The gemara (B.K. 94b) writes that if a parent dies and the heirs discover easily identifiable stolen goods in the parent's estate, they must return the goods to the rightful owner because of the mitzvah of kibud av. Asks the gemara, but if their parent was a rasha, a thief, there is no mitzvah of kibud av! The gemara answers that the case being spoken of is where the parent did teshuvah. Again, asks the gemara, if the parent did teshuvah, why does he/she still have stolen goods in his/her possession-- he/she should have returned them! Answers the gemara, indeed the parent intended to return the stolen property, but died before being able to do so.
How does the gemara's answer resolve the question --- bottom line, the parent did not actually return the stolen goods?! Would've, should've, could've -- but he/she didn't actually make good and restore the property to its rightful owner? What good is regret alone without making restitution?
There is a classic Brisker sevara (see Koveitz Shiurim) that postulates 2 dinim in teshuva: 1) To remove the status of rasha; 2) To rehabilitate the relationship between the person and Hashem.
An example: The Rambam implies at the beginning of Hil Teshuva that viduy is a necessary component to teshuvah. Yet, the gemara (Kid 49) tells us that kiddushin done on condition that a person is a tzadik works at least m'safeik because "shema hirheir teshuvah." Minchas Chinuch asks: How does hirhur accomplish anything without viduy? The Rogatchover answers that viduy is indeed necessary for the full rehabilitative effect of teshuvah to work, but simply to avoid the status of rasha, hirhur alone is sufficient.
Here too, so long as the parent is not categorized as a rasha, which can be accomplished by regret alone, there is a chiyuv of kibud av. Perhaps full teshuvah is lacking, but that is not the sugya's concern.
Rav Baruch Ber (see intro to Gittin) formulates the idea a little differently. (The Birchas Shmuel is cryptic; take a look at this piece in the Birchas Avraham for a clearer explanation.) Why is every lav in the Torah not considered a lav hanitak l'aseh, as each and every lav requires that a mitzvas aseh of teshuvah be done to rectify the error? R' B"B answers that there are two different "flavors" or two different types of teshuvah: 1) Teshuvah on the ma'aseh aveira, the wrongful act; 2) Teshuva on the cheit itself, the spiritual pollution that is a result of the act of aveira. The definition of a lav hanitak l'aseh is a lav that is accompanied by an aseh which undoes the ma'aseh avierah. When the Torah commands us to do teshuvah, the intent is more than that -- it is to undo the spiritual contamination that results from the aveira.
Returning to the sugya in B"K, even though the guilty parent failed to actually return the stolen property -- the undoing of the ma'aseh aveira was incomplete -- the parent still has properly fulfilled the mitzvah of teshuvah, as his/her regret over the wrongful act restores his neshoma and his/her relationship with Hashem to its prior state. The sugya is not speaking about undoing the ma'aseh aveirah; it is speaking about undoing damage to the soul.
I don't like the way the gemara reads using either sevara. If regret alone is sufficient to remove the status of rasha, what does the gemara mean by the follow-up question of why the parent still has stolen goods in his/her possession? The implication is that getting rid of the goods is part and parcel of teshuvah, not just regret. And the answer -- that the parent died before being able to make restitution -- does not seem to indicate that return of the property is not necessary, but rather it implies that even though return of the property is necessary, the fact that the parent died before being able to do so absolves him/her of resposibility. According to either of the approaches above you would have to learn the shakla v'terya as a siman, not a sibah -- regret alone removes the status of rasha, and the gemara is questioning the sincerity of that regret given that the parent failed to return the stolen goods.
On a completly different note: The Sefas Emes writes that there is a parallel between the 10 days of teshuvah, the 10 ma'amaros of creation, the 10 dibros, and the 10 makkos. I saw quoted in the name of R' Yonansan Eibshitz that we can understand very well why the first 2 days of Rosh haShana are considered kedusha achas based on the fact that the first two dibros of aseres hadibros were said together (this was a chiddush to me, but see Rashi Makos 24a "Achas dvar Elokeim...")
(If you are interested in this sort of thing R' Yosef Engel in his derashos has pages and pages which I've never managed to read through on the parallel between the aseres y'mei yeshuvah and the 10 dibros.)