As a part of the general redesign of files, all files, not just small
files, can inline some data directly in the metadata log. Originally,
this was a single piece of inlined data or an inlined tree (shrub) that
effectively acted as an overlay over the block/btree data.
This is now changed so that when we have a block/btree, the root of the
btree is inlined. In effect making a full btree a sort of extended
shrub.
I'm currently calling this a "geoxylic btree", since that seems to be a
somewhat related botanical term. Geoxylic btrees have, at least on
paper, a number of benefits:
- There is a single lookup path instead of two, this simplifies code a
bit and decreases lookup costs.
- One data structure instead of two also means lfsr_file_t requires
less RAM, since all of the on-disk variants can go into one big union.
Though I'm not sure this is very significant vs stack/buffer costs.
- The write path is much simpler and has less duplication (it was
difficult to deduplicate the shrub/btree code because of how the
shrub goes through the mdir).
In this redesign, lfsr_btree_commit_ leaves root attrs uncommitted,
allowing lfsr_bshrub_commit to finish the job via lfsr_mdir_commit.
- We don't need to maintain a shrub estimate, we just lazily evict trees
during mdir compaction. This has a side-effect of allowing shrubs to
temporarily grow larger than shrub_size before eviction.
NOTE THIS (fundamentally?) DOESN'T WORK
- There is no awkwardly high overhead for small btrees. The btree root
for two-block files should be able to comfortably fit in the shrub
portion of the btree, for example.
- It may be possible to also make the mtree geoxylic, which should
reduce storage overhead of small mtrees and make better use of the
mroot.
All of this being said, things aren't working yet. Shrub eviction during
compaction runs into a problem with a single pcache -- how do we write
the new btrees without dropping the compaction pcache? We can't evict
btrees in a separate pass becauce their number is unbounded...
This did not turn out to be useful, mainly because type-agnostic
inlining requires unnecessary encoding/decoding and risks a higher RAM
allocation than is really needed. It's better to just reserve a bit in
the weight field and allow higher-level operations to use
operation-specific unions.
code stack
before: 31580 2072
after: 31160 (-1.3%) 2072 (+0.0%)
This is based on how bench.py/bench_runners have actually been used in
practice. The main changes have been to make the output of bench.py more
readibly consumable by plot.py/plotmpl.py without needing a bunch of
hacky intermediary scripts.
Now instead of a single per-bench BENCH_START/BENCH_STOP, benches can
have multiple named BENCH_START/BENCH_STOP invocations to measure
multiple things in one run:
BENCH_START("fetch", i, STEP);
lfsr_rbyd_fetch(&lfs, &rbyd_, rbyd.block, CFG->block_size) => 0;
BENCH_STOP("fetch");
Benches can also now report explicit results, for non-io measurements:
BENCH_RESULT("usage", i, STEP, rbyd.eoff);
The extra iter/size parameters to BENCH_START/BENCH_RESULT also allow
some extra information to be calculated post-bench. This infomation gets
tagged with an extra bench_agg field to help organize results in
plot.py/plotmpl.py:
- bench_meas=<meas>+amor, bench_agg=raw - amortized results
- bench_meas=<meas>+div, bench_agg=raw - per-byte results
- bench_meas=<meas>+avg, bench_agg=avg - average over BENCH_SEED
- bench_meas=<meas>+min, bench_agg=min - minimum over BENCH_SEED
- bench_meas=<meas>+max, bench_agg=max - maximum over BENCH_SEED
---
Also removed all bench.tomls for now. This may seem counterproductive in
a commit to improve benchmarking, but I'm not sure there's actual value
to keeping bench cases committed in tree.
These were alway quick to fall out of date (at the time of this commit
most of the low-level bench.tomls, rbyd, btree, etc, no longer
compiled), and most benchmarks were one-off collections of scripts/data
with results too large/cumbersome to commit and keep updated in tree.
I think the better way to approach benchmarking is a seperate repo
(multiple repos?) with all related scripts/state/code and results
committed into a hopefully reproducible snapshot. Keeping the
bench.tomls in that repo makes more sense in this model.
There may be some value to having benchmarks in CI in the future, but
for that to make sense they would need to actually fail on performance
regression. How to do that isn't so clear. Anyways we can always address
this in the future rather than now.
This gives the mtree a dedicated type, with direct mptrs (single mdirs)
being stored decoded, instead of encoding into leb128s. This avoids
encoding/decoding in some cases.
This change is currently a net downgrade, but only because we still have
all of the inlined btree code. Eventually this inlined btree code should
be removed:
code stack
before: 31316 2064
after: 31480 (+0.5%) 2072 (+0.4%)
Also tweaked the tests to no longer test dropping the mtree down to
zero size. Thanks to root bookmarks, we never actually do this, and it
simplifies lfsr_mdir_commit to not support this.
- Ripped out outdated file-data representation. We don't need this.
- Changed lfsr_data_add/read/cmp to just assert when data is
concatenated data. Theoretically this is possible to implement, but
it's complicated and we never use it, so all it is is a waste of
code size...
- Added implicitly zero-filled hole representation, though this isn't
adopted in the code yet.
- Added lfsr_data_truncate/fruncate, these are really useful for
shrub/tree carving/coalescing.
---
New lfsr_data_t encoding, sign(size) indicates if the data is
on-disk/in-device, and a mode field indicates how in-device data should
be parsed:
sign(size)=1 => on-disk:
.---+---+---+---. .....
|1| size | ..'' ''..
+---+---+---+---+ : : :
| block ------+->| ..:|
+---+---+---+---+ | |......( )::::::|
| off -------' |:::' : |
'---+---+---+---' :' : :
''.. :.''
'''''
sign(size)=0, mode=0 => in-device buffer:
.---+---+---+---. .---+---+---+---.
|0| size | .>| data... |
+---+---+---+---+ | ' . '
|m=0| | | ' . '
+---+---+---+---+ | ' '
| ptr -------' ' '
'---+---+---+---' '---+---+---+---'
sign(size)=0, mode=1 => hole
.---+---+---+---.
|0| size |
+---+---+---+---+
|m=1| |
+---+ +
| |
'---+---+---+---'
sign(size)=0, mode=2 => inlined
.---+---+---+---.
|0| size |
+---+---+---+---+
|m=2| inlined d |
+---+ +
| ata... |
'---+---+---+---'
sign(size)=0, mode=3 => concatenated datas:
.---+---+---+---. .---+---+---+---.
|0| size | .>| data |
+---+---+---+---+ | + +
|m=3| c | | | | |
+---+---+---+---+ | + +
| ptr -------' | |
'---+---+---+---' +---+---+---+---+
| data |
+ +
| |
+ +
| |
+---+---+---+---+
' . '
' . '
' . '
' '
' '
'---+---+---+---'
---
Code/RAM changes:
code stack
before: 31952 2056
after: 31396 (-1.7%) 2064 (+0.4%)
I think the increased RAM cost is due to lfsr_data_add/truncate/fruncate
passing lfsr_data_t around by value, and GCC not being able to optimize
this very well since it's 3 words. I think most move optimizations stop
after 2-words...
The biggest change here is the breaking up of the FLAGS config into
RFLAGS/WFLAGS/OFLAGS. This is directly inspired by, and honestly not
much more than a renaming, of the compat/ro_compat/incompat flags found
in Linux/Unix/POSIX filesystems.
I think these were first introduced in ext2? But I need to do a bit more
research on that.
RFLAGS/WFLAGS/OFLAGS provide a much more flexible, and extensible,
feature flag mechanism than the previous minor version bumps.
