If a -pos argument is specified, a 'callgraph' query reports only the
functions within the query package. This produces a far more manageable
amount of information, and because we don't need to package-qualify the
names, the result is easier to read.
Added tests:
- callgraph query with/without -pos
(The test driver was extended to allow "nopos" queries.)
- callers and callees queries don't return wrappers
Also, in go/callgraph:
- (*Node).String, (*Edge).String
- (*Graph).DeleteSyntheticNodes eliminates synthetic wrapper functions,
preserving topology. Used in all four oracle "call*" queries.
- (*Graph).DeleteNode
LGTM=crawshaw
R=crawshaw
CC=golang-codereviews
https://golang.org/cl/66240044
- clearer separation between package-level and object-level state
- lazy allocation of various package-level maps
- NewPackage automatically allocates a package scope
- steps towards enabling incremental checking of extra (test) files
after the (non-test) package is type-checked (not yet exposed)
LGTM=adonovan
R=adonovan
CC=golang-codereviews
https://golang.org/cl/66230044
1) We remove context sensitivity from API. The pointer analysis is
not sufficiently context-sensitive for the context information to
be worth exposing. (The actual analysis precision still benefits
from being context-sensitive, though.) Since all clients would
discard the context info, we now do that for them.
2) Make the graph doubly-linked. Edges are now shared by the Nodes
at both ends of the edge so it's possible to navigate more easily
(e.g. to the callers).
3) Graph and Node are now concrete, not interfaces.
Less code in every file!
LGTM=crawshaw
R=crawshaw
CC=golang-codereviews
https://golang.org/cl/66460043
Previously, each {Indirect,}Query would return a set of Pointers, one per context; now it returns (at most) one Pointer combining information from all contexts.
The old API was more faithful to the implementation concepts, but the analysis is not sufficiently context-sensitive that it makes sense: all existing clients simply throw away the context information---so now we do that for them.
(I may remove the context-sensitivity from the callgraph too, but I'll benchmark that first to see if it reduces precision.)
LGTM=crawshaw
R=crawshaw
CC=golang-codereviews
https://golang.org/cl/66130044
- pre-poluate typIndex and typList using a predefined list of types
- no need to handle basic types explicitly anymore
- removed basicTag
- go/types: exported UniverseByte and UniverseRune for now
LGTM=adonovan
R=adonovan
CC=cmang, golang-codereviews
https://golang.org/cl/65920044
If this flag is set, (*Config).Load will not return an
error even if some packages had type errors. Each individual
PackageInfo can be queried for its error state, now exposed as
TypeError.
In addition, each PackageInfo exposes whether it is
"transitively error-free". ssa.Create skips packages without
this flag since it is required for SSA construction.
+ Test.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/62000045
Observation: not all alias facts are interesting.
- A channel-peers query also cares about pointers of kind chan.
- An oracle "points-to" query on an expression of kind map
only cares about maps.
- We always care about func, interface and reflect.Value,
since they're needed for sound analysis of dynamic dispatch.
We needn't bother collecting alias information for
uninteresting pointers, and this massively reduces the number
of labels flowing in to the constraint system.
The only constraints that create new labels are addressOf
and offsetAddr; both are now selectively emitted by type.
We compute the set of type kinds to track, based on the
{Indirect,}Query types. (We could enable tracking at an
even finer grain if we want.)
This requires that we can see all the {Indirect,}Query
value types a priori, which is not the case for the PrintCalls
mechanism used in the tests, so I have rewritten the latter
to use {Indirect,}Query instead.
This reduces the solver-phase time for the entire standard
library and tests from >8m to <2m. Similar speedups are
obtained on small and medium-sized programs.
Details:
- shouldTrack inspects the flattened form of a type to see if
it contains fields we must track. It memoizes the result.
- added precondition checks to (*Config).Add{,Indirect}Query.
- added (*ssa.Program).LookupMethod convenience method.
- added Example of how to use the Query mechanism.
- removed code made dead by a recent invariant:
the only pointerlike Const value is nil.
- don't generate constraints for any functions in "reflect".
(we had forgotten to skip synthetic wrappers too).
- write PTA warnings to the log.
- add annotations for more intrinsics.
LGTM=gri, crawshaw
R=crawshaw, gri
CC=golang-codereviews
https://golang.org/cl/62540043
Method Signatures (types.Signatures with a non-nil receiver) behave
like ordinary func Signatures in the hash function/equivalence relation
used by typemap.M, which leads to surprising incomplete traversal
over the type graph if an M is used to remember types already
visited.
This change avoids ever putting method Signatures in a typemap.
% go test -v code.google.com/p/go.tools/go/ssa
now repeatedly shows the exact same number of functions and
instructions.
(We should discuss how to avoid this problem more generally.)
Also:
- recur over the params/results of all the methods of
each type when computing necessary method sets.
- there's no longer any need to treat declarations of unexported
methods as a root for traversal. Added test case.
- enable SourceImports flag in stdlib_test, which was dropped
during a recent refactoring (d'oh).
- doc tweaks
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/65450043
This CL adds no-op stubs for intrinsics now required by tests:
runtime.Goexit
sync.runtime_Sem{acquire,release}
sync/atomic.AddUint{32,64}
Goexit needs more thought; for now I've disabled the interpreted
tests of the "testing" package to make the build green again.
TBR=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/65480044
- Add math.{Log,Ldexp} to externals
- Test now uses FromArgs
- Scan all initial packages for tests, don't assume Created[0] exists.
