Re: my review of the Device Driver Hardening Design Spec

[Greg KH]

On Sat, Sep 21, 2002 at 11:52:19PM +0200, Francois Romieu wrote:
> [Cc list trimmed]
> 
> Greg KH <[e-mail]> :
> [...]
> > Oh, there's lots of code:
> > 	A "hardened" binary kernel driver:
> > 		http://unc.dl.sourceforge.net/sourceforge/hardeneddrivers/sampledriver-0.1-1.i386.rpm
> > 	(um people, why a binary?  Where's the source for this?)
> 
> In the cvs. See:
> http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/hardeneddrivers/sample_driver/src/

Thanks for pointing this out, I missed it.

Hm, if this is the code that the CG group is proposing for reliable
drivers, we are all in trouble.  See:
	http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/*checkout*/hardeneddrivers/sample_driver/src/sampledriver.h?rev=1.1.1.1

as a very small example of what not to do :)

I'd be glad to provide concrete criticism of the other files in this
directory, if I thought people would actually change their programming
style to follow what their own spec says to do...

{sigh}

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/*checkout*/hardeneddrivers/sample_driver/src/sampledriver_init.c?rev=1.1.1.1
contains so many examples of bad style, and real bugs...

greg k-h