How to replicate value based on distinct column values from a different df pyspark

I have a df like:

df1 = 
AA	BB	CC	DD
1	X	Y	Z
2	M	N	O
3	P	Q	R

I have another df like:

df2 =
BB	CC	DD
G	K	O
H	L	P
I	M	Q

I want to copy all the columns and rows of df2 for every distinct value of 'AA' column of df1 and get the resultant df as:

df = 
AA	BB	CC	DD
1	X	Y	Z
1	G	K	O
1	H	L	P
1	I	M	Q
2	M	N	O
2	G	K	O
2	H	L	P
2	I	M	Q
3	P	Q	R
3	G	K	O
3	H	L	P
3	I	M	Q

What I am doing right now is:

AAs = df1.select("AA").distinct().rdd.flatMap(lambda x: x).collect()
out= []
for i in AAs:
  dff = df1.filter(col('AA')==i)
  temp_df = (df1.orderBy(rand())
        .withColumn('AA', lit(i))
        )
  out.append(temp_df)
df = reduce(DataFrame.unionAll, out)

Which is taking extremely long time and failing the cluster as these are mock dataframes, actual dataframes are quite large in dimension. Any Pysparky way of doing it? Thanks in advance.

This would work:

resultDf= df.select("AA")\
.crossJoin(df2)\
.union(df)

# No Need to order the actual result, this is just for displaying this example.
resultDf.orderBy("AA").show()

Although, this would still be a huge operation and can be expensive on the cluster.

Input

DF1:

+---+---+---+---+
| AA| BB| CC| DD|
+---+---+---+---+
|  1|  X|  Y|  Z|
|  2|  M|  N|  O|
|  3|  P|  Q|  R|
+---+---+---+---+

DF2:

+---+---+---+
| BB| CC| DD|
+---+---+---+
|  G|  K|  O|
|  H|  L|  P|
|  I|  M|  Q|
+---+---+---+

Output:

+---+---+---+---+
| AA| BB| CC| DD|
+---+---+---+---+
|  1|  G|  K|  O|
|  1|  X|  Y|  Z|
|  1|  I|  M|  Q|
|  1|  H|  L|  P|
|  2|  M|  N|  O|
|  2|  I|  M|  Q|
|  2|  H|  L|  P|
|  2|  G|  K|  O|
|  3|  P|  Q|  R|
|  3|  I|  M|  Q|
|  3|  H|  L|  P|
|  3|  G|  K|  O|
+---+---+---+---+