Matt Ridley has a good post on his Rational Optimist blog, where he does some back-of-the-envelope sums to figure out what the impact extra CO2 would (or will) have on crop yields over the course of the century. Taking the standard estimate that CO2 levels will probably increase by 300ppm (parts per million) by 2070, he says that:

There have now been 235 studies of what happens to wheat yields when you increase CO2 levels by 300ppm.

Answer yields increase by 32.1% +/- 1.8% (SE).

There have now been 182 studies of what happens to rice yields when you increase CO2 levels by 300ppm.

Answer: yields increase by 34.4% +/-1.8%

There have now been 179 studies of what happens to soybean yields when you increase CO2 levels by 300ppm.

Answer yields increase by 46.5% +/- 2.8% (SE).

This doesn't account for the longer warm seasons we can expect, nor the technological progress that we're likely to make between now and 2070. As Ridley says, CO2 isn't always the limiting factor, but there's a lot of cause for optimism here – bigger crop yields mean cheaper food, fewer people involved in farming, and more land that can be put aside for recreational use. Points like this are reminders that we don't really know if reducing carbon emissions is worth the cost, and it's silly and dangerous for politicians to pretend otherwise.