Files
thirdparty-littlefs/testbd/lfs_testbd.c
Christopher Haster 52ef0c1c9e Fixed a crazy consistency issue in test.py
The root of the problem was the notorious Python quirk with mutable
default parameters. The default defines for the TestSuite class ended
up being mutated as the class determined the permutations to test,
corrupting other test's defines.

However, the only define that was mutated this way was the CACHE_SIZE
config in test_entries.

The crazy thing was how this small innocuous change would cause
"./scripts/test.py -nr test_relocations" and "./scripts/test.py -nr"
to drift out of sync only after a commit spanning the different cache
sizes would be written out with a different number of prog calls. This
offset the power-cycle counter enough to cause one case to make it to
an erase, and the other to not.

Normally, the difference between a successful/unsuccessful erase
wouldn't change the result of a test, but in this case it offset the
revision count used for wear-leveling, causing one run run expand the
superblock and the other to not.

This change to the filesystem would then propogate through the rest of
the test, making it difficult to reproduce test failures.

Fortunately the fix was to just make a copy of the default define
dictionary. This should also prevent accidently mutating of dicts
belonging to our caller.

Oh, also fixed a buffer overflow in test_files.
2020-01-26 23:53:53 -06:00

290 lines
9.3 KiB
C

/*
* Testing block device, wraps filebd and rambd while providing a bunch
* of hooks for testing littlefs in various conditions.
*
* Copyright (c) 2017, Arm Limited. All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
*/
#include "testbd/lfs_testbd.h"
#include <stdlib.h>
int lfs_testbd_createcfg(const struct lfs_config *cfg, const char *path,
const struct lfs_testbd_config *bdcfg) {
LFS_TRACE("lfs_testbd_createcfg(%p {.context=%p, "
".read=%p, .prog=%p, .erase=%p, .sync=%p, "
".read_size=%"PRIu32", .prog_size=%"PRIu32", "
".block_size=%"PRIu32", .block_count=%"PRIu32"}, "
"\"%s\", "
"%p {.erase_value=%"PRId32", .erase_cycles=%"PRIu32", "
".badblock_behavior=%"PRIu8", .power_cycles=%"PRIu32", "
".buffer=%p, .wear_buffer=%p})",
(void*)cfg, cfg->context,
(void*)(uintptr_t)cfg->read, (void*)(uintptr_t)cfg->prog,
(void*)(uintptr_t)cfg->erase, (void*)(uintptr_t)cfg->sync,
cfg->read_size, cfg->prog_size, cfg->block_size, cfg->block_count,
path, (void*)bdcfg, bdcfg->erase_value, bdcfg->erase_cycles,
bdcfg->badblock_behavior, bdcfg->power_cycles,
bdcfg->buffer, bdcfg->wear_buffer);
lfs_testbd_t *bd = cfg->context;
bd->cfg = bdcfg;
// setup testing things
bd->persist = path;
bd->power_cycles = bd->cfg->power_cycles;
if (bd->cfg->erase_cycles) {
if (bd->cfg->wear_buffer) {
bd->wear = bd->cfg->wear_buffer;
} else {
bd->wear = lfs_malloc(sizeof(lfs_testbd_wear_t) * cfg->block_count);
if (!bd->wear) {
LFS_TRACE("lfs_testbd_createcfg -> %d", LFS_ERR_NOMEM);
return LFS_ERR_NOMEM;
}
}
memset(bd->wear, 0, sizeof(lfs_testbd_wear_t) * cfg->block_count);
}
// create underlying block device
if (bd->persist) {
bd->u.file.cfg = (struct lfs_filebd_config){
