Bug#422712: gs-esp: pstoraster fails with signal 11 when printing cups testpage

[Jonathan Nieder <jrnieder@gmail.com>]

severity 422712 wishlist
retitle 422712 pstoraster segfaults with ancient cups
tags 422712 - moreinfo
tags 422712 + upstream
quit
(-cc: the world since I have veered off topic)

Troy Rollo wrote:

> I do not recall the precise reasoning behind this, other than that at the time 
> the patch seemed to make sense mathematically and without it there was a 
> segmentation violation
[...]
> The code patched seems to attempt to constrain "lasty" to a position within 
> the passed in "btile", vertically (that is, "btile->size.y").

That helps.  If I understand correctly then

	by = lasty % btile->size.y;
	if (by < 0)
		by += btile->size.y;

would be another way to convey the same thing.

The context is beyond my depth (and undercommented, I suppose).
wrap_shifted_cursor seems to do something complicated involving
tile_shift, so if I had to guess, the fix would be something like:

	ptc->tile.shift = btile->shift;
	by = lasty + btile->size.y;
	if (by < 0)
		by += btile->size.y;
	bx = endx;
	if (ptc->tile.shift) {
		int quotient = lasty / btile->size.y;
		if (lasty < 0)
			quotient--;
		bx += quotient * ptc->tile.shift;
	}
	bx %= tw;
	if (bx < 0)
		bx += tw;

What I had been hoping for was something like "lasty gets set using
this library call from CUPS, which used to return this out-of-range
but meaningful value but always returns nonnegative values now".
Oh, well.

Maybe it would make sense to do

	if (by < 0)
		... error out with a helpful message ...

I'm marking figuring this out as a wishlist bug for now.

Thanks again.
Jonathan