I have a basic question about hashing. It is said that hashing is one way. I have a doubt that if we simply reverse the steps in program/algorithm/logic then can't we find at least one input which hashes to the given output hash value?.
I found 2 related posts, but I am still not completely clear:
How is one way hashing possible?
How do one-way hash functions work? (Edited)
I have the same question as the comment to the accepted answer in the first post:
"Well, but if I want to bypass a password check it suffices to find one string that hashes to the same value as the original password". Does this comment hold water?.
What you're thinking of is called "hash collisions".
And you're right to think, that if one could find an efficient method to determined inputs for a given hash functions that produce a desired output, this would break a lot of systems (https://en.wikipedia.org/wiki/Preimage_attack)
That's there the bones and meat of cryptographically secure hash functions come in. Those are built in a way, that it is very, very difficult to find a preimage that produces a desired hash.
Over time mathamaticians and cryptologists are chipping away on those hashes and quite a number of hash functions that were used for securing thing have been broken (MD4, MD5, SHA-1).
Also it's important to differentiate between hashes that are intended to check the integrity of messages, and hashes that are intended to protect secrets.
For integrety checking you want fast hashes, so that you can put a lot of data through them with minimal effort. MD5, SHA-1, SHA-2 are such hashes.
For secret keeping you want SLOW -er than molasses hashes, so that one can't easily brute force through dictionaries of other predicable patterns of a secret. SCrypt, BCrypt, Argon and many-round PBKDF schemes are such hashes.