Github/littlefs
1
Fork 0
Code Actions Activity

bmap: Added low-level bmap set algorithm and related tests

Browse Source 
The neat thing about the on-disk bmap is that it's a range tree. We can
leverage order-statistic properties to compactly represent ranges of
similar blocks.

However, this does make updating the bmap slightly more complicated...
This commit is contained in:
Christopher Haster
2025-07-25 13:12:54 -05:00
parent 98f016b07e
commit 732d6079e3
3 changed files with 551 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lfs3.c
+169 -9
View File
@@ -5588,9 +5588,9 @@ static int lfs3_btree_commit_(lfs3_t *lfs3,
// are we root?
} else if (child.blocks[0] == btree->r.blocks[0]) {
// mark btree as unerased in case of failure, our btree rbyd and
// mark btree as unfetched in case of failure, our btree rbyd and
// root rbyd can diverge if there's a split, but we would have
// marked the old root as unerased earlier anyways
// marked the old root as unfetched earlier anyways
lfs3_btree_claim(btree);
// need to lookup child's parent
@@ -6920,15 +6920,19 @@ static int lfs3_bshrub_commitroot_(lfs3_t *lfs3, lfs3_bshrub_t *bshrub,
static int lfs3_bshrub_commit(lfs3_t *lfs3, lfs3_bshrub_t *bshrub,
lfs3_bid_t bid, const lfs3_rattr_t *rattrs, lfs3_size_t rattr_count) {
#ifndef LFS3_2BONLY
// TODO why are we marking bshrubs as unfetched here? we should either
// be marking all btrees in lfs3_btree_commit, or move this to
// lfs3_file_commit for finer control...
//
// before we touch anything, we need to mark all other btree references
// as unerased
// as unfetched
if (lfs3_bshrub_isbtree(bshrub)) {
for (lfs3_handle_t *h = lfs3->handles; h; h = h->next) {
if (lfs3_o_isbshrub(h->flags)
&& h != &bshrub->h
&& ((lfs3_bshrub_t*)h)->shrub.r.blocks[0]
== bshrub->shrub.r.blocks[0]) {
// mark as unerased
// mark as unfetched
lfs3_btree_claim(&((lfs3_bshrub_t*)h)->shrub);
}
}
@@ -9282,7 +9286,7 @@ static int lfs3_mdir_commit(lfs3_t *lfs3, lfs3_mdir_t *mdir,
// update our mtree
} else {
// mark as unerased in case of failure
// mark as unfetched in case of failure
lfs3_btree_claim(&lfs3->mtree);
err = lfs3_mtree_commit(lfs3, &mtree_,
@@ -9321,7 +9325,7 @@ static int lfs3_mdir_commit(lfs3_t *lfs3, lfs3_mdir_t *mdir,
// how can we drop if we have no mtree?
LFS3_ASSERT(lfs3->mtree.r.weight != 0);
// mark as unerased in case of failure
// mark as unfetched in case of failure
lfs3_btree_claim(&lfs3->mtree);
// update our mtree
@@ -9358,7 +9362,7 @@ static int lfs3_mdir_commit(lfs3_t *lfs3, lfs3_mdir_t *mdir,
// update our mtree
} else {
// mark as unerased in case of failure
// mark as unfetched in case of failure
lfs3_btree_claim(&lfs3->mtree);
err = lfs3_mtree_commit(lfs3, &mtree_,
@@ -10598,6 +10602,162 @@ static int lfs3_data_readgbmap(lfs3_t *lfs3, lfs3_data_t *data) {
}
#endif
// on-disk block map operations
//
// note these take any btree, since we sometimes update the gbmap, and
// sometimes the gbatc
#ifdef LFS3_BMAP
static int lfs3_bmap_lookupnext(lfs3_t *lfs3, lfs3_btree_t *bmap,
lfs3_bid_t bid,
lfs3_bid_t *bid_, lfs3_bid_t *weight_) {
return lfs3_btree_lookupnext(lfs3, bmap, bid,
bid_, weight_, NULL);
}
#endif
// this is the same as lfs3_btree_commit, but we set the inbmap flag
// for debugging reasons
#ifdef LFS3_BMAP
static int lfs3_bmap_commit(lfs3_t *lfs3, lfs3_btree_t *bmap,
lfs3_bid_t bid, const lfs3_rattr_t *rattrs, lfs3_size_t rattr_count) {
