| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This replaces the OpCode enum with a new Op enum which is guaranteed to fit in a
single byte. Instead of carrying enum variants with data, every variant that has
runtime data encodes it into the `Vec<u8>` that a `Chunk` now carries.
This has several advantages:
* Less stack space is required at runtime, and fewer allocations are required
while compiling.
* The OpCode doesn't need to carry "weird" special-cased data variants anymore.
* It is faster (albeit, not by much). On my laptop, results consistently look
approximately like this:
Benchmark 1: ./before -E '(import <nixpkgs> {}).firefox.outPath' --log-level ERROR --no-warnings
Time (mean ± σ): 8.224 s ± 0.272 s [User: 7.149 s, System: 0.688 s]
Range (min … max): 7.759 s … 8.583 s 10 runs
Benchmark 2: ./after -E '(import <nixpkgs> {}).firefox.outPath' --log-level ERROR --no-warnings
Time (mean ± σ): 8.000 s ± 0.198 s [User: 7.036 s, System: 0.633 s]
Range (min … max): 7.718 s … 8.334 s 10 runs
See notes below for why the performance impact might be less than expected.
* It is faster while at the same time dropping some optimisations we previously
performed.
This has several disadvantages:
* The code is closer to how one would write it in C or Go.
* Bit shifting!
* There is (for now) slightly more code than before.
On performance I have the following thoughts at the moment:
In order to prepare for adding GC, there's a couple of places in Tvix where I'd
like to fence off certain kinds of complexity (such as mutating bytecode, which,
for various reaons, also has to be part of data that is subject to GC). With
this change, we can drop optimisations like retroactively modifying existing
bytecode and *still* achieve better performance than before.
I believe that this is currently worth it to pave the way for changes that are
more significant for performance.
In general this also opens other avenues of optimisation: For example, we can
profile which argument sizes actually exist and remove the copy overhead of
varint decoding (which does show up in profiles) by using more adequately sized
types for, e.g., constant indices.
Known regressions:
* Op::Constant is no longer printing its values in disassembly (this can be
fixed, I just didn't get around to it, will do separately).
Change-Id: Id9b3a4254623a45de03069dbdb70b8349e976743
Reviewed-on: https://cl.tvl.fyi/c/depot/+/12191
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
|
|
* Treewide: re-run depotfmt
* //third_party/nixpkgs:html5validator: build with Python 3.11,
dependency openstackdocstheme doesn't support 3.12
* //users/sterni/machines/ingeborg: adapt to poorly handled fcgiwrap
module API change: https://github.com/NixOS/nixpkgs/pull/318599
* //tvix/*-go: regenerate protobuf files
* //third_party/nixpkgs:treefmt: Remove patch for merged pull request
* //users/flokli/ipu6-softisp: rebase, drop upstreamed kernel patches
Change-Id: Ie4e0df007c287e8cd6207683a9a25838aa5bd39a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11971
Autosubmit: sterni <sternenseemann@systemli.org>
Reviewed-by: sterni <sternenseemann@systemli.org>
Reviewed-by: flokli <flokli@flokli.de>
Reviewed-by: aspen <root@gws.fyi>
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
Reviewed-by: Ilan Joselevich <personal@ilanjoselevich.com>
|
|
Previously, OpConstant would display some detail about its
ConstantIdx: whether it's a thunk or closure, and what its address
is. This has been expanded to also show when the ConstantIdx is a
blueprint, along with the blueprint's address, and to the other
opcodes that use a ConstantIdx.
Currently, it seems like blueprint addresses don't correspond to the
address of the thunk listed in the bytecode output, but it's still
useful to see that the constant being grabbed is a blueprint, and
maybe this pointer can be made to make more sense in the future.
