net: Move NET_TCP/UDP_HAVE_STACK to netconfig.h

[wengzhe]

Summary

Now (after #14634 and #14650) the HAVE_PFINET(6)_SOCKETS depends on NET_TCP/UDP_HAVE_STACK, which is previously defined in net/ folder and cannot be included. Considering many places use this check, maybe moving them to netconfig.h could be better.

Impact

Before: IPv4 + TCP, without ICMP -> HAVE_PFINET_SOCKETS is undefined, we cannot use TCP socket (error is EAFNOSUPPORT)
After: IPv4 + TCP, without ICMP -> HAVE_PFINET_SOCKETS is defined, we can use TCP socket

Testing

QEMU with CONFIG_NET_IPv4 & CONFIG_NET_TCP enabled but CONFIG_NET_ICMP_SOCKET disabled.

[nuttxpr]

[Experimental Bot, please feedback here]

This PR does not fully meet the NuttX requirements. Here's why and how to fix it:

Missing Information:

• Summary: While the summary explains the "why", it lacks clarity on the "what" and "how". Specifically:

  • What functional part is changed? Clearly state which files are modified (e.g., netconfig.h).
  • How does the change work? Instead of describing the before/after behavior, explain how moving the definitions to netconfig.h solves the problem. For example, "This change moves the definitions of HAVE_PFINET(6)_SOCKETS to netconfig.h, making them accessible to code that previously couldn't include the necessary headers from the net/ directory."

• Impact: Several sections need more detail:

  • Is new feature added? Is existing feature changed? Be explicit. This is likely an "existing feature changed" due to a bug fix.
  • Impact on user: While the before/after behavior is described, clarify how a user might observe this change. For example, "Applications that previously failed to create TCP sockets when ICMP was disabled will now function correctly."
  • Impact on build: If only header files are changed, the build process itself is unlikely to change significantly. Clarify this.
  • Impact on hardware: Be explicit. Likely "NO".
  • Impact on documentation: If the behavior changed, documentation might need updating. Even if no changes are needed, explicitly state "NO".
  • Impact on security: Be explicit. Likely "NO".
  • Impact on compatibility: This is the most important part needing clarification. Does this change break anything? Is it backwards compatible? Explain thoroughly.
  • Anything else to consider: Be explicit. Likely "NO".

• Testing:

  • Build Host(s): Provide details about your build host OS, CPU architecture, and compiler version.
  • Target(s): Specify the QEMU architecture (e.g., qemu-system-x86_64). Also, provide the full NuttX configuration name (e.g., nsh).
  • Testing logs: The logs are missing entirely. Include actual output demonstrating the failure before the change and the success after the change. Show commands used and their output. For example:

Before (without the patch):
$ make ARCH=sim BOARD=qemu-i486 CONFIG=nsh
... build output ...
$ nsh
nsh> netstat -a
... output showing no TCP socket available ...
nsh> telnet google.com 80
EAFNOSUPPORT

After (with the patch):
$ make ARCH=sim BOARD=qemu-i486 CONFIG=nsh
... build output ...
$ nsh
nsh> netstat -a
... output showing TCP socket available ...
nsh> telnet google.com 80
Connected to google.com.
Escape character is '^]'.

By providing the missing information and clarifying the existing details, the PR will better meet the NuttX requirements and be easier for reviewers to evaluate.