[9fans] rotzoomer -- another xscreensaver hack

[9fans] rotzoomer -- another xscreensaver hack

[andrey mirtchovski]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 544 bytes --]

saw this on somebody else's machine and just wanted to have it...

the best way to see it is to start in a small window and then resize to full
screen as soon as it has grabbed the screenshot.

it didn't take that long to convert all the X calls to draw ones, but much
longer to clean it up into viewable form. i'm sure i haven't cleaned
everything :)

the non-plan9 code is gnu, so have that in mind if you decide to play with
it.

andrey

ps: other xscreensaver stuff for plan9:

    http://www.acl.lanl.gov/plan9/xscreensaver

[-- Attachment #2: Type: APPLICATION/octet-stream, Size: 99848 bytes --]

Re: [9fans] rotzoomer -- another xscreensaver hack

[geoff]

> do you want me to add your modifications to www.acl.lanl.gov's
> xscreensaver archive?

Sure, that would be fine.

> i just tried your web site and the link to rotzoomer.bun appears to be
> broken...

Sorry about that; I had the link working and then broke it.  It's
fixed now.  I've also incorporated your suggested changes into my
rotzoomer bundle.

Re: [9fans] rotzoomer -- another xscreensaver hack

[rob pike, esq.]

[-- Attachment #1: Type: text/plain, Size: 324 bytes --]

On a related note, here's a screen locker. I intended to wrap it in a shell
script to watch the idle time in vgactl but bounced off a bug in an early
version of that code and never returned.  I just run it by hand.

/lib/bunny.bit is a cropped version of glenda from the web site and too
large to attach here.

-rob

[-- Attachment #2: lock.c --]
[-- Type: text/plain, Size: 5802 bytes --]

#include <u.h>
#include <libc.h>
#include <libsec.h>
#include <draw.h>

char pic[] = "/lib/bunny.bit";

int vgactl;
int debug;

uchar	storedhash[SHA1dlen];
uchar	thishash[SHA1dlen];
char user[256];
char home[256];
char hashfile[256];

void
error(char *fmt, ...)
{
	Fmt f;
	char buf[64];
	va_list arg;

	fmtfdinit(&f, 1, buf, sizeof buf);
	fmtprint(&f, "lock: ");
	va_start(arg, fmt);
	fmtvprint(&f, fmt, arg);
	va_end(arg);
	fmtprint(&f, "\n");
	fmtfdflush(&f);
	postnote(PNGROUP, getpid(), "die");
	exits("fatal error");
}

void
usage(void)
{
	fprint(2, "usage: lock [-p]\n");
	exits("usage");
}


void
readfile(char *name, char *buf, int nbuf, int addnul)
{
	int fd;

	fd = open(name, OREAD);
	if(fd == -1)
		error("%s: can't open: %r", name);
	nbuf = read(fd, buf, nbuf-addnul);
	close(fd);
	if(nbuf == -1)
		error("%s: can't can't read: %r", name);
	if(addnul)
		buf[nbuf] = '\0';
}

int
readstoredhash(int mustexist)
{
	if(access(hashfile, AEXIST) < 0){
		if(!mustexist)
			return 0;
		error("no password set; use lock -p to set password");
	}
	readfile(hashfile, (char*)storedhash, sizeof storedhash, 0);
	return 1;
}

void
readline(char *buf, int nbuf)
{
	char c;
	int i;

	i = 0;
	while(i < nbuf-1){
		if(read(0, &c, 1) != 1 || c == '\04' || c == '\177'){
			buf[0] = '\0';
			return;
		}
		if(c == '\n'){
			buf[i] = '\0';
			return;
		}
		if(c == '\b' && i > 0){
			--i;
			continue;
		}
		if(c == '\025'){
			i = 0;
			continue;
		}
		buf[i++] = c;
	}
}

void
checkpassword(int must)
{
	char buf[256];
	static int opened;
	int fd, consctl;

	if(!opened){
		fd = open("/dev/cons", OREAD);
		if(fd == -1)
			error("can't open cons: %r");
		dup(fd, 0);
		close(fd);
		fd = open("/dev/cons", OWRITE);
		if(fd == -1)
			error("can't open cons: %r");
		dup(fd, 1);
		dup(1, 2);
		close(fd);
		consctl = open("/dev/consctl", OWRITE);
		if(consctl == -1)
			error("can't open consctl: %r");
		if(write(consctl, "rawon", 5) != 5)
			error("can't turn off echo\n");
		opened = 1;
	}

