lib: fix unity builds with BearSSL, MSH3, Quiche, OmniOS

[vszakats]

• fix MSH3 static symbol clash.
• fix Quiche static symbol clash.
• fix local macro clash with BearSSL header.
• fix local macro clash with OmniOS system header.

  In file included from ../../lib/urldata.h:197,
                     from ../../lib/altsvc.c:32,
                     from libcurlall.c:2:
    ../../lib/cf-socket.h:55:25: error: expected ':', ',', ';', '}' or '__attribute__' before '.' token
       55 | #define sa_addr _sa_ex_u.addr
          |                         ^
    In file included from ../../lib/urldata.h:197,
                     from ../../lib/altsvc.c:32,
                     from libcurlall.c:2:
    ../../lib/cf-socket.h:55:25: error: expected ':', ',', ';', '}' or '__attribute__' before '.' token
       55 | #define sa_addr _sa_ex_u.addr
          |                         ^
  

  Ref: https://github.com/curl/curl/actions/runs/10738314933/job/29781644299?pr=14772#step:3:6115

Discovered while adding support for "unity" builds for autotools.

Required-by: #14922
Cherry-picked from #14815

---

w/o whitespace: https://github.com/curl/curl/pull/14932/files?w=1

[vszakats]

Proposing it for 8.10.1.

[bagder]

fix local macro clash with OmniOS system header.

sa_addr has been defined like this since 2011 and I'm pretty sure it has been built for OmniOS many times since. How come this becomes a problem in a unity build?

[vszakats]

fix local macro clash with OmniOS system header.

sa_addr has been defined like this since 2011 and I'm pretty sure it has been built for OmniOS many times since. How come this becomes a problem in a unity build?

In "unity" all code ends up in a single source, making it catch more symbol clashes and mixups.

IOW it raises the risk for clashes when using non-namespaced local symbols because the namespace is filled with all the symbols used by libcurl, not just those pulled into an individual source file. Then, depending on the order of define/declare and use, it causes a redefinition issue, or messes up the user code that comes after the definition/declaration.

I think what happens here is that sa_addr gets defined in libcurl (via altsvc.c → urldata.h → cf-socket.h) before including an OmniOS header that happens to use this symbol, and breaks it. With non-unity this doesn't happen because the scope of the definition is more limited, and it's defined after including system headers.

sa_addr seems to be used in net/if.h in OmniOS:
https://github.com/omniosorg/illumos-omnios/blob/1a0b3ccac61c4543fa90d8e90de1a882e5491521/usr/src/uts/common/net/if.h#L415

edit: The issue would have surfaced earlier if the OmniOS job used cmake, but it happens to use autotools, so it only surfaced when unity mode arrived to autotools.

[dfandrich]

Analysis of PR #14932 at f4037249:

Test 301 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 306 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 300 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 304 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 309 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 325 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

Test 303 failed, which has NOT been flaky recently, so there could be a real issue in the PR.

There are more failures, but that's enough from Gha.

Generated by Testclutch