.erase_value = bd->cfg->erase_value,
};
int err = lfs_filebd_createcfg(cfg, path, &bd->u.file.cfg);
LFS_TRACE("lfs_testbd_createcfg -> %d", err);
return err;
} else {
bd->u.ram.cfg = (struct lfs_rambd_config){
.erase_value = bd->cfg->erase_value,
.buffer = bd->cfg->buffer,
};
int err = lfs_rambd_createcfg(cfg, &bd->u.ram.cfg);
LFS_TRACE("lfs_testbd_createcfg -> %d", err);
return err;
}
}
int lfs_testbd_create(const struct lfs_config *cfg, const char *path) {
LFS_TRACE("lfs_testbd_create(%p {.context=%p, "
".read=%p, .prog=%p, .erase=%p, .sync=%p, "
".read_size=%"PRIu32", .prog_size=%"PRIu32", "
".block_size=%"PRIu32", .block_count=%"PRIu32"}, "
"\"%s\")",
(void*)cfg, cfg->context,
(void*)(uintptr_t)cfg->read, (void*)(uintptr_t)cfg->prog,
(void*)(uintptr_t)cfg->erase, (void*)(uintptr_t)cfg->sync,
cfg->read_size, cfg->prog_size, cfg->block_size, cfg->block_count,
path);
static const struct lfs_testbd_config defaults = {.erase_value=-1};
int err = lfs_testbd_createcfg(cfg, path, &defaults);
LFS_TRACE("lfs_testbd_create -> %d", err);
return err;
}
int lfs_testbd_destroy(const struct lfs_config *cfg) {
LFS_TRACE("lfs_testbd_destroy(%p)", (void*)cfg);
lfs_testbd_t *bd = cfg->context;
if (bd->cfg->erase_cycles && !bd->cfg->wear_buffer) {
lfs_free(bd->wear);
}
if (bd->persist) {
int err = lfs_filebd_destroy(cfg);
LFS_TRACE("lfs_testbd_destroy -> %d", err);
return err;
} else {
int err = lfs_rambd_destroy(cfg);
LFS_TRACE("lfs_testbd_destroy -> %d", err);
return err;
}
}
/// Internal mapping to block devices ///
static int lfs_testbd_rawread(const struct lfs_config *cfg, lfs_block_t block,
lfs_off_t off, void *buffer, lfs_size_t size) {
lfs_testbd_t *bd = cfg->context;
if (bd->persist) {
return lfs_filebd_read(cfg, block, off, buffer, size);
} else {
return lfs_rambd_read(cfg, block, off, buffer, size);
}
}
static int lfs_testbd_rawprog(const struct lfs_config *cfg, lfs_block_t block,
lfs_off_t off, const void *buffer, lfs_size_t size) {
lfs_testbd_t *bd = cfg->context;
if (bd->persist) {
return lfs_filebd_prog(cfg, block, off, buffer, size);
} else {
return lfs_rambd_prog(cfg, block, off, buffer, size);
}
}
static int lfs_testbd_rawerase(const struct lfs_config *cfg,
lfs_block_t block) {
lfs_testbd_t *bd = cfg->context;
if (bd->persist) {
return lfs_filebd_erase(cfg, block);
} else {
return lfs_rambd_erase(cfg, block);
}
}
static int lfs_testbd_rawsync(const struct lfs_config *cfg) {
lfs_testbd_t *bd = cfg->context;
if (bd->persist) {
return lfs_filebd_sync(cfg);
} else {
return lfs_rambd_sync(cfg);
}
}
/// block device API ///
int lfs_testbd_read(const struct lfs_config *cfg, lfs_block_t block,
lfs_off_t off, void *buffer, lfs_size_t size) {
LFS_TRACE("lfs_testbd_read(%p, 0x%"PRIx32", %"PRIu32", %p, %"PRIu32")",
(void*)cfg, block, off, buffer, size);
lfs_testbd_t *bd = cfg->context;
// check if read is valid
LFS_ASSERT(off % cfg->read_size == 0);
LFS_ASSERT(size % cfg->read_size == 0);
LFS_ASSERT(block < cfg->block_count);
// block bad?
if (bd->cfg->erase_cycles &&
bd->cfg->badblock_behavior == LFS_TESTBD_BADBLOCK_NOREAD &&
bd->wear[block] >= bd->cfg->erase_cycles) {