The (re)naming of these flags is intended to make their requirements
more clear. In order to do the relevant operation, you must understand
every flag set in the relevant flag:
- RFLAGS / incompat flags - All flags must be understood to read the
filesystem, if not understood the only possible behavior is to fail.
- WFLAGS / ro-compat flags - All flags must be understood to write to the
filesystem, if not understood the filesystem may be mounted read-only.
- OFLAGS / compat flags - Optional flags, if not understood the relevant
flag must be cleared before the filesystem can be written to, but other
than that these flags can mostly be ignored.
Some hypothetical littlefs examples:
- RFLAGS / incompat flags - Transparent compression
Is this the same as a major disk-version break? Yes kinda? An
implementation that doesn't understand compression can't read the
filesystem.
On the other hand, it's useful to have a filesystem that can read both
compressed and uncompressed variants.
- WFLAGS / ro-compat flags - Closed block-map
The idea behind a closed block-map (currently planned), is that
littlefs maintains in global space a complete mapping of all blocks in
use by the filesystem.
For such a mapping to remain consistent means that if you write to the
filesystem you must understand the closed block-map. Or in other
words, if you don't understand the closed block-map you must not write
to the filesystem.
Reading, on the other hand, can ignore many such write-related
auxiliary features, so the filesystem can still be read from.
- OFLAGS / compat flags - Global checksums
Global checksums (currently planned) are extra checksums attached to
each mdir that when combined self-validate the filesystem.
But if you don't understand global checksums, you can still read and
write the filesystem without them. The only catch is that when you write
to the filesystem, you may end up invalidating the global checksum.
Clearing the global checksum bit in the OFLAGS is a cheap way to
signal that the global checksum is no longer valid, allowing you to
still write to the filesystem without this optional feature.
Other tweaks to note:
- Renamed BLOCKLIMIT/DISKLIMIT -> BLOCKSIZE/BLOCKCOUNT
Note these are still the _actual_ block_size/block_count minus 1. The
subtle difference here was the original reason for the name change,
but after working with it for a bit, I just don't think new, otherwise
unused, names are worth it.
The minus 1 stays, however, since it avoids overflow issues at
extreme boundaries of powers of 2.
- Introduces STAGLIMIT/SATTRLIMIT, sys-attribute parallels to
UTAGLIMIT/UATTRLIMIT.
These may be useful if only uattrs are supported, or vice-versa.
- Dropped UATTRLIMIT/SATTRLIMIT to 255 bytes.
This feels extreme, but matches NAMELIMIT. These _should_ be small,
and limiting the uattr/sattr size to a single-byte leads to really
nice packing of the utag+uattrsize in a single integer.
This can always be expanded in the future if this limit proves to be a
problem.
- Renamed MLEAFLIMIT -> MDIRLIMIT and (re?)introduced MTREELIMIT.
These may be useful to limiting the mtree when needed, though it's not
clear the exact use case quite yet.
The original name was a bit of a mouthful.
Also dropped the default crystal_size in the test/bench runners
block_size/4 -> block_size/8. I'm already noticing large amounts of
inflation when blocks are fragmented, though I am experimenting with a
rather small fragment_size right now.
Future benchmarks/experimentation is required to figure out good values
for these.
The attempt to implement in-rbyd data slicing, being lazily coalesced
during rbyd compaction, failed pretty much completely.
Slicing is a very enticing write strategy, getting both minimal overhead
post-compaction and fast random write speeds, but the idea has some
fundamental conflicts with how we play out attrs post-compaction.
This idea might work in a more powerful filesystem, but brings back the
need to simulate rbyds in RAM, which is something I really don't want to
do (complex, bug-prone, likely adds code cost, may not even be tractable).
So, third time's the charm?
---
This new write strategy writes only datas and bptrs, and avoids dagging
by completely rewriting any regions of data larger than a configurable
crystallization threshold.
This loses most of the benefits of data crystallization, random writes
will now usually need to rewrite a full block, but as a tradeoff our
data at rest is always stored with optimal overhead.
And at least data crystallization still saves space when our data isn't
block aligned, or in sparse files. From reading up on some other
filesystem designs it seems this is a desirable optimization sometimes
referred to as "tail-packing" or "block suballocation"
Some other changes from just having more time to think about the
problem:
1. Instead of scanning to figure out our current crystal size, we can
use a simple heuristic of 1. look up left block, 2. look up right
block, 3. assume any data between these blocks contribute to our
current crystal.
This is just a heuristic, so worst case you write the first and last
byte of a block which is enough to trigger compaction into a block.
But on the plus side this avoids issues with small holes preventing
blocks from being formed.
This approach brings the number of btree lookups down from
O(crystallize_size) to 2.
2. I've gone ahead and dropped the previous scheme of coalesce_size
+ fragment_size and instead adopted a single fragment_size that
controls the size of, well, fragments, i.e. data elements stored
directly in trees.
This affects both the inlined shrub as well as fragments stored in
the inner nodes of the btree. I believe it's very similar to what is
often called "pages" in logging filesystems, though I'm going to
avoid that term for now because it's a bit overloaded.
Previously, neighboring writes that, when combined, would exceed our
coalesce_size, they just weren't combined. Now they are combined up
to our fragment size, potentially splitting the right fragment.
Before (fragment_size=8):
.---+---+---+---+---+---+---+---.
| 8 bytes |
'---+---+---+---+---+---+---+---'
+
.---+---+---+---+---.
| 5 bytes |
'---+---+---+---+---'
=
.---+---+---+---+---+---+---+---+---+---.
| 5 bytes | 5 bytes |
'---+---+---+---+---+---+---+---+---+---'
After:
.---+---+---+---+---+---+---+---.
| 8 bytes |
'---+---+---+---+---+---+---+---'
+
.---+---+---+---+---.
| 5 bytes |
'---+---+---+---+---'
=
.---+---+---+---+---+---+---+---+---+---.
| 8 bytes |2 bytes|
'---+---+---+---+---+---+---+---+---+---'
This leads to better fragment alignment (much like our block
strategy), and minimizes tree overhead.
Any neighboring data to the right is only coalesced if it fits in the
current fragment, or would be rewritten (carved) anyways, to avoid
unnecessary data rewriting.
For example (fragment_size=8):
.---+---+---+---+---+---+---+---+---+---+---+---+---+---.
| 6 bytes | 6 bytes |2 bytes|
'---+---+---+---+---+---+---+---+---+---+---+---+---+---'
+
.---+---+---+---+---.
| 5 bytes |
'---+---+---+---+---'
=
.---+---+---+---+---+---+---+---+---+---+---+---+---+---.
| 8 bytes | 4 bytes |2 bytes|
'---+---+---+---+---+---+---+---+---+---+---+---+---+---'
Other than these changes this commit is mostly a bunch of carveshrub
rewriting again, which continues to be nuanced and annoying to get
bug free.
- -> lfsr_shrub_t
- -> lfsr_tree_t
The idea here is to adopt "shrub" as an umbrella term for the
shrub/sprout union, and "tree" as an umbrella term for the bptr/btree
union. I think this is a bit better than calling shrub/sprout "inlined"
which is a _very_ overloaded term in this codebase (inlined in the tree?
the mdir? inlined in the C struct?).
This is a pretty big rewrite, but is necessary to avoid "dagging".
"Dagging" (I just made this term up) is when you transform a pure tree
into a directed acyclic graph (DAG). Normally DAGs are perfectly fine in
a copy-on-write system, but in littlefs's cases, it creates havoc for
future block allocator plans, and it's interaction with parity blocks
raises some uncomfortable questions.
