httpauth: http digest challenge request using RFC 7616

[cHuberCoffee]

No description provided.

[cHuberCoffee]

It is the first of my PR regarding HTTP digest auth. The upcoming one will handle the response creation and the verification steps.

[sreimers]

Unsure why there is a segfault for mingw now (the stack trace looks broken after this):

13: test_httpauth_digest_request (\home\runner\work\re\re\baresip-win32\re\test\httpauth.c:446)

Nothing obvious to me, at first look.

[cHuberCoffee]

Currently i'm setting up the mingw on my local machine to run the tests.

[cspiel1]

I suggest to call this function:
httpauth_digest_chall_request_full()
and the short form
httpauth_digest_chall_request().

Don't forget to update the doxygen!

[sreimers]

Maybe simply use uint8_t hash[SHA256_DIGEST_LENGTH]; I see no need for heap usage here.

[cHuberCoffee]

The hashlen variable was the problematic piece. Using the hashlen variable in mbuf_printf(mb, "%w", hash, hashlen) produced a wrong behavior inside the printing function re_vhprintf. Changing the hashlen variable to size_t (required by %w) it works fine. Also the cast of time_t is now a fixed size integer type uint64_t with its corresponding PRIxx macro.

I can't tell why this phenomena is only observed if it was cross-compiled in the workflow. I cross-compiled it on my local machine and executed the binary on an windows machine without this behavior. In case i downloaded the binary from the workflow, i got the same result as the workflow itself.

[sreimers]

Nice, but there is no need for hash heap usage, you can use a stack hash array variable for better performance, since its only 32 bytes (no need for mem_zalloc). So you can use simply sizeof(hash) instead of hashlen.