Fix x86

Signed-off-by: Heinz N. Gies <heinz@licenser.net>

Author: Licenser

src/impls/avx2/stage1.rs

@@ -1,6 +1,6 @@
 #![allow(dead_code)]
 use crate::{
-    Stage1Parse,
+    SIMDINPUT_LENGTH, Stage1Parse,
     macros::{static_cast_i32, static_cast_i64, static_cast_u32},
 };
 #[cfg(target_arch = "x86")]
@@ -11,9 +11,9 @@ use std::arch::x86_64 as arch;
 
 use arch::{
     __m256i, _mm_clmulepi64_si128, _mm_set_epi64x, _mm_set1_epi8, _mm256_add_epi32,
-    _mm256_and_si256, _mm256_cmpeq_epi8, _mm256_loadu_si256, _mm256_max_epu8, _mm256_movemask_epi8,
-    _mm256_set_epi32, _mm256_set1_epi8, _mm256_setr_epi8, _mm256_setzero_si256,
-    _mm256_shuffle_epi8, _mm256_srli_epi32, _mm256_storeu_si256,
+    _mm256_and_si256, _mm256_cmpeq_epi8, _mm256_load_si256, _mm256_loadu_si256, _mm256_max_epu8,
+    _mm256_movemask_epi8, _mm256_set_epi32, _mm256_set1_epi8, _mm256_setr_epi8,
+    _mm256_setzero_si256, _mm256_shuffle_epi8, _mm256_srli_epi32, _mm256_storeu_si256,
 };
 
 macro_rules! low_nibble_mask {

src/impls/native/stage1.rs

@@ -1,6 +1,6 @@
 #![allow(clippy::cast_lossless, clippy::cast_sign_loss)]
 
-use crate::{Stage1Parse, macros::static_cast_i32};
+use crate::{SIMDINPUT_LENGTH, Stage1Parse, macros::static_cast_i32};
 
 type V128 = [u8; 16];
 

src/impls/neon/stage1.rs

@@ -42,9 +42,6 @@ pub unsafe fn neon_movemask_bulk(
 
 // /NEON-SPECIFIC
 
-//pub const SIMDJSON_PADDING: usize = mem::size_of::<uint8x16_t>() * 4;
-//pub const SIMDINPUT_LENGTH: usize = 64;
-
 #[derive(Debug)]
 pub(crate) struct SimdInput {
     v0: uint8x16_t,

src/impls/portable/stage1.rs

@@ -1,6 +1,6 @@
 use std::simd::{ToBitMask, prelude::*};
 
-use crate::{Stage1Parse, macros::static_cast_i32};
+use crate::{SIMDINPUT_LENGTH, Stage1Parse, macros::static_cast_i32};
 #[derive(Debug)]
 pub(crate) struct SimdInput {
     v: u8x64,

src/impls/simd128/stage1.rs

@@ -1,4 +1,4 @@
-use crate::Stage1Parse;
+use crate::{Stage1Parse, SIMDINPUT_LENGTH};
 use std::arch::wasm32::{
     i8x16_splat, u8x16, u8x16_bitmask, u8x16_eq, u8x16_le, u8x16_shr, u8x16_splat, u8x16_swizzle,
     u32x4, u32x4_add, u32x4_splat, v128, v128_and, v128_load, v128_store,

src/impls/sse42/stage1.rs

@@ -1,5 +1,5 @@
 use crate::{
-    Stage1Parse,
+    SIMDINPUT_LENGTH, Stage1Parse,
     macros::{static_cast_i32, static_cast_i64, static_cast_u32},
 };
 #[cfg(target_arch = "x86")]
@@ -10,16 +10,17 @@ use std::arch::x86_64 as arch;
 
 #[cfg(target_arch = "x86")]
 use arch::{
-    __m128i, _mm_add_epi32, _mm_and_si128, _mm_cmpeq_epi8, _mm_cmpgt_epi8, _mm_loadu_si128,
-    _mm_max_epu8, _mm_movemask_epi8, _mm_or_si128, _mm_set_epi32, _mm_set1_epi8, _mm_setr_epi8,
-    _mm_setzero_si128, _mm_shuffle_epi8, _mm_srli_epi32, _mm_storeu_si128, _mm_testz_si128,
+    __m128i, _mm_add_epi32, _mm_and_si128, _mm_cmpeq_epi8, _mm_cmpgt_epi8, _mm_load_si128,
+    _mm_loadu_si128, _mm_max_epu8, _mm_movemask_epi8, _mm_or_si128, _mm_set_epi32, _mm_set1_epi8,
+    _mm_setr_epi8, _mm_setzero_si128, _mm_shuffle_epi8, _mm_srli_epi32, _mm_storeu_si128,
+    _mm_testz_si128,
 };
 