Change-Id: Ia212b0d52b004c87051542c093274e7106ee08e4
Reviewed-on: https://cl.tvl.fyi/c/depot/+/12044
Tested-by: BuildkiteCI
Reviewed-by: aspen <root@gws.fyi>
Autosubmit: chickadee <matthewktromp@gmail.com>
|
|
In the compiler, skip emitting an OpForce if the last op was an
OpConstant for a non-thunk constant. This gives a small (~1% on my
machine) perf boost, eg when evaluating hello.outPath:
❯ hyperfine \
"./before --no-warnings -E '(import <nixpkgs> {}).hello.outPath'" \
"./after --no-warnings -E '(import <nixpkgs> {}).hello.outPath'"
Benchmark 1: ./before --no-warnings -E '(import <nixpkgs> {}).hello.outPath'
Time (mean ± σ): 1.151 s ± 0.022 s [User: 1.003 s, System: 0.151 s]
Range (min … max): 1.123 s … 1.184 s 10 runs
Benchmark 2: ./after --no-warnings -E '(import <nixpkgs> {}).hello.outPath'
Time (mean ± σ): 1.140 s ± 0.022 s [User: 0.989 s, System: 0.152 s]
Range (min … max): 1.115 s … 1.175 s 10 runs
Summary
./after --no-warnings -E '(import <nixpkgs> {}).hello.outPath' ran
1.01 ± 0.03 times faster than ./before --no-warnings -E '(import <nixpkgs> {}).hello.outPath'
Change-Id: I2105fd431d4bad699087907e16c789418e9a4062
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10714
Reviewed-by: sterni <sternenseemann@systemli.org>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
|
|
There's some more left, but they've been renamed/refactored out of
sight.
Change-Id: I41579dedc74342b4c5f8cb39d2995b5b0c90b0f4
Reviewed-on: https://cl.tvl.fyi/c/depot/+/9372
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Autosubmit: flokli <flokli@flokli.de>
|
|
This method extends the contents of one chunk with that of another,
effectively merging the thunks together.
This will be used for the upcoming "unthunking" functionality.
Change-Id: I6ad74232cd7f3eca198ed921e455205e00d76e6b
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7958
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
|
|
This switches out the previous compressed representation (count of
instructions per span) with a representation where the chunk's span
list stores the index of the first operation that belongs to a span,
and finds the right span by using a binary search when looking them
up.
This improves the lookup complexity from O(n) to O(log n).
This improvement was suggested and (mostly) implemented by GPT-4. I
only fixed up some names and updated the logic for deleting
spans (which it only did not do because I didn't tell it about that).
The code was verified by producing a complex error before/after the
change and ensuring that all spans in the error match exactly.
Co-Authored-By: GPT-4
Change-Id: Ibfa12cc6973af1c9b0ae55bb464d1975209771f5
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8385
Reviewed-by: ezemtsov <eugene.zemtsov@gmail.com>
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
|
|
This reports the span
1. of the code within a thunk,
2. of the place where the thunk was instantiated,
3. of the place where the thunk was first forced,
4. of the place where the thunk was forced again,
when yielding an infinite recursion error, which hopefully makes it
easier to debug them.
The spans are tracked in the ThunkRepr::Blackhole variant when putting
a thunk under evaluation.
Note that we currently have some loss of span precision in the VM loop
when switching between frame types, so spans 3/4 are currently a bit
wonky. Working on it.
Change-Id: Icbd2a9df903d00e8c2545b3fc46dcd2a9e3e3e55
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8270
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
|
|
This adds addresses of thunk and closure chunks to the debug output
displayed when dumping bytecode.
This makes it possible to see in the dump which thunks are referenced
by constants in other thunks.
Change-Id: I2c98de5227e7cb415666cd3134c947a56979dc80
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8137
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
|
|
The codebase contains a lot of complexity and odd roundabout
handling for shadowing globals. I'm pretty sure none of this is
necessary, and all of it disappears if you simply make the globals
part of the ordinary identifier resolution chain, with their own
scope up above the root scope. Then the ordinary shadowing routines
do the right thing, and no special cases or new terminology are
required.
This commit does that.
Note by tazjin: This commit was originally abandoned when Adam decided
not to take away reviewer bandwidth for this at the time (eval was
still in a much earlier stage). As we've recently done some
significant refactoring of globals initialisation this came up again,
and it seems we can easily cover the use-cases of the poison tracking
in other ways now, so I've rebased, updated and resurrected the CL.
Co-Authored-By: Vincent Ambo <tazjin@tvl.su>
Signed-off-by: Adam Joseph <adam@westernsemico.com>
Change-Id: Ib3309a47a7b31fa5bf10466bade0d876b76ae462
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7089
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
|
|
It isn't possible to implement PartialEq properly for Value, because
any sensible implementation needs to force() thunks, which cannot be
done without a `&mut VM`.
The existing derive(PartialEq) has false negatives, which caused the
bug which cl/7142 fixed. Fortunately that bug was easy to find, but
a silent false negative deep within the bowels of nixpkgs could be a
real nightmare to hunt down.
Let's just remove the PartialEq impl for Value, and the other
derive(PartialEq)'s that depend on it.
