Augment the ir.Inline container to include an entry for function
properties (currently serialized as a string), and if
GOEXPERIMENT=newinliner is set, compute and store function
properties for all inline candidates processed by the inliner.
The idea here is that if the function properties are going to drive
inlining decisions, we'd like to have the same info from non-local /
imported functions as for local / in-package functions, hence we need
to include the properties in the export data.
Hand testing on the compiler itself and with k8s kubelet shows that
this increases the size of export data overall by about 2-3 percent,
so a pretty modest increase.
Updates #61502.
Change-Id: I9d1c311aa8418d02ffea3629c3dd9d8076886d15
Reviewed-on: https://go-review.googlesource.com/c/go/+/511562
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
We don't actually depend on Inl.Body anywhere, except it implicitly
serves to indicate whether Inl.Dcl has been populated. So replace it
with a boolean so we don't need to keep a useless copy of every
inlinable function body in memory.
While here, also add some Fatalfs to make sure there are no unused
local variables. The unified frontend now omits unreachable code
during export data writing, so there shouldn't be unused local
variables.
Also, since unified IR uses the same code/data to construct the
original function as inlined and/or imported functions, the Dcl list
should always be the same, which addresses the real root issue (i.e.,
that export/import could skew the Dcl lists).
Change-Id: I6e3435f3a0352f6efbae787344006efac1891e84
Reviewed-on: https://go-review.googlesource.com/c/go/+/523315
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The types.RecvsParamsResults, etc. helpers existed to make it "easier"
to iterate over all parameters, or recvs+params, or params+results;
but they end up still being quite clumsy to use due to the design goal
of not allocating temporary slices.
Now that recvs+params+results are stored in a single consecutive slice
anyway, we can just return different subslices and simplify the loops.
Change-Id: I84791b80dc099dfbfbbe6eddbc006135528c23b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/521375
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
This CL removes a lot of the redundant methods for accessing struct
fields and signature parameters. In particular, users never have to
write ".Slice()" or ".FieldSlice()" anymore; the exported APIs just do
what you want.
Further internal refactorings to follow.
Change-Id: I45212f6772fe16aad39d0e68b82d71b0796e5639
Reviewed-on: https://go-review.googlesource.com/c/go/+/521295
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Due to a missing "&& !alias" check, the unified linker was treating
type aliases the same as defined types for the purpose of exporting
method bodies. The methods will get exported anyway alongside the
aliased type, so this mistake is normally harmless.
However, if multiple type aliases instantiated the same generic type
but with different type arguments, this could result in the
same (generic) method body being exported multiple times under
different symbol names. Further, because bodies aren't expected to be
exported multiple times, we were sorting them simply based on index.
And consequently, the sort wasn't total and is sensitive to the map
iteration order used while ranging over linker.bodies.
The fix is simply to add the missing "&& !alias" check, so that we
don't end up with duplicate bodies in the first place.
Thanks rsc@ for providing a minimal repro case.
Fixes#59571.
Change-Id: Iaa55968cc7110b601e2f0f9b620901c2d55f7014
Reviewed-on: https://go-review.googlesource.com/c/go/+/484155
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Go programs can now use the //go:wasmimport module_name function_name
directive to import functions from the WebAssembly runtime.
For now, the directive is restricted to the runtime and syscall/js
packages.
* Derived from CL 350737
* Original work modified to work with changes to the IR conversion code.
* Modification of CL 350737 changes to fully exist in Unified IR path (emp)
* Original work modified to work with changes to the ABI configuration code.
* Fixes#38248
Co-authored-by: Vedant Roy <vroy101@gmail.com>
Co-authored-by: Richard Musiol <mail@richard-musiol.de>
Co-authored-by: Johan Brandhorst-Satzkorn <johan.brandhorst@gmail.com>
Change-Id: I740719735d91c306ac718a435a78e1ee9686bc16
Reviewed-on: https://go-review.googlesource.com/c/go/+/463018
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Johan Brandhorst-Satzkorn <johan.brandhorst@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Johan Brandhorst-Satzkorn <johan.brandhorst@gmail.com>
Reviewed-by: Johan Brandhorst-Satzkorn <johan.brandhorst@gmail.com>
An important optimization in the existing export data format is the
pruning of unreachable inline bodies. That is, when re-exporting
transitively imported types, omitting the inline bodies for methods
that can't actually be needed due to importing that package.
