Changed callbacks to take user-provided context directly

This is a style change to make littlefs's callbacks consistent with most
callback declarations found in C. That is, taking in a user-provided
`void*`.

Previously, these callbacks took a pointer to the config struct itself,
which indirectly contained a user provided context, and this gets the
job done, but taking in a callback with a `void*` is arguably more
expected, has a better chance of integrating with C++/OS-specific code,
and is more likely to be optimized out by a clever compiler.

---

As a part of these changes, the geometry for the test bds needed to be
moved into bd specific configuration objects. This is a good change as
it also allows for testing situations where littlefs's geometry does not
match the underlying bd.
This commit is contained in:
Christopher Haster
2020-11-26 09:16:20 -06:00
parent a549413077
commit a7cdd563f6
15 changed files with 383 additions and 397 deletions

View File

@@ -19,15 +19,15 @@ code = '''
lfs_dir_close(&lfs, &dir) => 0;
lfs_unmount(&lfs) => 0;
uint8_t bbuffer[LFS_BLOCK_SIZE];
cfg.read(&cfg, block, 0, bbuffer, LFS_BLOCK_SIZE) => 0;
lfs_testbd_read(&bd, block, 0, bbuffer, LFS_BLOCK_SIZE) => 0;
int off = LFS_BLOCK_SIZE-1;
while (off >= 0 && bbuffer[off] == LFS_ERASE_VALUE) {
off -= 1;
}
memset(&bbuffer[off-3], LFS_BLOCK_SIZE, 3);
cfg.erase(&cfg, block) => 0;
cfg.prog(&cfg, block, 0, bbuffer, LFS_BLOCK_SIZE) => 0;
cfg.sync(&cfg) => 0;
lfs_testbd_erase(&bd, block) => 0;
lfs_testbd_prog(&bd, block, 0, bbuffer, LFS_BLOCK_SIZE) => 0;
lfs_testbd_sync(&bd) => 0;
lfs_mountcfg(&lfs, &cfg) => 0;
lfs_stat(&lfs, "parent/orphan", &info) => LFS_ERR_NOENT;