Signed-off-by: Adam Joseph <adam@westernsemico.com>
Change-Id: Iacd3726fefc7fc1edadcd7e9b586e04cf8466775
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7144
Reviewed-by: kanepyork <rikingcoding@gmail.com>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
|
|
This commit deduplicates the Thunk-like functionality from Closure
and unifies it with Thunk.
Specifically, we now have one and only one way of breaking reference
cycles in the Value-graph: Thunk. No other variant contains a
RefCell. This should make it easier to reason about the behavior of
the VM. InnerClosure and UpvaluesCarrier are no longer necessary.
This refactoring allowed an improvement in code generation:
`Rc<RefCell<>>`s are now created only for closures which do not have
self-references or deferred upvalues, instead of for all closures.
OpClosure has been split into two separate opcodes:
- OpClosure creates non-recursive closures with no deferred
upvalues. The VM will not create an `Rc<RefCell<>>` when executing
this instruction.
- OpThunkClosure is used for closures with self-references or
deferred upvalues. The VM will create a Thunk when executing this
opcode, but the Thunk will start out already in the
`ThunkRepr::Evaluated` state, rather than in the
`ThunkRepr::Suspeneded` state.
To avoid confusion, OpThunk has been renamed OpThunkSuspended.
Thanks to @sterni for suggesting that all this could be done without
adding an additional variant to ThunkRepr. This does however mean
that there will be mutating accesses to `ThunkRepr::Evaluated`,
which was not previously the case. The field `is_finalised:bool`
has been added to `Closure` to ensure that these mutating accesses
are performed only on finalised Closures. Both the check and the
field are present only if `#[cfg(debug_assertions)]`.
Change-Id: I04131501029772f30e28da8281d864427685097f
Signed-off-by: Adam Joseph <adam@westernsemico.com>
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7019
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
|
|
Lambda has a quite large and variable-sized runtime representation,
unlike Rc<Lambda>. It would be easy to accidentally call clone() on
this and create input-dependent performance regressions.
Nothing in the codebase is currently using Lambda.clone(). Let's
remove the derived instance. If it's really needed it is very easy
to add it back in, but whoever does that will have to look at the
struct they're adding Clone to, which will hopefully prompt them to
think about whether or not that's really what they want to do.
Change-Id: I7806a741862ab4402229839ce3f4ea75aafd6d12
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7029
Autosubmit: Adam Joseph <adam@westernsemico.com>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
|
|
This type hides away the lower-level handling of most codemap data
structures, especially to library consumers (see corresponding changes
in tvixbolt).
This will help with implement `import` by giving us central control
over how the codemap works.
Change-Id: Ifcea36776879725871b30c518aeb96ab5fda035a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6855
Tested-by: BuildkiteCI
Reviewed-by: wpcarro <wpcarro@gmail.com>
|
|
Using rust's PartialEq trait to implement Nix equality semantics is
reasonably fraught with peril, both because the actual laws are
different than what nix expects, and (more importantly) because certain
things actually require extra context to compare for equality (for
example, thunks need to be forced). This converts the manual PartialEq
impl for Value (and all its descendants) to a *derived* PartialEq
impl (which requires a lot of extra PartialEq derives on miscellanious
other types within the codebase), and converts the previous
nix-semantics equality comparison into a new `nix_eq` method. This
returns an EvalResult, even though it can't currently return an error,
to allow it to fail when eg forcing thunks (which it will do soon).
Since the PartialEq impls for Value and NixAttrs are now quite boring,
this converts the generated proptests for those into handwritten ones
that cover `nix_eq` instead
Change-Id: If3da7171f88c22eda5b7a60030d8b00c3b76f672
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6650
Autosubmit: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
|
|
This is pretty boring at the moment, but mostly serves as a foot in the
door in the direction of writing more tests
Change-Id: Id88eb4ec7e53ebb2d5b5c254c8f45ff750238811
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6637
Autosubmit: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
|
|
This is used to drop an already emitted operation from a chunk again
and clean up its span tracking. This is required in cases where the
compiler has to backtrack.
Change-Id: I8112da9427688bb2dec96a2ddd12390f6e9734c3
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6499
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
|
|
As pointed out by grfn on cl/6091
Change-Id: I28308577b7cf99dffb4a4fd3cc8783eb9ab4d0d6
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6460
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
|
|
Change-Id: Ic6710c609ed647bfa47d673aaf22c4da96c0f319
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6451
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
This type implements an observer that is called whenever the compiler
emits a chunk (after the toplevel, thunks, or lambdas) and prints the
output of the disassembler to its internal writer.
