michael/sailfish
michael
/
sailfish
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Speed up feature detection for escaping 

- Remove sailfish::runtime::escape::escape_with
This commit is contained in:
Kogia-sima 2020-06-10 10:27:13 +09:00
parent 655bcb9e2f
commit 53f0db024c
5 changed files with 68 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
sailfish/src/runtime/escape/avx2.rs
View File

@ -8,21 +8,22 @@ use std::slice;
use super::{naive, sse2};
use super::{ESCAPED, ESCAPED_LEN, ESCAPE_LUT};
use super::super::Buffer;
const VECTOR_BYTES: usize = std::mem::size_of::<__m256i>();
const VECTOR_ALIGN: usize = VECTOR_BYTES - 1;
#[target_feature(enable = "avx2")]
pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
pub unsafe fn escape(buffer: &mut Buffer, bytes: &[u8]) {
let len = bytes.len();
let mut start_ptr = bytes.as_ptr();
let end_ptr = start_ptr.add(len);
if len < VECTOR_BYTES {
if len < 16 {
naive::escape(writer, start_ptr, start_ptr, end_ptr);
naive::escape(buffer, start_ptr, start_ptr, end_ptr);
} else {
sse2::escape(writer, bytes);
sse2::escape(buffer, bytes);
}
return;
}
@ -56,9 +57,9 @@ pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
if start_ptr < ptr2 {
let slc =
slice::from_raw_parts(start_ptr, ptr2 as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(c));
buffer.push_str(*ESCAPED.get_unchecked(c));
start_ptr = ptr2.add(1);
mask ^= 1 << trailing_zeros;
}
@ -77,9 +78,9 @@ pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
if start_ptr < ptr2 {
let slc =
slice::from_raw_parts(start_ptr, ptr2 as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(c));
buffer.push_str(*ESCAPED.get_unchecked(c));
start_ptr = ptr2.add(1);
mask ^= 1 << trailing_zeros;
}
@ -88,5 +89,5 @@ pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
next_ptr = next_ptr.add(VECTOR_BYTES);
}
sse2::escape_aligned(writer, start_ptr, ptr, end_ptr);
sse2::escape_aligned(buffer, start_ptr, ptr, end_ptr);
}

17
sailfish/src/runtime/escape/fallback.rs
View File

@ -1,6 +1,7 @@
#![allow(clippy::cast_ptr_alignment)]
use super::naive;
use super::super::Buffer;
#[cfg(target_pointer_width = "16")]
const USIZE_BYTES: usize = 2;
@ -37,13 +38,13 @@ fn contains_key(x: usize) -> bool {
contains_zero_byte(z1) || contains_zero_byte(z2)
}
pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
pub unsafe fn escape(buffer: &mut Buffer, bytes: &[u8]) {
let len = bytes.len();
let mut start_ptr = bytes.as_ptr();
let end_ptr = start_ptr.add(len);
if bytes.len() < USIZE_BYTES {
naive::escape(writer, start_ptr, start_ptr, end_ptr);
naive::escape(buffer, start_ptr, start_ptr, end_ptr);
return;
}
@ -54,14 +55,14 @@ pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
let chunk = (ptr as *const usize).read_unaligned();
if contains_key(chunk) {
start_ptr = naive::proceed(writer, start_ptr, ptr, aligned_ptr);
start_ptr = naive::proceed(buffer, start_ptr, ptr, aligned_ptr);
}
escape_aligned(writer, start_ptr, aligned_ptr, end_ptr);
escape_aligned(buffer, start_ptr, aligned_ptr, end_ptr);
}
pub unsafe fn escape_aligned<F: FnMut(&str)>(
writer: &mut F,
pub unsafe fn escape_aligned(
buffer: &mut Buffer,
mut start_ptr: *const u8,
mut ptr: *const u8,
end_ptr: *const u8,
@ -73,11 +74,11 @@ pub unsafe fn escape_aligned<F: FnMut(&str)>(
eprintln!("# {:x}", chunk);
if contains_key(chunk) {
eprintln!("true!");
start_ptr = naive::proceed(writer, start_ptr, ptr, ptr.add(USIZE_BYTES))
start_ptr = naive::proceed(buffer, start_ptr, ptr, ptr.add(USIZE_BYTES))
}
ptr = ptr.add(USIZE_BYTES);
}
debug_assert!(ptr <= end_ptr);
debug_assert!(start_ptr <= ptr);
naive::escape(writer, start_ptr, ptr, end_ptr);
naive::escape(buffer, start_ptr, ptr, end_ptr);
}

