mirrors/tracy
1
0
Fork 0
mirror of https://github.com/wolfpld/tracy synced 2025-04-30 12:53:51 +00:00
Code Issues Releases Wiki Activity

Wrap rpmalloc in tracy namespace.

Browse Source
This commit is contained in:
Bartosz Taudul 2017-10-14 16:50:08 +02:00
parent 709d86ad0c
commit b117c56257
2 changed files with 26 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
client/tracy_rpmalloc.cpp
View File

@ -9,7 +9,7 @@
* *
*/ */
#include "tracy_rpmalloc.h" #include "tracy_rpmalloc.hpp"
// Build time configurable limits // Build time configurable limits
@ -77,6 +77,7 @@
# if ENABLE_VALIDATE_ARGS # if ENABLE_VALIDATE_ARGS
# include <Intsafe.h> # include <Intsafe.h>
# endif # endif
# include <intrin.h>
#else #else
# define ALIGNED_STRUCT(name, alignment) struct __attribute__((__aligned__(alignment))) name # define ALIGNED_STRUCT(name, alignment) struct __attribute__((__aligned__(alignment))) name
# define FORCEINLINE inline __attribute__((__always_inline__)) # define FORCEINLINE inline __attribute__((__always_inline__))
@ -114,6 +115,9 @@
# define assert(x) # define assert(x)
#endif #endif
namespace tracy
{
// Atomic access abstraction // Atomic access abstraction
ALIGNED_STRUCT(atomic32_t, 4) { ALIGNED_STRUCT(atomic32_t, 4) {
int32_t nonatomic; int32_t nonatomic;
@ -417,7 +421,7 @@ _memory_deallocate_deferred(heap_t* heap, size_t size_class);
static heap_t* static heap_t*
_memory_heap_lookup(int32_t id) { _memory_heap_lookup(int32_t id) {
uint32_t list_idx = id % HEAP_ARRAY_SIZE; uint32_t list_idx = id % HEAP_ARRAY_SIZE;
heap_t* heap = atomic_load_ptr(&_memory_heaps[list_idx]); heap_t* heap = (heap_t*)atomic_load_ptr(&_memory_heaps[list_idx]);
while (heap && (heap->id != id)) while (heap && (heap->id != id))
heap = heap->next_heap; heap = heap->next_heap;
return heap; return heap;
@ -636,7 +640,7 @@ use_active:
//Happy path, we have a span with at least one free block //Happy path, we have a span with at least one free block
span_t* span = heap->active_span[class_idx]; span_t* span = heap->active_span[class_idx];
count_t offset = class_size * active_block->free_list; count_t offset = class_size * active_block->free_list;
uint32_t* block = pointer_offset(span, SPAN_HEADER_SIZE + offset); uint32_t* block = (uint32_t*)pointer_offset(span, SPAN_HEADER_SIZE + offset);
assert(span); assert(span);
--active_block->free_count; --active_block->free_count;
@ -712,7 +716,7 @@ use_active:
} }
else { else {
//Step 6: All caches empty, map in new memory pages //Step 6: All caches empty, map in new memory pages
span = _memory_map(size_class->page_count); span = (span_t*)_memory_map(size_class->page_count);
} }
//Mark span as owned by this heap and set base data //Mark span as owned by this heap and set base data
@ -786,7 +790,7 @@ _memory_allocate_large_from_heap(heap_t* heap, size_t size) {
} }
else { else {
//Step 6: All caches empty, map in new memory pages //Step 6: All caches empty, map in new memory pages
span = _memory_map(SPAN_MAX_PAGE_COUNT); span = (span_t*)_memory_map(SPAN_MAX_PAGE_COUNT);
} }
//Mark span as owned by this heap and set base data //Mark span as owned by this heap and set base data
@ -851,7 +855,7 @@ use_cache:
} }
else { else {
//Step 4: Map in more memory pages //Step 4: Map in more memory pages
span = _memory_map(num_spans * SPAN_MAX_PAGE_COUNT); span = (span_t*)_memory_map(num_spans * SPAN_MAX_PAGE_COUNT);
} }
//Mark span as owned by this heap //Mark span as owned by this heap
atomic_store32(&span->heap_id, heap->id); atomic_store32(&span->heap_id, heap->id);