	for(;;){
		fprint(2, "password: ");
		readline(buf, sizeof buf);
		fprint(2, "\n");
		if(buf[0] == '\0' || buf[0] == '\04'){
			if(must)
				continue;
			error("no password typed");
		}
		sha1((uchar*)buf, strlen(buf), thishash, nil);
		memset(buf, 0, sizeof buf);
		if(memcmp(thishash, storedhash, sizeof storedhash) == 0)
			break;
		fprint(2, "password mismatch\n");
	}
}

void
changepassword(void)
{
	int fd;
	char buf[256];

	if(readstoredhash(0))
		checkpassword(0);
	for(;;){
		fprint(2, "password: ");
		readline(buf, sizeof buf);
		fprint(2, "\n");
		if(buf[0] == '\0' || buf[0] == '\04')
			exits("no password typed");
		sha1((uchar*)buf, strlen(buf), thishash, nil);
		memset(buf, 0, sizeof buf);
		fprint(2, "re-type password: ");
		readline(buf, sizeof buf);
		fprint(2, "\n");
		if(buf[0] == '\0' || buf[0] == '\04')
			exits("no password typed");
		sha1((uchar*)buf, strlen(buf), storedhash, nil);
		memset(buf, 0, sizeof buf);
		if(memcmp(storedhash, thishash, sizeof storedhash) != 0){
			fprint(2, "password mismatch\n");
			continue;
		}

		fd = create(hashfile, OWRITE, 0600);
		if(fd < 0)
			error("can't create hashfile: %r");
		if(write(fd, storedhash, sizeof storedhash) != sizeof storedhash)
			error("error writing hashfile: %r");
		break;
	}
}

void
grabmouse(void)
{
	int fd;
	char ibuf[256], obuf[256];

	if(debug)
		return;
	fd = open("/dev/mouse", ORDWR);
	if(fd < 0)
		error("can't open /dev/mouse: %r");
	switch(fork()){
	case -1:
		error("can't fork mouse process: %r");
	default:
		return;
	case 0:
		break;
	}

	snprint(obuf, sizeof obuf, "m %d %d",
		screen->r.min.x+Dx(screen->r)/2,
		screen->r.min.y+Dy(screen->r)/2);
	while(read(fd, ibuf, sizeof ibuf) > 0)
		fprint(fd, "%s", obuf);
}

void
lockscreen(void)
{
	enum { Nfld=5, Fldlen = 12 };
	char buf[Nfld*Fldlen], *flds[Nfld], newcmd[128];
	enum { Cursorlen=2*4+2*2*16 };
	char cbuf[Cursorlen];
	int fd, dx, dy;
	Image *i;
	Rectangle r;

	fd = open("/dev/screen", OREAD);
	if(fd < 0)
		error("can't open /dev/screen: %r");
	if(read(fd, buf, Nfld*Fldlen) != Nfld*Fldlen)
		error("can't read /dev/screen: %r");
	close(fd);
	buf[sizeof buf-1] = 0;
	if(tokenize(buf, flds, Nfld) != Nfld)
		error("can't tokenize /dev/screen header");
	snprint(newcmd, sizeof newcmd, "-r %s %s %d %d",
		flds[1], flds[2],
		atoi(flds[3])-1, atoi(flds[4])-1);
	newwindow(newcmd);
	initdraw(nil, nil, "lock");

	/* screen is now open and covered. grab mouse and hold on tight */
	grabmouse();
	fd = open(pic, OREAD);
	if(fd > 0){
		i = readimage(display, fd, 0);
		if(i){
			r = screen->r;
			dx = (Dx(screen->r)-Dx(i->r))/2;
			r.min.x += dx;
			r.max.x -= dx;
			dy = (Dy(screen->r)-Dy(i->r))/2;
			r.min.y += dy;
			r.max.y -= dy;
			draw(screen, r, i, nil, i->r.min);
			flushimage(display, 1);
		}
		close(fd);
	}

	/* clear the cursor */
	fd = open("/dev/cursor", OWRITE);
	if(fd > 0){
		memset(cbuf, 0, sizeof cbuf);
		write(fd, cbuf, sizeof cbuf);
		/* leave it open */
	}
	
	fprint(vgactl, "blank");
}

void
main(int argc, char *argv[])
{
	readfile("#c/user", user, sizeof user, 1);
	readfile("#e/home", home, sizeof home, 1);
	snprint(hashfile, sizeof hashfile, "%s/lib/lockhash", home);

	vgactl = open("#v/vgactl", OWRITE);
	if(vgactl == -1)
		error("can't open vgactl: %r");