60
sailfish/src/runtime/escape/mod.rs
View File

@ -25,33 +25,34 @@ static ESCAPE_LUT: [u8; 256] = [
const ESCAPED: [&str; 4] = ["&quot;", "&amp;", "&lt;", "&gt;"];
const ESCAPED_LEN: usize = 4;
/// write the escaped contents with custom function
///
/// This function is soft-deprecated because using this function causes a large binary size.
#[inline]
pub fn escape_with<F: FnMut(&str)>(mut writer: F, feed: &str) {
unsafe {
#[cfg(target_feature = "avx2")]
{
avx2::escape(&mut writer, feed.as_bytes());
}
#[cfg(not(target_feature = "avx2"))]
{
if is_x86_feature_detected!("avx2") {
avx2::escape(&mut writer, feed.as_bytes());
} else if is_x86_feature_detected!("sse2") {
sse2::escape(&mut writer, feed.as_bytes());
} else {
fallback::escape(&mut writer, feed.as_bytes());
}
}
}
}
#[doc(hidden)]
pub fn escape_to_buf(feed: &str, buf: &mut Buffer) {
escape_with(|e| buf.push_str(e), feed);
use std::mem;
use std::sync::atomic::{AtomicPtr, Ordering};
type FnRaw = *mut ();
static FN: AtomicPtr<()> = AtomicPtr::new(detect as FnRaw);
fn detect(buffer: &mut Buffer, bytes: &[u8]) {
let fun = if is_x86_feature_detected!("avx2") {
avx2::escape as FnRaw
} else if is_x86_feature_detected!("sse2") {
sse2::escape as FnRaw
} else {
fallback::escape as FnRaw
};
FN.store(fun as FnRaw, Ordering::Relaxed);
unsafe {
mem::transmute::<FnRaw, fn(&mut Buffer, &[u8])>(fun)(buffer, bytes);
}
}
unsafe {
let fun = FN.load(Ordering::Relaxed);
mem::transmute::<FnRaw, fn(&mut Buffer, &[u8])>(fun)(buf, feed.as_bytes());
}
}
/// write the escaped contents into `String`
@ -152,13 +153,8 @@ mod tests {
unsafe {
escape_to_buf(&*s, &mut buf1);
fallback::escape(&mut |s| buf2.push_str(s), s.as_bytes());
naive::escape(
&mut |s| buf3.push_str(s),
s.as_ptr(),
s.as_ptr(),
s.as_ptr().add(s.len()),
);
fallback::escape(&mut buf2, s.as_bytes());
naive::escape(&mut buf3, s.as_ptr(), s.as_ptr(), s.as_ptr().add(s.len()));
}
assert_eq!(buf1.as_str(), buf3.as_str());

