fix(linux): prevent infinite re-exec loop under AppArmor

[anishesg]

The MaybeReenterWithoutASLR() function in src/benchmark.cc caused infinite execv() loops when running benchmarks under AppArmor-enabled LSMs. The existing fix from #1985 only checked whether personality(ADDR_NO_RANDOMIZE) succeeded in the current process before calling execv(). However, some LSMs like AppArmor can silently reset personality flags during the execve() system call transition, even though the flag was successfully set in the parent process.

This fix implements a fork-based verification: before committing to re-exec the parent process, we fork a test child, exec it with a special argument, and have it report back whether ADDR_NO_RANDOMIZE actually survived the exec boundary. Only if the child confirms the flag persists do we proceed with re-executing the parent. This prevents infinite loops in environments where the LSM strips the personality flag during exec.

The test child is invoked with --benchmark_aslr_test_child=<fd>, checks its personality after exec, writes the result back through the provided file descriptor, and exits. This approach maintains hermeticity by using process arguments rather than environment variables.

Fixes #2184

[LebedevRI]

Thank you for taking a look!
Can you repro the original issue?

This does roughly look how i imagined it would look.
But i'm not yet confident the fix is safe.

[anishesg]

I haven't personally reproduced the original issue reported in #2184, but the reporter confirmed the infinite loop occurs on Ubuntu 24.04 with AppArmor enabled. The fork-based verification approach should be safe - if the personality doesn't survive exec, we just skip re-exec and continue normally.

[LebedevRI]

Could you please make the CI happy?

[LebedevRI]

@dmah42 thoughts?