458fe16f38
Metastability is a rather nasty error condition where successive reads
to a memory location may return different values, either due to bus
issues or a failed prog. It's a tricky error condition to detect, and
one that ckreads was, in theory, supposed to help with.
To help test metastability (and other single-bit errors), emubd gained
several new features:
- LFS_EMUBD_BADBLOCK_PROGFLIP - Prog flips a bit
- LFS_EMUBD_BADBLOCK_READFLIP - Read flips a bit sometimes
- LFS_EMUBD_POWERLOSS_METASTABLE - Reads may flip a bit
These only affect a single bit in a given block, but by randomizing
which bit during every erase (and exhaustive bit testing in test_ck) we
should still see some fairly interesting bit-error patterns over time.
It's a bit difficult to test with more than a single bit error because
you can quickly find checksum/parity collisions when fuzz testing. But
there may be other interesting error patterns to look at in the future?
Also the erase_cycles implementation got a bit of a rework since it was
lopsided previously (progs/reads would always error before erases). And
since I was messing with emubd's internals I added lfs_emubd_markbad/
markgood and a few other convenience functions that seem useful:
- lfs_emubd_seed - Manually set the prng, needed in test_ck actually
- lfs_emubd_markbad - Mark block as bad, same as wear=-1
- lfs_emubd_markgood - Mark block as good, same as wear=0
- lfs_emubd_badbit - Get which big failed
- lfs_emubd_setbadbit - Set which bit will fail
- lfs_emubd_randomizebadbit - Randomize bad bit on erase
- lfs_emubd_markbadbit - Mark bit as bad, same as setbadbit+markbad
---
The intention of this new metastability emulation was to extend test_ck
to test ckreads/ckprogs. This went... interestingly.
The good news, the new emulation and tests worked quite well. They were
able to quite quickly show that ckreads is fundamentally not able to
detect all single-bit errors in our current design.
The problem boils down to the fact that the location of our parity bits
depends on the tag's leb128-encoded size. If a bit flip changes this
size field, we end up with a new parity bit, which 50/50 may or may not
detect the error.
For example, one bit flip:
40 0c 00 12 80 0d ff ff
'----.----' ^--------------------.
'- altble 0xc w0 -18 parity=1
40 0c 80 12 80 0d ff ff
'-------.-------' ^----------------------.
'- altble 0xc w2304 -1664 parity=1
This doesn't make ckreads _completely_ useless, just mostly useless. We
can still use it to check parity bits, but without a systematic proof.
But there's enough problems with ckreads: performance, RAM, code, etc,
that I think it may just be an interesting proof-of-concept and not
something users should actually use. Checking reads in the bd-layer
solves all of these problems...
---
At the very least ckprogs gets better testing, thanks to new tests in
test_ck and the addition of LFS_EMUBD_BADBLOCK_PROGFLIP in
test_badblocks.
The extra testing also found a ckprog/ckread hole in that we don't
ckprog/ckread during lfsr_format! I fixed this by making lfsr_format
always use ckprogs/ckreads if available, but maybe lfsr_format should
take its own set of flags?
Funnily enough this had no impact on code size since it probably just
changed the constant in a constant pool:
code stack
before: 37872 3048
after: 37872 (+0.0%) 3048 (+0.0%)
268 lines
8.6 KiB
C
268 lines
8.6 KiB
C
/*
|
|
* Emulating block device, wraps filebd and rambd while providing a bunch
|
|
* of hooks for testing littlefs in various conditions.
|
|
*
|
|
* Copyright (c) 2022, The littlefs authors.
|
|
* Copyright (c) 2017, Arm Limited. All rights reserved.
