Github/littlefs
1
Fork 0
Code Actions Activity

Dropped dsize estimates from attrs that don't end up in shrubs

Browse Source 
The only requirement we have for dsize estimates is to help calculate
shrub estimates. So why bother with dsize for attrs that are never
committed to shrubs?

At the moment, the only tags we commit to shrubs are:

- LFSR_TAG_DATA
- LFSR_TAG_BLOCK
- LFSR_TAG_BRANCH

This saves a decent chunk of code:

           code          stack
  before: 35556           2440          636
  after:  35488 (-0.2%)   2440 (+0.0%)  636 (+0.0%)
This commit is contained in:
Christopher Haster
2025-02-11 03:56:51 -06:00
parent bb4d1c517d
commit b7af2b7a24
1 changed files with 20 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lfs.c
+20 -21
View File
@@ -2143,14 +2143,14 @@ typedef struct lfsr_rattr {
#define LFSR_RATTR_LE32(_tag, _weight, _le32) \
((lfsr_rattr_t){ \
.tag=_tag, \
.data_count=LFSR_LE32_DSIZE, \
.data_count=0, \
.weight=_weight, \
.u.le32=_le32})
#define LFSR_RATTR_LEB128(_tag, _weight, _leb128) \
((lfsr_rattr_t){ \
.tag=_tag, \
.data_count=LFSR_LEB128_DSIZE, \
.data_count=0, \
.weight=_weight, \
.u.leb128=_leb128})
@@ -2164,6 +2164,7 @@ typedef struct lfsr_name {
#define LFSR_RATTR_NAME_(_tag, _weight, _name) \
((lfsr_rattr_t){ \
.tag=_tag, \
.data_count=0, \
.weight=_weight, \
.u.etc=(const lfsr_name_t*){_name}})
@@ -4372,7 +4373,7 @@ static int lfsr_rbyd_appendcksum_(lfs_t *lfs, lfsr_rbyd_t *rbyd,
LFSR_TAG_ECKSUM, 0,
(&(lfsr_ecksum_t){
.cksize=lfs->cfg->prog_size,
.cksum=ecksum}), LFSR_ECKSUM_DSIZE));
.cksum=ecksum}), 0));
if (err) {
return err;
}
@@ -7577,12 +7578,10 @@ static int lfsr_mdir_commit__(lfs_t *lfs, lfsr_mdir_t *mdir,
return err;
}
// write our new shrb tag
// write our new shrub tag
err = lfsr_rbyd_appendrattr(lfs, &mdir->rbyd,
rid - lfs_smax(start_rid, 0),
LFSR_RATTR(
LFSR_TAG_BSHRUB, 0,
&shrub, LFSR_SHRUB_DSIZE));
LFSR_RATTR(LFSR_TAG_BSHRUB, 0, &shrub, 0));
if (err) {
return err;
}
@@ -7917,7 +7916,7 @@ static int lfsr_mdir_compact__(lfs_t *lfs, lfsr_mdir_t *mdir_,
// write the new shrub tag
err = lfsr_rbyd_appendcompactrattr(lfs, &mdir_->rbyd,
LFSR_RATTR(tag, weight, &shrub, LFSR_SHRUB_DSIZE));
LFSR_RATTR(tag, weight, &shrub, 0));
if (err) {
LFS_ASSERT(err != LFS_ERR_RANGE);
return err;
@@ -8363,13 +8362,13 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
0, LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_MDIR, +(1 << lfs->mdir_bits),
mdir_[0].rbyd.blocks, LFSR_MPTR_DSIZE),
mdir_[0].rbyd.blocks, 0),
LFSR_RATTR_DATA(
LFSR_TAG_NAME, +(1 << lfs->mdir_bits),
&split_name),
LFSR_RATTR(
LFSR_TAG_MDIR, 0,
mdir_[1].rbyd.blocks, LFSR_MPTR_DSIZE)));
mdir_[1].rbyd.blocks, 0)));
if (err) {
goto failed;
}