@ -873,7 +877,7 @@ _memory_allocate_heap(void) {
//Try getting an orphaned heap //Try getting an orphaned heap
atomic_thread_fence_acquire(); atomic_thread_fence_acquire();
do { do {
heap = atomic_load_ptr(&_memory_orphan_heaps); heap = (heap_t*)atomic_load_ptr(&_memory_orphan_heaps);
if (!heap) if (!heap)
break; break;
next_heap = heap->next_orphan; next_heap = heap->next_orphan;
@ -886,7 +890,7 @@ _memory_allocate_heap(void) {
} }
//Map in pages for a new heap //Map in pages for a new heap
heap = _memory_map(2); heap = (heap_t*)_memory_map(2);
memset(heap, 0, sizeof(heap_t)); memset(heap, 0, sizeof(heap_t));
//Get a new heap ID //Get a new heap ID
@ -900,7 +904,7 @@ _memory_allocate_heap(void) {
//Link in heap in heap ID map //Link in heap in heap ID map
size_t list_idx = heap->id % HEAP_ARRAY_SIZE; size_t list_idx = heap->id % HEAP_ARRAY_SIZE;
do { do {
next_heap = atomic_load_ptr(&_memory_heaps[list_idx]); next_heap = (heap_t*)atomic_load_ptr(&_memory_heaps[list_idx]);
heap->next_heap = next_heap; heap->next_heap = next_heap;
} }
while (!atomic_cas_ptr(&_memory_heaps[list_idx], heap, next_heap)); while (!atomic_cas_ptr(&_memory_heaps[list_idx], heap, next_heap));
@ -1021,7 +1025,7 @@ _memory_deallocate_to_heap(heap_t* heap, span_t* span, void* p) {
void* blocks_start = pointer_offset(span, SPAN_HEADER_SIZE); void* blocks_start = pointer_offset(span, SPAN_HEADER_SIZE);
count_t block_offset = (count_t)pointer_diff(p, blocks_start); count_t block_offset = (count_t)pointer_diff(p, blocks_start);
count_t block_idx = block_offset / (count_t)size_class->size; count_t block_idx = block_offset / (count_t)size_class->size;
uint32_t* block = pointer_offset(blocks_start, block_idx * size_class->size); uint32_t* block = (uint32_t*)pointer_offset(blocks_start, block_idx * size_class->size);
*block = block_data->free_list; *block = block_data->free_list;
block_data->free_list = (uint16_t)block_idx; block_data->free_list = (uint16_t)block_idx;
} }
@ -1097,7 +1101,7 @@ _memory_deallocate_deferred(heap_t* heap, size_t size_class) {
do { do {
void* next = *(void**)p; void* next = *(void**)p;
//Get span and check which type of block //Get span and check which type of block
span_t* span = (void*)((uintptr_t)p & SPAN_MASK); span_t* span = (span_t*)(void*)((uintptr_t)p & SPAN_MASK);
if (span->size_class < SIZE_CLASS_COUNT) { if (span->size_class < SIZE_CLASS_COUNT) {
//Small/medium block //Small/medium block
got_class |= (span->size_class == size_class); got_class |= (span->size_class == size_class);
@ -1139,7 +1143,7 @@ _memory_allocate(size_t size) {
size_t num_pages = size / PAGE_SIZE; size_t num_pages = size / PAGE_SIZE;
if (size % PAGE_SIZE) if (size % PAGE_SIZE)
++num_pages; ++num_pages;
span_t* span = _memory_map(num_pages); span_t* span = (span_t*)_memory_map(num_pages);
atomic_store32(&span->heap_id, 0); atomic_store32(&span->heap_id, 0);
//Store page count in next_span //Store page count in next_span
span->next_span = (span_t*)((uintptr_t)num_pages); span->next_span = (span_t*)((uintptr_t)num_pages);
@ -1154,7 +1158,7 @@ _memory_deallocate(void* p) {
return; return;
//Grab the span (always at start of span, using 64KiB alignment) //Grab the span (always at start of span, using 64KiB alignment)
span_t* span = (void*)((uintptr_t)p & SPAN_MASK); span_t* span = (span_t*)(void*)((uintptr_t)p & SPAN_MASK);
int32_t heap_id = atomic_load32(&span->heap_id); int32_t heap_id = atomic_load32(&span->heap_id);
heap_t* heap = _memory_thread_heap; heap_t* heap = _memory_thread_heap;