|
|
* SPDX-License-Identifier: BSD-3-Clause
|
|
*/
|
|
#ifndef LFS_EMUBD_H
|
|
#define LFS_EMUBD_H
|
|
|
|
#include "lfs.h"
|
|
#include "lfs_util.h"
|
|
#include "bd/lfs_rambd.h"
|
|
#include "bd/lfs_filebd.h"
|
|
|
|
#ifdef __cplusplus
|
|
extern "C"
|
|
{
|
|
#endif
|
|
|
|
|
|
// Block device specific tracing
|
|
#ifndef LFS_EMUBD_TRACE
|
|
#ifdef LFS_EMUBD_YES_TRACE
|
|
#define LFS_EMUBD_TRACE(...) LFS_TRACE(__VA_ARGS__)
|
|
#else
|
|
#define LFS_EMUBD_TRACE(...)
|
|
#endif
|
|
#endif
|
|
|
|
// Mode determining how "bad-blocks" behave during testing. This simulates
|
|
// some real-world circumstances such as progs not sticking (prog-noop),
|
|
// a readonly disk (erase-noop), ECC failures (read-error), and of course,
|
|
// random bit failures (prog-flip, read-flip)
|
|
typedef enum lfs_emubd_badblock_behavior {
|
|
LFS_EMUBD_BADBLOCK_PROGERROR = 0, // Error on prog
|
|
LFS_EMUBD_BADBLOCK_ERASEERROR = 1, // Error on erase
|
|
LFS_EMUBD_BADBLOCK_READERROR = 2, // Error on read
|
|
LFS_EMUBD_BADBLOCK_PROGNOOP = 3, // Prog does nothing silently
|
|
LFS_EMUBD_BADBLOCK_ERASENOOP = 4, // Erase does nothing silently
|
|
LFS_EMUBD_BADBLOCK_PROGFLIP = 5, // Prog flips a bit
|
|
LFS_EMUBD_BADBLOCK_READFLIP = 6, // Read flips a bit sometimes
|
|
} lfs_emubd_badblock_behavior_t;
|
|
|
|
// Mode determining how power-loss behaves during testing.
|
|
typedef enum lfs_emubd_powerloss_behavior {
|
|
LFS_EMUBD_POWERLOSS_NOOP = 0, // Progs are atomic
|
|
LFS_EMUBD_POWERLOSS_SOMEBITS = 1, // One bit is progged
|
|
LFS_EMUBD_POWERLOSS_MOSTBITS = 2, // All-but-one bit is progged
|
|
LFS_EMUBD_POWERLOSS_OOO = 3, // Blocks are written out-of-order
|
|
LFS_EMUBD_POWERLOSS_METASTABLE = 4, // Reads may flip a bit
|
|
} lfs_emubd_powerloss_behavior_t;
|
|
|
|
// Type for measuring read/program/erase operations
|
|
typedef uint64_t lfs_emubd_io_t;
|
|
typedef int64_t lfs_emubd_sio_t;
|
|
|
|
// Type for measuring wear
|
|
typedef uint32_t lfs_emubd_wear_t;
|
|
typedef int32_t lfs_emubd_swear_t;
|
|
|
|
// Type for tracking power-cycles
|
|
typedef uint32_t lfs_emubd_powercycles_t;
|
|
typedef int32_t lfs_emubd_spowercycles_t;
|
|
|
|
// Type for delays in nanoseconds
|
|
typedef uint64_t lfs_emubd_sleep_t;
|
|
typedef int64_t lfs_emubd_ssleep_t;
|
|
|
|
// emubd config, this is required for testing
|
|
struct lfs_emubd_config {
|
|
// 8-bit erase value to use for simulating erases. -1 simulates a noop
|
|
// erase, which is faster than simulating a fixed erase value.
|
|
int32_t erase_value;
|
|
|
|
// Number of erase cycles before a block becomes "bad". The exact behavior
|
|
// of bad blocks is controlled by badblock_behavior.
|
|
uint32_t erase_cycles;
|
|
|
|
// The mode determining how bad-blocks fail
|
|
lfs_emubd_badblock_behavior_t badblock_behavior;
|
|
|
|
// Number of write operations (erase/prog) before triggering a power-loss.
|
|
// power_cycles=0 disables this. The exact behavior of power-loss is
|
|
// controlled by a combination of powerloss_behavior and powerloss_cb.
