const struct timespec[],int)        int(int,const char*,const struct timespec[2],int)       int utimensat(int, const char *, const struct timespec [2], int);       int utimensat(int fd, const char *path, const struct timespec times[2], int flag);
+header utimes  sys/time.h      sys_stat.h sys_time.h
+proto  utimes  sys/time.h      int(const char*,const struct timeval[]) int(const char*,const struct timeval[2])        int utimes (const char *, const struct timeval [2]);    int utimes(const char *path, const struct timeval times[2]);
+header wait    sys/wait.h      sys_wait.h
+header waitid  sys/wait.h      sys_wait.h
+header waitpid sys/wait.h      sys_wait.h
+header wcrtomb wchar.h stdio.h
+header writev  sys/uio.h       sys_uio.h

diff --git a/makeall.sh b/makeall.sh
new file mode 100755 (executable)
index 0000000..99aa757
--- /dev/null
+++ b/makeall.sh
@@ -0,0 +1,64 @@
+#!/bin/sh
+
+export LC_ALL=C
+
+[ -f data/musl.tags.proto -a -f data/posix2008.ok.proto ] || ./makeproto.sh
+
+{
+       awk -F'\t' '{print $1}' data/musl.tags data/posix2008.ok # data/c99
+       awk -F'\t' '{print $2}' data/musl.syms
+} |sort |uniq |awk -F'\t' '
+BEGIN {
+       posix = "data/posix2008.ok.proto"
+       syms = "data/musl.syms"
+       tags = "data/musl.tags.proto"
+
+       while (getline < syms == 1)
+               sym[$2] = $1
+       while (getline < tags == 1)
+               tag[$1] = $2 "\t" $3 "\t" $5 "\t" $6
+       while (getline < posix == 1)
+               pos[$1] = $2 "\t" $5 "\t" $6
+}
+{
+       s = ""
+       if (sym[$1])
+               s = s " obj"
+       if (tag[$1])
+               s = s " inc"
+       if (pos[$1])
+               s = s " posix"
+
+       print $1 "\t" substr(s,2) "\t" sym[$1] "\t" tag[$1] "\t" pos[$1]
+}' >data/musl.all
+
+awk -F'\t' '
+$2 == "obj" || $2 == "obj posix" {
+       # not declared
+       if ($1 !~ /^_/)
+               print "nodecl\t" $1 "\t" $3
+}
+$2 ~ /inc posix/ && $4 != $8 {
+       # different header
+       n = split($8, a, " ")
+       for (i = 1; i <= n; i++)
+               if ($4 == a[i])
+                       break
+       if (i > n)
+               print "header\t" $1 "\t" $4 "\t" $8
+}
+$2 ~ /inc posix/ && $7 != $10 && $5 == "p" {
+       # different prototype
+
+       # todo: move to type.sh
+       gsub(/restrict const/, "const", $10)
+       gsub(/restrict/, "", $10)
+#      gsub(/\[[0-9]+\]/, "[]", $10)
+       gsub(/unsigned int/, "unsigned", $7)
+       gsub(/long int/, "long", $7)
+
+       if ($7 != $10)
+               print "proto\t" $1 "\t" $4 "\t" $7 "\t" $10 "\t" $6 "\t" $9
+}
+' data/musl.all >data/musl.problems
+

diff --git a/makeproto.sh b/makeproto.sh
new file mode 100755 (executable)
index 0000000..326e6dd
--- /dev/null
+++ b/makeproto.sh
@@ -0,0 +1,29 @@
+#!/bin/sh
+
+export LC_ALL=C
+MUSL=${MUSL:-../musl}
+
+awk -F'\t' -v inc=$MUSL/include/ '
+$3 == "p" && $2 !~ /^bits\// {
+       cmd = "awk '\''NR == " $4 "{s=$0;if(s!~/;/){getline; s=s " " $0} print s; exit}'\'' " inc $2
+       cmd | getline proto
+       close(cmd)
+       gsub(/\t/, " ", proto)
+}
+{
+       print $0 "\t" proto
+       proto = ""
+}
+' data/musl.tags >/tmp/tags.proto
+
+awk -F'\t' '{print $5}' /tmp/tags.proto |./type.sh >/tmp/tags.type
+awk -F'\t' '{print $5}' data/posix2008.ok |./type.sh >/tmp/posix2008.type
+
+join() {
+       awk -v f=$2 '{getline s < f; print $0 "\t" s}' $1
+}
+
+join /tmp/tags.proto /tmp/tags.type >data/musl.tags.proto
+join data/posix2008.ok /tmp/posix2008.type >data/posix2008.ok.proto
+
+rm -f /tmp/tags.proto /tmp/tags.type /tmp/posix2008.type

diff --git a/type.sh b/type.sh
new file mode 100755 (executable)
index 0000000..3aa5a9c
--- /dev/null
+++ b/type.sh
@@ -0,0 +1,80 @@
+#!/bin/sh
+
+export LC_ALL=C
+
+# drop names from a declaration (hack to make prototypes comparable)
+awk '
+BEGIN {
+       # type is not typedefed so next unknown id is probably a variable name
+       split("void char short int long float double signed unsigned _Bool _Complex", a)
+       for (i in a)
+               tok[a[i]] = "type"
+
+       # next token is an id, type is not typedefed
+       split("struct union enum", a)
+       for (i in a)
+               tok[a[i]] = "struct"
+
+       # decoratoin that can be skipped
+       split("static extern auto register inline const volatile restrict", a)
+       for (i in a)
+               tok[a[i]] = "decor"
+
+       # punctuators
+       split("( ) [ ] , ... *", a)
+       for (i in a)
+               tok[a[i]] = "punct"
+}
+
+function put(tok) {
+       if (tok ~ /^[a-zA-Z_]/) {
+               s = s sep tok
+               sep = " "
+       } else {
+               s = s tok
+               sep = ""
+       }
+}
+
+{
+       # eat comments
+       gsub(/\/\*[^/]*\*\//, "")
+       gsub(/\/\/.*/, "")
+
+       gsub(/[^a-zA-Z0-9_.]/," & ")
+       gsub(/\.\.\./, " & ")
+
+       state = "type"
+       sep = ""
+       s = ""
+       for (i = 1; i <= NF; i++) {
+               if ($i == ";")
+                       break
+               if (state == "type") {
+                       put($i)
+                       if (!tok[$i] || tok[$i] == "type")
+                               state = "id"
+                       if (tok[$i] == "struct") {
+                               i++
+                               put($i)
+                               state = "id"
+                       }
+               } else if (state == "id") {
+                       if (!tok[$i]) {
+                               state = "idfound"
+                               continue
+                       }
+                       put($i)
+                       if ($i == ")")
+                               state = "idfound"
+                       if ($i == ",")
+                               state = "type"
+               } else if (state == "idfound") {
+                       put($i)
+                       if ($i == "(" || $i == ",")
+                               state = "type"
+               }
+       }
+       print s
+}
+'