//Check if block belongs to this heap or if deallocation should be deferred //Check if block belongs to this heap or if deallocation should be deferred
@ -1179,7 +1183,7 @@ static void*
_memory_reallocate(void* p, size_t size, size_t oldsize, unsigned int flags) { _memory_reallocate(void* p, size_t size, size_t oldsize, unsigned int flags) {
if (p) { if (p) {
//Grab the span (always at start of span, using 64KiB alignment) //Grab the span (always at start of span, using 64KiB alignment)
span_t* span = (void*)((uintptr_t)p & SPAN_MASK); span_t* span = (span_t*)(void*)((uintptr_t)p & SPAN_MASK);
int32_t heap_id = atomic_load32(&span->heap_id); int32_t heap_id = atomic_load32(&span->heap_id);
if (heap_id) { if (heap_id) {
if (span->size_class < SIZE_CLASS_COUNT) { if (span->size_class < SIZE_CLASS_COUNT) {
@ -1235,7 +1239,7 @@ _memory_reallocate(void* p, size_t size, size_t oldsize, unsigned int flags) {
static size_t static size_t
_memory_usable_size(void* p) { _memory_usable_size(void* p) {
//Grab the span (always at start of span, using 64KiB alignment) //Grab the span (always at start of span, using 64KiB alignment)
span_t* span = (void*)((uintptr_t)p & SPAN_MASK); span_t* span = (span_t*)(void*)((uintptr_t)p & SPAN_MASK);
int32_t heap_id = atomic_load32(&span->heap_id); int32_t heap_id = atomic_load32(&span->heap_id);
if (heap_id) { if (heap_id) {
if (span->size_class < SIZE_CLASS_COUNT) { if (span->size_class < SIZE_CLASS_COUNT) {
@ -1344,7 +1348,7 @@ rpmalloc_finalize(void) {
//Free all thread caches //Free all thread caches
for (size_t list_idx = 0; list_idx < HEAP_ARRAY_SIZE; ++list_idx) { for (size_t list_idx = 0; list_idx < HEAP_ARRAY_SIZE; ++list_idx) {
heap_t* heap = atomic_load_ptr(&_memory_heaps[list_idx]); heap_t* heap = (heap_t*)atomic_load_ptr(&_memory_heaps[list_idx]);
while (heap) { while (heap) {
_memory_deallocate_deferred(heap, 0); _memory_deallocate_deferred(heap, 0);
@ -1510,7 +1514,7 @@ rpmalloc_thread_finalize(void) {
//Orphan the heap //Orphan the heap
heap_t* last_heap; heap_t* last_heap;
do { do {
last_heap = atomic_load_ptr(&_memory_orphan_heaps); last_heap = (heap_t*)atomic_load_ptr(&_memory_orphan_heaps);
heap->next_orphan = last_heap; heap->next_orphan = last_heap;
} }
while (!atomic_cas_ptr(&_memory_orphan_heaps, heap, last_heap)); while (!atomic_cas_ptr(&_memory_orphan_heaps, heap, last_heap));
@ -1725,7 +1729,7 @@ rpmalloc_thread_statistics(rpmalloc_thread_statistics_t* stats) {
void* p = atomic_load_ptr(&heap->defer_deallocate); void* p = atomic_load_ptr(&heap->defer_deallocate);
while (p) { while (p) {
void* next = *(void**)p; void* next = *(void**)p;
span_t* span = (void*)((uintptr_t)p & SPAN_MASK); span_t* span = (span_t*)(void*)((uintptr_t)p & SPAN_MASK);
stats->deferred += _memory_size_class[span->size_class].size; stats->deferred += _memory_size_class[span->size_class].size;
p = next; p = next;
} }
@ -1776,3 +1780,5 @@ rpmalloc_global_statistics(rpmalloc_global_statistics_t* stats) {
stats->cached_large += list_bytes; stats->cached_large += list_bytes;
} }
} }
}

7
client/tracy_rpmalloc.hpp
View File

@ -13,9 +13,8 @@
#include <stddef.h> #include <stddef.h>
#ifdef __cplusplus namespace tracy
extern "C" { {
#endif
#if defined(__clang__) || defined(__GNUC__) #if defined(__clang__) || defined(__GNUC__)
# define RPMALLOC_ATTRIBUTE __attribute__((__malloc__)) # define RPMALLOC_ATTRIBUTE __attribute__((__malloc__))
@ -114,6 +113,4 @@ rpposix_memalign(void **memptr, size_t alignment, size_t size);
extern size_t extern size_t
rpmalloc_usable_size(void* ptr); rpmalloc_usable_size(void* ptr);
#ifdef __cplusplus
} }
#endif