Github/littlefs
1
Fork 0
Code Actions Activity
Files
9b803f962504813d30ca0227149d6018ff0cf78f
littlefs/lfs_util.h
T
Christopher Haster a511696bad Added ability to bypass rbyd fetch during B-tree lookups 
This is an absurd optimization that stems from the observation that the
branch encoding for the inner-rbyds in a B-tree is enough information to
jump directly to the trunk of the rbyd without needing an lfsr_rbyd_fetch.

This results in a pretty ridiculous performance jump from O(m log_m(n/m))
to O(log(m) log_m(n/m)).

If the complexity analysis isn't impressive enough, look at some rough
benchmarking of read operations for 4KiB-block, 1K-entry B-trees:

   12KiB ^     ::  :. :: .: .: :. : .: :. : : .. : : . : .: : : :
         |    .:: .::.::.:: ::.::::::::::::.::::::::.::::::::::::.
         |    : :::':: ::'::'::':: :' :':: :'::::::::': ::::::': :
before   |  ::: ::' :' :' :: :' '' '  ' '' : : : '' ' ' '
         | :::            ''
         |:
      0B :'------------------------------------------------------>

  .17KiB ^               ............:::::::::::::::::::::::::::::
         |   .   .....:::::'''''''''  '         '          '
         |  .::::::::::::
after    |  :':''
         |.::
         .:'
      0B :------------------------------------------------------->
         0                                                      1K

In order for this to work, the branch encoding did need to be tweaked
slightly. Before it stored block+off, now it stores block+trunk where
"trunk" is the offset of the entry point into the rbyd tree. Both off
and trunk are enough info to know when to stop fetching, if necessary,
but trunk allows lookups to jump directly into the branches rbyd tree
without a fetch.

With the change to trunk, lfsr_rbyd_fetch has also be extended to allow
fetching of any internal trunks, not just the last trunk in the commit.
This is very useful for dbgrbyd.py, but doesn't currently have a use in
littlefs itself. But it's at least valuable to have the feature available
in case it does become useful.

Note that two cases still requires the slower O(m log_m(n/m)) lookup
with lfsr_rbyd_fetch:

1. Name lookups, since we currently use a linear-search O(m) to find names.

2. Validating B-tree rbyd's, which requires a linear fetch O(m) to
   validate the checksums. We will need to do this at least once
   after mount.

It's also worth mentioning this will likely have a large impact on B-tree
traversal speed. Which is huge as I am expecting B-tree traversal to be
the main bottleneck once garbage-collection (or its replacement) is
involved.
2023-04-14 00:51:34 -05:00

