mem: Allow overriding the default buffer pool.

[vanja-p]

The default buffer pool only has a only a few tiers, which hurts performance in some applications. In #8770, it was decided that instead of adding more tiers to the default pool, we should allow changing the default pool.

RELEASE NOTES:

• mem: Allow changing the default buffer pool

[vanja-p]

I didn't know if you internal.SetDefaultBufferPool should be moved from internal/internal.go to internal/experimental.go so I kept it as is for now.

[codecov]

Codecov Report

❌ Patch coverage is 50.00000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.00%. Comparing base (bccbb10) to head (03954f1).
⚠️ Report is 24 commits behind head on master.

Files with missing lines	Patch %	Lines
experimental/experimental.go	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8806      +/-   ##
==========================================
- Coverage   83.30%   83.00%   -0.31%     
==========================================
  Files         418      414       -4     
  Lines       33004    32779     -225     
==========================================
- Hits        27493    27207     -286     
- Misses       4102     4115      +13     
- Partials     1409     1457      +48     

Files with missing lines	Coverage Δ
internal/internal.go	50.00% <ø> (-10.00%)	⬇️
internal/leakcheck/leakcheck.go	90.04% <100.00%> (ø)
mem/buffer_pool.go	96.61% <100.00%> (ø)
experimental/experimental.go	25.00% <0.00%> (-8.34%)	⬇️

... and 42 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[vanja-p]

I can't figure out how to get the PR validation to pass. My initial PR title didn't start with "mem:", but now it does.

[eshitachandwani]

@vanja-p can you check if the test failing is related to the changes or not?
https://github.com/grpc/grpc-go/actions/runs/20763701462/job/59624817820?pr=8806

[vanja-p]

@vanja-p can you check if the test failing is related to the changes or not? https://github.com/grpc/grpc-go/actions/runs/20763701462/job/59624817820?pr=8806

@eshitachandwani, looks like it's not related to my change. Rerunning them passed, and also, my 2 new lines don't have any coverage.

[arjan-bal]

I didn't know if you internal.SetDefaultBufferPool should be moved from internal/internal.go to internal/experimental.go so I kept it as is for now.

I think it's better to move SetDefaultBufferPool to the internal/experimental.go file to maintain consistency with other non-test functions exposed by the mem package. Otherwise, LGTM.

[vanja-p]

I think it's better to move SetDefaultBufferPool to the internal/experimental.go file to maintain consistency with other non-test functions exposed by the mem package. Otherwise, LGTM.

Done.

[arjan-bal]

Just one minor comment; otherwise, LGTM. Adding a second reviewer.

[arjan-bal]

Instead of requiring users to set this before making RPCs, we recommend calling these functions within an init() function. This prevents data races that can occur when multiple packages invoke the same method simultaneously. For example, the balancer.Register function includes the following requirement:

NOTE: this function must only be called during initialization time (i.e., in an init() function) and is not thread-safe...

Can you please add a similar note here?

[vanja-p]

yep, done.

[easwars]

Nit: Can the extra parenthesis be avoided? Can this be
internal.SetDefaultBufferPool.(func(mem.BufferPool))(&globalPool) ?

[vanja-p]

Yep. done.