Around 19 o'clock on Nov 2, Jim Gettys wrote:

> I had about 5 minutes to look into this this afternoon: BIG-REQUESTS
> postdates me, but it sort of looks like XDrawString doesn't implement
> it, from what I could tell in a very quick look.

That's right, none of the text requests even attempt to use BIG-REQUESTS.  
It's hard to imagine a legitimate case where a PolyText request could be 
larger than 256K -- the coordinate space is only 16 bits, so you'd either be 
placing many glyphs on top of one another or clipping many of them.

> causing the poor viewer of that page's mozilla to crash
> (by the request stream getting FUBAR'ed beyond belief).

Yeah; there's no way to recover from this error; one possibility is to 
have Xlib generate an invalid request which will cause a Length error to 
come back from the X server; that at least would have the happy result of 
Mozilla surviving.

>  Mozilla was
> worried this behavior was a potential security problem (and maybe it is,
> but I haven't thought about it....).

Shouldn't be a problem; the X server code has been pretty well reviewed 
and tested (the X test suite checks these and other cases).  It did manage 
to toast xscope...

> Whether we should bother to fix this one is less than clear, but I thought
> it should be at least thought about.

Yes, we might review the remaining Xlib functions at the same time and 
check whether they can generate illegal protocol as well.

[EMAIL PROTECTED]        XFree86 Core Team              SuSE, Inc.


_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert

Reply via email to