There is always a change in phase of the carrier wave upon reflection. At normal incidence the phase changes by pi radians. However, most experiments are not sensitive to the phase. Here is a good animation of the process: http://www.acs.psu.edu/drussell/Demos/reflect/reflect.html

Using nanosecond (narrow bandwidth) pulses there should be negligible change in the pulse duration following a reflection off an appropriate laser mirror.

However, with few-cycle femtosecond pulses the story is much different. Femtosecond pulses have a broad bandwidth and the effect of dispersion off a single reflection can produce observable differences in amplitude, duration and pulse shape. This is especially true for reflections off dielectric mirrors, where the optical coating can introduce significant dispersion. The specific result really depends on the way the coating is produced. Some mirrors are designed with coatings to produce negative dispersion and compress the pulse shorter. Other coatings will stretch the pulse longer. Many optical companies produce zero-dispersion mirrors for femtosecond pulses that balance the dispersion over a wavelength range.

Assuming a perfectly flat surface, the duration and shape of long (~ps or longer) pulses will not change.

For few-cycle fs pulses it will depend on whether it is a metallic (and which metal) or a dielectric mirror, and how the dielectric mirror is designed. Dielectric mirrors can increase or reduce the chirp on any fs pulse.