Hi Nikolay,
I'd put it to kernel_main.c, new file is probably too much. Note that it's not really right to keep calling it semi-stub, or leaving a fixme in, because once you forward to W variant there is nothing else to do for A call. Another thing is that you could get away with static sized buffer, like e.g. OpenMutexA() does, if tests show similar name length limit.
I just remembered that the functions in namespaceapi are dependent on one another. So I'm also going to have to add CreatePrivateNamespaceA and OpenPrivateNamespaceA to kernel32.
Also, I think I need to add a new test called security for these functions. In this case, would a new file still be too much? Kernel_main says it's for kernel initialization so these functions don't seem like they belong there.
For the subject, would it be better to say it's an implementation, e.g.
kernel32: Implement CreateBoundaryDescriptorA.
Regarding the static sized buffer, I did a test on my Windows 10 machine. It does not seem to have a name length limit. But, I also tested other functions, including one that has a MAX_PATH limit according to Microsoft Docs. None of them have a limit on my machine. I'm guessing it differs depending on whether it's 32-bit or 64-bit.
But shouldn't the ANSI function check if the argument exceeds MAX_PATH before it's forwarded to the Unicode function?
In any case, given the nature of the function, MAX_PATH should be more than enough to accommodate a name. I'm guessing I should submit a test patch for CreateBoundaryDescriptorA so that the testbot can verify if it doesn't have a limit on other configurations.
I've attached the test. It has some extra traces that I added temporarily. I'll remove them when I submit the patch and of course create a new file. Also, please let me know if the test is incorrect.
Another thing I forgot to mention. I personally find documentation headers before functions quite useless, when they don't describe anything more than argument types that you can already see right below.
Okay, I'll remove the documentation.
-- Kind regards, Mohamad