17
sailfish/src/runtime/escape/naive.rs
View File

@ -1,25 +1,26 @@
use core::slice;
use super::super::Buffer;
use super::{ESCAPED, ESCAPED_LEN, ESCAPE_LUT};
#[inline]
pub(super) unsafe fn escape<F: FnMut(&str)>(
writer: &mut F,
pub(super) unsafe fn escape(
buffer: &mut Buffer,
mut start_ptr: *const u8,
ptr: *const u8,
end_ptr: *const u8,
) {
start_ptr = proceed(writer, start_ptr, ptr, end_ptr);
start_ptr = proceed(buffer, start_ptr, ptr, end_ptr);
if end_ptr > start_ptr {
let slc = slice::from_raw_parts(start_ptr, end_ptr as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
}
#[inline]
pub(super) unsafe fn proceed<F: FnMut(&str)>(
writer: &mut F,
pub(super) unsafe fn proceed(
buffer: &mut Buffer,
mut start_ptr: *const u8,
mut ptr: *const u8,
end_ptr: *const u8,
@ -32,9 +33,9 @@ pub(super) unsafe fn proceed<F: FnMut(&str)>(
if ptr > start_ptr {
let slc =
slice::from_raw_parts(start_ptr, ptr as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(idx));
buffer.push_str(*ESCAPED.get_unchecked(idx));
start_ptr = ptr.add(1);
}
ptr = ptr.add(1);

25
sailfish/src/runtime/escape/sse2.rs
View File

@ -8,19 +8,20 @@ use std::slice;
use super::naive;
use super::{ESCAPED, ESCAPED_LEN, ESCAPE_LUT};
use super::super::Buffer;
const VECTOR_BYTES: usize = std::mem::size_of::<__m128i>();
const VECTOR_ALIGN: usize = VECTOR_BYTES - 1;
#[target_feature(enable = "sse2")]
#[inline]
pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
pub unsafe fn escape(buffer: &mut Buffer, bytes: &[u8]) {
let len = bytes.len();
let mut start_ptr = bytes.as_ptr();
let end_ptr = start_ptr.add(len);
if bytes.len() < VECTOR_BYTES {
naive::escape(writer, start_ptr, start_ptr, end_ptr);
naive::escape(buffer, start_ptr, start_ptr, end_ptr);
return;
}
@ -53,20 +54,20 @@ pub unsafe fn escape<F: FnMut(&str)>(writer: &mut F, bytes: &[u8]) {
if start_ptr < ptr2 {
let slc =
slice::from_raw_parts(start_ptr, ptr2 as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(c));
buffer.push_str(*ESCAPED.get_unchecked(c));
start_ptr = ptr2.add(1);
mask ^= 1 << trailing_zeros;
}
}
ptr = aligned_ptr;
escape_aligned(writer, start_ptr, ptr, end_ptr);
escape_aligned(buffer, start_ptr, ptr, end_ptr);
}
pub unsafe fn escape_aligned<F: FnMut(&str)>(
writer: &mut F,
pub unsafe fn escape_aligned(
buffer: &mut Buffer,
mut start_ptr: *const u8,
mut ptr: *const u8,
end_ptr: *const u8,
@ -95,9 +96,9 @@ pub unsafe fn escape_aligned<F: FnMut(&str)>(
if start_ptr < ptr2 {
let slc =
slice::from_raw_parts(start_ptr, ptr2 as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(c));
buffer.push_str(*ESCAPED.get_unchecked(c));
start_ptr = ptr2.add(1);
mask ^= 1 << trailing_zeros;
}
@ -118,9 +119,9 @@ pub unsafe fn escape_aligned<F: FnMut(&str)>(
if start_ptr < ptr2 {
let slc =
slice::from_raw_parts(start_ptr, ptr2 as usize - start_ptr as usize);
writer(std::str::from_utf8_unchecked(slc));
buffer.push_str(std::str::from_utf8_unchecked(slc));
}
writer(*ESCAPED.get_unchecked(c));
buffer.push_str(*ESCAPED.get_unchecked(c));
start_ptr = ptr2.add(1);
mask ^= 1 << trailing_zeros;
}
@ -130,5 +131,5 @@ pub unsafe fn escape_aligned<F: FnMut(&str)>(
debug_assert!(ptr <= end_ptr);
debug_assert!(start_ptr <= ptr);
naive::escape(writer, start_ptr, ptr, end_ptr);
naive::escape(buffer, start_ptr, ptr, end_ptr);
}