How does dagging happen?
Consider an innocent little btree with a single block:
.-----.
|btree|
| |
'-----'
|
v
.-----.
|abcde|
| |
'-----'
Say we wanted to write a small amount of data in the middle of our
block. Since the data is so small, the previous scheme would simply
inline the data, carving the left and right sibling (in the case the
same block) to make space:
.-----.
|btree|
| |
'-----'
.' v '.
| c' |
'. .'
v v
.-----.
|ab de|
| |
'-----'
Oh no! A DAG!
With the potential for multiple pointers to reference the same block in
our btree, some invariants break down:
- Blocks no longer have a single reference
- If you remove a reference you can no longer assume the block is free
- Knowing when a block is free requires scanning the whole btree
- This split operation effectively creates two blocks, does that mean
we need to rewrite parity blocks?
---
To avoid this whole situation, this commit adopts a new crystallization
algorithm.
Instead of allowing crystallization data to be arbitrarily fragmented,
we eagerly coalesce any data under our crystallization threshold, and if
we can't coalesce, we compact everything into a block.
Much like a Knuth heap, simply checking both siblings to coalesce has
the effect that any data will always coalesce up to the maximum size
where possible. And when checking for siblings, we can easily find the
block alignment.
This also has the effect of always rewriting blocks if we are writing a
small amount of data into a block. Unfortunately I think this is just
necessary in order to avoid dagging.
At the very least crystallization is still useful for files not quite
block aligned at the edges, and sparse files. This also avoids concerns
of random writes inflating a file via sparse crystallization.
Still needs testing, though the byte-level fuzz tests were already causing
blocks to crystallize. I noticed this because of test failures which are
fixed now.
Note the block allocator currently doesn't understand file btrees. To
get the current tests passing requires -DDISK_SIZE=16777216 or greater.
It's probably also worth noting there's a lot that's not implemented
yet! Data checksums and write validation for one. Also ecksums. And we
should probably have some sort of special handling for linear writes so
linear writes (the most common) don't end up with a bunch of extra
crystallizing writes.
Also the fact that btrees can become DAGs now is an oversight and a bit
concerning. Will that work with a closed allocator? Block parity?
Added lfsr_bptr_t to represent block pointers (maybe we should rename
mblocks back to mptr), added fetching of btrees/bptrs in
lfsr_file_opencfg, added estimate tracking to our shrubs so we actually
know when to create a btree, and implemented most of the high-level
btree logic.
It's not working yet, but the biggest idea introduced here is how we
handle block alignment.
See, we really don't want awkward btree topologies to form where small
amounts of data get stuck between blocks:
.-----.--.-----.
| | | |
| | | |
'-----'--'-----'
This is wasteful, as the middle bit of data either gets represented as a
full block with its data partially covered, or as data inlined in the
btree, which comes with ~2x overhead.
The solution here is to scan for a block on either the left or right to
derive our block alignment from.
Unfortunately, since our sibling blocks could have been carved, this
requires scanning all the way from pos-2*B+1 to pos+2*B-1, a total of
4*B-2, to make sure we find a sibling if there is one.
worst case left worst case right
.-----.-----. .-----.-----.
| xxxx| | |p |xxxxx|
|xxxxx| p| | |xxxx |
'-----'-----' '-----'-----'
'----+----' '----+----'
pos-2*bs+1 pos+2*bs-1
Fortunately, at this stage, data should have had many chances to
coalesce, so hopefully the actual scan overhead should be much smaller
in practice.
Writing data to a file linearly, for example, only needs a single lookup
to find the previous block.
Turns out it's hard to test file holes without seek.
It's interesting to note most of seek's buffer flush work actually
occurs lazily in lfsr_file_write, so lfsr_file_seek turns out to be a
relatively simple function.
- coalesce_size - The amount of data allowed to coalesce into single
data entries.
- crystallize_size - How much data is allowed to be written to btree
inner nodes before needing to be compacted into a block.
Also deduplicated the test config is something I've been wanting to do
for a while. It doesn't make sense to need to modify several different
instantiations of lfs_config every time a config option is added or
removed...
The main purpose of this change is to introduce LFSR_DATA_CAT, a
generalized way to concatenated various data references internally.
As a side-effect lfsr_data_t has been completely restructured. Now,
lfsr_data_t can be in one of 4 modes:
If the size field's sign bit=0, the lfsr_data_t points in-device. A new,
count field, determines the encoding:
sign(size)=0, count=0 => inlined:
.---+---+---+---.
| size |
|---+---+---+---|
|c=0| inlined d | note inlined data is just enough to hold
|---+ | one encoded leb128
| ata... |
'---------------'
sign(size)=1, count=1 => direct:
.---+---+---+---. .---+---+---+---.
| size | .>| data... |
|---+---+---+---| | | . |
|c=1| | | . . .
|---+---+---+---| | . . .
| direct ptr -----' . .
'---------------'
sign(size)=1, count>=2 => indirect:
.---+---+---+---. .---+---+---+---. .---+---+---+---.
| size | .>| size | .>| data... |
|---+---+---+---| | |---+---+---+---| | | . |
|c>1| | | |c=1| | | . . .
|---+---+---+---| | |---+---+---+---| | . . .
| indirect ptr ---' | direct ptr -----' . .
'---------------' '---------------' .---+---+---+---.
| size | .>| data... |
|---+---+---+---| | | . |
|c=1| | | . . .
|---+---+---+---| | . . .
| direct ptr -----' . .
'---+---+---+---'
| . |
| . |
. . .
. .
. .
note only one indirect layer is allowed due to no recursion
If the size field's sign bit=1, the lfsr_data_t points on-disk:
sign(size)=0 => on-disk:
.---+---+---+---. .....
| size | ..'' ''..
|---+---+---+---| : : :
| block ------+->| ..:|
|---+---+---+---| | |......( )::::::|
| off -------' |:::' : |
'---------------' :' : :
''.. :.''
'''''
My goal with this commit was to test the new implementation and see how
it would impact code/RAM size before adopting it in the actual file
handling code, and the results are... not great...
code stack
before: 24668 1840
after: 25552 (+3.5%) 1920 (+4.2%)
I think most of the new cost comes from the now correct handling of
read/cmp with concatentated datas, which previously would just assert.
This change gives us LFSR_DATA_CAT, so I will be working with it for
now, but this may be worth looking at again in the future. Maybe the
correct handling of read/cmp should just be reverted to an assert...
The *_is* functions for testing for bit flags have proven useful, maybe
mostly due to C's bad bitwise operator precedence, but the
*_set*/*_clear* functions just add extra code without much benefit.
Relying on explicit bitwise operations also lets us use the |=/&= shortcut
operators, which are nice.
Get it? Because they're small trees!
Joking aside, having a new term for these helps structure and describe
the filesystem at a high-level without needing to say "inlined trees"
all the time.
Shrub trees are small rbyd trees inlined directly in a file's mdir.
The main improvement is moving the special inlined-file compaction logic
up into lfsr_mdir_compact__. We only need this logic for files stored in
mdirs, and thanks to its recursive nature, we weren't getting any
benefit from handling this at a lower level anyways.
This is a nice logical restructuring that probably saves a bit of code
cost in the end.
Another significant improvement is moving the staging copy of the
inlined tree's state up into the file struct itself. This solves the
problem of needed N copies of temporary inlined state when you have N
open files.
