How do I do error analysis with a least-squares fit that puts error bars in the fit?

[Simfish]

With MATLAB or something. Basically, I just have a bunch of data points where I should do a linear least squares fit, but each of the points have error bars around them.

[I like Serena]

With MATLAB or something. Basically, I just have a bunch of data points where I should do a linear least squares fit, but each of the points have error bars around them.

What you would need is a "weighted least squares fit".
The points must be weighted by 1/variance.

According to google:
In MATLAB, the LSCOV function can perform weighted-least-square regression.