LFS_TRACE("lfs_testbd_read -> %d", LFS_ERR_CORRUPT);
return LFS_ERR_CORRUPT;
}
// read
int err = lfs_testbd_rawread(cfg, block, off, buffer, size);
LFS_TRACE("lfs_testbd_read -> %d", err);
return err;
}
int lfs_testbd_prog(const struct lfs_config *cfg, lfs_block_t block,
lfs_off_t off, const void *buffer, lfs_size_t size) {
LFS_TRACE("lfs_testbd_prog(%p, 0x%"PRIx32", %"PRIu32", %p, %"PRIu32")",
(void*)cfg, block, off, buffer, size);
lfs_testbd_t *bd = cfg->context;
// check if write is valid
LFS_ASSERT(off % cfg->prog_size == 0);
LFS_ASSERT(size % cfg->prog_size == 0);
LFS_ASSERT(block < cfg->block_count);
// block bad?
if (bd->cfg->erase_cycles &&
bd->cfg->badblock_behavior == LFS_TESTBD_BADBLOCK_NOPROG &&
bd->wear[block] >= bd->cfg->erase_cycles) {
LFS_TRACE("lfs_testbd_prog -> %d", LFS_ERR_CORRUPT);
return LFS_ERR_CORRUPT;
}
// prog
int err = lfs_testbd_rawprog(cfg, block, off, buffer, size);
if (err) {
LFS_TRACE("lfs_testbd_prog -> %d", err);
return err;
}
// lose power?
if (bd->power_cycles > 0) {
bd->power_cycles -= 1;
if (bd->power_cycles == 0) {
// sync to make sure we persist the last changes
assert(lfs_testbd_rawsync(cfg) == 0);
// simulate power loss
exit(33);
}
}
LFS_TRACE("lfs_testbd_prog -> %d", 0);
return 0;
}
int lfs_testbd_erase(const struct lfs_config *cfg, lfs_block_t block) {
LFS_TRACE("lfs_testbd_erase(%p, 0x%"PRIx32")", (void*)cfg, block);
lfs_testbd_t *bd = cfg->context;
// check if erase is valid
LFS_ASSERT(block < cfg->block_count);
// block bad?
if (bd->cfg->erase_cycles) {
if (bd->wear[block] >= bd->cfg->erase_cycles) {
if (bd->cfg->badblock_behavior == LFS_TESTBD_BADBLOCK_NOERASE) {
LFS_TRACE("lfs_testbd_erase -> %d", LFS_ERR_CORRUPT);
return LFS_ERR_CORRUPT;
}
} else {
// mark wear
bd->wear[block] += 1;
}
}
// erase
int err = lfs_testbd_rawerase(cfg, block);
if (err) {
LFS_TRACE("lfs_testbd_erase -> %d", err);
return err;
}
// lose power?
if (bd->power_cycles > 0) {
bd->power_cycles -= 1;
if (bd->power_cycles == 0) {
// sync to make sure we persist the last changes
assert(lfs_testbd_rawsync(cfg) == 0);
// simulate power loss
exit(33);
}
}
LFS_TRACE("lfs_testbd_prog -> %d", 0);
return 0;
}
int lfs_testbd_sync(const struct lfs_config *cfg) {
LFS_TRACE("lfs_testbd_sync(%p)", (void*)cfg);
int err = lfs_testbd_rawsync(cfg);
LFS_TRACE("lfs_testbd_sync -> %d", err);
return err;
}
/// simulated wear operations ///
lfs_testbd_swear_t lfs_testbd_getwear(const struct lfs_config *cfg,
lfs_block_t block) {
LFS_TRACE("lfs_testbd_getwear(%p, %"PRIu32")", (void*)cfg, block);
lfs_testbd_t *bd = cfg->context;
// check if block is valid
LFS_ASSERT(bd->cfg->erase_cycles);
LFS_ASSERT(block < cfg->block_count);
LFS_TRACE("lfs_testbd_getwear -> %"PRIu32, bd->wear[block]);
return bd->wear[block];
}
int lfs_testbd_setwear(const struct lfs_config *cfg,
lfs_block_t block, lfs_testbd_wear_t wear) {
LFS_TRACE("lfs_testbd_setwear(%p, %"PRIu32")", (void*)cfg, block);
lfs_testbd_t *bd = cfg->context;
// check if block is valid
LFS_ASSERT(bd->cfg->erase_cycles);
LFS_ASSERT(block < cfg->block_count);
bd->wear[block] = wear;
LFS_TRACE("lfs_testbd_setwear -> %d", 0);
return 0;
}