Re: [CHECKER] 24 more buffer overruns in 2.5.48

[Petr Vandrovec]

On  2 Jan 03 at 19:06, Andy Chou wrote:
> > > void do_install_cmap(int con, struct fb_info *info)
> > > {
> > >     if (con != info->currcon)
> > >     return;
> > 
> > currcon can never be -1. I don't think the compiler can ever deduce that
> > detail though.

With matroxfb (and dozen of other fbdevs) currcon can be -1 (with matroxfb
currcon can be even -2 during probe, so con != info->currcon tests always
fail...) when there is no VT attached to the fbdev. That's why matroxfb
uses my_install_cmap instead of generic one.

It should be no issue in 2.5.54 though, as there is no install_cmap 
anymore...
                                                    Petr Vandrovec