The gemara says that there in the development of birds there is a stage where the feathers begin begin to get yellow but aren't yet golden which is excluded by the pasuk because it is too old to qualify as bnei yona, but too young to qualify as torim. The gemara asks whether we consider the birds at that stage to have a separate status and would therefore not qualify as being part of either group, or whether it is a safeik so it cannot be sacrificed as torim because it may be too young, and cannnot be sacrificed as b'nei yona because it may be too old. According to this second approach that it is actually one or the other, just that we have no way to determine, Tosafos points out that if the kohein would sacrifice both, it would certainly be effective to fulfill the persons neder (although the other would be a violation of chulin b'azara). The gemara responds that this second option is not possible because if it is a safeik, "would we need a pasuk to exclude a safeik!" The Turei Even in Chagiga points out that it would seem that this is a proof to the concept that safeik d'oraysa is l'kula on a Torah level, because if the Torah would demand being machmir for a safeik, it would make sense that we would need a pasuk to exclude the safeik. However, this proof can be easily pushed off based on the interpretation of what the gemara means to say that we don't need a pasuk to exclude a safeik. Tosafos explains that in our gemara, the explanation could simply be that it wouldn't make sense that the Torah would exclude it mi'safeik. Meaning, that it is entirely possible that we would need a pasuk to exclude a safeik case and say that it is definitely not included, but here the gemara is suggesting that even after it is excluded by the pasuk it remains a safeik. Therefore, the gemara responds that we never find that a pasuk would exclude something and have it remain a safeik even after being excluded.
Tosafos points out that this approach although it would work in our sugya, wouldn't work in various other places where the gemara uses the same phrase. Therefore, Tosafos suggests that there are different types of sfei'kos. When the safeik is based on the metzius of the object, such as a tumtum where we don't know if he is male or female, but each one is different, the pasuk could exclude a safeik. However, when the safeik is a safeik in din such as an androginus where they are all the same, just that we don't know what the din is on such a person, the pasuk wouldn't exlcude a safeik. The rationale for this distinction is that the pasuk may come and tell me what to do in a situation that I can't clarify the facts, but when there is a safeik in din, the Torah wouldn't ever deal with that as a safeik since the Torah doesn't have a safeik in din - קמי שמיא גליא אם הם קטנים או גדולים וכו הלכך לא אתא גרא למעוטי מטעם שהם ספק שאינו ספקא קמי שמיא
According to both of these approaches, there would be no proof at all to the concept of safeik d'oraysa because the gemara is not saying that we don't need a pasuk to exclude a safeik d'oraysa, Rather, the gemara is saying that although we may need a pasuk to exclude a safeik d'oraysa, that cannot be what the Torah is doing. However, Rashi seems to use the language of Tosafos - מי איכא ספקא קמי קודשא בריך הוא, but ends with a strange phrase - וכיון דלא פשיטא לן לא הא ולא הא, מהיכא תיסק אדעתין לאכשורינהו. This seems to indicate that since the Torah isn't dealing with the safeik and isn't including them into either category, we are forced to exclude them from both categories mi'safeik. This would seem to imply the opposite of the Turei Even's suggestion and that since the Torah doesn't include them, we are forced to exclude them mi'safeik, which would indicate ספק דאורייתא לחומרא.