zlib: fix pointer alignment

[jhofstee]

The function AllocForBrotli prefixes the allocated memory with its size, and returns a pointer to the region after it. This pointer can however no longer be suitably aligned. Correct this by allocating the maximum of the the size of the size_t and the max alignment.

On Arm 32bits the size_t is 4 bytes long, but the alignment is 8 for some NEON instructions. When Brotli is compiled with optimizations enabled newer GCC versions will use the NEON instructions and trigger a bus error killing node.

see google/brotli#1159

I don't think there is any additional test needed, since existing test will crash node already.
Not sure about the notable-change label.

[targos]

/cc @nodejs/cpp-reviewers

I can't review this, but I'm very interested as it might fix crashes that we have with the latest V8 update: https://ci.nodejs.org/job/node-test-commit-arm/57862/

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.24%. Comparing base (1c2d98d) to head (152058a).
Report is 85 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #57727      +/-   ##
==========================================
+ Coverage   90.22%   90.24%   +0.01%     
==========================================
  Files         630      630              
  Lines      185073   185075       +2     
  Branches    36222    36222              
==========================================
+ Hits       166990   167017      +27     
+ Misses      11044    11034      -10     
+ Partials     7039     7024      -15     

Files with missing lines	Coverage Δ
src/node_zlib.cc	78.80% <100.00%> (+0.43%)	⬆️

... and 24 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[lemire]

I believe that alignof(std::max_align_t)should always be as large assizeof(size_t)` so the code might be unnecessarily complicated, but it has the benefit of being clear, so I recommend merging as-is.

[jhofstee]

I believe that alignof(std::max_align_t)should always be as large assizeof(size_t)` so the code might be unnecessarily complicated, but it has the benefit of being clear, so I recommend merging as-is.

Actually I initially wrote that, only using alignof(std::max_align_t), but that might be confusing to read, since it is casted to a <size_t*> thereafter and a reader might be led to believe it is wrong and put a size_t back. In practice the alignment is often, if not always, twice the sizeof(size_t) and guaranteed by the standard not to be less if I am not mistaken. But like this, the code clearly states it intend, make room for a size_t, while keeping the data after it suitable aligned. So it is clearer like this if you ask me and the compiler will remove the Max for us.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/66155/

[nodejs-github-bot]

Landed in dc035bb

[jhofstee]

@targos are there any plans to backport this to LTS versions?