 #[cfg(target_arch = "x86_64")]
 use arch::{
-    __m128i, _mm_add_epi32, _mm_and_si128, _mm_cmpeq_epi8, _mm_loadu_si128, _mm_max_epu8,
-    _mm_movemask_epi8, _mm_set_epi32, _mm_set1_epi8, _mm_setr_epi8, _mm_setzero_si128,
-    _mm_shuffle_epi8, _mm_srli_epi32, _mm_storeu_si128,
+    __m128i, _mm_add_epi32, _mm_and_si128, _mm_cmpeq_epi8, _mm_load_si128, _mm_loadu_si128,
+    _mm_max_epu8, _mm_movemask_epi8, _mm_set_epi32, _mm_set1_epi8, _mm_setr_epi8,
+    _mm_setzero_si128, _mm_shuffle_epi8, _mm_srli_epi32, _mm_storeu_si128,
 };
 
 macro_rules! low_nibble_mask {

src/lib.rs

@@ -335,7 +335,8 @@ type ParseStrFn = for<'invoke, 'de> unsafe fn(
     any(target_arch = "x86_64", target_arch = "x86"),
 ))]
 type FindStructuralBitsFn = unsafe fn(
-    input: &[u8],
+    input: &AlignedBuf,
+    len: usize,
     structural_indexes: &mut Vec<u32>,
 ) -> std::result::Result<(), ErrorType>;
 
@@ -611,7 +612,8 @@ impl Deserializer<'_> {
         any(target_arch = "x86_64", target_arch = "x86"),
     ))]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
+        len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {
         unsafe {
@@ -636,18 +638,19 @@ impl Deserializer<'_> {
 
             #[cfg_attr(not(feature = "no-inline"), inline)]
             unsafe fn get_fastest(
-                input: &[u8],
+                input: &AlignedBuf,
+                len: usize,
                 structural_indexes: &mut Vec<u32>,
             ) -> core::result::Result<(), error::ErrorType> {
                 unsafe {
                     let fun = get_fastest_available_implementation();
                     FN.store(fun as FnRaw, Ordering::Relaxed);
-                    (fun)(input, structural_indexes)
+                    (fun)(input, len, structural_indexes)
                 }
             }
 
             let fun = FN.load(Ordering::Relaxed);
-            mem::transmute::<FnRaw, FindStructuralBitsFn>(fun)(input, structural_indexes)
+            mem::transmute::<FnRaw, FindStructuralBitsFn>(fun)(input, len, structural_indexes)
         }
     }
 
@@ -664,7 +667,7 @@ impl Deserializer<'_> {
     )))]
     #[cfg_attr(not(feature = "no-inline"), inline)]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
         len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {
@@ -683,7 +686,7 @@ impl Deserializer<'_> {
     #[cfg(all(feature = "portable", not(feature = "runtime-detection")))]
     #[cfg_attr(not(feature = "no-inline"), inline)]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
         len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {
@@ -703,7 +706,7 @@ impl Deserializer<'_> {
     ))]
     #[cfg_attr(not(feature = "no-inline"), inline)]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
         len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {
@@ -720,7 +723,7 @@ impl Deserializer<'_> {
     ))]
     #[cfg_attr(not(feature = "no-inline"), inline)]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
         len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {
@@ -744,7 +747,7 @@ impl Deserializer<'_> {
     #[cfg(all(target_feature = "simd128", not(feature = "portable")))]
     #[cfg_attr(not(feature = "no-inline"), inline)]
     pub(crate) unsafe fn find_structural_bits(
-        input: &[u8],
+        input: &AlignedBuf,
         len: usize,
         structural_indexes: &mut Vec<u32>,
     ) -> std::result::Result<(), ErrorType> {