[kbd] New vt_mode tool for kbd?

[kbd] New vt_mode tool for kbd?

[Michael Schutte]

[-- Attachment #1.1: Type: text/plain, Size: 858 bytes --]

Hi Alexey,

The console-setup package in Debian needs a way to check for the
text/graphics mode of the active VT (from a shell script).  This can be
done with “setfont -v -o /dev/null”, for example, but that isn’t very
elegant.  Julien Cristau has proposed a small utility (which I’m calling
“vt_mode” for now) to report this information in a canonical fashion.

I’ve adapted it to kbd; see attachment.  Do you think this code is
suitable for inclusion?  If so, I am going to integrate it into the
build system and write a short man page for it, so I can send you a
proper patch.  I’d like to hear your opinion!

All the best,
-- 
Michael Schutte                   | michi@{uiae.at,debian.org}
Innsbruck, Austria                | happily accepting encrypted mail
OpenPGP: 0x16fb 517b a866 c3f6 8f11 1485 f3e4 122f 1D8C 261A

[-- Attachment #1.2: vt_mode.c --]
[-- Type: text/x-csrc, Size: 1902 bytes --]

/*
 * vt_mode: report VT mode (text or graphics)
 *
 * Contributed by Julien Cristau <jcristau@debian.org> in
 *   http://bugs.debian.org/618573;
 * adapted to kbd by Michael Schutte <michi@debian.org>.
 */

#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <linux/kd.h>
#include <getopt.h>
#include "getfd.h"
#include "nls.h"
#include "version.h"

static void attr_noreturn
usage(int code) {
        fprintf(stderr, _("Usage: vt_mode [-C DEVICE]\n"));
        exit(code);
}

int
main(int argc, char *argv[]) {
        int fd, rc, mode, c;
        char *console = NULL;

        set_progname(argv[0]);
#ifndef __klibc__
        setlocale(LC_ALL, "");
        bindtextdomain(PACKAGE_NAME, LOCALEDIR);
        textdomain(PACKAGE_NAME);
#endif

        while ((c = getopt(argc, argv, "C:hV")) != EOF) {
                switch (c) {
                case 'C':
                        if (optarg == NULL || optarg[0] == '\0')
                                usage(1);
                        console = optarg;
                        break;
                case 'h':
                        usage(0);
                        break;
                case 'V':
                        print_version_and_exit();
                        break;
                default:
                        usage(1);
                }
        }

        fd = getfd(console);

        rc = ioctl(fd, KDGETMODE, &mode);
        if (rc < 0) {
                fprintf(stderr, _("%s: Error reading VT mode.\n"), progname);
                return 1;
        }

        switch (mode) {
        case KD_TEXT:
                printf("text\n");
                break;
        case KD_GRAPHICS:
                printf("graphics\n");
                break;
        default:
                printf("other (%d)\n", mode);
                break;
        }

        return 0;
}

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Dmitry V. Levin]

[-- Attachment #1: Type: text/plain, Size: 913 bytes --]

On Mon, Mar 21, 2011 at 10:42:36PM +0100, Michael Schutte wrote:
> Hi Alexey,
> 
> The console-setup package in Debian needs a way to check for the
> text/graphics mode of the active VT (from a shell script).  This can be
> done with “setfont -v -o /dev/null”, for example, but that isn’t very
> elegant.  Julien Cristau has proposed a small utility (which I’m calling
> “vt_mode” for now) to report this information in a canonical fashion.

I wonder whether you actually need to know what's the KD_ mode, or
maybe your scripts just need yes/no answers on simple two questions:
- whether the active VT is in the text mode?
- whether the active VT is in the graphics mode?

In the latter case, I suggest to implement an utility with an option or
argument specifying what mode to query, and a tty(1)-like exit status,
so its use would be as simple as "if vt_mode text; then ...; fi".


-- 
ldv

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Michael Schutte]

[-- Attachment #1: Type: text/plain, Size: 1517 bytes --]

Hey,

On Tue, Mar 22, 2011 at 01:13:54AM +0300, Dmitry V. Levin wrote:
> On Mon, Mar 21, 2011 at 10:42:36PM +0100, Michael Schutte wrote:
> > The console-setup package in Debian needs a way to check for the
> > text/graphics mode of the active VT (from a shell script).  This can be
> > done with “setfont -v -o /dev/null”, for example, but that isn’t very
> > elegant.  Julien Cristau has proposed a small utility (which I’m calling
> > “vt_mode” for now) to report this information in a canonical fashion.
> 
> I wonder whether you actually need to know what's the KD_ mode, or
> maybe your scripts just need yes/no answers on simple two questions:
> - whether the active VT is in the text mode?
> - whether the active VT is in the graphics mode?
> 
> In the latter case, I suggest to implement an utility with an option or
> argument specifying what mode to query, and a tty(1)-like exit status,
> so its use would be as simple as "if vt_mode text; then ...; fi".

