9:30 a.m.
Compiling on Linux/i686 I get these warnings:
In file included from /usr/src/cvs/release-21-2/src/callproc.c:31:
/usr/src/cvs/release-21-2/src/process.h:143: warning: `PTY_ITERATION' redefined
s/linux.h:216: warning: this is the location of the previous definition
/usr/src/cvs/release-21-2/src/process.h:146: warning: `PTY_OPEN' redefined
s/linux.h:228: warning: this is the location of the previous definition
/usr/src/cvs/release-21-2/src/process.h:148: warning: `PTY_TTY_NAME_SPRINTF'
redefined
s/linux.h:253: warning: this is the location of the previous definition
process.h has:
#ifdef HAVE_GETPT
#define PTY_ITERATION
#define PTY_OPEN \
if ((fd = getpt()) < 0 || grantpt (fd) < 0 || unlockpt (fd) < 0) \
return -1;
#define PTY_NAME_SPRINTF
#define PTY_TTY_NAME_SPRINTF strcpy (pty_name, ptsname (fd));
#endif
Compare this with linux.h:
#define PTY_ITERATION for (i = 0; i < 1; i++)
/* no iteration at all */
#define PTY_OPEN \
do { \
fd = getpt(); \
if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
fatal ("could not set master PTY to non-block mode"); \
} while (0)
#define PTY_TTY_NAME_SPRINTF \
{ \
char *ptsname(), *ptyname; \
\
EMACS_BLOCK_SIGNAL (SIGCHLD); \
if (grantpt(fd) == -1) \
fatal("could not grant slave pty"); \
if (unlockpt(fd) == -1) \
fatal("could not unlock slave pty"); \
if (!(ptyname = ptsname(fd))) \
fatal ("could not enable slave pty"); \
strncpy(pty_name, ptyname, sizeof(pty_name)); \
pty_name[sizeof(pty_name) - 1] = 0; \
EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
}
IMO we should unify this. Could some folks please try the appended
patch? It works for me on Linux 2.4.0.../glibc 2.1.3/i686 but I'd
like to give it some more testing before it gets applied.
Andreas
2000-09-20 Andreas Jaeger <aj(a)suse.de>
* src/s/linux.h: Move PTY handling to process.h
* src/process.h: Unify PTY handling with version from s/linux.h.
Index: src/process.h
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/process.h,v
retrieving revision 1.8.2.10
diff -u -r1.8.2.10 process.h
--- src/process.h 2000/06/12 04:18:21 1.8.2.10
+++ src/process.h 2000/09/20 08:00:01
＠＠ -139,13 +139,43 ＠＠
#endif /* emacs */
-#ifdef HAVE_GETPT
+#if defined(HAVE_GRANTPT) && defined(HAVE_UNLOCKPT) &&
defined(HAVE_PTSNAME)
+/* No iteration at all: */
#define PTY_ITERATION
-#define PTY_OPEN \
- if ((fd = getpt()) < 0 || grantpt (fd) < 0 || unlockpt (fd) < 0) \
- return -1;
+
+#ifdef HAVE_GETPT
+/* Use getpt() if it's available. */
+
+#define PTY_OPEN \
+ do { \
+ fd = getpt(); \
+ if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
+ fatal ("could not set master PTY to non-block mode"); \
+ } while (0)
+
#define PTY_NAME_SPRINTF
-#define PTY_TTY_NAME_SPRINTF strcpy (pty_name, ptsname (fd));
+
+#else
+/* the master PTY device */
+#define PTY_NAME_SPRINTF strcpy (pty_name, "/dev/ptmx");
+#endif
+
+#define PTY_TTY_NAME_SPRINTF \
+ { \
+ char *ptsname(), *ptyname; \
+ \
+ EMACS_BLOCK_SIGNAL (SIGCHLD); \
+ if (grantpt(fd) == -1) \
+ fatal("could not grant slave pty"); \
+ if (unlockpt(fd) == -1) \
+ fatal("could not unlock slave pty"); \
+ if (!(ptyname = ptsname(fd))) \
+ fatal ("could not enable slave pty"); \
+ strncpy(pty_name, ptyname, sizeof(pty_name)); \
+ pty_name[sizeof(pty_name) - 1] = 0; \
+ EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
+ }
+
#endif
#endif /* INCLUDED_process_h_ */
Index: src/s/linux.h
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/s/linux.h,v
retrieving revision 1.11.2.4
diff -u -r1.11.2.4 linux.h
--- src/s/linux.h 2000/09/14 06:31:09 1.11.2.4
+++ src/s/linux.h 2000/09/20 08:00:01
＠＠ -208,48 +208,3 ＠＠
/* XEmacs: removed setpgrp() definition because we use setpgid() when
it's available, and autodetect it. */
-#if defined(HAVE_GRANTPT) && defined(HAVE_UNLOCKPT) &&
defined(HAVE_PTSNAME)
-/* UNIX98 PTYs are available.
