Tosafos is looking for a case that is not a hasra'as safeik but would still be a la'av sh'ein bo ma'aseh. Tosafos explains that if one would swear to eat a loaf of bread and then throw it into the sea - it would be a definite warning since the throwing of the loaf into the sea will definitely result in the violation of the oath, but it is still a passive violation since the nature of the violation is "not eating" (although it is done through an action, it is considered passive - see steipler in kehilas yacov). Tosafos then undermines this premise and explains that this too would only be hasra'as safeik (according to the opinion who holds bitlu v'lo bitlu), because it is similar to breaking the wings of a bird by shiluach ha'kan which is considered a hasra'as safeik. The violation of the la'av is by taking the birds, but the malkus is pending on destroying the possibility of fulfilling the a'seh by breaking the wings. Warning can't be given at the time of the la'av since he may fulfill the aseh so its hara'as safeik, and also cannot be given at the time of destroying the aseh because warning must be at the time of violating the la'av.
R. Akiva Eiger (gilyon ha'shas) challenges Tosafos assumption. In the situation of shiluach hakan it is clearly a problem of hasra'as safeik - the aveira is at the time of taking the mother, not the breaking of the wings, but the breaking of the wings ruins the possibility of fulfilling the aseh. There is no point to give hasra'ah because if the warning is given prior to breaking the wings, it is not a valid warning since that isn't the time of violation. However, in the context of swearing to eat the bread, the point of violation is not when the shavuah is made, rather it is when the shavuah is violated - the moment of violation is when the loaf is thrown into the ocean by not being able to eat it - warning at this point should be considered a definite warning at the time of violation, not a hasra'as safeik?
The assumption of R. Akiva Eiger is that even though we require warning immediately proceeding the violation of the la'av, the warning at the moment of throwing the bread into the sea is considered the point of violation of the la'av. It is not clear to me why R. Akiva Eiger makes this assumption. When Rashi in Makos 16 writes that warning must be at the point of the violation of the la'av, he doesn't mean that warning must be at the moment of violation, rather that it must be when the l'av is transgressed. It is true that throwing the bread into the sea is an issur since it inevitably causes the la'av to be violated, but it is only a cause of the violation not the actual violation. For example, if he swears to eat the bread today and throws it into the sea at midday, he is doing an issur at midday but only violating the la'av at sundown (which is not possible to be precise about - ritva), therefore warning at the time of throwing into the sea can only be considered a hasra'as safeik.