Yes, that would certainly work as well; it could even do both things,
depending on the presence of an argument.  I don’t really care either
way (it isn’t even my script, I’m just relaying for the console-setup
developers).  I just figured it would make sense to stick to the way
kbd_mode works.

Cheers,
-- 
Michael Schutte                   | michi@{uiae.at,debian.org}
Innsbruck, Austria                | happily accepting encrypted mail
OpenPGP: 0x16fb 517b a866 c3f6 8f11 1485 f3e4 122f 1D8C 261A

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

[-- Attachment #1: Type: text/plain, Size: 563 bytes --]

22.03.2011 00:42, Michael Schutte wrote:
> Hi Alexey,
> 
> The console-setup package in Debian needs a way to check for the
> text/graphics mode of the active VT (from a shell script).  This can be
> done with “setfont -v -o /dev/null”, for example, but that isn’t very
> elegant.  Julien Cristau has proposed a small utility (which I’m calling
> “vt_mode” for now) to report this information in a canonical fashion.

May be something like this (see attachment) ?

I thing this utility is more script-friendly and more extensible.

-- 
Rgrds, legion


[-- Attachment #2: kbdinfo.c --]
[-- Type: text/plain, Size: 2187 bytes --]

#include <stdio.h>
#include <errno.h>
#include <error.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <linux/kd.h>
#include <getopt.h>
#include "getfd.h"
#include "nls.h"
#include "version.h"

#define RETURN_BASE 100
#define GETMODE     10
#define GKBMODE     20
#define GKBMETA     30

static void attr_noreturn
usage(int code) {
	fprintf(stderr, _("Usage: kbdinfo [-C DEVICE] GETMODE | GKBMODE | GKBMETA\n"));
	exit(code);
}

int
main(int argc, char *argv[]) {
	int fd, mode, c, rc;
	int quiet = 0;
	char *console = NULL;

	set_progname(argv[0]);

	setlocale(LC_ALL, "");
	bindtextdomain(PACKAGE_NAME, LOCALEDIR);
	textdomain(PACKAGE_NAME);

	while ((c = getopt(argc, argv, "C:qhV")) != EOF) {
		switch (c) {
		case 'C':
			if (optarg == NULL || optarg[0] == '\0')
				usage(1);
			console = optarg;
			break;
		case 'q':
			quiet = 1;
			break;
		case 'V':
			print_version_and_exit();
			break;
		case 'h':
			usage(0);
			break;
		default:
			usage(1);
		}
	}

	if (argc == optind)
		usage(1);

	fd = getfd(console);

	rc = EXIT_FAILURE;

#define answer(typ,text,code) { \
	if (!quiet) \
		printf("%s\n", text); \
	rc = (RETURN_BASE + typ + code); \
};

	if (!strcasecmp("GETMODE", argv[optind])) {
		if (ioctl(fd, KDGETMODE, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
		case KD_TEXT:		answer(GETMODE, "text",     1);	break;
		case KD_GRAPHICS:	answer(GETMODE, "graphics", 2); break;
		}
		return rc;
	}

	if (!strcasecmp("GKBMODE", argv[optind])) {
		if (ioctl(fd, KDGKBMODE, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
		case K_RAW:		answer(GKBMODE, "raw",       1); break;
		case K_XLATE:		answer(GKBMODE, "xlate",     2); break;
		case K_MEDIUMRAW:	answer(GKBMODE, "mediumraw", 3); break;
		case K_UNICODE:		answer(GKBMODE, "unicode",   4); break;
		}
		return rc;
	}

	if (!strcasecmp("GKBMETA", argv[optind])) {
		if (ioctl(fd, KDGKBMETA, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
		case K_METABIT:		answer(GKBMETA, "metabit",   1); break;
		case K_ESCPREFIX:	answer(GKBMETA, "escprefix", 2); break;
		}
		return rc;
	}

	return rc;
}

Re: [kbd] New vt_mode tool for kbd?

[Michael Schutte]

[-- Attachment #1: Type: text/plain, Size: 870 bytes --]

On Fri, Mar 25, 2011 at 04:20:19PM +0300, Alexey Gladkov wrote:
> 22.03.2011 00:42, Michael Schutte wrote:
> > The console-setup package in Debian needs a way to check for the
> > text/graphics mode of the active VT (from a shell script).  This can be
> > done with “setfont -v -o /dev/null”, for example, but that isn’t very
> > elegant.  Julien Cristau has proposed a small utility (which I’m calling
> > “vt_mode” for now) to report this information in a canonical fashion.
> 
> May be something like this (see attachment) ?
> 
> I thing this utility is more script-friendly and more extensible.