|
|
lfs_emubd_powercycles_t power_cycles;
|
|
|
|
// The mode determining how power-loss affects disk
|
|
lfs_emubd_powerloss_behavior_t powerloss_behavior;
|
|
|
|
// Function to call to emulate power-loss. The exact behavior of power-loss
|
|
// is up to the runner to provide.
|
|
void (*powerloss_cb)(void*);
|
|
|
|
// Data for power-loss callback
|
|
void *powerloss_data;
|
|
|
|
// Seed for prng, which may be used for emulating failed progs. This does
|
|
// not affect normal operation.
|
|
uint32_t seed;
|
|
|
|
// Path to file to use as a mirror of the disk. This provides a way to view
|
|
// the current state of the block device.
|
|
const char *disk_path;
|
|
|
|
// Artificial delay in nanoseconds, there is no purpose for this other
|
|
// than slowing down the simulation.
|
|
lfs_emubd_sleep_t read_sleep;
|
|
|
|
// Artificial delay in nanoseconds, there is no purpose for this other
|
|
// than slowing down the simulation.
|
|
lfs_emubd_sleep_t prog_sleep;
|
|
|
|
// Artificial delay in nanoseconds, there is no purpose for this other
|
|
// than slowing down the simulation.
|
|
lfs_emubd_sleep_t erase_sleep;
|
|
};
|
|
|
|
// A reference counted block
|
|
typedef struct lfs_emubd_block {
|
|
uint32_t rc;
|
|
lfs_emubd_wear_t wear;
|
|
bool metastable;
|
|
// sign(bad_bit)=0 => randomized on erase
|
|
// sign(bad_bit)=1 => fixed
|
|
lfs_size_t bad_bit;
|
|
|
|
uint8_t data[];
|
|
} lfs_emubd_block_t;
|
|
|
|
// Disk mirror
|
|
typedef struct lfs_emubd_disk {
|
|
uint32_t rc;
|
|
int fd;
|
|
uint8_t *scratch;
|
|
} lfs_emubd_disk_t;
|
|
|
|
// emubd state
|
|
typedef struct lfs_emubd {
|
|
// array of copy-on-write blocks
|
|
lfs_emubd_block_t **blocks;
|
|
|
|
// some other test state
|
|
lfs_emubd_io_t readed;
|
|
lfs_emubd_io_t proged;
|
|
lfs_emubd_io_t erased;
|
|
uint32_t prng;
|
|
lfs_emubd_powercycles_t power_cycles;
|
|
lfs_emubd_block_t **ooo_before;
|
|
lfs_emubd_block_t **ooo_after;
|
|
lfs_emubd_disk_t *disk;
|
|
|
|
const struct lfs_emubd_config *cfg;
|
|
} lfs_emubd_t;
|
|
|
|
|
|
/// Block device API ///
|
|
|
|
// Create an emulating block device using the geometry in lfs_config
|
|
//
|
|
// Note that filebd is used if a path is provided, if path is NULL
|
|
// emubd will use rambd which can be much faster.
|
|
int lfs_emubd_create(const struct lfs_config *cfg, const char *path);
|
|
int lfs_emubd_createcfg(const struct lfs_config *cfg, const char *path,
|
|
const struct lfs_emubd_config *bdcfg);
|
|
|
|
// Clean up memory associated with block device
|
|
int lfs_emubd_destroy(const struct lfs_config *cfg);
|
|
|
|
// Read a block
|
|
int lfs_emubd_read(const struct lfs_config *cfg, lfs_block_t block,
|
|
lfs_off_t off, void *buffer, lfs_size_t size);
|
|
|
|
// Program a block
|
|
//
|
|
// The block must have previously been erased.
|
|
int lfs_emubd_prog(const struct lfs_config *cfg, lfs_block_t block,
|
|
lfs_off_t off, const void *buffer, lfs_size_t size);
|
|
|
|
// Erase a block
|
|
//
|
|
// A block must be erased before being programmed. The
|
|
// state of an erased block is undefined.