The existing logic (implemented in typecheck/crawler.go) is fairly
sophisticated, but also relies on actually expanding inline bodies in
the process, which is undesirable. However, including all inline
bodies is also prohibitive for testing GOEXPERIMENT=unified against
very large Go code bases that impose size limits on build action
inputs.
As a short-term solution, this CL implements a simple heuristic for
GOEXPERIMENT=unified: include the inline bodies for all
locally-declared functions/methods, and for any imported
functions/methods that were inlined into this package.
Change-Id: I686964a0cd9262b77d3d5587f89cfbcfe8b2e521
Reviewed-on: https://go-review.googlesource.com/c/go/+/419675
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
This adds more documentation throughout the core Unified IR logic and
removes their UNREVIEWED notices.
Updates #48194.
Change-Id: Iddd30edaee1c6ea8a05a5a7e013480e02be00d29
Reviewed-on: https://go-review.googlesource.com/c/go/+/411917
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Element indices are very common in the pkgbits API, so introduce a new
defined type to help make that clearer.
Change-Id: I8f9493e2335601c740eb403d1fdcd11183122907
Reviewed-on: https://go-review.googlesource.com/c/go/+/407435
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Replace `pkg.Path == ""` check with `pkg == types.LocalPkg`. This is a
prep refactoring for CL 393715, which will properly initialize
types.LocalPkg.
Updates #51734.
Change-Id: I7a5428ef1f422de396762b6bc6d323992834b27c
Reviewed-on: https://go-review.googlesource.com/c/go/+/406056
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
This logic is needed for the go/types unified IR importer, so extract
it into a separate internal package so we can reuse a single copy.
Change-Id: I5f734b76e580fdb69ee39e45ac553c22d01c5909
Reviewed-on: https://go-review.googlesource.com/c/go/+/386000
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
This CL makes two changes:
1. It moves object symbols and code tags into a new "relocName"
relocation, which should eventually allow getting rid of objStub.
2. It moves the type parameter data into the relocObjDict relocation,
so everything related to writing out dictionaries is contained there.
Change-Id: If0f7ff7d9384e8664957c3180bf6f20e97bcff6e
Reviewed-on: https://go-review.googlesource.com/c/go/+/336051
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
We already use -lang=go1.18 to control the types2 type checker
behavior. This CL does the same for the parser.
Also, disable an assertion in the unified IR linker that depended on
the -G flag. This assertion was more useful during initial
bootstrapping of that code, but it's less critical now.
With these two changes, "GOEXPERIMENT=unified ./make.bash" is enough
to get a fully functional generics-enabled toolchain. There's no need
to continue specifying custom compiler flags later on.
Change-Id: I7766381926f3bb17eee2e5fcc182a38a39e937e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/332373
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
This CL updates the unified IR export data serialization to explicitly
and separately record the derived types used by a declaration. The
readers currently just use this data to construct types/IR the same as
before, but eventually we can use it for emitting GC-shape
dictionaries.
Change-Id: I7d67ad9b3f1fbe69664bf19e056bc94f73507220
Reviewed-on: https://go-review.googlesource.com/c/go/+/331829
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Matthew Dempsky <mdempsky@google.com>
While initially building out unified IR, I didn't have any indexing
scheme. Everything was written out in order. Consequently, if I wanted
to write A before B, I had to compute A before B.
One particular example of this is handling closure variables: the
reader needs the list of closure variables before it can start reading
the function body, so I had to write them out first, and so I had to
compute them first in a separate, dedicated pass.
However, that constraint went away a while ago. For example, it's now
possible to replace the two-pass closure variable capture with a
single pass. We just write out the function body earlier, but then
wait to write out its index.
I anticipate this approach will make it easier to implement
dictionaries: rather than needing a separate pass to correctly
recognize and handle all of the generics cases, we can just hook into
the existing logic.
