[Xorg-driver-geode] xf86-video-geode: Changes to 'master'

[Matthew Garrett]

On Tue, Oct 12, 2010 at 09:23:35AM +0800, Huang, FrankR wrote:

> The reason why valid modes are being pruned is due to the MODE_BAD 
> return in this function. In the following function 
> xf86PruneInvalidModes, the modes that is not MODE_OK will be deleted. 
> It is very clear. Just look at the code and you can get the answer. So 
> I don't know "introduces more bugs" means what? What bug exactly will 
> be introduced? If you take the example Ajax gave(1920x1440 160), we 
> can add code to filter out that mode in the mode_valid function.

What your patch does is:

if (something)
	return MODE_OK

if (something_else)
	return MODE_OK

if (something_misc)
	return MODE_OK

return MODE_OK

This is clearly and obviously wrong. If you're always going to return 
MODE_OK, then what are the other checks for? The rational analysis is 
that they're the bits of code that are intended to determine whether a 
mode is ok, and if they don't succeed then the mode is bad. Your problem 
is that they're failing to pass for valid modes. Figure out why not and 
you'll know how to fix this bug properly.

-- 
Matthew Garrett | mjg59 at srcf.ucam.org