It also provides a central place to stage changes when compacting
inlined trees, which happens across several different places in the mdir
commit logic. Though some may see this as more a hack than a feature.
Also note-worthy, but minor: these changes required an additional
opened-mdir linked-list to know when the mdir is a file and may contain
an inlined tree.
Ran into an interesting macro-related bug. Turns out the way we are
doing implicit prefixing in TAG/ATTR macros sort of breaks how C macros
work a bit. The following does not compile:
lfsr_mdir_commit(lfs, &file->m.mdir, LFSR_ATTRS(
LFSR_ATTR(file->m.mdir.mid, DEFER, 0, DEFER(
(lfsr_rbyd_t*)&file->inlined,
LFSR_ATTR(file->buffer_pos,
DEFERRED(INLINED), +file->buffer_size, BUF(
file->buffer, file->buffer_size))))));
Or to distill it down, this does not compile:
#define LFSR_ATTR(_data) (LFSR_##_data)
#define LFSR_DEFER(_data) (LFSR_##_data)
#define LFSR_DATA(_data) (_data)
int a = LFSR_ATTR(DEFER(ATTR(DATA(1))));
But this does:
#define LFSR_ATTR(_data) (_data)
#define LFSR_DEFER(_data) (_data)
#define LFSR_DATA(_data) (_data)
int a = LFSR_ATTR(LFSR_DEFER(LFSR_ATTR(LFSR_DATA(1))));
Why? Well it turns out the whole way nested C macro's work is a big
hack.
A very reasonable design decision in C is to disallow recursive macro
expansions. Unlike C++, we don't want our preprocessor to suddenly stack
overflow. This rule is enforced by stopping macro expansion when a macro
contains itself. For example:
#define A() B()
#define B() A()
A()
Expands to:
A()
-> B()
-> A() (stops, probably erroring with 'A' undeclared)
But it _is_ common to want to recursively expand macro arguments. Macros
are a part of C's syntax after all, and users usually expect
expressions, such as arguments, to be context-free:
#define A(x) (x) + 1
A(A(A(A(A(0)))))
Naively this would expand to:
A(A(A(A(A(0)))))
-> (A(A(A(A(0))))) + 1 (stops)
The big hack that makes this work in C's preprocessor is the "Argument
prescan". Instead of expanding the "called" macro first, we expand any macro
inside our argument list, _then_ expand the "called" macro, and _then_
expand any new macros produced as a result of the expansion again just
for good measure.
So the above actually expands to:
A(A(A(A(A(0)))))
-> A(A(A(A((0) + 1))))
-> A(A(A(((0) + 1) + 1)))
-> A(A((((0) + 1) + 1) + 1))
-> A(((((0) + 1) + 1) + 1) + 1)
-> (((((0) + 1) + 1) + 1) + 1) + 1
This is still recursive actually! But the recursion is limited to the
actual length of the source code, so the developers likely thought this
was a reasonable tradeoff.
But what does this mean for our implicit prefixing?
#define P_A(x) P_##x
#define P_B(x) P_##x
#define P_C(x) (x)
P_A(B(A(C(0))))
None of A, B, C are in scope without prefixes, so they get expanded
after the "called" macro's expansion:
P_A(B(A(C)))
-> P_B(A(C(0)))
-> P_A(C(0)) (stops)
But this breaks when we hit the nested P_A macro.
---
For now I've gone with the temporary, and extra hacky, solution of
introducing a second LFSR_ATTR_ macro. This nesting of ATTR macros only
happens because of shrubs, and only ever goes 2 layers deep.
In the future maybe we should move away from implicit prefixing. They
have a few rough corners and may be a bit confusing for anyone new to
the code.
Currently limited to inlined files and only simpler truncate-writes.
But still this lets us test file creation/deletion.
This is also enough logic to make it clear that, even though we have
some powerful high-level primitives, mapping file operations onto these
is still going to be non-trivial.
Now, instead of storing a single contiguous block of config data, config
is stored as tagged metadata like any other attribute.
This allows more flexibility towards adding/removing config in the
future, without cluttering up the config with deprecated entries (see
ATA's "IDENTIFY DEVICE" response).
Most of the config entries are single leb128 limits on various integer
types, with the exception of the magic string and version (major/minor
pair).
---
Note this also includes some semantic changes to the config:
- Limits are stored as size-1. This avoid issues with integer overflow
at extreme ranges.
This was also adopted for block size (block limit) and block count
(disk limit). This deviation between on-disk config and user-facing
config risks confusion, but allows the potential for the full 2^31 range
for these values.
- The default cksum type, crc32c, has been changed to 0.
Originally this was 2 to allow the type to map to the crc width for
crc8, crc16, crc32c, crc64, etc. But dropping this idea and numbering
checksums as they are implemented simplifies things.
May come back to this.
- Storing these configs as attributes opens up of the option of on-disk
defaults when configs are missing.
I'm being a bit conservative with this one, as it's not clear to me if
we should prefer default configs (less code/storage, risk of untested
config parsing) or prefer explicit on-disk configs.
Currently the following have defaults since they seem the most obvious
to me:
- cksum type => defaults to crc32c
- redund type => defaults to parity (TODO, should this default to
no redund?)
- utag_limit => defaults to 0x7f (no special tag decoding)
- uattr_limit => defaults to block_limit (implicit)
- mbits -> mleaf_bits
- mlimit -> mleaf_limit
- mweight -> mleaf_weight
- lfsr_mridmask -> lfsr_midrmask
- lfsr_mbidmask -> lfsr_midbmask
This is a bit tricky to name, since we want to clarify it's not the
mtree limit and not the mdir's actual rbyd weight. But this also risks
confusing around the difference between mdirs/mleaves (mdirs are
mtree's leaves).
- Fixed LFSR_GRM_DSIZE upper bound, since our mids now fit in a single
leb128.
- Renamed pgrm -> ggrm. To be honest I don't have a great name for this
variable.
Adopted lfsr_rid/bid/mid/did_t where appropriate. This includes using
lfsr_rid_t for tag/rbyd weights. Although I am using lfsr_srid_t for
rbyd weights now, since it both captures the use of the sign bit and
reduces the number of casts a bit in the code.
I learned recently Zig has any-bit integers (e.g. uint31_t), and I'm
realizing how nice it would be to have those in this codebase.
Also tried to use lfs_size_t/lfs_off_t more correctly. In Linux/BSD,
only off_t is used for file-size-related operations and is usually much
larger than size_t. These were used interchangably in littlefs and their
original meaning kind of fell by the wayside. Getting their use right
will be important if littlefs ever supports different integer widths.
This reverts the big hack of treating the lfsr_dir_t as an mdir array in
lfsr_mdir_commit in an effort to deduplicate the bookmark/pos mid
updates.
It worked, but lets be honest, it was a big hack and probably not very
maintainable. It made other opened-mdir updates, such as propagation of
unerases more complex, and is made the code a bit unreadable.
We also don't really need a full mdir for the dir's bookmark, since
rewinds really aren't that common, and a single mtree lookup in that
case gets the job done. Removing the bookmark mdir (though we still need
the bookmark mid to adjust the dir pos correctly) saves 24 bytes from
every lfsr_dir_t.
It would be nice to deduplicate some of the mid logic here, but that's
been difficult because of mid-related side-effects, such as updating the
mdir's pos. There may be room for improvement here.
