Implement CRL monitor for Windows

[julek-wolfssl]

• Add crl monitor unit testing
• Add a threading and conditions interface for easier porting to new OS
• wolfSSL_CTX_LoadCRL() now returns NOT_COMPILED_IN when monitoring is requested but it is not compiled in

[julek-wolfssl]

retest this please

[dgarske]

Can we avoid these 1 second sleep? Notice the testsuite/testsuite.test his this 6 times, so adds 6 seconds to the run...

[julek-wolfssl]

Adding a way to shortcut this wait. Let's try to remove the file for two seconds. If we succeed then we continue with the test.

[dgarske]

Not a fan of hard coded 2500 size.

I think it is to accommodate these:

ls -la certs/crl/crl.pem
-rw-r--r--  1 davidgarske  staff  2229 Feb  1 09:03 certs/crl/crl.pem
davidgarske@Davids-MBP-2 wolfssl % ls -la certs/crl/crl.revoked
-rw-r--r--  1 davidgarske  staff  2330 Feb  1 09:03 certs/crl/crl.revoked

Why not just use heap?

[julek-wolfssl]

There is no cross-platform way to check the file size (that I could find). I'll see if there is a better way to do this.

[julek-wolfssl]

Changed to reading/writing in 100 byte chunks. That should be fine for stack size.

[dgarske]

Wow, thread helpers!
Note: We have some API's for this already in async here. Feel free to pull over and leverage any of it: https://github.com/wolfSSL/wolfAsyncCrypt/blob/master/wolfcrypt/src/async.c#L865

[dgarske]

Can we make it VERY clear this is only required for CRL monitor? We don't normally need to port threading, but this will be very handy. I'd also like to see this for WOLFSSL_CMSIS_RTOSv2 eventually.

[julek-wolfssl]

Ideally I would like to use this threading interface in our unit tests. This will make them easier to port to new platforms and allow us to test TLS on more platforms.

I kept this threading interface extremely simple to make it easier to port. The async code make much more advanced usage of threading. I think it makes sense to keep them separate for now. Maybe in the future it would make sense to have an additional THREAD_OPTIONS parameter to allow for setting all of the advanced options being consumed by async. For now I want to keep it simple and port the tests to this common interface.

[dgarske]

Should WOLFSSL_CRL_MONITORS_LEN be overridable? Seems like the existing CRL monitor code only support .der and .pem?

[julek-wolfssl]

I didn't like having the magic number 2 all around the code. There is no benefit of having this variable be overridable by the user so I didn't guard it with a #ifndef.

[dgarske]

By the way @julek-wolfssl very nice work on this. It all tests out perfectly on Linux and Windows.