|
|
int lfs_emubd_erase(const struct lfs_config *cfg, lfs_block_t block);
|
|
|
|
// Sync the block device
|
|
int lfs_emubd_sync(const struct lfs_config *cfg);
|
|
|
|
|
|
/// Additional extended API for driving test features ///
|
|
|
|
// Set the current prng state
|
|
int lfs_emubd_seed(const struct lfs_config *cfg, uint32_t seed);
|
|
|
|
// A checksum of a block for debugging purposes
|
|
int lfs_emubd_cksum(const struct lfs_config *cfg,
|
|
lfs_block_t block, uint32_t *cksum);
|
|
|
|
// A checksum of the entire block device for debugging purposes
|
|
int lfs_emubd_bdcksum(const struct lfs_config *cfg, uint32_t *cksum);
|
|
|
|
// Get total amount of bytes read
|
|
lfs_emubd_sio_t lfs_emubd_readed(const struct lfs_config *cfg);
|
|
|
|
// Get total amount of bytes programmed
|
|
lfs_emubd_sio_t lfs_emubd_proged(const struct lfs_config *cfg);
|
|
|
|
// Get total amount of bytes erased
|
|
lfs_emubd_sio_t lfs_emubd_erased(const struct lfs_config *cfg);
|
|
|
|
// Manually set amount of bytes read
|
|
int lfs_emubd_setreaded(const struct lfs_config *cfg, lfs_emubd_io_t readed);
|
|
|
|
// Manually set amount of bytes programmed
|
|
int lfs_emubd_setproged(const struct lfs_config *cfg, lfs_emubd_io_t proged);
|
|
|
|
// Manually set amount of bytes erased
|
|
int lfs_emubd_seterased(const struct lfs_config *cfg, lfs_emubd_io_t erased);
|
|
|
|
// Get simulated wear on a given block
|
|
lfs_emubd_swear_t lfs_emubd_wear(const struct lfs_config *cfg,
|
|
lfs_block_t block);
|
|
|
|
// Manually set simulated wear on a given block
|
|
int lfs_emubd_setwear(const struct lfs_config *cfg,
|
|
lfs_block_t block, lfs_emubd_wear_t wear);
|
|
|
|
// Mark a block as bad, this is equivalent to setting wear to maximum
|
|
int lfs_emubd_markbad(const struct lfs_config *cfg, lfs_block_t block);
|
|
|
|
// Clear any simulated wear on a given block
|
|
int lfs_emubd_markgood(const struct lfs_config *cfg, lfs_block_t block);
|
|
|
|
// Get which bit failed, this changes on erase/power-loss unless manually set
|
|
lfs_ssize_t lfs_emubd_badbit(const struct lfs_config *cfg,
|
|
lfs_block_t block);
|
|
|
|
// Set which bit should fail in a given block
|
|
int lfs_emubd_setbadbit(const struct lfs_config *cfg,
|
|
lfs_block_t block, lfs_size_t bit);
|
|
|
|
// Randomize the bad bit on erase (the default)
|
|
int lfs_emubd_randomizebadbit(const struct lfs_config *cfg,
|
|
lfs_block_t block);
|
|
|
|
// Mark a block as bad and which bit should fail
|
|
int lfs_emubd_markbadbit(const struct lfs_config *cfg,
|
|
lfs_block_t block, lfs_size_t bit);
|
|
|
|
// Get the remaining power-cycles
|
|
lfs_emubd_spowercycles_t lfs_emubd_powercycles(
|
|
const struct lfs_config *cfg);
|
|
|
|
// Manually set the remaining power-cycles
|
|
int lfs_emubd_setpowercycles(const struct lfs_config *cfg,
|
|
lfs_emubd_powercycles_t power_cycles);
|
|
|
|
// Create a copy-on-write copy of the state of this block device
|
|
int lfs_emubd_copy(const struct lfs_config *cfg, lfs_emubd_t *copy);
|
|
|
|
|
|
#ifdef __cplusplus
|
|
} /* extern "C" */
|
|
#endif
|
|
|
|
#endif
|