---
This looks pretty bad, with the additional loop over the attr-list to
update just the dir's bookmark, but it's really not that bad when
compiled, and probably worth the code readability:
code stack structs
before: 20958 1744 864
after: 21052 (+0.4%) 1744 (+0.0%) 840 (-2.9%)
This is a tricky nuance of how rbyd's erased state interacts with
possible errors during commits.
- If an rbyd passes its ecksum during rbyd-fetch, it's erased and we can
write to it.
- If an rbyd is committed to successfully and still has erased space
remaining, it's erased and we can write to it.
- But if we fail to commit to the rbyd, we can't be sure the trailing
data is still erased. It most likely isn't, and we would need to fetch
again to check the ecksum. And since errors are exceptional here, we
might as well just mark any failed commits as unerased, triggering a
compaction on the next write to the rbyd.
To make things more annoying, changing state in all error routes is
tricky to get right, and trickier to test. To keep this relatively
simple and robust, all rbyd/btree/mdir operations mark the original copy
as unerased until the commit succeeds, and then clears the unerased
state. This fits in well with how we make copies of the rbyd/btree/mdir
structs in the relevant functions.
Note this needs to affect _all_ copies of the rbyd, including any opened
mdirs, mroots, etc. This will probably still lead to some bugs in the
future...
- Added lfsr_mdir_lookupnext, for iteration through only a single mid.
This is useful for MOVE attributes.
- Renamed LFSR_MDIR_MROOTANCHOR -> LFSR_MROOTANCHOR.
- Renamed functions that operate on mdir blocks lfsr_mdir_* ->
lfsr_mblocks_*.
- Reordered arguments in lfsr_mdir_fetch.
- Renamed mrid_bits/mbid_weight -> mbits/mweight.
This adopts a previously discarded idea for compressed mids with a few
tweaks to avoiding decoding the bid/rid portions as much as possible.
The idea of compressed mids is to shove both the mid bid and mid rid
into a single integer, saving RAM and potentially helping filesystem
integration where a unique per-file integer is useful.
Unfortunately this has proven tricky. littlefs fundamentally needs two
ids, one "bid" to lookup which mdir our entry resides on, and one "rid"
to lookup the entry in the mdir. It's tempting to use two half-sized
integers (16-bit for example), but this risks surprising limitations
around the number of files when blocks are either really large or
really small.
Optimally, we'd limit the number of bits reserved for the rid to the
upper bound of number of rids that can fit in a single mdir. This would
allows for more bids when the block size is small, and more rids when
the block size is large. This should roughly approximate the limits of
a per-file integer.
With a bit of math we can estimate the upper bound to be <=block_size/16
with our current compaction strategy.
This idea was previously discarded due to the overhead of extracting the
bids/rids when we need them, but the RAM savings and file-to-integer
mapping is too useful to give up. When it became clear half-width
integers wasn't really going to work, compressed mids became the new
plan:
0bbbbbbb bbbbbbbb bbbbbbbb rrrrrrrr
^'-----------+-----------' '---+--'
'------------|-----------------|---- sign-bit, reserved for driver
| '---- nlog2(bs/16) bits for rid
| (8-bits for 4KiB blocks)
'---------------------- remaining bits for bid
(23-bits for 4KiB blocks)
To reduce the overhead of encoding/decode bids/rids a few extra features
were added to the internal mdir APIs:
1. The mtree has been changed to store mids directly. Giving each mdir
the upper bound as a weight. This allows direct lookup of mids
without any sort of bid decoding, though does bake the upper bound
estimate into the metadata of the filesystem, which isn't the
cleanest design, but if it works it works.
On the plus side, with this upper bound baked in to the filesystems,
GRMs can be encoded in a single leb128, which is nice. This may have
other savings if we ever store mids anywhere else in the filesystem.
2. rids are now mid relative in lfsr_mdir_lookup when non-negative. This
is implemented with a simple condition that is hopefully optimized
out when inlined, though there may be some room for improvement here.
3. rids are now mid relative in lfsr_mdir_commit. This was a bit tricky,
but we can leverage the existing mechanisms for bid-relative rids
used in the btree implementation.
The above changes make it so you can pass the mid around directly for
most of the mdir functions, hopefully reducing the mid decoding
overhead. This savings should only grow as more high-level filesystem
APIs are added.
Here is the resulting code/RAM changes for this entire change (from
before we adopted the mroot bit):
code stack structs
before: 20590 1784 908
after: 20890 (+1.4%) 1744 (-2.3%) 864 (-5.1%)
More on this when explaining compressed mids, but basically the idea is
instead of just storing all mdirs in our mtree as single element
entries, store each mdir in as a weighted entry, where the weight is a
known upper bound on the possible number of mid entries in a single
mdir.
With the current mid representation, this just complicates things
without much benefits. But with compressed mids it allows us to lookup
mdirs with the mid directly, and avoid decoding the bid from the mid in
some cases.
The mid-per-mdir upper bound is derived from the block size. We know:
1. Each tag needs <=2 alts+null with our current compaction strategy
2. Each tag/alt encodes to a minimum of 4 bytes
This gives us ~4*4 or ~16 bytes per mid at minimum. If we cram an mdir
with the smallest possible mids, this gives us at most ~block_size/16
mids in a single mdir before the mdir runs out of space.
Note we can't assume ~1/2 block utilization here, as an mdir may
temporarily fill with more mids before compaction occurs.
This is an intermediate commit as a part of a tangent into compressed
mids.
The idea here, is instead of using bid=-1 as a special value for mroots,
use only the top bit to indicate mroots. This allows you to compare
against the grm/other uninlined mids by masking instead of signed
comparison.
This is valuable for compressed mids since extracting bids relies on
knowledge of the block size, and becomes quite a bit more expensive.
mroot bid mroot cmp
before: 0xffffffff lfs_smax32(a, 0) == lfs_smax32(b, 0)
after: 0x80000000 (a & 0x7fffffff) == (b & 0x7fffffff)
The implementation here is a bit clumsy. I think GCC may be not that
great at optimizing out copies of structs being passed around via
inlined functions. But this is only a proof-of-concept.
The possibility of 16-bit mbid/mrids being a problematic limit is too
high for me to be able to confidently move forward with this internal
encoding.
Consider a filesystem with small blocks and/or large amounts of metadata
per file. If a single file almost fills up an mdir, it risks an
effective limit on the filesystem of 2^16 files. Not a deal-breaker, but
certainly a surprising limit on a supposed "32-bit" filesystem.
We can add more granular integer-limit configurations, but with simple
configurations, the option to increase the integer-limit filesystem-wide
to 64-bits would cost more RAM than just increasing the mbid/mrids
limits.
Though this is always up for reconsideration in the future.
code stack
before: 20762 1720
after: 20590 (-0.8%) 1784 (+3.7%)
It's interesting to not the code/RAM tradeoff here. RAM sees a
significant hit, but code improves, likely because of better instruction
sequences for 32-bit operations (this is targeting ARM thumb,
32-bit MCUs).
Though the code savings likely varies widely across instruction sets,
and I would guess swings negative on 16/8-bit MCUs.
The idea here: Instead of having unique functionality for each
individual btree operation (push/set/pop/split), we treat btrees sort of
like rbyds, with a single commit entry point that operates on attr-lists.
This adds code cost, due to needing to parse the attr-list for properties
that can affect inlined btrees (tag changes mostly), but, in theory, comes
with some advantages:
1. A single btree commit entry point with all of the inlined/uninlining
logic should offer better chances for code deduplication, vs
spreading this logic out in each btree operation.
2. Higher-levels should know what the current weight of the branch is,
so we may be able to avoid the implicit math needed to calculate
deltas.
