Port getprogname module to SCO OpenServer
I ported the getprogname module to SCO OpenServer 6 (should also work on
OSR5 and UnixWare). It prevents several OSS packages from building.
I just made it read from /proc//cmdline to get the command name.
The patch is below. Comments are welcome. Thanks!
-Benji
diff --git a/lib/getprogname.c b/lib/getprogname.c
index 744466ea9..9ee4c226d 100644
--- a/lib/getprogname.c
+++ b/lib/getprogname.c
@@ -51,6 +51,11 @@
# include
#endif
+#ifdef __SCO_VERSION__
+# include
+# include
+#endif
+
#include "basename-lgpl.h"
#ifndef HAVE_GETPROGNAME /* not Mac OS X, FreeBSD, NetBSD,
OpenBSD >= 5.4, Cygwin */
@@ -245,6 +250,39 @@ getprogname (void)
}
}
return NULL;
+# elif defined __SCO_VERSION__ /* SCO
OpenServer/UnixWare */
+ char buf[50];
+ char *ret;
+ int fd;
+ int pathlen;
+ sprintf (buf, "/proc/%d/cmdline", (int)getpid());
+ fd = open (buf, O_RDONLY);
+ if (0 <= fd)
+ {
+ size_t n = read (fd, buf, 49);
+ if (n > 0)
+ {
+ buf[49] = '\0'; /* Guarantee null-termination */
+ pathlen = strlen (buf);
+ ret = malloc (pathlen + 1);
+ if (ret)
+ {
+ int i;
+ int pathsep = 0;
+ for (i = 0; i < pathlen; i++)
+ {
+ if (buf[i] == '/')
+ {
+ pathsep = i;
+ }
+ }
+ strcpy (ret, buf + pathsep + 1);
+ return ret;
+ }
+ }
+ close (fd);
+ }
+ return "?";
# else
# error "getprogname module not ported to this OS"
# endif
Re: hard-locale.c: SETLOCALE_NULL_MAX
Hi Bruno, thank you for your reply. I could finally track down the error. It was in my build configuration. Somehow, the wrong "-I ..." flags were set up during the compilation of the Gnulib modules. It's working now. So, everything is alright with Gnulib and I can retract my bug report. Please excuse me for bothering you. Initially, when I found the bug on Gentoo's bug tracker, I didn't think of such a simple cause. As for your questions, it was probably only (6) that caused the problem. Marc Am Mi., 30. Sept. 2020 um 18:38 Uhr schrieb Bruno Haible : > > Hi Marc, > > > The compiler throws the following error at me: > > > > lib/hard-locale.c: In function 'hard_locale': > > lib/hard-locale.c:29:15: error: 'SETLOCALE_NULL_MAX' undeclared (first > > use in this function); did you mean 'SETLOCALE_NULL_ALL_MTSAFE'? > >29 | char locale[SETLOCALE_NULL_MAX]; > > | ^~ > > Does the error occur always, or only with "make -j"? > > Are the following statements true in your build? > (1) I use the module 'hard-locale'. > (2) I use the module 'setlocale-null'. > (3) config.status defines GNULIB_SETLOCALE_NULL to 1. > (4) The file lib/setlocale_null.h exists. > (5) The file lib/locale.h includes setlocale_null.h. > (6) The -I options passed to GCC make sure that setlocale_null.h gets found. > > Bruno >
Re: Port getprogname module to SCO OpenServer
Hi Benji,
> I just made it read from /proc//cmdline to get the command name.
> The patch is below. Comments are welcome. Thanks!
Thanks for the patch. I have a couple of improvement suggestions, though:
> + char buf[50];
> + char *ret;
> + int fd;
> + int pathlen;
Can you try to minimize the scope of local variables? Listing all the
local variables upfront is BSD style and leads to code that is hard to
understand.
> +{
> + size_t n = read (fd, buf, 49);
> + if (n > 0)
> +{
> + buf[49] = '\0'; /* Guarantee null-termination */
> + pathlen = strlen (buf);
If n < 49, this call to strlen may read uninitialized memory, no?
Better put a NUL in buf[n], not buf[49], then.
> + ret = malloc (pathlen + 1);
> + if (ret)
> +{
> + int i;
> + int pathsep = 0;
> + for (i = 0; i < pathlen; i++)
> +{
> + if (buf[i] == '/')
> +{
> + pathsep = i;
> +}
> +}
> + strcpy (ret, buf + pathsep + 1);
Can't this code be simplified by calling strrchr (buf, '/') ?
> + return ret;
Is the size pathlen + 1 really needed for ret? It looks like you need
only strlen (buf + pathsep + 1) + 1 bytes.
