I did the exercises (last weekend actually).
Didn't take me long, it is more about confirming that you understood the principles. For the 3rd one I had to do some re-reading, though, as I had forgotten some details.
I'll share my solutions here:
2.1:
► Show Spoiler
In the first table, I put a + in the middle of the 2nd row, in the second table, in the 1st row left and in the 3rd row right.
Regarding the loan words, Sumerian {b}, {d}, {g} and Akkadian {p}, {t}, {k}, respectively, are the same phonemes and therefore correspond to each other.
Akkadian does not have phonemic aspiration, so it treats Sumerian aspirated and unaspirated alike as simple voiceless.
Sumerian does not have phonemic voicing or emphasis, so it treats any Akkadian plosive as unaspirated voiceless.
2.2:
► Show Spoiler
I didn't bother reading up on Akkadian phonology (I know pretty much nothing about it), so my solution might be wrong.
But, Sumerian words are stressed on the final syllable, and the loan words into Akkadian have added a syllable afterwards, so I would assume that this is because Akkadian normally stresses the penultimate.
2.3:
► Show Spoiler
There is actually not enough information given to fully solve this one.
In a) the 3rd and 5th word can certainly take that allomorph since a is always a low vowel. The 1st and 4th one however contain an u, and according to the lesson, it depends on the word in question whether that's a high or a low u (the same with e, but there is no example with e in the exercise). The 2nd one contains an i, which is always high, so it would take the other allomorph of the prefix.
In b), the 1st is fitting, the 2nd and 5th perhaps, the rest not.
In c), the 2nd and 5th are fitting, the 1st, 2nd and 6th perhaps, the 4th not.
In d), the 1st, 3rd and 5th are fitting, the 2nd perhaps, the 4th not.