3. Higher-levels have more knowledge about the state of the btree in
general, so there may be other shortcuts. The mtree, for example,
only operates on weight=1 entries, which greatly simplifies a lot of
the related math.
Note that btrees still have strict limits in what's possible in an
attr-list. Btree operations can't cross leaf-rbyd boundaries for
example.
---
A notable omission in this change is the loss of reinlining btrees.
This wase dropped for a couple reasons. It may be worth adding back at a
later time, maybe after we actually have files implemented, but for now
does not seem worth it:
1. Reinlining adds code cost. Reinlining is more complex than you might
expect because we only reinline on compaction. And because we compact
before playing out our attr-list, we need to know if a commit makes
the btree inlinable before committing to the btree.
This is still doable with our attr-lists. We already derive the
change in tags, since we need this to know when to uninline. But it
adds a kind of complex bailing out of btree commits.
2. The benefits of reinlining may not be that great. In most systems, a
tree that is uninlined once is likely to be uninlined again. It's
only if there is a bigger state change in a system that it makes
sense to reinline.
Though, to be fair, waiting for compaction to reinline handled this
quite well. Only reinlining when all erased storage is used up...
3. Thanks to our roots did entry, our mtree can never reinline.
It would be nice to change this, but this would require explicit
handling in lfsr_mdir_commit. Future work?
4. Files are another can of worms, with more complex interactions with
inlinability thanks to (at least on paper right now) always having
inlined data even when uninlined.
If reinlining is valuable for files this can change during that work.
5. Even if files never support reinlinability, truncating files (via
either lfsr_file_truncate or LFSR_O_TRUNC) should give the file a
blank slate, effectively reinlining the file in that case.
---
The current implementation also changes the attr-list to be mutable so
we can adjust attr-list based on the current btree node. This is a
temporary hack! We should add the appropriate functionality to our rbyd
utilities to revert this eventually.
- lfsr_mid_cmp no longer uses a union. This was undefined behavior and
the lfsr_mid_t type isn't word aligned, so this could break pretty badly
on machine/compiler change.
Also dropped ordering based on endianness, since we need to marshal
these into an int for the comparison anyways.
- Changed lfsr_mdir_cmp to use min/max functions as part of the
comparison. The result is also "ordered" now, though the ordering
is nonsensical. I guess the mrootanchor is less than all other mdirs?
Also considered only comparing a single min/max block, since it would
be an error for mdirs to share blocks, but note we rely on
lfsr_mdir_cmp to check for relocations in lfsr_mdir_commit. These
relocations can end up being partial in the case of bad block
detection.
Generally the more creative you get with C macros, the more
unmaintainable your codebase becomes, but in this case I think a small
bit of macro sugar for the attribute lists in littlefs goes a long way
for making the internals flexible and readable.
Attribute lists generally look like this:
LFSR_ATTRS(
LFSR_ATTR(id, TAG, delta, DATA(data)),
LFSR_ATTR(id, TAG, delta, DATA(data)),
...
LFSR_ATTR(id, TAG, delta, DATA(data)))
Which more-or-less gets expanded to this:
((const lfsr_attr_t[]){
((lfsr_attr_t){id, LFSR_TAG_TAG, delta, LFSR_DATA_DATA(data)}),
((lfsr_attr_t){id, LFSR_TAG_TAG, delta, LFSR_DATA_DATA(data)}),
...
((lfsr_attr_t){id, LFSR_TAG_TAG, delta, LFSR_DATA_DATA(data)}),}),
attr_count
Note the use of preprocessor concatenation to put the TAG and DATA
identifiers in their respective namespaces. These can end up invoking
other macros, which allows attrs to be rather extensible.
Previously there were also LFSR_ATTR_ (note the trailing underscore)
macros to allow passing of variable tags/datas. This is replaced with
redundant macros which sort of "unwrap" themselves as a part of macro
expansion. This avoids a bunch of duplicate macro definitions.
#define LFSR_TAG_TAG(tag) (tag)
#define LFSR_DATA_DATA(data) (data)
So:
LFSR_ATTR(id, TAG(tag), delta, DATA(data))
Becomes:
((lfsr_attr_t){id, LFSR_TAG_TAG(tag), delta, LFSR_DATA_DATA(data)})
Becomes:
((lfsr_attr_t){id, tag, delta, data})
This checksum is used to keep track of if we have erased, and not yet
touched, the unused bytes trailing our current commit in the rbyd.
The working theory is that if any prog attempt is made, it will, most
likely, change the checksum of the contents, allowing littlefs to
determine if trailing erased-state is safe to use, even under powerloss.
littlefs can also perturb future data by a single bit, to force this
checksum to always be invalidated during normal operation.
The original name, "forward erased-state checksums (fcksum)", came from the
idea that the checksum "looks forward" into the next commit.
But after using them for a bit, I think the name is unnecessarily
confusing. It, uh, also looks a lot like a swear word. I think
shortening the name to just "erased-state checksums (ecksum)", even
though the previous name is already in use in a release, is reasonable.
---
It's probably hard to believe but the name change from fcrc -> ecrc
really was unrelated to the crc -> cksum change. But boy is it
convenient for avoiding an awkward name. A lot of these name changes
involved sed scripts, so I didn't notice how awkward fcksum would be to
use until writing this commit message.
The reason for this is to move away from the idea that littlefs is
strictly bound to CRCs and make the code more welcoming to other
checksum types, such as SHA256, etc.
Of course, changing the name doesn't really do anything. littlefs
actually _is_ strictly bound to CRCs in a couple ways that other
filesystems aren't. These would need to have workarounds for other
checksum types:
- We leverage the parity-preserving nature of (some) CRCs to not have
to also calculate the parity of metadata in rbyd commits.
- We leverage the linearity of CRCs to retroactively flip the
perturb bit in the cksum tag without needing to recalculate the
checksum. Though the fact we need to do this is because of how we
use parity above, so this may just not be needed for non-CRC
checksums.
- The plans for global-CRCs (not yet implemented) rely heavily on the
mathematical properties of CRC polynomials. This doesn't mean
global-CRCs can't work with other checksums, you would just need to
find a different type of polynomial.
For a couple reasons:
1. Organizing the overlaps this way avoid potential undefined behavior.
It turns out C does define the overlap the "initial sequence" of
union members, as long as the types are the same. But when we
overlapped the block with the size/tag fields in lfsr_btree_t, it was
probably undefined behavior.
At the very least, it would introduce a need for quite a bit of
preprocessing to make it work with different integer sizes and
redundancy levels.
2. Overlapping the blocks at the end of the rbyd struct means our block
array is natural ordered such that the first block is the "active"
block, i.e. the block with the most recent revision count that passes
checksums.
This has been useful as a debugging tool, so I would like to continue
the pattern. It is possible to mostly preserve this order with the
previous method by intentional reversing the block array when
logging or writing to disk, but it's a bit cumbersome.
2. It's unlikely we'll be able to use readonly variants of the rbyd/mdir
structs for RAM savings. Unfortunately C makes this too cumbersome.
Though if we do this should be revisited.
Here are the new overlaps. Note it's no longer possible to truncate the
types when readonly. If readonly struct are useful this will need to be
revisited again:
lfsr_rbyd_t lfsr_btree_t lfsr_mdir_t
8b 8b 8b 8b
.----+----+----+----.