	ARGBEGIN{
	case 'd':
		debug++;
		break;
	case 'p':
		changepassword();
		exits(nil);
	default:
		usage();
	}ARGEND

	if(argc != 0)
		usage();

	readstoredhash(1);
	rfork(RFNOTEG);
	lockscreen();
	checkpassword(1);
	postnote(PNGROUP, getpid(), "die");
	exits(nil);
}

Re: [9fans] rotzoomer -- another xscreensaver hack

[Russ Cox]

> Let me put in a vote for the opposite sort of thing: using DPMS to powersave
> the monitor.

every time i have tried to do dpms by the vesa spec,
the screens never come back correctly
when i turn them back on.  the code is in
/sys/src/9/pc/vga.c:/^vgablank, but it breaks
so many setups that it's not used.

some video cards have their own ways to get at dpms
(see /sys/src/9/pc/vgasavage.c:/^savageblank, also
s3, mga6xx, nvidia) and that seems to work better.
it's still not perfect though.  mach64xx, for example,
doesn't behave right even using the mach64xx-specific
blanking.

it's just a nightmare.  the lcds tend to work better because
they have an explicit signal to do blanking rather than
infer something from the horizontal and vertical sync pulses.
both laptops and digital cards like t2r4 seem to work well.

Re: [9fans] rotzoomer -- another xscreensaver hack

[Scott Schwartz]

Russ writes:
> if you were into that sort of thing.

Let me put in a vote for the opposite sort of thing: using DPMS to powersave
the monitor.

Re: [9fans] rotzoomer -- another xscreensaver hack

[Russ Cox]

you don't need to write to the screen.  allocate a new
window as big as the screen.

Re: [9fans] rotzoomer -- another xscreensaver hack

[andrey mirtchovski]

I don't know how to write over the entire screen instead of just in a
window...

plus, I'm not sure whether rio will be forgiving enough if I sidestep it, or
whether you guys will accept modified rio sources (like adding extra
controls to rio itself)

i wouldn't mind converting all those hacks into screensavers, just never
really thought about it seriously enough..

On Tue, 21 Jan 2003, Russ Cox wrote:

> you know, you could read /dev/vgactl in a daemon
> to figure out how long the terminal has been idle
> and then invoke the screensaver automatically.
>
> if you were into that sort of thing.
>

Re: [9fans] rotzoomer -- another xscreensaver hack

[Russ Cox]

you know, you could read /dev/vgactl in a daemon 
to figure out how long the terminal has been idle
and then invoke the screensaver automatically.

if you were into that sort of thing.

Re: [9fans] rotzoomer -- another xscreensaver hack

[andrey mirtchovski]

that's how it originally operated. i didn't find it too fast on my own
machine and opted for simplicity...

do you want me to add your modifications to www.acl.lanl.gov's xscreensaver
archive?

i just tried your web site and the link to rotzoomer.bun appears to be broken...



On Tue, 21 Jan 2003 geoff@collyer.net wrote:

> I've got a sped-up version of rotzoomer at
> http://collyer.net/~geoff/9, along with faster (though still slow)
> versions of maze and mandel.  I also found a few bugs in maze by
> inspection and fixed them.
>
> I suspect that rotzoomer should build up new images in memory and push
> them out in large fractions of a window, rather than pixel by pixel.
>

Re: [9fans] rotzoomer -- another xscreensaver hack

[geoff]

I've got a sped-up version of rotzoomer at
http://collyer.net/~geoff/9, along with faster (though still slow)
versions of maze and mandel.  I also found a few bugs in maze by
inspection and fixed them.

I suspect that rotzoomer should build up new images in memory and push
them out in large fractions of a window, rather than pixel by pixel.

[9fans] rotzoomer -- another xscreensaver hack

[andrey mirtchovski]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 534 bytes --]

i saw this one on somebody else's computer and just wanted to have it :)

start it up in a small window and then resize to full screen for
complete satisfaction...

it doesn't take all that much to convert the X drawing calls to draw ones,
took me longer to clean it up into some sort of viewable form. i'm sure i
haven't managed to do so completely.

non-plan9 code is GNU, so have that in mind if you decide to play with it.

andrey

ps: other xscreensaver hacks here:

    http://www.acl.lanl.gov/plan9/xscreensaver/

[-- Attachment #2: Type: APPLICATION/octet-stream, Size: 99848 bytes --]