Hi Eric, Thanks for the review. I agree that a separate test for each sysctl bound does not scale and that this focused tcp_rmem test is not the right fit. I won't send a v2 of it.
Your points about the non-portable 4096 assumption and partial updates are also noted. The generic network-sysctl test you described is broader work; I'll treat it separately rather than expanding this patch. Thanks, Sahaj