8b 8b 8b 8b 8b 8b 8b 8b | mid.bid | mid.rid |
.----+----+----+----. .----+----+----+----. |----+----+----+----|
| weight |.>| weight | | weight |
|----+----+----+----| |----+----+----+----| |----+----+----+----|
| trunk | | tag | size | | trunk |
|----+----+----+----| |----+----+----+----| |----+----+----+----|
| off | | inlined data | | off |
|----+----+----+----| | | | |----+----+----+----|
| crc | | v | | crc |
|----+----+----+----| | | |----+----+----+----|
| block |..| |.>| blocks |
'----+----+----+----' '----+----+----+----' | |
| |
'----+----+----+----'
This turned out to be tricky.
At littlefs's core, we have the lfsr_rbyd_t struct. It is really
important this is as small as possible since littlefs creates many rbyd
copies in order to track state of metadata on disk.
Wrapping rbyd, we have the lfsr_btree_t struct, which can alternatively
contain a single inlined entry, accomplished by overlapping the width
field in both cases. And the lfsr_mdir_t struct, which tracks any redundant
blocks, and would be nice if the blocks lined up as neighbors so all blocks
involved in the mdir could be passed around as an array. Both of these
wrappers attempt to overlap fields of the lfsr_rbyd_t struct, which presents
a bit of a problem.
The solution here is to put the rbyd block field at the beginning of the
lfsr_rbyd_t struct, and use exactly 32-bits of padding in lfsr_btree_t
to overlap the width field even though it is not at the beginning of the
struct. To avoid inflating the lfsr_btree_t size, we sneak the inlined
size and tag into the overlapping padding. This will need special
handling if the size of these fields change, but saves a decent amount
of RAM:
lfsr_rbyd_t lfsr_btree_t lfsr_mdir_t
8b 8b 8b 8b
.----+----+----+----.
| mid.bid | mid.rid |
|----+----+----+----|
8b 8b 8b 8b 8b 8b 8b 8b | blocks |
.----+----+----+----. .----+----+----+----. | |
| block |..| tag |size|padd|.>| |
|----+----+----+----| |----+----+----+----| |----+----+----+----|
| weight |.>| weight | | weight |
|----+----+----+----| |----+----+----+----| |----+----+----+----|
| trunk | | inlined data | | trunk |
|----+----+----+----| | | | |----+----+----+----|
| off | | v | | off |
|----+----+----+----| | | |----+----+----+----|
| crc | | | | crc |
'----+----+----+----' '----+----+----+----' '----+----+----+----'
Also tried to reduce the amount of mdir usage in lfsr_mdir_commit by
better using only the arrays of relevant mdir blocks, to limited success.
- Updated LFSR_BTREE_INLINESIZE to properly include the overhead for
mdir pointers, which need 2 block addresses instead of 1. This adds
4 bytes to the lfsr_btree_t struct.
- Changed code that marks rbyds as "needing compaction" to use -1
instead of block_size. This can use a cheaper constant and helps
debugging.
- Changed the mid representation of root to 0.0 from ?.-1. The mid 0.0
is always reserved for the roots dstart, so it shouldn't be used for
any actual file. This disambiguates root vs special metadata mids and
is a step towards making mids unsigned.
It also saves a tiny bit of code since 0 comparisons are generally
cheaper and we can leverage the order-preserving conversion of mid
to an integer.
This didn't really work out as well as I had hoped. There were a few
ideas on how to encode the bid/rid tuple without sacrificing the
(currently 31-bit) integer limit, but these just introduced too much
complexity.
Ideas:
1. In theory, as the mdirs increase in size, the quantity of mdirs needed
for a given number of files decreases. If we say the number of files
fits in an integer of a given size, than we can model the mapping to
mdirs and rids roughly as the number of bits in that integer split
between the two.
Since the block_size is known, the we can find a rather conservative,
yet useful, estimate of the upper bound of rids, which ends up
being ~16 bytes ((2 alts + 1 null + 1 tag) * 4 bytes).
And since our btrees are perfectly balanced, this encoding should only
waste 1 or 2 bits due to rounding to rounding and sign encoding for
special values.
bbbbbbbb bbbbbbbb bbbbbbbr rrrrrrrr
'-----------+-----------''----+---'
| '-- log2(block_size/32)-bit rid
'-------------------- remaining-bit bid
Unfortunately, while this works ok on paper, and maximize the use of
the bits we have available for the mid, the implementation ended up
awkward and difficult to use.
We need to either calculate the relatively complciated log2 of the
block_size on the fly, or cache the value, and use it to shift the
mid around to extract the bid/rid when needed.
Unfortunately, perhaps due to the it being easy to use the bid/rid
directly, we use and mutate the bid/rid quite a bit. We mutate when
updating the mdirs, when decoding grms, when seeking mdirs, etc. If
anything, updating the mid in total is rarer than updating the
bid/rid component in complicated situations.
Note to mention this required access to the lfs config to even begin
decoding, complicating the API and making the result less efficient.
Initial (unoptimized, and not even tested) code size showed ~+800
bytes. So I decided to scrap this.
Maybe it will be worth investigating dynamic rid sizes later, to
increase the possible mtree size for a given mid width. Not sure.
2. Probably one of the worst ideas I've had so far, but it would solve
the mid encoding problem, is to use some form a floating point to
encode the bid/rid pair:
.----------.
v .+-.
bbbbbbbb bbbbbbbb bbbrrrrr rrrrssss
'-----------+-------''----+---''-+'
| | '-- rid bits
| '--------- variable rid
'----------------------- variable bid
An even worse idea would be to use IEEE floating point here. Yes it
would work, and probably work annoyingly well, but we it risk
bringing in a lot of standard conforming backbending that we really
don't care about.
The idea here is to sacrifice some bits to encode the ratio of rid
bits to bid bits. The value of this over the using the block_size is
that we can decode the bid and rid using all of the bits in the
integer alone. Avoiding memory access (and worse debugging) to load
any external constants.
As a plus, all mids in the system would have the same exponent,
simplifying comparisons and other operations.
But this is just trying to solve complexity by adding more
complexity, so I'm not even going to try implementing it.
Still, it's an interesting idea...
In the end I've gone with the KISS implementation. Use half-width
integers, in this case uint16s, for both the bid and rid:
bbbbbbbb bbbbbbbb rrrrrrrr rrrrrrrr
'-------+-------' '-------+-------'
| '-- 16-bit rid
'-------------------- 16-bit bid
This suffers from weakened limits around the number of rids in a block
and number of mdirs in the mtree, which is unfortunate. Still it is
probably worth the tradeoff for the RAM savings and encoding simplicity.
If the mdir is reasonably sized, this does probably approach a decent
distribution of rids and bids in 32-bits. But for outlier cases with
very small and very large mdirs, it risks premature out of bounds
errors.
To protect against mtree errors, we will probably need an additional
configuration option in the form of an mdir limit. Conveniently this
would also provide a way to enforce 2-block mode.
rid errors, on the other hand, depend on block_size/32, so we may not
need another configuration option and can rely on the block_size
to determine if the rids can overflow.
This is probably worth revisiting in the future. Fortunately, with
mdir_limit and block_size configuration options, it should be possible
to increase these limits in the future if this mid bid/rid design
changes.
code stack
before: 22126 2136
after: 22326 (+0.9%) 2088 (-2.2%)
This code size increase was unexpected. Maybe non-32-bit-aligned integers
cost more to load in thumb? Unsure.
The main intention here was to make the tracking of opened mdirs,
mostly opened lfsr_dir_t structs, simpler and more resilient to weird
corner cases. I'm not entirely sure this was successful.