That’s much better, of course!

Cheers,
-- 
Michael Schutte                   | michi@{uiae.at,debian.org}
Innsbruck, Austria                | happily accepting encrypted mail
OpenPGP: 0x16fb 517b a866 c3f6 8f11 1485 f3e4 122f 1D8C 261A

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Dmitry V. Levin]

[-- Attachment #1: Type: text/plain, Size: 479 bytes --]

On Fri, Mar 25, 2011 at 04:20:19PM +0300, Alexey Gladkov wrote:
[...]
> I thing this utility is more script-friendly and more extensible.
[...]
> #define RETURN_BASE 100
> #define GETMODE     10
> #define GKBMODE     20
> #define GKBMETA     30

Multi-value exit status is not quite script-friendly, especially for
values greater than 128, because exit codes greater than 128 are used
by the shell to indicate that a program died due to a fatal signal.


-- 
ldv

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

27.03.2011 01:33, Dmitry V. Levin wrote:
> Multi-value exit status is not quite script-friendly, especially for
> values greater than 128, because exit codes greater than 128 are used
> by the shell to indicate that a program died due to a fatal signal.

I do not really like the idea of ​​creating for each ioctl by the utility.

Does anyone have any suggestions on how combine all the requests in a
single utility?

On the other hand 127 return codes are sufficient to describe all the
values​​.

-- 
Rgrds, legion

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

27.03.2011 13:33, Alexey Gladkov wrote:
> On the other hand 127 return codes are sufficient to describe all the
> values​​.

According this:

http://tldp.org/LDP/abs/html/exitcodes.html

we have 123 non-reserved exit codes.

-- 
Rgrds, legion

Re: [kbd] New vt_mode tool for kbd?

[Dmitry V. Levin]

[-- Attachment #1: Type: text/plain, Size: 1035 bytes --]

On Sun, Mar 27, 2011 at 01:33:23PM +0400, Alexey Gladkov wrote:
> 27.03.2011 01:33, Dmitry V. Levin wrote:
> > Multi-value exit status is not quite script-friendly, especially for
> > values greater than 128, because exit codes greater than 128 are used
> > by the shell to indicate that a program died due to a fatal signal.
> 
> I do not really like the idea of ​​creating for each ioctl by the utility.
> 
> Does anyone have any suggestions on how combine all the requests in a
> single utility?

Usage: kbdinfo [-C DEVICE] getmode [text|graphics]
       kbdinfo [-C DEVICE] gkbmode [raw|xlate|mediumraw|unicode]
       kbdinfo [-C DEVICE] gkbmeta [metabit|escprefix]

If only one argument is given, then just print the current mode to stdout
and exit 0 (unless ioctls returned an error).

If two arguments are given, then print nothing to stdout and exit 0 iff
the current mode matches the mode described by these two arguments.

This interface would be both script-friendly and extensible.


-- 
ldv

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

28.03.2011 15:19, Dmitry V. Levin wrote:
> Usage: kbdinfo [-C DEVICE] getmode [text|graphics]
>        kbdinfo [-C DEVICE] gkbmode [raw|xlate|mediumraw|unicode]
>        kbdinfo [-C DEVICE] gkbmeta [metabit|escprefix]
> 
> If only one argument is given, then just print the current mode to stdout
> and exit 0 (unless ioctls returned an error).
> 
> If two arguments are given, then print nothing to stdout and exit 0 iff
> the current mode matches the mode described by these two arguments.
> 
> This interface would be both script-friendly and extensible.

This is very good idea. Thanks!

-- 
Rgrds, legion

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

[-- Attachment #1: Type: text/plain, Size: 170 bytes --]

28.03.2011 15:19, Dmitry V. Levin wrote:
> This interface would be both script-friendly and extensible.

Ok. Another implementation. How about this ?