358 lines
9.8 KiB
C
Raw Blame History

/*
* lfs utility functions
*
* Copyright (c) 2022, The littlefs authors.
* Copyright (c) 2017, Arm Limited. All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
*/
#ifndef LFS_UTIL_H
#define LFS_UTIL_H
// Users can override lfs_util.h with their own configuration by defining
// LFS_CONFIG as a header file to include (-DLFS_CONFIG=lfs_config.h).
//
// If LFS_CONFIG is used, none of the default utils will be emitted and must be
// provided by the config file. To start, I would suggest copying lfs_util.h
// and modifying as needed.
#ifdef LFS_CONFIG
#define LFS_STRINGIZE(x) LFS_STRINGIZE2(x)
#define LFS_STRINGIZE2(x) #x
#include LFS_STRINGIZE(LFS_CONFIG)
#else
// System includes
#include <stdint.h>
#include <stdbool.h>
#include <sys/types.h>
#include <string.h>
#include <inttypes.h>
#ifndef LFS_NO_MALLOC
#include <stdlib.h>
#endif
#ifndef LFS_NO_ASSERT
#include <assert.h>
#endif
#if !defined(LFS_NO_DEBUG) || \
!defined(LFS_NO_WARN) || \
!defined(LFS_NO_ERROR) || \
defined(LFS_YES_TRACE)
#include <stdio.h>
#endif
#ifdef __cplusplus
extern "C"
{
#endif
// Macros, may be replaced by system specific wrappers. Arguments to these
// macros must not have side-effects as the macros can be removed for a smaller
// code footprint
// Logging functions
#ifndef LFS_TRACE
#ifdef LFS_YES_TRACE
#define LFS_TRACE_(fmt, ...) \
printf("%s:%d:trace: " fmt "%s\n", __FILE__, __LINE__, __VA_ARGS__)
#define LFS_TRACE(...) LFS_TRACE_(__VA_ARGS__, "")
#else
#define LFS_TRACE(...)
#endif
#endif
#ifndef LFS_DEBUG
#ifndef LFS_NO_DEBUG
#define LFS_DEBUG_(fmt, ...) \
printf("%s:%d:debug: " fmt "%s\n", __FILE__, __LINE__, __VA_ARGS__)
#define LFS_DEBUG(...) LFS_DEBUG_(__VA_ARGS__, "")
#else
#define LFS_DEBUG(...)
#endif
#endif
#ifndef LFS_WARN
#ifndef LFS_NO_WARN
#define LFS_WARN_(fmt, ...) \
printf("%s:%d:warn: " fmt "%s\n", __FILE__, __LINE__, __VA_ARGS__)
#define LFS_WARN(...) LFS_WARN_(__VA_ARGS__, "")
#else
#define LFS_WARN(...)
#endif
#endif
#ifndef LFS_ERROR
#ifndef LFS_NO_ERROR
#define LFS_ERROR_(fmt, ...) \
printf("%s:%d:error: " fmt "%s\n", __FILE__, __LINE__, __VA_ARGS__)
#define LFS_ERROR(...) LFS_ERROR_(__VA_ARGS__, "")
#else
#define LFS_ERROR(...)
#endif
#endif
// Runtime assertions
#ifndef LFS_ASSERT
#ifndef LFS_NO_ASSERT
#define LFS_ASSERT(test) assert(test)
#else
#define LFS_ASSERT(test)
#endif
#endif
#ifndef LFS_UNREACHABLE
#ifndef LFS_NO_ASSERT
#define LFS_UNREACHABLE() __builtin_unreachable()
#else
#define LFS_UNREACHABLE() LFS_ASSERT(false)
#endif
#endif
// Builtin functions, these may be replaced by more efficient
// toolchain-specific implementations. LFS_NO_INTRINSICS falls back to a more
// expensive basic C implementation for debugging purposes
// Min/max functions for unsigned 32-bit numbers
static inline uint32_t lfs_max(uint32_t a, uint32_t b) {
return (a > b) ? a : b;
}
static inline uint32_t lfs_min(uint32_t a, uint32_t b) {
return (a < b) ? a : b;
}
static inline uint32_t lfs_max32(uint32_t a, uint32_t b) {
return (a > b) ? a : b;
}
static inline uint32_t lfs_min32(uint32_t a, uint32_t b) {
return (a < b) ? a : b;
}
static inline int32_t lfs_smax32(int32_t a, int32_t b) {
return (a > b) ? a : b;
}
static inline int32_t lfs_smin32(int32_t a, int32_t b) {
return (a < b) ? a : b;
}
static inline uint16_t lfs_max16(uint16_t a, uint16_t b) {
return (a > b) ? a : b;
}
static inline uint16_t lfs_min16(uint16_t a, uint16_t b) {
return (a < b) ? a : b;
}
// Absolute value of signed numbers
static inline int32_t lfs_abs32(int32_t a) {
return a < 0 ? -a : a;
}
// TODO how many of these do we actually need
// Swap two 16-bit numbers
static inline void lfs_swap16(uint16_t *a, uint16_t *b) {
uint16_t t = *a;
*a = *b;
*b = t;
}
static inline void lfs_sswap16(int16_t *a, int16_t *b) {
int16_t t = *a;
*a = *b;
*b = t;
}
// Swap two 32-bit numbers
static inline void lfs_swap32(uint32_t *a, uint32_t *b) {
uint32_t t = *a;
*a = *b;
*b = t;
}
static inline void lfs_sswap32(int32_t *a, int32_t *b) {
int32_t t = *a;
*a = *b;
*b = t;
}
// Align to nearest multiple of a size
static inline uint32_t lfs_aligndown(uint32_t a, uint32_t alignment) {
return a - (a % alignment);
}
static inline uint32_t lfs_alignup(uint32_t a, uint32_t alignment) {
return lfs_aligndown(a + alignment-1, alignment);
}
// Find the smallest power of 2 greater than or equal to a
static inline uint32_t lfs_npw2(uint32_t a) {
#if !defined(LFS_NO_INTRINSICS) && (defined(__GNUC__) || defined(__CC_ARM))
return 32 - __builtin_clz(a-1);
#else
uint32_t r = 0;
uint32_t s;
a -= 1;
s = (a > 0xffff) << 4; a >>= s; r |= s;
s = (a > 0xff ) << 3; a >>= s; r |= s;
s = (a > 0xf ) << 2; a >>= s; r |= s;
