Apply LLVM patches for deterministic operand order and correct peephole optimization for minsize (#500)

## Description

Applies LLVM patches that have not been backported to the major version
used in our submodule:

* [LLVM PR #163679](https://github.com/llvm/llvm-project/pull/163679)
  * *"[RISCV] Allow large div peephole optimization for minsize"*
* Fixes peephole optimization of large integers when `minsize` function
attribute is set.
* (Our revive codebase used a workaround that can be removed once this
500 PR is merged and we publish a new LLVM release.)
* [LLVM PR #190219](https://github.com/llvm/llvm-project/pull/190219)
* *"[TargetLowering] Speculative fix for a non-determinism issue between
different compilers"*
* Operand ordering of `mul i256` were different depending on the target
ABI when building LLVM, producing different RISC-V and PVM bytecode for
our Windows vs macOS/linux builds.
* The bytecode divergence (yet functionally equivalent) affected
Solidity contracts containing a value [multiplied by
itself](https://github.com/llvm/llvm-project/issues/190148) (e.g. via
explicit squaring or exponentiation), manifesting primarily when using
optimization level `-O0`.

crates/llvm-builder/src/lib.rs

@@ -82,6 +82,8 @@ pub fn build(
         )
     }
 
+    utils::apply_patches()?;
+
     std::fs::create_dir_all(llvm_path::DIRECTORY_LLVM_TARGET.get().unwrap())?;
 
     if cfg!(target_arch = "x86_64") {

crates/llvm-builder/src/llvm_path.rs

@@ -20,6 +20,9 @@ impl LLVMPath {
     /// The Emscripten SDK source directory.
     pub const DIRECTORY_EMSDK_SOURCE: &'static str = "./emsdk/";
 
+    /// The directory containing patches to apply to the LLVM source.
+    pub const DIRECTORY_PATCHES: &'static str = "./patches/llvm/";
+
     /// Returns the path to the `llvm` stage 1 host LLVM source module directory.
     pub fn llvm_host_module_llvm() -> anyhow::Result<PathBuf> {
         let mut path = PathBuf::from(Self::DIRECTORY_LLVM_HOST_SOURCE);

crates/llvm-builder/src/utils.rs

@@ -1,6 +1,6 @@
 //! The LLVM builder utilities.
 
-use std::fs::File;
+use std::fs::{read_dir, File};
 use std::path::Path;
 use std::path::PathBuf;
 use std::process::Command;
@@ -10,12 +10,6 @@ use std::time::Duration;
 use anyhow::Context;
 use path_slash::PathBufExt;
 
-/// The LLVM host repository URL.
-pub const LLVM_HOST_SOURCE_URL: &str = "https://github.com/llvm/llvm-project";
-
-/// The LLVM host repository tag.
-pub const LLVM_HOST_SOURCE_TAG: &str = "llvmorg-18.1.8";
-
 /// The minimum required XCode version.
 pub const XCODE_MIN_VERSION: u32 = 11;
 
@@ -220,6 +214,57 @@ pub fn install_emsdk() -> anyhow::Result<()> {
     Ok(())
 }
 
+/// Applies all LLVM `.patch` files to the LLVM source.
+///
+/// Patches that have already been applied are skipped.
+pub fn apply_patches() -> anyhow::Result<()> {
+    let patches_directory = PathBuf::from(crate::LLVMPath::DIRECTORY_PATCHES);
+    if !patches_directory.exists() {
+        log::debug!("no patches directory found, skipping");
+        return Ok(());
+    }
+
+    let llvm_source = PathBuf::from(crate::LLVMPath::DIRECTORY_LLVM_SOURCE);
+    let mut patches: Vec<_> = read_dir(&patches_directory)?
+        .filter_map(|entry| entry.ok())
+        .map(|entry| entry.path())
+        .filter(|path| {
+            path.extension()
+                .is_some_and(|extension| extension == "patch")
+        })
+        .collect();
+    patches.sort();
+
+    for patch in patches {
+        let patch = absolute_path(&patch)?;
+        let already_applied = Command::new("git")
+            .args(["apply", "--reverse", "--check"])
+            .arg(&patch)
+            .current_dir(&llvm_source)
+            .stdout(Stdio::null())
+            .stderr(Stdio::null())
+            .status()
+            .map(|status| status.success())
+            .unwrap_or(false);
+
+        if already_applied {
+            log::info!("patch already applied, skipping: {}", patch.display());
+            continue;
+        }
+
+        log::info!("applying patch: {}", patch.display());
+        command(
+            Command::new("git")
+                .args(["apply"])
+                .arg(&patch)
+                .current_dir(&llvm_source),
+            &format!("applying patch {}", patch.display()),
+        )?;
+    }
+
+    Ok(())
+}
+
 /// The LLVM target directory default path.
 pub fn directory_target_llvm(target_env: crate::target_env::TargetEnv) -> PathBuf {
     crate::llvm_path::DIRECTORY_LLVM_TARGET