[lkml]   [1997]   [Feb]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Linux and virili (was BOUNCED something, something
On Tue, 11 Feb 1997 wrote:

> On Sun, 9 Feb 1997, W. Reilly Cooley wrote:
> This brings to mind a long standing complaint I have regarding the way the
> VGA registers are handled. In an attempt to assure security, the VGA
> device is designed so only root can alter the card's registers.
> Unfortunately, this forces folks who want or need to do direct graphics to
> make their programs suid root. Making a program suid root is, as Mr.
> Cooley points out, a huge security risk. So, what started as a feature for
> improving system security, ends up leaving it much less secure.
> What would be the risks in making the registers writable by anyone who has
> the rights to open the VGA device? The only two I can think of are 1) that
> someone might be able to peek at the contents of the console screen and 2)
> that someone could write a hostile program to block others from using the
> console. It seems to me that neither of these risks are as bad as
> necessitating the use of suid root programs.

There are big risks involved with just allowing user programs to write
directly to the VGA registers. Consider the case of an ISA- or VL-bus
board which supports linear-mapped memory. All ISA boards, and many VL
boards, map the the framebuffer at 14M. If your system has more than 14M
of physical RAM, the user program can easily crash the whole system just
by enabling the linear frame buffer.

So we can just allow a user program to write anything to VGA registers.
What we really need is to have drivers for graphics cards in the kernel.
In other words, GGI.

| Nathan Bryant | Unsolicited commercial e-mail WILL be |
| | charged an $80/hr proofreading fee. |

 \ /
  Last update: 2005-03-22 13:38    [W:0.041 / U:0.040 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site