s = (a > 0x3 ) << 1; a >>= s; r |= s;
return (r | (a >> 1)) + 1;
#endif
}
// TODO we should eventually adopt this as the new name for npw2
// Find the ceiling of log base 2 of the given number
static inline uint32_t lfs_nlog2(uint32_t a) {
return lfs_npw2(a);
}
// Count the number of trailing binary zeros in a
// lfs_ctz(0) may be undefined
static inline uint32_t lfs_ctz(uint32_t a) {
#if !defined(LFS_NO_INTRINSICS) && defined(__GNUC__)
return __builtin_ctz(a);
#else
return lfs_npw2((a & -a) + 1) - 1;
#endif
}
// Count the number of binary ones in a
static inline uint32_t lfs_popc(uint32_t a) {
#if !defined(LFS_NO_INTRINSICS) && (defined(__GNUC__) || defined(__CC_ARM))
return __builtin_popcount(a);
#else
a = a - ((a >> 1) & 0x55555555);
a = (a & 0x33333333) + ((a >> 2) & 0x33333333);
return (((a + (a >> 4)) & 0xf0f0f0f) * 0x1010101) >> 24;
#endif
}
// Find the sequence comparison of a and b, this is the distance
// between a and b ignoring overflow
static inline int lfs_scmp(uint32_t a, uint32_t b) {
return (int)(unsigned)(a - b);
}
// Convert between 32-bit little-endian and native order
static inline uint32_t lfs_fromle32(uint32_t a) {
#if (defined( BYTE_ORDER ) && defined( ORDER_LITTLE_ENDIAN ) && BYTE_ORDER == ORDER_LITTLE_ENDIAN ) || \
(defined(__BYTE_ORDER ) && defined(__ORDER_LITTLE_ENDIAN ) && __BYTE_ORDER == __ORDER_LITTLE_ENDIAN ) || \
(defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
return a;
#elif !defined(LFS_NO_INTRINSICS) && ( \
(defined( BYTE_ORDER ) && defined( ORDER_BIG_ENDIAN ) && BYTE_ORDER == ORDER_BIG_ENDIAN ) || \
(defined(__BYTE_ORDER ) && defined(__ORDER_BIG_ENDIAN ) && __BYTE_ORDER == __ORDER_BIG_ENDIAN ) || \
(defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__))
return __builtin_bswap32(a);
#else
return (((uint8_t*)&a)[0] << 0) |
(((uint8_t*)&a)[1] << 8) |
(((uint8_t*)&a)[2] << 16) |
(((uint8_t*)&a)[3] << 24);
#endif
}
static inline uint32_t lfs_tole32(uint32_t a) {
return lfs_fromle32(a);
}
// Convert between 32-bit big-endian and native order
static inline uint32_t lfs_frombe32(uint32_t a) {
#if !defined(LFS_NO_INTRINSICS) && ( \
(defined( BYTE_ORDER ) && defined( ORDER_LITTLE_ENDIAN ) && BYTE_ORDER == ORDER_LITTLE_ENDIAN ) || \
(defined(__BYTE_ORDER ) && defined(__ORDER_LITTLE_ENDIAN ) && __BYTE_ORDER == __ORDER_LITTLE_ENDIAN ) || \
(defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__))
return __builtin_bswap32(a);
#elif (defined( BYTE_ORDER ) && defined( ORDER_BIG_ENDIAN ) && BYTE_ORDER == ORDER_BIG_ENDIAN ) || \
(defined(__BYTE_ORDER ) && defined(__ORDER_BIG_ENDIAN ) && __BYTE_ORDER == __ORDER_BIG_ENDIAN ) || \
(defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
return a;
#else
return (((uint8_t*)&a)[0] << 24) |
(((uint8_t*)&a)[1] << 16) |
(((uint8_t*)&a)[2] << 8) |
(((uint8_t*)&a)[3] << 0);
#endif
}
static inline uint32_t lfs_tobe32(uint32_t a) {
return lfs_frombe32(a);
}
// Convert to/from 16-bit little-endian
static inline void lfs_tole16_(uint16_t word, void *buffer) {
((uint8_t*)buffer)[0] = word >> 0;
((uint8_t*)buffer)[1] = word >> 8;
}
static inline uint16_t lfs_fromle16_(const void *buffer) {
return (((uint8_t*)buffer)[0] << 0)
| (((uint8_t*)buffer)[1] << 8);
}
// Convert to/from 32-bit little-endian
static inline void lfs_tole32_(uint32_t word, void *buffer) {
((uint8_t*)buffer)[0] = word >> 0;
((uint8_t*)buffer)[1] = word >> 8;
((uint8_t*)buffer)[2] = word >> 16;
((uint8_t*)buffer)[3] = word >> 24;
}
static inline uint32_t lfs_fromle32_(const void *buffer) {
return (((uint8_t*)buffer)[0] << 0)
| (((uint8_t*)buffer)[1] << 8)
| (((uint8_t*)buffer)[2] << 16)
| (((uint8_t*)buffer)[3] << 24);
}
// Convert to/from leb128 encoding
// TODO should we really be using ssize_t here and not lfs_ssize_t?
ssize_t lfs_toleb128(uint32_t word, void *buffer, size_t size);
ssize_t lfs_fromleb128(uint32_t *word, const void *buffer, size_t size);
// Calculate CRC-32 with polynomial = 0x04c11db7
uint32_t lfs_crc(uint32_t crc, const void *buffer, size_t size);
// Calculate crc32c incrementally
//
// polynomial = 0x11edc6f41
// init = 0xffffffff
// fini = 0xffffffff
//
uint32_t lfs_crc32c(uint32_t crc, const void *buffer, size_t size);
// Allocate memory, only used if buffers are not provided to littlefs
// Note, memory must be 64-bit aligned
static inline void *lfs_malloc(size_t size) {
#ifndef LFS_NO_MALLOC
return malloc(size);
#else
(void)size;
return NULL;
#endif
}
// Deallocate memory, only used if buffers are not provided to littlefs
static inline void lfs_free(void *p) {
#ifndef LFS_NO_MALLOC
free(p);
#else
(void)p;
#endif
}
#ifdef __cplusplus
} /* extern "C" */
#endif
#endif
#endif
View Git Blame Copy Permalink