This replaces half of the uses of the `disassembler` feature, which
has been removed from the Cargo configuration.
Note that at this commit runtime tracing is not yet implemented as an
observer.
Change-Id: I7894ca1ba445761aba4ad51d98e4a7b6445f1aea
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6449
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
This is a step towards hiding the internal fields of thunk, and making
the interface of the type more predictable.
Part of the preparation for implementing observers.
Change-Id: I1a88a96419c72eb9e2332b56a2dd94afa47e6f88
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6447
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
There can be different spans on the same line, so the previous
implementation would duplicate line numbers unnecessarily.
Change-Id: I8d8db77177aee0d834a6ec3584641e1bd5f31c3e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6434
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
If the disassembler feature is enabled, make sure that an Rc of the
codemap is available through the chunk.
Change-Id: I700f27ab665a704f73457b19bd2d7efc93828a16
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6414
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
As of this commit, the source spans of all emitted bytecode are fully
tracked.
Change-Id: I4c83deee0fc3f5e6fd6acad5a39047aec693b388
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6403
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
|
|
These are not actually used yet; this is in preparation for a
multi-commit chain for emitting all the right spans in the right
locations.
Change-Id: Ie99d6add2696c1cc0acb9ab928917a10237159de
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6379
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
|
|
This adds a new vector to the chunk data structure which tracks spans
into a codemap. The compiler will emit this information to the chunk
when adding instructions.
The internal representation of the spans is slightly optimised to
avoid storing duplicate spans, as there are cases where many
instructions might be derived from the same span.
Change-Id: I336f8c912e7eb50ea02ed71e6164f651ca3ca790
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6376
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
|
|
Changes the internal compiler plumbing to not just return a chunk of
code, but the same chunk wrapped inside of a lambda value.
This is one more step towards compiling runtime lambdas.
Change-Id: If0035f8e65a2970c5ae123fc068a2396e1d8fd72
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6240
Tested-by: BuildkiteCI
Reviewed-by: grfn <grfn@gws.fyi>
|
|
This makes for a much nicer view of an execution if `--feature
disassembler` is set, for example:
tvix-repl> let value = [ 1 2 { a = 1; } ]; in value ++ [ 1 ]
=== compiled bytecode (11 operations) ===
02 OpConstant(1)
02 OpConstant(2)
02 OpConstant("a")
02 OpConstant(1)
02 OpAttrs(1)
02 OpList(3)
02 OpGetLocal(0)
02 OpConstant(1)
02 OpList(1)
02 OpConcat
02 OpCloseScope(1)
=== runtime trace ===
0001 OpConstant(ConstantIdx(0)) [ 1 ]
0002 OpConstant(ConstantIdx(1)) [ 1 2 ]
0003 OpConstant(ConstantIdx(2)) [ 1 2 "a" ]
0004 OpConstant(ConstantIdx(3)) [ 1 2 "a" 1 ]
0005 OpAttrs(1) [ 1 2 { a = 1; } ]
0006 OpList(3) [ [ 1 2 { a = 1; } ] ]
0007 OpGetLocal(0) [ [ 1 2 { a = 1; } ] [ 1 2 { a = 1; } ] ]
0008 OpConstant(ConstantIdx(4)) [ [ 1 2 { a = 1; } ] [ 1 2 { a = 1; } ] 1 ]
0009 OpList(1) [ [ 1 2 { a = 1; } ] [ 1 2 { a = 1; } ] [ 1 ] ]
0010 OpConcat [ [ 1 2 { a = 1; } ] [ 1 2 { a = 1; } 1 ] ]
0011 OpCloseScope(1) [ [ 1 2 { a = 1; } 1 ] ]
=> [ 1 2 { a = 1; } 1 ] :: list
Change-Id: If79c7fd1f0f18255ddb3763c1ba585fda8041b1b
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6195
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
|
|
grfn pointed out in cl/6069 that naming them like this makes it clear
that things are being added to the end of the state.
Change-Id: I6a23215c4fef713869a3c85b0dde1ebbda7637e9
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6179
Tested-by: BuildkiteCI
Reviewed-by: grfn <grfn@gws.fyi>
|
|
Change-Id: I53202e93938bede421c8f1c98901e4c67544e257
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6069
Tested-by: BuildkiteCI
Reviewed-by: grfn <grfn@gws.fyi>
|
