fix(ext/http): avoid misaligned sockaddr_in access in SocketAddr constructor

[Shubhammehta2008]

Fixes #4307

Changes

Avoids undefined behavior caused by binding a sockaddr_in & to m_data through reinterpret_cast in the SocketAddr(u_long, uint16_t) constructor.

The fix is to declare m_data as sockaddr_storage.

For significant contributions please make sure you have completed the following items:

• CHANGELOG.md updated for non-trivial changes
• Unit tests have been added
• Changes in public API reviewed

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: Shubhammehta2008 / name: shubham mehta (0f208c6)

[marcalff]

Thanks for the patch.

Please see comments, the proper fix is to use sockaddr_storage to represent a socket in an arbitrary protocol.

Other places using m_data may need some cleanup.

[codecov]

Codecov Report

❌ Patch coverage is 91.66667% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 86.67%. Comparing base (107a1b5) to head (77c6861).

Files with missing lines	Patch %	Lines
...clude/opentelemetry/ext/http/server/socket_tools.h	91.67%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4629      +/-   ##
==========================================
+ Coverage   86.66%   86.67%   +0.01%     
==========================================
  Files         525      525              
  Lines       20543    20542       -1     
==========================================
+ Hits        17802    17803       +1     
+ Misses       2741     2739       -2     

Files with missing lines	Coverage Δ
...clude/opentelemetry/ext/http/server/socket_tools.h	95.66% <91.67%> (-0.02%)	⬇️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Shubhammehta2008]

Updated the tests to use ss_family for sockaddr_storage to resolve the Windows build failure. Please approve the workflow runs and take another look when you get a chance.

[marcalff]

Looks much better.

See comments.

[marcalff]

See remaining cleanup

[marcalff]

This is no longer true, now that m_data is aligned, but bigger than sockaddr_in.

I suspect this is what makes the test fail on MacOS, providing a length greater that sizeof(sockaddr_in).

Please implement:

• a method addr() to return &m_data
• a method addr_len() to return sizeof(sockaddr_in), since only AF_INET is supported anyway.

Use these methods when calling bind / connect, to pass the proper addr.addr() and addr.addr_len() to the networking code.

[marcalff]

See MacOS runtime failures, and possible fix.