* getopt_long_only bug @ 2018-04-26 10:46 Szabolcs Nagy 2018-04-26 14:50 ` Rich Felker 0 siblings, 1 reply; 3+ messages in thread From: Szabolcs Nagy @ 2018-04-26 10:46 UTC (permalink / raw) To: musl i got the following bug report: $ cat a.c #include <stdio.h> #include <getopt.h> int main() { struct option lo[] = { {"emit-relocs", no_argument, NULL, 'q'}, {"qmagic", no_argument, NULL, 'Q'}, { 0 } }; int li = -1; char* a[] = { "", "-q", NULL }; int c = getopt_long_only(2, a, "q", lo, &li); printf("c=%c/%i\n", c, c); return 0; } $ ./a.musl c=Q/81 $ ./a.glibc c=q/113 ^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: getopt_long_only bug 2018-04-26 10:46 getopt_long_only bug Szabolcs Nagy @ 2018-04-26 14:50 ` Rich Felker 2018-04-26 20:17 ` Szabolcs Nagy 0 siblings, 1 reply; 3+ messages in thread From: Rich Felker @ 2018-04-26 14:50 UTC (permalink / raw) To: musl [-- Attachment #1: Type: text/plain, Size: 1131 bytes --] On Thu, Apr 26, 2018 at 12:46:31PM +0200, Szabolcs Nagy wrote: > i got the following bug report: > > $ cat a.c > #include <stdio.h> > #include <getopt.h> > > int main() > { > struct option lo[] = > { > {"emit-relocs", no_argument, NULL, 'q'}, > {"qmagic", no_argument, NULL, 'Q'}, > { 0 } > }; > > int li = -1; > char* a[] = { "", "-q", NULL }; > int c = getopt_long_only(2, a, "q", lo, &li); > > printf("c=%c/%i\n", c, c); > > return 0; > } > $ ./a.musl > c=Q/81 > $ ./a.glibc > c=q/113 OK, this is a weird corner case. Apparently when in longonly mode, short options need to be counted in addition to long ones for determining if a partial match is unique. The attached patch should solve the problem. Any review/comments would be helpful. It uses an ugly inline strstr of sorts, because to use strstr we'd need to copy to a temp buffer, and we wouldn't even benefit since the string being searched is so short (usually 1 byte, at most 4). The only reason it's there at all is because we committed to supporting multibyte option chars in getopt.c so getopt_long_only needs to handle them consistently. Rich [-- Attachment #2: glo.diff --] [-- Type: text/plain, Size: 1219 bytes --] diff --git a/src/misc/getopt_long.c b/src/misc/getopt_long.c index 008b747..ddcef94 100644 --- a/src/misc/getopt_long.c +++ b/src/misc/getopt_long.c @@ -1,5 +1,7 @@ #define _GNU_SOURCE #include <stddef.h> +#include <stdlib.h> +#include <limits.h> #include <getopt.h> #include <stdio.h> #include <string.h> @@ -58,10 +60,10 @@ static int __getopt_long_core(int argc, char *const *argv, const char *optstring { int colon = optstring[optstring[0]=='+'||optstring[0]=='-']==':'; int i, cnt, match; - char *arg, *opt; + char *arg, *opt, *start = argv[optind]+1; for (cnt=i=0; longopts[i].name; i++) { const char *name = longopts[i].name; - opt = argv[optind]+1; + opt = start; if (*opt == '-') opt++; while (*opt && *opt != '=' && *opt == *name) name++, opt++; @@ -74,6 +76,17 @@ static int __getopt_long_core(int argc, char *const *argv, const char *optstring } cnt++; } + if (cnt==1 && longonly && arg-start == mblen(start, MB_LEN_MAX)) { + int l = arg-start; + for (i=0; optstring[i]; i++) { + int j; + for (j=0; j<l && start[j]==optstring[i+j]; j++); + if (j==l) { + cnt++; + break; + } + } + } if (cnt==1) { i = match; opt = arg; ^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: getopt_long_only bug 2018-04-26 14:50 ` Rich Felker @ 2018-04-26 20:17 ` Szabolcs Nagy 0 siblings, 0 replies; 3+ messages in thread From: Szabolcs Nagy @ 2018-04-26 20:17 UTC (permalink / raw) To: musl * Rich Felker <dalias@libc.org> [2018-04-26 10:50:37 -0400]: > OK, this is a weird corner case. Apparently when in longonly mode, > short options need to be counted in addition to long ones for > determining if a partial match is unique. > > The attached patch should solve the problem. Any review/comments would > be helpful. It uses an ugly inline strstr of sorts, because to use > strstr we'd need to copy to a temp buffer, and we wouldn't even > benefit since the string being searched is so short (usually 1 byte, > at most 4). The only reason it's there at all is because we committed > to supporting multibyte option chars in getopt.c so getopt_long_only > needs to handle them consistently. looks ok to me. > diff --git a/src/misc/getopt_long.c b/src/misc/getopt_long.c > index 008b747..ddcef94 100644 > --- a/src/misc/getopt_long.c > +++ b/src/misc/getopt_long.c > @@ -1,5 +1,7 @@ > #define _GNU_SOURCE > #include <stddef.h> > +#include <stdlib.h> > +#include <limits.h> > #include <getopt.h> > #include <stdio.h> > #include <string.h> > @@ -58,10 +60,10 @@ static int __getopt_long_core(int argc, char *const *argv, const char *optstring > { > int colon = optstring[optstring[0]=='+'||optstring[0]=='-']==':'; > int i, cnt, match; > - char *arg, *opt; > + char *arg, *opt, *start = argv[optind]+1; > for (cnt=i=0; longopts[i].name; i++) { > const char *name = longopts[i].name; > - opt = argv[optind]+1; > + opt = start; > if (*opt == '-') opt++; > while (*opt && *opt != '=' && *opt == *name) > name++, opt++; > @@ -74,6 +76,17 @@ static int __getopt_long_core(int argc, char *const *argv, const char *optstring > } > cnt++; > } > + if (cnt==1 && longonly && arg-start == mblen(start, MB_LEN_MAX)) { > + int l = arg-start; > + for (i=0; optstring[i]; i++) { > + int j; > + for (j=0; j<l && start[j]==optstring[i+j]; j++); > + if (j==l) { > + cnt++; > + break; > + } > + } > + } > if (cnt==1) { > i = match; > opt = arg; ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2018-04-26 20:17 UTC | newest] Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2018-04-26 10:46 getopt_long_only bug Szabolcs Nagy 2018-04-26 14:50 ` Rich Felker 2018-04-26 20:17 ` Szabolcs Nagy
Code repositories for project(s) associated with this public inbox https://git.vuxu.org/mirror/musl/ This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).