@@ -8383,13 +8382,13 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
lfsr_mid_bid(lfs, mdir->mid), LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_MDIR, 0,
mdir_[0].rbyd.blocks, LFSR_MPTR_DSIZE),
mdir_[0].rbyd.blocks, 0),
LFSR_RATTR_DATA(
LFSR_TAG_NAME, +(1 << lfs->mdir_bits),
&split_name),
LFSR_RATTR(
LFSR_TAG_MDIR, 0,
mdir_[1].rbyd.blocks, LFSR_MPTR_DSIZE)));
mdir_[1].rbyd.blocks, 0)));
if (err) {
goto failed;
}
@@ -8446,7 +8445,7 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
0, LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_MDIR, +(1 << lfs->mdir_bits),
mdir_[0].rbyd.blocks, LFSR_MPTR_DSIZE)));
mdir_[0].rbyd.blocks, 0)));
if (err) {
goto failed;
}
@@ -8460,7 +8459,7 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
lfsr_mid_bid(lfs, mdir->mid), LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_MDIR, 0,
mdir_[0].rbyd.blocks, LFSR_MPTR_DSIZE)));
mdir_[0].rbyd.blocks, 0)));
if (err) {
goto failed;
}
@@ -8515,7 +8514,7 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
-1, LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_SUB | LFSR_TAG_MTREE, 0,
&mtree_, LFSR_BTREE_DSIZE),
&mtree_, 0),
// were we committing to the mroot? include any -1 rattrs
(mdir->mid == -1)
? LFSR_RATTR(
@@ -8566,7 +8565,7 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
-1, LFSR_RATTRS(
LFSR_RATTR(
LFSR_TAG_MROOT, 0,
mrootchild_.rbyd.blocks, LFSR_MPTR_DSIZE)));
mrootchild_.rbyd.blocks, 0)));
if (err) {
LFS_ASSERT(err != LFS_ERR_RANGE);
LFS_ASSERT(err != LFS_ERR_NOENT);
@@ -8614,7 +8613,7 @@ static int lfsr_mdir_commit(lfs_t *lfs, lfsr_mdir_t *mdir,
"littlefs", 8),
LFSR_RATTR(
LFSR_TAG_MROOT, 0,
mrootchild_.rbyd.blocks, LFSR_MPTR_DSIZE)));
mrootchild_.rbyd.blocks, 0)));
if (err) {
LFS_ASSERT(err != LFS_ERR_RANGE);
LFS_ASSERT(err != LFS_ERR_NOENT);
@@ -12382,12 +12381,12 @@ int lfsr_file_sync(lfs_t *lfs, lfsr_file_t *file) {
rattrs[rattr_count++] = LFSR_RATTR(
LFSR_TAG_SUB | LFSR_TAG_BSHRUB, 0,
// note we use the staged trunk here
&file->b.shrub_, LFSR_SHRUB_DSIZE);
&file->b.shrub_, 0);
// btree?
} else if (lfsr_bshrub_isbtree(&file->b)) {
rattrs[rattr_count++] = LFSR_RATTR(
LFSR_TAG_SUB | LFSR_TAG_BTREE, 0,
&file->b.shrub, LFSR_BTREE_DSIZE);
&file->b.shrub, 0);
} else {
LFS_UNREACHABLE();
}
@@ -13810,7 +13809,7 @@ static int lfsr_formatinited(lfs_t *lfs) {
LFSR_TAG_GEOMETRY, 0,
(&(lfsr_geometry_t){
lfs->cfg->block_size,
lfs->cfg->block_count}), LFSR_GEOMETRY_DSIZE),
lfs->cfg->block_count}), 0),
LFSR_RATTR_LEB128(
LFSR_TAG_NAMELIMIT, 0,
lfs->name_limit),
@@ -14341,7 +14340,7 @@ int lfsr_fs_grow(lfs_t *lfs, lfs_size_t block_count_) {
LFSR_TAG_GEOMETRY, 0,
(&(lfsr_geometry_t){
lfs->cfg->block_size,
block_count_}), LFSR_GEOMETRY_DSIZE)));
block_count_}), 0)));
if (err) {
goto failed;
}