- Added by Florian Weimer <Florian.Weimer(a)RUS.Uni-Stuttgart.DE>,
- RUS-CERT, University of Stuttgart. Based on Emacs code for DGUX. */
-
-#define PTY_ITERATION for (i = 0; i < 1; i++)
-/* no iteration at all */
-
-/* Use getpt() if it's available, because it provides Unix98 PTY
- emulation for kernels which doesn't support it natively. */
-
-#ifdef HAVE_GETPT
-#define PTY_OPEN \
- do { \
- fd = getpt(); \
- if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
- fatal ("could not set master PTY to non-block mode"); \
- } while (0)
-
-#else
-/* the master PTY device */
-#define PTY_NAME_SPRINTF strcpy (pty_name, "/dev/ptmx");
-#endif
-
-/* This sets the name of the slave side of the PTY. grantpt(3) and
- unlockpt(3) may fork a subprocess, so keep sigchld_handler() from
- intercepting that death. */
-
-#define PTY_TTY_NAME_SPRINTF \
- { \
- char *ptsname(), *ptyname; \
- \
- EMACS_BLOCK_SIGNAL (SIGCHLD); \
- if (grantpt(fd) == -1) \
- fatal("could not grant slave pty"); \
- if (unlockpt(fd) == -1) \
- fatal("could not unlock slave pty"); \
- if (!(ptyname = ptsname(fd))) \
- fatal ("could not enable slave pty"); \
- strncpy(pty_name, ptyname, sizeof(pty_name)); \
- pty_name[sizeof(pty_name) - 1] = 0; \
- EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
- }
-
-#endif
--
Andreas Jaeger
SuSE Labs aj(a)suse.de
private aj(a)arthur.inka.de
http://www.suse.de/~aj
10:02 a.m.
Please see my other emails on this subject.
If you're going to fix this (and it's on my very long todo list),
let's do it right. The grantpt, unlockpt, etc. should be in the code
for OPENing the pty, not the code for printing it!
We'll have to fix similar code in s/hpux.h and probably elsewhere.
Consider throwing most of the current code away and starting fresh
with Expect's pty_termios.c, or some other source of excellent
pty-handling code.
Martin
>>>> "AJ" == Andreas Jaeger <aj(a)suse.de>
writes:
AJ> Compiling on Linux/i686 I get these warnings:
AJ> In file included from /usr/src/cvs/release-21-2/src/callproc.c:31:
AJ> /usr/src/cvs/release-21-2/src/process.h:143: warning: `PTY_ITERATION'
redefined
AJ> s/linux.h:216: warning: this is the location of the previous definition
AJ> /usr/src/cvs/release-21-2/src/process.h:146: warning: `PTY_OPEN' redefined
AJ> s/linux.h:228: warning: this is the location of the previous definition
AJ> /usr/src/cvs/release-21-2/src/process.h:148: warning: `PTY_TTY_NAME_SPRINTF'
redefined
AJ> s/linux.h:253: warning: this is the location of the previous definition
AJ> process.h has:
AJ> #ifdef HAVE_GETPT
AJ> #define PTY_ITERATION
AJ> #define PTY_OPEN \
AJ> if ((fd = getpt()) < 0 || grantpt (fd) < 0 || unlockpt (fd) < 0) \
AJ> return -1;
AJ> #define PTY_NAME_SPRINTF
AJ> #define PTY_TTY_NAME_SPRINTF strcpy (pty_name, ptsname (fd));
AJ> #endif
AJ> Compare this with linux.h:
AJ> #define PTY_ITERATION for (i = 0; i < 1; i++)
AJ> /* no iteration at all */
AJ> #define PTY_OPEN \
AJ> do { \
AJ> fd = getpt(); \
AJ> if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
AJ> fatal ("could not set master PTY to non-block mode"); \
AJ> } while (0)
AJ> #define PTY_TTY_NAME_SPRINTF \
AJ> { \
AJ> char *ptsname(), *ptyname; \
AJ> \
AJ> EMACS_BLOCK_SIGNAL (SIGCHLD); \
AJ> if (grantpt(fd) == -1) \
AJ> fatal("could not grant slave pty"); \
AJ> if (unlockpt(fd) == -1) \
AJ> fatal("could not unlock slave pty"); \
AJ> if (!(ptyname = ptsname(fd))) \
AJ> fatal ("could not enable slave pty"); \
AJ> strncpy(pty_name, ptyname, sizeof(pty_name)); \
AJ> pty_name[sizeof(pty_name) - 1] = 0; \
AJ> EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
AJ> }
AJ> IMO we should unify this. Could some folks please try the appended
AJ> patch? It works for me on Linux 2.4.0.../glibc 2.1.3/i686 but I'd
AJ> like to give it some more testing before it gets applied.
