Hi move,

I don't believe there's a way to detect at runtime if the max_fragment_length has been negotiated, but you are able to check the current values of the max_fragment_length extension in the ssl object by using the wolfSSL_GetMaxOutputSize API. It checks the SSL object and returns what the current max_fragment_length extension value is.

Hope this helps!

Hi move,

The two header files you're referring to aren't really there for detecting features, they're there to tell the wolfSSL library which features it should build with. You can search the files to see if features are being used or not though. Depending on the configuration of the wolfSSL library, it can use different settings files under different names.

The options.h file is generated and used when the wolfSSL library is configured and built with automake tools and the configure script. The settings.h file is used for setting OS-specific options that the wolfSSL library will use. Every application using the wolfSSL library should be including the file wolfssl/wolfcrypt/settings.h before including any other wolfSSL headers, and I see that it is being included in examples/client/client.c around line 27. Is this the client you're referring to?

If you had any other questions or information that was needed, please feel free to let me know.

Hi m0ot,

I couldn't find anything immediately wrong with the certificate or your code after a first glance, except for one thing - when you're passing the buffer to wolfSSL_CTX_load_verify_buffer, you should be passing the length of the buffer instead of the size of the pointer variable. When performing sizeof calls, if you pass a pointer variable like that it'll only return the size of the pointer and not the size of the buffer the pointer is referring to.

If changing this doesn't resolve your issue, then it may help debug the issue if you enable wolfSSL debugging, and share the error codes that are being encountered when you are attempting to use the modified example.

Hi valcroft,

Sorry for the delay in my reply. I looked at the example code you had shared and there didn't seem to be any obvious problems with it, so I looked at the certificates you had mentioned. The certificate you were defining in your code was an issued certificate for Google, which to me does not appear as if it would work, but as you're successfully connecting to the webserver and receiving the 400 response, it obviously succeeds when loaded in with trustidrootx3_chain.pem, so this may not be the issue but it is something to keep in mind.

In response to actually receiving the blank response from the webserver, the client is actually receiving something from the server - the "400 bad request" response (79 bytes long). Would you be able to pass in the "-g" option to the client application when connecting to the sandnox domain? This should resolve that "bad request"/empty response issue.

Thank you!

Hi zeta,

The Makefile is actually just an auto-generated file that has more information about the wolfSSL configure options than regular compiler options. Compiler options can be added when building however by using the "CFLAGS" variable. Additionally, using the "--enable-debug" option should perform the same function as the "-g" option. I've shown both of these options below:

./autogen.sh
./configure CFLAGS="-g"    #Option A
./configure --enable-debug #Option B

More information about building wolfSSL can be found in chapter 2 of the wolfSSL manual. Please let me know how this suggestion works out for you!

Thank you!

Hi david.ruiz,

I checked internally on some applications that could be run on some GPUs and just found out that we actually DO NOT support CUDA and OpenCL frameworks. We have had interest in the past on running wolfSSL on GPUs and so we had done a small amount of work porting wolfSSL to GPUs, but had never actually completed any ports.

Do you mind if I ask what you're interested in using wolfSSL for on a GPU?

Sorry about that misinformation before.
Thank you!

Hi kporenta,

In order to determine if this is the issue that you are truly experiencing, could you please share a test application that we are able to compile so we can reproduce the issue for testing purposes? Additionally, would you be able to share the openssl command you are using?

Thank you!

Hi zeta,

I'm glad to hear that it worked out for you! And yes, you can ask questions regarding general information about TLS here, but you will tend see best response times if you email support@wolfssl.com!

Hi Jim,

Glad to hear that it started working out for you! And just in case, if you did want to install to a different directory from the default directory that wolfSSL installs to, then all that's needed is to specify that directory while running the "configure" script. An example would look like this:

$ ./configure --prefix=/home/alex/some_directory/

Please let us know if you have any other questions at all.

Thank you!
Alex Abrahamson