Auto merge of rust-lang#134866 - osiewicz:write-rlib-bufwriter, r=bjorn3

rustc_codegen_ssa: Buffer file writes in link_rlib

This makes this step take ~25ms on my machine (M3 Max 64GB) for Zed repo instead of ~150ms (on editor crate). Additionally it takes down the time needed for a clean cargo build of ripgrep from ~6.1s to 5.9s.

This change is mostly relevant for dev builds of crates with multiple large CGUs.
I imagine it could be quite relevant for dev scenarios on Windows, but sadly I have no way to measure that myself.

compiler/rustc_codegen_ssa/src/back/archive.rs

@@ -2,7 +2,7 @@ use std::env;
 use std::error::Error;
 use std::ffi::OsString;
 use std::fs::{self, File};
-use std::io::{self, Write};
+use std::io::{self, BufWriter, Write};
 use std::path::{Path, PathBuf};
 
 use ar_archive_writer::{
@@ -509,16 +509,19 @@ impl<'a> ArArchiveBuilder<'a> {
                 io_error_context("couldn't create a directory for the temp file", err)
             })?;
         let archive_tmpfile_path = archive_tmpdir.path().join("tmp.a");
-        let mut archive_tmpfile = File::create_new(&archive_tmpfile_path)
+        let archive_tmpfile = File::create_new(&archive_tmpfile_path)
             .map_err(|err| io_error_context("couldn't create the temp file", err))?;
 
+        let mut archive_tmpfile = BufWriter::new(archive_tmpfile);
         write_archive_to_stream(
             &mut archive_tmpfile,
             &entries,
             archive_kind,
             false,
             /* is_ec = */ self.sess.target.arch == "arm64ec",
         )?;
+        archive_tmpfile.flush()?;
+        drop(archive_tmpfile);
 
         let any_entries = !entries.is_empty();
         drop(entries);