Change-Id: Iab1e07f9202cd5d2b6864eef10116960456214df
Reviewed-on: https://go-review.googlesource.com/c/go/+/330851
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
CL 328051 introduced new syncImplicitTypes, but forgot to add a sync
after syncAddBody in linker.relocFuncExt, cause the compiler crashes
when reading in package data.
Adding missing w.sync(syncImplicitTypes) call fixes this.
While at it, also run go generate to update code generated for
syncImplicitTypes, which is also missed in CL 328051.
Change-Id: Ic65092f69f8d8e63de15989c7f15b6e5633d8f9e
Reviewed-on: https://go-review.googlesource.com/c/go/+/328054
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
This CL adds a new unified IR construction mode to the frontend. It's
purely additive, and all files include "UNREVIEWED" at the top, like
how types2 was initially imported. The next CL adds a -d=unified flag
to actually enable unified IR mode.
See below for more details, but some highlights:
1. It adds ~6kloc (excluding enum listings and stringer output), but I
estimate it will allow removing ~14kloc (see CL 324670, including its
commit message);
2. When enabled by default, it passes more tests than -G=3 does (see
CL 325213 and CL 324673);
3. Without requiring any new code, it supports inlining of more code
than the current inliner (see CL 324574; contrast CL 283112 and CL
266203, which added support for inlining function literals and type
switches, respectively);
4. Aside from dictionaries (which I intend to add still), its support
for generics is more complete (e.g., it fully supports local types,
including local generic types within generic functions and
instantiating generic types with local types; see
test/typeparam/nested.go);
5. It supports lazy loading of types and objects for types2 type
checking;
6. It supports re-exporting of types, objects, and inline bodies
without needing to parse them into IR;
7. The new export data format has extensive support for debugging with
"sync" markers, so mistakes during development are easier to catch;
8. When compiling with -d=inlfuncswithclosures=0, it enables "quirks
mode" where it generates output that passes toolstash -cmp.
--
The new unified IR pipeline combines noding, stenciling, inlining, and
import/export into a single, shared code path. Previously, IR trees
went through multiple phases of copying during compilation:
1. "Noding": the syntax AST is copied into the initial IR form. To
support generics, there's now also "irgen", which implements the same
idea, but takes advantage of types2 type-checking results to more
directly construct IR.
2. "Stenciling": generic IR forms are copied into instantiated IR
forms, substituting type parameters as appropriate.
3. "Inlining": the inliner made backup copies of inlinable functions,
and then copied them again when inlining into a call site, with some
modifications (e.g., updating position information, rewriting variable
references, changing "return" statements into "goto").
4. "Importing/exporting": the exporter wrote out the IR as saved by
the inliner, and then the importer read it back as to be used by the
inliner again. Normal functions are imported/exported "desugared",
while generic functions are imported/exported in source form.
These passes are all conceptually the same thing: make a copy of a
function body, maybe with some minor changes/substitutions. However,
they're all completely separate implementations that frequently run
into the same issues because IR has many nuanced corner cases.
For example, inlining currently doesn't support local defined types,
"range" loops, or labeled "for"/"switch" statements, because these
require special handling around Sym references. We've recently
extended the inliner to support new features like inlining type
switches and function literals, and they've had issues. The exporter
only knows how to export from IR form, so when re-exporting inlinable
functions (e.g., methods on imported types that are exposed via
exported APIs), these functions may need to be imported as IR for the
sole purpose of being immediately exported back out again.
By unifying all of these modes of copying into a single code path that
cleanly separates concerns, we eliminate many of these possible
issues. Some recent examples:
1. Issues #45743 and #46472 were issues where type switches were
mishandled by inlining and stenciling, respectively; but neither of
these affected unified IR, because it constructs type switches using
the exact same code as for normal functions.
2. CL 325409 fixes an issue in stenciling with implicit conversion of
values of type-parameter type to variables of interface type, but this
issue did not affect unified IR.
Change-Id: I5a05991fe16d68bb0f712503e034cb9f2d19e296
Reviewed-on: https://go-review.googlesource.com/c/go/+/324573
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
