From 71e468c19641edd869a44dc0632c9181e30bdb16 Mon Sep 17 00:00:00 2001
From: Lukasz Anforowicz <lukasza@chromium.org>
Date: Fri, 25 Oct 2024 16:42:49 +0000
Subject: [PATCH] safety: introduce `ALIGN_MASK` based on `core::mem::align_of`

Fixes #194, Closes #197
---
 src/ascii.rs          | 5 +++--
 src/byteset/scalar.rs | 7 +++----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/src/ascii.rs b/src/ascii.rs
index 464da96..7ef6b5c 100644
--- a/src/ascii.rs
+++ b/src/ascii.rs
@@ -24,6 +24,8 @@
 #[cfg(any(test, miri, not(target_arch = "x86_64")))]
 const USIZE_BYTES: usize = core::mem::size_of::<usize>();
 #[cfg(any(test, miri, not(target_arch = "x86_64")))]
+const ALIGN_MASK: usize = core::mem::align_of::<usize>() - 1;
+#[cfg(any(test, miri, not(target_arch = "x86_64")))]
 const FALLBACK_LOOP_SIZE: usize = 2 * USIZE_BYTES;
 
 // This is a mask where the most significant bit of each byte in the usize
@@ -53,7 +55,6 @@ pub fn first_non_ascii_byte(slice: &[u8]) -> usize {
 
 #[cfg(any(test, miri, not(target_arch = "x86_64")))]
 fn first_non_ascii_byte_fallback(slice: &[u8]) -> usize {
-    let align = USIZE_BYTES - 1;
     let start_ptr = slice.as_ptr();
     let end_ptr = slice[slice.len()..].as_ptr();
     let mut ptr = start_ptr;
@@ -69,7 +70,7 @@ fn first_non_ascii_byte_fallback(slice: &[u8]) -> usize {
             return first_non_ascii_byte_mask(mask);
         }
 
-        ptr = ptr_add(ptr, USIZE_BYTES - (start_ptr as usize & align));
+        ptr = ptr_add(ptr, USIZE_BYTES - (start_ptr as usize & ALIGN_MASK));
         debug_assert!(ptr > start_ptr);
         debug_assert!(ptr_sub(end_ptr, USIZE_BYTES) >= start_ptr);
         if slice.len() >= FALLBACK_LOOP_SIZE {
diff --git a/src/byteset/scalar.rs b/src/byteset/scalar.rs
index 3e0c7ed..24dcf37 100644
--- a/src/byteset/scalar.rs
+++ b/src/byteset/scalar.rs
@@ -5,6 +5,7 @@
 use core::{cmp, usize};
 
 const USIZE_BYTES: usize = core::mem::size_of::<usize>();
+const ALIGN_MASK: usize = core::mem::align_of::<usize>() - 1;
 
 // The number of bytes to loop at in one iteration of memchr/memrchr.
 const LOOP_SIZE: usize = 2 * USIZE_BYTES;
@@ -22,7 +23,6 @@ pub fn inv_memchr(n1: u8, haystack: &[u8]) -> Option<usize> {
     let vn1 = repeat_byte(n1);
     let confirm = |byte| byte != n1;
     let loop_size = cmp::min(LOOP_SIZE, haystack.len());
-    let align = USIZE_BYTES - 1;
     let start_ptr = haystack.as_ptr();
 
     unsafe {
@@ -38,7 +38,7 @@ pub fn inv_memchr(n1: u8, haystack: &[u8]) -> Option<usize> {
             return forward_search(start_ptr, end_ptr, ptr, confirm);
         }
 
-        ptr = ptr.add(USIZE_BYTES - (start_ptr as usize & align));
+        ptr = ptr.add(USIZE_BYTES - (start_ptr as usize & ALIGN_MASK));
         debug_assert!(ptr > start_ptr);
         debug_assert!(end_ptr.sub(USIZE_BYTES) >= start_ptr);
         while loop_size == LOOP_SIZE && ptr <= end_ptr.sub(loop_size) {
@@ -62,7 +62,6 @@ pub fn inv_memrchr(n1: u8, haystack: &[u8]) -> Option<usize> {
     let vn1 = repeat_byte(n1);
     let confirm = |byte| byte != n1;
     let loop_size = cmp::min(LOOP_SIZE, haystack.len());
-    let align = USIZE_BYTES - 1;
     let start_ptr = haystack.as_ptr();
 
     unsafe {
@@ -78,7 +77,7 @@ pub fn inv_memrchr(n1: u8, haystack: &[u8]) -> Option<usize> {
             return reverse_search(start_ptr, end_ptr, ptr, confirm);
         }
 
-        ptr = ptr.sub(end_ptr as usize & align);
+        ptr = ptr.sub(end_ptr as usize & ALIGN_MASK);
         debug_assert!(start_ptr <= ptr && ptr <= end_ptr);
         while loop_size == LOOP_SIZE && ptr >= start_ptr.add(loop_size) {
             debug_assert_eq!(0, (ptr as usize) % USIZE_BYTES);