On the previous Daf we tried to explain the opinion of Tosafos who says that according to R. Yochanan that a chatzi shiur is assur from the Torah, a shavua not to eat a half kezayis of neveila would be binding, but a shavua to eat it wouldn't be binding. The Ritva holds that even if one swears not to eat a half kezayis, it wouldn't be binding according to R. Yochanan. However, the Rambam takes the exact opposite opinion. The Rambam holds that even a shavuah to eat a half kezayis of neveila would be binding. The Shulchan Aruch (238:4) paskens like the Rambam that if one would swear to eat less than a kezayis fo neveila or treifa the shavua would be binding. It is true that Rashi 24a d.h. bishlama, also says that a shavua to eat a chatzi shiur of issur is binding (as is apparent from the gemara), but rashi may only be saying this in the opinion of reish lakish who holds that chatzi shiur is not assur m'doraysa (because the entire statement of the gemara is anyway according to reish lakish). It is therefore strange that the GR"A quotes this rashi as the source of the shulchan aruch saying that a shavua to eat a chatzi shiur of issur is binding, since we need a source according to R. Yochanan who holds chatzi shiur is assur m'doraysa, not just according to reish lakish.
The difficulty is that the Shulchan Aruch 239:6 writes that a shavua isn't even binding if one swears to violate an issur d'rabonon. How can the S.A. hold that a shavua on an issur d'rabonon isn't binding, yet a shavua on a chatzi shiur which we paksen is d'oraysa IS binding? The Shach (239:20) quotes the Maharal ben chaviv (103) who says based on the Ran that an issur d'rabonon is more stringent than a chatzi shiur. A chatzi shiur has no clear issur in the torah therefore a shavua is binding on chatzi shiur, but an issur d'rabonon is well sourced in the pasuk of lo tasur, therefore a shavua wouldn't be binding on an issur d'rabonon (although this seems very counter intuitive).
See Pischei Teshuva (239:4) where he quotes a machlokes Radvaz and Ran if one would swear to violate a cheirem d'rabbeinu gershom, such as he swears to divorce his wife against her will. Radvaz holds that it is not as strict as a d'rabonon and therefore binding, but the Ran holds that it is more severe than a d'rabonon and therefore not binding.
Another point addressed by Tosafos 24a is that the gemara seems to hold that the concept of kollel (including heter with the preexisting issur) would help to assur what was already assur, but it won't help if one uses it to swear to violate the Torah i.e. swearing to eat neveilos and shechutos. Tosafos asks that the Yerushalmi says that if one makes a shavua that he will not eat matzah the entire year, the shavua would be binding and forbid him from eating matzah on pesach. To reconcile the contradiction, Tosafos writes in the name of the R"I that if the Torah demands one to be passive and not do something i.e. not eat neveila, a shavua to eat neveila wouldn't be binding. But if the Torah demands that one be active and do something i.e. eat matzah on pesach, a shavua is can be binding to not eat matzah. The question is, Why is the shavua using kollel more binding to uproot a mitzvas aseh (which requires action), than a lo ta'aseh (which requires being passive)? Perhaps the sevara would be that when it comes to a mitzvas aseh, an outside factor can prevent me from doing it and by default it won't get done. Therefore, in regard to a shavua we view it as if he is anyway going to maintain his status quo of not eating matzah and allow it to be binding. But when it comes to a lo ta'aseh, no outside factor can force the violation of a lo ta'aseh (even if it were an o'nes, he would be patur for o'nes), therefore the default is that he will fulfill the mitzvah and not eat neveila, so the oath that he takes now cannot uproot his present status of not violating the mitzvah even using the concept of kollel (including heter with the issur).