dplyr::mutate new columns based on a second (filtered) dataframe

Question

I have 2 data frames:

df1:

a
s1
s2
s3
s4

df2:

a    b
s1    w
s1    x
s4    y
s2    z
s4    x

I want to