#ifdef LFS3_REVDBG
lfs3->flags |= LFS3_i_INBMAP;
#endif
int err = lfs3_btree_commit(lfs3, bmap, bid, rattrs, rattr_count);
if (err) {
goto failed;
}
#ifdef LFS3_REVDBG
lfs3->flags &= ~LFS3_i_INBMAP;
#endif
return 0;
failed:;
#ifdef LFS3_REVDBG
lfs3->flags &= ~LFS3_i_INBMAP;
#endif
return err;
}
#endif
#ifdef LFS3_BMAP
static int lfs3_bmap_set(lfs3_t *lfs3, lfs3_btree_t *bmap,
lfs3_block_t block, lfs3_tag_t tag) {
// lookup bmap range
lfs3_bid_t bid__;
lfs3_bid_t weight__;
lfs3_stag_t tag__ = lfs3_bmap_lookupnext(lfs3, bmap, block,
&bid__, &weight__);
if (tag__ < 0) {
LFS3_ASSERT(tag__ != LFS3_ERR_NOENT);
return tag__;
}
// wait, already set to expected type? guess we're done
if (tag__ == tag) {
return 0;
}
// temporary copy, we definitely _don't_ want to leave this in a
// weird state on error
lfs3_btree_t bmap_ = *bmap;
// TODO should we mark the gbmaps as unfetched as well?
// TODO should we just claim all matching btrees in lfs3_btree_commit?
// mark as unfetched in case of error
lfs3_btree_claim(bmap);
// TODO should these use builder pattern?
// first delete block from range, this may split the range into
// multiple neighbors
//
// note this is never unnecessary work, the resulting neighbors
// can't share a type or else we would've already returned
int err = lfs3_bmap_commit(lfs3, &bmap_, bid__, LFS3_RATTRS(
(bid__-(weight__-1) < block)
? LFS3_RATTR(LFS3_TAG_GROW, -((bid__+1) - block))
: LFS3_RATTR(LFS3_TAG_RM, -((bid__+1) - block)),
(bid__ > block)
? LFS3_RATTR(tag__, +(bid__ - block))
: LFS3_RATTR_NOOP()));
if (err) {
return err;
}
// weight of new range
lfs3_bid_t weight = 1;
// can we merge with right neighbor?
if (block < lfs3->block_count-1) {
// note the use of the old bmap to try to leverage leaf caching
tag__ = lfs3_bmap_lookupnext(lfs3, bmap, block+1,
&bid__, &weight__);
if (tag__ < 0) {
LFS3_ASSERT(tag__ != LFS3_ERR_NOENT);
return tag__;
}
if (tag__ == tag) {
LFS3_ASSERT(weight__ == bid__ - block);
// merge
weight += weight__;
// delete to prepare merge
//
// note the shifted bid because of the previous delete
err = lfs3_bmap_commit(lfs3, &bmap_, bid__-1, LFS3_RATTRS(
LFS3_RATTR(LFS3_TAG_RM, -weight__)));
if (err) {
return err;
}
}
}
// can we merge with left neighbor?
if (block > 0) {
// note the use of the old bmap to try to leverage leaf caching
tag__ = lfs3_bmap_lookupnext(lfs3, bmap, block-1,
&bid__, &weight__);
if (tag__ < 0) {
LFS3_ASSERT(tag__ != LFS3_ERR_NOENT);
return tag__;
}
if (tag__ == tag) {
LFS3_ASSERT(bid__ == block-1);
// we can merge everything in one commit here
err = lfs3_bmap_commit(lfs3, &bmap_, bid__, LFS3_RATTRS(
LFS3_RATTR(LFS3_TAG_GROW, +weight)));
if (err) {
return err;
}
// done!
*bmap = bmap_;
return 0;
}
}
// needs a new range
err = lfs3_bmap_commit(lfs3, &bmap_, block, LFS3_RATTRS(
LFS3_RATTR(tag, +weight)));
if (err) {
return err;
}
// done!
*bmap = bmap_;
return 0;
}
#endif
// TODO lfs3_bmap_relocate
// TODO lfs3_bmap_mdirdiff
// TODO lfs3_bmap_btreediff
/// Block allocator ///
@@ -10878,9 +11038,9 @@ static lfs3_sblock_t lfs3_alloc(lfs3_t *lfs3, uint32_t flags) {
while (lfs3->bmap.known > 0
&& lfs3->lookahead.ckpoint > 0) {
lfs3_bid_t bid;
lfs3_stag_t tag = lfs3_btree_lookupnext(lfs3, &lfs3->bmap.gbatc,
lfs3_stag_t tag = lfs3_bmap_lookupnext(lfs3, &lfs3->bmap.gbatc,
lfs3->bmap.cursor,
&bid, NULL, NULL);
&bid, NULL);
if (tag < 0) {
return tag;
}