> + }
> +}
> + close (fd);
> +}
> + return "?";
> # else
Bruno
Re: Port getprogname module to SCO OpenServer
Hi Benji, On Wed, 30 Sep 2020, Benji Wiebe wrote: > I ported the getprogname module to SCO OpenServer 6 (should also work on OSR5 > and UnixWare). It prevents several OSS packages from building. No proc filesystem on Openserver 5 so it will not work there. Would need a different "#ifdef" for 5 anyway. > I just made it read from /proc//cmdline to get the command name. The > patch is below. Comments are welcome. Thanks! Note reading /proc//cmdline will limit you to a 79 char pathname. And your program limits to 49. > > +# elif defined __SCO_VERSION__ /* SCO OpenServer/UnixWare */ While __SCO_VERSION__ covers Openserver 6 and UnixWare 7, what is normally used for 6 and 7 is __USLC__ for the native compiler and __sysv5__ for gcc Ie. # elif defined __USLC__ || defined __sysv5__ -- Tim RiceMultitalents t...@multitalents.net
Re: Port getprogname module to SCO OpenServer
Oh and I forgot to mention, the cast from strrchr is needed to silence a warning from SCO's CC: UX:acomp: WARNING: "gpn_test.c", line 16: improper pointer/integer combination: op "="
another stdio patch for UnixWare
I have attached an additional patch for the stdio parts 2020-09-30 Tim Rice * lib/stdio-impl.h: Add support for UnixWare * lib/freadahead.c: Use __fpending on UnixWare * lib/fflush.c: Update comments for UnixWare * lib/fpending.c: Likewise. * lib/freadable.c: Likewise. * lib/freadptr.c: Likewise. * lib/freadseek.c: Likewise. * lib/fseterr.c: Likewise. * lib/fwritable.c: Likewise. * lib/fwriting.c: Likewise. THe hunk in lib/freadahead.c could potentially be used for other system that define HAVE__FPENDING (Solaris and others?) but I have not tested them so limited to UniWare (and OpenServer 6). With either of the nap() patches in https://lists.gnu.org/archive/html/bug-gnulib/2020-09/msg00126.html and the stdio_ext patch in https://lists.gnu.org/archive/html/bug-gnulib/2020-09/msg00127.html and the stdio_ext_m4 patch in https://lists.gnu.org/archive/html/bug-gnulib/2020-09/msg00132.html and the attached patch, I am down to 2 failures in the stdio tests. I used this from one of Bruno's posts ./gnulib-tool --create-testdir --dir=/var/tmp --with-tests \ --single-configure --avoid=havelib-tests fseterr freadable fwritable \ fbufmode freading fwriting freadptr freadseek freadahead fpurge fseeko ftello fpending fflush gltests/test-suite.log looks like this. - = dummy 0: gltests/test-suite.log = # TOTAL: 118 # PASS: 116 # SKIP: 0 # XFAIL: 0 # FAIL: 2 # XPASS: 0 # ERROR: 0 .. contents:: :depth: 2 FAIL: test-fflush2.sh = test-fflush2.c:93: assertion 'c == '!'' failed Abort - core dumped FAIL test-fflush2.sh (exit status: 134) FAIL: test-sigaction test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed [snip 12150 lines] test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed test-sigaction.c:69: assertion '(sa.sa_flags & SA_SIGINFO) == 0' failed FAIL test-sigaction (exit status: 139) - Looks like test-fflush2 failure may be an ungetc issue. I'll see if I can get a Xinuos engineer to fix ungetc/ungetwc While I was working on this I spotted a mistake in the extended stdio bits on UnixWare. __fbufsiz should have been __fbufsize. I will work on getting a Xinuos engineer to fix this. And then there is the test-sigaction.c failure. Ah, more debugging to do. I am looking forward to seeing the patches make it onto the tree. Thanks for your consideration. -- Tim RiceMultitalents t...@multitalents.net2020-09-30 Tim Rice * lib/stdio-impl.h: Add support for UnixWare * lib/freadahead.c: Use __fpending on Unixware * lib/fflush.c: Update comments for UnixWare * lib/fpending.c: Likewise. * lib/freadable.c: Likewise. * lib/freadptr.c: Likewise. * lib/freadseek.c: Likewise. * lib/fseterr.c: Likewise. * lib/fwritable.c: Likewise. * lib/fwriting.c: Likewise. diff --git a/lib/stdio-impl.h b/lib/stdio-impl.h index 067b95ebd..f745a170b 100644 --- a/lib/stdio-impl.h +++ b/lib/stdio-impl.h @@ -175,7 +175,8 @@ # define fp_ fp # endif -# if defined _SCO_DS/* OpenServer */ +/* OpenServer, UnixWare (best not to use __base on these platforms) */ +# if defined _SCO_DS || defined __USLC__ || defined __sysv5__ # define _cnt __cnt # define _ptr __ptr # define _base __base diff --git a/lib/freadahead.c b/lib/freadahead.c index be14a3dab..aaa2b9b57 100644 --- a/lib/freadahead.c +++ b/lib/freadahead.c @@ -62,10 +62,14 @@ freadahead (FILE *fp) if ((fp_->_flags & _IOWRITING) != 0) return 0; return fp_->_count; -#elif defined _IOERR/* AIX, HP-UX, IRIX, OSF/1, Solaris, OpenServer, mingw, MSVC, NonStop Kernel, OpenVMS */ +#elif defined _IOERR/* AIX, HP-UX, IRIX, OSF/1, Solaris, OpenServer, mingw, MSVC, NonStop Kernel, OpenVMS, UnixWare */ if ((fp_->_flag & _IOWRT) != 0) return 0; +# if defined __USLC__ || defined __sysv5__ + return __fpending(fp_); +# else return fp_->_cnt; +# endif #elif defined __UCLIBC__/* uClibc */ # ifdef __STDIO_BUFFERS if (fp->__modeflags & __FLAG_WRITING) diff --git a/lib/fflush.c b/lib/fflush.c index b3a40e86a..4e437a623 100644 --- a/lib/fflush.c +++ b/lib/fflush.c @@ -64,7 +64,7 @@ clear_ungetc_buffer (FILE *fp) fp->_ungetc_count = 0; fp->_rcount = - fp->_rcount; } -# elif defined _