The main changes:
- lfsr_dir_t now contains a full mdir for the dstart entry.
This makes it so that dstarts are not a special case when it comes
to mdir updates, though the fact that directories have 2 mdirs is
still an awkward case on its own.
I considered using two entries in the opened linked-list for this, but
it wouldn't have worked out that well. Both entries need to update the
directory position, so it would have required a third file type. We
would also have needed to make sure removed mdirs mark both mdirs as
removed, otherwise the position mdir would move around arbitrary into
possibly erronous values.
Instead the current solution treats the directory mdirs as a small
array of 2 mdirs, which is as hacky as it is hacky, but does get the
job done with little code duplication.
- Directory positions are updated a bit more intellegently.
Instead of checking if in range before updating, which requires access
to both mdirs and duplicate mid/rid comparison logic, position is
updated without regard for the beginning of the directory, and
un-updated if it was actually out of range of the directory.
This means we only need to compare the mids/rids for each mdir once.
This changes make it so that lfsr_dir_rewind is much cheaper, and
doesn't even need to go to disk. Though I'm not sure it's worth the RAM
increase...
Expanding the lfsr_dir_t dstart entry to a full mdir does a lot for
making mdir updates more consistent, but increases the lfsr_dir_t size
from 52 bytes to 76 bytes (+46.2%).
When updating any opened mdirs to keep things in sync, we need to know
what rid the mdir is targeting in order to know which on-disk mdir it
should follow in the case of splits. Making this rid an actual member of
the mdir struct simplifies things.
This adds some RAM cost, though the plan is to merge the mid/rid into a
single integer, which requires this change and should actually save RAM
in the long run.
code stack
before: 22342
after: 22204 (-0.6%) 2144 (+1.1%)
It turned out our dir-read-idempotent test never created non-dstart
neighbors. This was a bit of a problem since we relied on dstart entries
to know when our dir read terminates. If we seek to an invalid position
(in theory undefined behavior, but easily possible with concurrent
modifications to the directory), we can end up reading an unrealted,
non-dstart entry, and incorrectly reporting that entry as in our current
dir.
This fix reintroduces the did into the lfsr_dir_t struct and uses the
did to determine end-of-dir. This adds some RAM cost, but is more
resilient to any seeks that overshoot the end of the directory.
Using did is also a stronger guarantee we will never accidentally report
unrelated entries as a part of the current directory.
This is entirely a pragmatic change, lfsr_mdir_commit already does
several hairy things with grm tags, decoding, fixing, reencoding, etc,
so it makes sense to move all the encoding logic into lfsr_mdir_commit.
This leads to a couple optimizations:
- We don't need to decode the grm to apply any last minute fixes.
- By allowing the grm arugment to be mutated (they are just sitting on
the stack anyways, we need a copy in case we back out of change due to
error), we can apply and save any grm fixes in the grm argument
itself.
This means we only need to fix the grm at most once, after any mtree
modifications.
Which in turn saves some code and stack cost:
code stack
before: 22930 2392
after: 22706 (-1.0%) 2344 (-2.0%)
Unfortunately the previous attempt to fix the dir seek system didn't
really work. Using a packed mid/rid integer for the offset is tempting,
but since mid/rid can change with any metadata id change in the
filesystem, dir tell offsets would become invalidated if you modified
files in unrelated directories, which isn't great and likely to catch
users by surprise.
This solution builds on the previous dir offset design, which tracks the
dstart-relative position independently from the current mid/rid in our
directory. To update this correctly when there are unrelated changes to
the filesystem, we need to know if metadata id changes are in the range
between our directories dstart and current mid/rid. This in turn means
we need to track our dstart. So our opened directories need three
separate pointers we need to update on every mdir commit:
dir->pos
|
.-------+-------.
a b c d e f g h i j k l m n o p
^ ^
| |
dir->dstart dir->mdir
This has quite a few moving parts, which I was hoping to avoid.
Fortunately we don't need a second mdir, so the RAM cost is pretty
small.
We can also drop dir->did, since the dstart mid/rid render it redundant,
which is interesting.
This is an attempt to fix issues with dir seeking in a filesystem
undergoing changes. The problem with the previous dstart-relative
position encoding is that if we deleted/created new entries outside of
our current directory, we didn't if they were inside or outside of the
current directory, so we couldn't always update our position correctly.
Instead of using a dstart-relative position, this solution crams both
the mid and rid into a single 31-bit integer. Things get a bit tight
here, so we use the current block_size as a heuristic for how many
possible rids we can ever have in a single mdir. The idea is the larger
the rid encoding needs to be, the smaller the mid encoding needs to be,
and we should, _roughly_, approach the same encoding limitation we would
have to dstart-relative position anyways.
Making some assumptions about the maximum possible number of rids in a
block gives us at most ~block_size/8 rids per mdir.
So for 4096 byte blocks (note the exact encoding is dynamic):
sbbbbbbb bbbbbbbb bbbbbbbr rrrrrrrr
^'-----------+----------''----+---'
'------------|----------------|----- sign bit (used for errors)
'----------------|----- 22-bit metadata bid
'----- 9-bit metadata rid
Note this introduced as new, significant limitation on the number of
total mdirs in the system. Normally I would be against this solution for
that reason, however if we adopt this encoding elsewhere in the system it
may improve some RAM cost and in general simplify things by being able to
store any mid in a single integer. More work needs to be done here...
This approach needs some fleshing out and has its own issues (the
offset returned by tell quickly becomes out of date if the filesystem
is modified, but is that really a problem?), but it improves over the
previous implementation by making tell always correct at that moment.
The idea here is to combine the current mtree size with the theoretical
upper bound on the number of directories in a single mdir, assuming our
block size, to give us a heuristic for did truncation that does not
require any extra state.
- Each directory needs 1 name tag, 1 did tag, and 1 dstart
- Each tag needs ~2 alts with our current compaction strategy
- Each tag/alt encodes to a minimum of 4 bytes
- We can also assume ~1/2 block utilization due to our split threshold
This gives us ~3*3*4*2 or ~72 bytes per directory at minimum, or
rounding down, ~block_size/32 directories per mdir.
This is a nice number because for common NOR flash geometry,
4096/32 = 128, so a filesystem with a single mdir encodes dids in a
single byte.
The biggest benefit though is being able to drop the mlimit state from
the lfs_t struct.
---
Unfortunately, this change revealed several bugs.
It turns out __builtin_clz in GCC is undefined at 0, which caused our
lfs_nlog2 function to return incorrect values at 1. This was causing
our dids to all collide when the mtree was inlined, which was resolved
by the linear scanning that resolves dids, but was severely limiting
what exactly our tests covered.
Now that this is fixed (with a simple if statement in lfs_nlog2,
lfs_nlog2 now always has defined behavior, even at 0), several bugs
needed fixing:
- We update the rid based on attrs in lfsr_mdir_commit before updating
the mdir. If we have multiple attrs this causes the assert on
rid-in-bounds to trigger incorrectly. Just removed that assert for now.
- We needed to adjust second grms if they are affected by the fixing
of the first grm.
- Directory position updates are incorrectly updated if an unrelated
weight change occurs before an opened directory, but is not a part of
that opened directory.
This is NOT fixed yet, the current implementation is just broken
enough that I've just ripped it out for now (it was causing the
read_with_rms test to fail because pos backed up into the "."/".."
entries).
This needs some thinking to fix.
Because of that last, unfixed bug, tests are not all passing at the
moment. To pass testing -DSEEK=0 is needed to disable the failing tests.
Christopher Haster