-- 
Rgrds, legion


[-- Attachment #2: kbdinfo.c --]
[-- Type: text/plain, Size: 3121 bytes --]

#include <stdio.h>
#include <errno.h>
#include <error.h>
#include <unistd.h>
#include <sys/ioctl.h>
#include <linux/kd.h>
#include <getopt.h>
#include "getfd.h"
#include "nls.h"
#include "version.h"

static char *action = NULL;
static char *value  = NULL;

static void attr_noreturn
usage(int code) {
	fprintf(stderr, _("Usage: %s [-C DEVICE] getmode [text|graphics]\n"), progname);
	fprintf(stderr, _("   or: %s [-C DEVICE] gkbmode [raw|xlate|mediumraw|unicode]\n"), progname);
	fprintf(stderr, _("   or: %s [-C DEVICE] gkbmeta [metabit|escprefix]\n"), progname);
	fprintf(stderr, _("   or: %s [-C DEVICE] gkbled  [scrolllock|numlock|capslock]\n"), progname);
	exit(code);
}

static int
answer(char *ans) {
	if (value)
		return !strcasecmp(value, ans) ? EXIT_SUCCESS : EXIT_FAILURE;

	printf("%s\n", ans);
	return EXIT_SUCCESS;
}

int
main(int argc, char **argv) {
	int fd, mode, c;
	int rc = EXIT_FAILURE;
	char flags;
	char *console = NULL;

	set_progname(argv[0]);

	setlocale(LC_ALL, "");
	bindtextdomain(PACKAGE_NAME, LOCALEDIR);
	textdomain(PACKAGE_NAME);

	while ((c = getopt(argc, argv, "C:hV")) != EOF) {
		switch (c) {
			case 'C':
				if (optarg == NULL || optarg[0] == '\0')
					usage(EXIT_FAILURE);
				console = optarg;
				break;
			case 'V':
				print_version_and_exit();
				break;
			case 'h':
				usage(EXIT_SUCCESS);
				break;
		}
	}

	if (optind == argc) {
		fprintf(stderr, _("Error: Not enough arguments.\n"));
		exit(EXIT_FAILURE);
	}

	action = argv[optind++];

	if (optind < argc)
		value = argv[optind++];

	fd = getfd(console);

	if (!strcasecmp("GETMODE", action)) {
		if (ioctl(fd, KDGETMODE, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
			case KD_TEXT:		rc = answer("text");		break;
			case KD_GRAPHICS:	rc = answer("graphics");	break;
		}

	} else if (!strcasecmp("GKBMODE", action)) {
		if (ioctl(fd, KDGKBMODE, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
			case K_RAW:		rc = answer("raw");		break;
			case K_XLATE:		rc = answer("xlate");		break;
			case K_MEDIUMRAW:	rc = answer("mediumraw");	break;
			case K_UNICODE:		rc = answer("unicode");		break;
		}

	} else if (!strcasecmp("GKBMETA", action)) {
		if (ioctl(fd, KDGKBMETA, &mode) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		switch (mode) {
			case K_METABIT:		rc = answer("metabit");		break;
			case K_ESCPREFIX:	rc = answer("escprefix");	break;
		}

	} else if (!strcasecmp("GKBLED", action)) {
		if (ioctl(fd, KDGKBLED, &flags) == -1)
			error(EXIT_FAILURE, errno, "ioctl");

		mode = (flags & 0x7);

		if (value) {
			if (((mode & LED_SCR) && !strcasecmp(value, "scrolllock")) ||
			    ((mode & LED_NUM) && !strcasecmp(value, "numlock"))    ||
			    ((mode & LED_CAP) && !strcasecmp(value, "capslock")))
				rc = EXIT_SUCCESS;
		} else {
			printf("scrolllock:%s ", (mode & LED_SCR) ? "on" : "off");
			printf("numlock:%s ",    (mode & LED_NUM) ? "on" : "off");
			printf("capslock:%s\n",  (mode & LED_CAP) ? "on" : "off");
			rc = EXIT_SUCCESS;
		}

	} else {
		fprintf(stderr, _("Error: Unrecognized action: %s\n"), action);
	}

	close(fd);
	return rc;
}

Re: [kbd] New vt_mode tool for kbd?

[Dmitry V. Levin]

[-- Attachment #1: Type: text/plain, Size: 552 bytes --]

On Tue, Mar 29, 2011 at 01:07:28AM +0400, Alexey Gladkov wrote:
> 28.03.2011 15:19, Dmitry V. Levin wrote:
> > This interface would be both script-friendly and extensible.
> 
> Ok. Another implementation. How about this ?

Fine with me.

> static int
> answer(char *ans) {

I'd change this and some others "char *" to "const char *", but it's not
important.

> 		return !strcasecmp(value, ans) ? EXIT_SUCCESS : EXIT_FAILURE;

I'd code it a bit differently:
		return strcasecmp(value, ans) ? EXIT_FAILURE : EXIT_SUCCESS;


-- 
ldv

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [kbd] New vt_mode tool for kbd?

[Alexey Gladkov]

30.03.2011 01:51, Dmitry V. Levin wrote:
> Fine with me.

http://git.altlinux.org/people/legion/packages/kbd.git?p=kbd.git;a=commitdiff;h=4578bbc6

Applied. Thanks, Dmitry for your advices!

-- 
Rgrds, legion