Use HashMap with faster hashing algorithm for packfile writer

Summary: We use HashMap to store the set of objects written to packfile to ensure we do not add duplicate objects to the pack. These collections can grow quite large in which case the performance of the hashing algorithm plays a role in the overall runtime. Since the input to these collections are known to be safe, we do not need a cryptographic hash and can instead use a much faster non-cryptographic hashing algorithm used within the rust compiler

Differential Revision: D54638708

fbshipit-source-id: 5053cf94598be3670ca2339878eeacb2a60eddd0

eden/mononoke/git/packfile/Cargo.toml

@@ -26,6 +26,7 @@ mononoke_types = { version = "0.1.0", path = "../../mononoke_types" }
 packfile_thrift = { version = "0.1.0", path = "if" }
 pin-project = "0.4.30"
 quickcheck = "1.0"
+rustc-hash = "1.1.0"
 sha1 = "0.10.5"
 thiserror = "1.0.49"
 tokio = { version = "1.29.1", features = ["full", "test-util", "tracing"] }

eden/mononoke/git/packfile/TARGETS

@@ -18,6 +18,7 @@ rust_library(
         "fbsource//third-party/rust:gix-pack",
         "fbsource//third-party/rust:pin-project",
         "fbsource//third-party/rust:quickcheck",
+        "fbsource//third-party/rust:rustc-hash",
         "fbsource//third-party/rust:sha1",
         "fbsource//third-party/rust:thiserror",
         "fbsource//third-party/rust:tokio",

eden/mononoke/git/packfile/src/pack.rs

@@ -6,6 +6,7 @@
  */
 
 use std::collections::HashMap;
+use std::hash::BuildHasherDefault;
 use std::io::Write;
 
 use anyhow::Context;
@@ -16,6 +17,8 @@ use gix_hash::ObjectId;
 use gix_pack::data::header;
 use gix_pack::data::output::Entry;
 use gix_pack::data::Version;
+use rustc_hash::FxHashMap;
+use rustc_hash::FxHasher;
 use thiserror::Error;
 use tokio::io::AsyncWrite;
 use tokio::io::AsyncWriteExt;
@@ -66,7 +69,7 @@ where
     /// The form of deltas that should be allowed in the packfile
     delta_form: DeltaForm,
     /// Mapping from Object Id to index in `object_offset_with_validity`
-    object_id_with_index: HashMap<ObjectId, usize>,
+    object_id_with_index: FxHashMap<ObjectId, usize>,
 }
 
 impl<T: AsyncWrite + Unpin> PackfileWriter<T> {
@@ -82,7 +85,10 @@ impl<T: AsyncWrite + Unpin> PackfileWriter<T> {
             // Git uses V2 right now so we do the same
             header_info: Some((Version::V2, count)),
             object_offset_with_validity: Vec::with_capacity(count as usize),
-            object_id_with_index: HashMap::with_capacity(count as usize),
+            object_id_with_index: HashMap::with_capacity_and_hasher(
+                count as usize,
+                BuildHasherDefault::<FxHasher>::default(),
+            ),
             delta_form,
         }
     }