AJ> Andreas
AJ> 2000-09-20 Andreas Jaeger <aj(a)suse.de>
AJ> * src/s/linux.h: Move PTY handling to process.h
AJ> * src/process.h: Unify PTY handling with version from s/linux.h.
AJ> Index: src/process.h
AJ> ===================================================================
AJ> RCS file: /usr/CVSroot/XEmacs/xemacs/src/process.h,v
AJ> retrieving revision 1.8.2.10
AJ> diff -u -r1.8.2.10 process.h
AJ> --- src/process.h 2000/06/12 04:18:21 1.8.2.10
AJ> +++ src/process.h 2000/09/20 08:00:01
AJ> ＠＠ -139,13 +139,43 ＠＠
AJ> #endif /* emacs */
AJ> -#ifdef HAVE_GETPT
AJ> +#if defined(HAVE_GRANTPT) && defined(HAVE_UNLOCKPT) &&
defined(HAVE_PTSNAME)
AJ> +/* No iteration at all: */
AJ> #define PTY_ITERATION
AJ> -#define PTY_OPEN \
AJ> - if ((fd = getpt()) < 0 || grantpt (fd) < 0 || unlockpt (fd) < 0) \
AJ> - return -1;
AJ> +
AJ> +#ifdef HAVE_GETPT
AJ> +/* Use getpt() if it's available. */
AJ> +
AJ> +#define PTY_OPEN \
AJ> + do { \
AJ> + fd = getpt(); \
AJ> + if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
AJ> + fatal ("could not set master PTY to non-block mode"); \
AJ> + } while (0)
AJ> +
AJ> #define PTY_NAME_SPRINTF
AJ> -#define PTY_TTY_NAME_SPRINTF strcpy (pty_name, ptsname (fd));
AJ> +
AJ> +#else
AJ> +/* the master PTY device */
AJ> +#define PTY_NAME_SPRINTF strcpy (pty_name, "/dev/ptmx");
AJ> +#endif
AJ> +
AJ> +#define PTY_TTY_NAME_SPRINTF \
AJ> + { \
AJ> + char *ptsname(), *ptyname; \
AJ> + \
AJ> + EMACS_BLOCK_SIGNAL (SIGCHLD); \
AJ> + if (grantpt(fd) == -1) \
AJ> + fatal("could not grant slave pty"); \
AJ> + if (unlockpt(fd) == -1) \
AJ> + fatal("could not unlock slave pty"); \
AJ> + if (!(ptyname = ptsname(fd))) \
AJ> + fatal ("could not enable slave pty"); \
AJ> + strncpy(pty_name, ptyname, sizeof(pty_name)); \
AJ> + pty_name[sizeof(pty_name) - 1] = 0; \
AJ> + EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
AJ> + }
AJ> +
AJ> #endif
AJ> #endif /* INCLUDED_process_h_ */
AJ> Index: src/s/linux.h
AJ> ===================================================================
AJ> RCS file: /usr/CVSroot/XEmacs/xemacs/src/s/linux.h,v
AJ> retrieving revision 1.11.2.4
AJ> diff -u -r1.11.2.4 linux.h
AJ> --- src/s/linux.h 2000/09/14 06:31:09 1.11.2.4
AJ> +++ src/s/linux.h 2000/09/20 08:00:01
AJ> ＠＠ -208,48 +208,3 ＠＠
AJ> /* XEmacs: removed setpgrp() definition because we use setpgid() when
AJ> it's available, and autodetect it. */
AJ> -#if defined(HAVE_GRANTPT) && defined(HAVE_UNLOCKPT) &&
defined(HAVE_PTSNAME)
AJ> -/* UNIX98 PTYs are available.
AJ> - Added by Florian Weimer <Florian.Weimer(a)RUS.Uni-Stuttgart.DE>,
AJ> - RUS-CERT, University of Stuttgart. Based on Emacs code for DGUX. */
AJ> -
AJ> -#define PTY_ITERATION for (i = 0; i < 1; i++)
AJ> -/* no iteration at all */
AJ> -
AJ> -/* Use getpt() if it's available, because it provides Unix98 PTY
AJ> - emulation for kernels which doesn't support it natively. */
AJ> -
AJ> -#ifdef HAVE_GETPT
AJ> -#define PTY_OPEN \
AJ> - do { \
AJ> - fd = getpt(); \
AJ> - if (fcntl (fd, F_SETFL, O_NDELAY) == -1) \
AJ> - fatal ("could not set master PTY to non-block mode"); \
AJ> - } while (0)
AJ> -
AJ> -#else
AJ> -/* the master PTY device */
AJ> -#define PTY_NAME_SPRINTF strcpy (pty_name, "/dev/ptmx");
AJ> -#endif
AJ> -
AJ> -/* This sets the name of the slave side of the PTY. grantpt(3) and
AJ> - unlockpt(3) may fork a subprocess, so keep sigchld_handler() from
AJ> - intercepting that death. */
AJ> -
AJ> -#define PTY_TTY_NAME_SPRINTF \
AJ> - { \
AJ> - char *ptsname(), *ptyname; \
AJ> - \
AJ> - EMACS_BLOCK_SIGNAL (SIGCHLD); \
AJ> - if (grantpt(fd) == -1) \
AJ> - fatal("could not grant slave pty"); \
AJ> - if (unlockpt(fd) == -1) \
AJ> - fatal("could not unlock slave pty"); \
AJ> - if (!(ptyname = ptsname(fd))) \
AJ> - fatal ("could not enable slave pty"); \
AJ> - strncpy(pty_name, ptyname, sizeof(pty_name)); \
AJ> - pty_name[sizeof(pty_name) - 1] = 0; \
AJ> - EMACS_UNBLOCK_SIGNAL (SIGCHLD); \
AJ> - }
AJ> -
AJ> -#endif
10:49 a.m.
>>>> "mb" == Martin Buchholz
<martin(a)xemacs.org> writes:
mb> Please see my other emails on this subject. If you're going
mb> to fix this (and it's on my very long todo list), let's do it
mb> right. The grantpt, unlockpt, etc. should be in the code for
mb> OPENing the pty, not the code for printing it!
mb> We'll have to fix similar code in s/hpux.h and probably
mb> elsewhere.
Also we should pay close attention to Linux/glibc-isms. The code in
question is conditioned on HAVE_GETPT, but this refers to a GNU
extension in glibc which appears to be a convenience routine. It's
quite possible that other systems will comply with the standard
(Unix9X?) and could take advantage of the apparently more careful
code in the Linux s/m file, but won't have getpt() -- or at least we
should check on that.
--
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091
_________________ _________________ _________________ _________________
What are those straight lines for? "XEmacs rules."
7:24 a.m.
I'm going to do some work on pty allocation, with the hope of doing
things more sensibly and removing warnings.
getpt() will be used where available.
Unix98 functions will be used where available.
Ideas (but little code) will be taken from Don Libes' expect.
If my new allocation function fails, I plan to fallback to the old
crufty one.
Martin
8624
days inactive
8625
days old
3 comments
3 participants
participants (3)
-
Andreas Jaeger -
Martin Buchholz -
Stephen J. Turnbull