(It doesn't if we import a package that has only in-package tests.)
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/62010044
Previously, each word could be a package import path or a
comma-separated list of *.go file names. Now, if the
first word ends with ".go", all words are assumed to be
Go source files. This makes it impossible to specify
two ad-hoc packages from source files, but no-one needs that.
FromArgs also takes a boolean indicating whether tests
are wanted or not.
Also: ssadump: add -test flag to set that boolean.
For the oracle it's always true.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/61470047
Method-set caching is now performed externally using a MethodSetCache (if desired), not by the Types themselves.
This a minor deoptimization due to the extra maps, but avoids a situation in which method-sets are computed and frozen prematurely. (See b/7114)
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/61430045
Move mutation of program to the final step.
+ 2 assertions;
+ switches.go: comment fix.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/61510044
The "static bool" parameter to Implements was confusing; typically we think about interface implementation and type assertion as separate but related concepts, but here they were merged.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/58540044
- Centralize "visited" check in check.objDecl.
- Assert that object-specific declX functions
are only called with objects that have no
type associated with them, yet.
- Added test case.
Thanks to Richard Musiol (neelance@gmail.com) for
finding the bug and providing an easy test case.
For a discussion of the bug see the issue.
Fixesgolang/go#7245.
TBR=adonovan
R=adonovan
CC=golang-codereviews
https://golang.org/cl/59210043
- consistently set underlying types of incoming named types in typInternal
- use underlying() helper function to resolve forward chains
- related consistency cleanups
LGTM=adonovan
R=adonovan
CC=golang-codereviews
https://golang.org/cl/53870044
This is (a) more efficient and (b) avoids the need for
constant error handling, since buffer writes can't fail.
Also:
- added WriteFunction and WritePackage functions,
similar to types.WriteFoo.
- *Function and *Package now implement io.WriterTo.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/57930044
This results in significant improvement to type-checking time:
it reduces by 4% the entire running time of ssa/stdlib_test
(GOMAXPROCS=8, n=7).
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/57770043
Since Put() makes a dynamic function call, this CL includes a long-overdue change to supply a *frame (and thus the call
stack and the interpreter) to intrinsics too.
This fixes the tests that were broken by (sound) revision e4a4cb47c141.
LGTM=gri
R=gri
CC=golang-codereviews
https://golang.org/cl/57350043
CL 49530047 made the (over-)simplifying assumption that the
Path of an ad-hoc (Created) package should default to its
Name, i.e. package declaration.
With this change, the Name is still always computed from the
package declaration (by go/types) but the Path may be
specified by the loader.Config. If "", the value of the Name
is used, which is not globally unique.
R=gri, axwalk
CC=golang-codereviews
https://golang.org/cl/55180043
Was: Now:
call.Graph callgraph.Graph
call.GraphNode callgraph.Node
call.Edge callgraph.Edge
Though call.Graph was cute, the original naming was a mistake:
'call' is too useful a var name to waste on a package.
R=gri, crawshaw
CC=golang-codereviews
https://golang.org/cl/53190043
The goal is to allow clients to distinguish function-local objects
from package-local objects and identify the specific function by
examining its scope. For example, a compiler may need to mangle
object names according to the outer function name.
R=gri
CC=golang-codereviews, golang-dev
https://golang.org/cl/52280043
Both gc and gccgo appear to use this size, so it seems appropriate
to use it here.
R=gri
CC=golang-codereviews, golang-dev, iant, rsc
https://golang.org/cl/52290043
The issue is addressed with the change in conversions.go.
Also:
- added corresponding API test
- made names for untyped ints, bools consistent with others
- use *Basic with names byte/rune instead of uint8/int32 for better output
- minor cleanups
Fixesgolang/go#6949.
R=adonovan
CC=golang-codereviews
https://golang.org/cl/50520043
Now that inner functions are processed "in line", usage errors
can be detected immediately after each function is processed.
R=adonovan
CC=golang-codereviews
https://golang.org/cl/49900044
Inner function bodies must be type-checked "in-place" for
them to see the correct state of the surrounding scopes.
Fixesgolang/go#7035.
R=adonovan
CC=golang-codereviews
https://golang.org/cl/49350043
This can import all of the standard library, and has been tested
by using gotype to type check libgo with gccgo's export data (this
would be nice to automate, but I can't see a good way to do it,
not least because system-specific source files cause errors which
I needed to identify manually).
It includes a builtin export locator. Unfortunately I can't see a
more reliable way to locate the builtin export files than to parse
the output of 'gccgo -###'.
R=gri, iant, gri
CC=golang-codereviews, golang-dev
https://golang.org/cl/31860043
- if a named type was imported before, read it again
and throw it away in favor of the existing type
(the old code did the same, but more circuitously)
- better tag name for int64 values
R=adonovan
CC=golang-codereviews
https://golang.org/cl/47650044
Anonymous field names are emitted as "" in the export data
since the actual name can be reconstructed easily from the
field's type name. But "" names are not exported names and
thus the respective qualified name emits complete package
information even if the actual field name is exported. Fix
the package upon import.
R=adonovan
CC=golang-codereviews
https://golang.org/cl/42090044
The package of a struct field is independent of the type of the
struct field - the old code was technically not correct. That said,
it does not seem possible (or very difficult) to create a test case
because for exported anonymous fields that field package doesn't matter
(it's not needed for name identity), and non-exported anonymous fields
cannot be accessed from an imported package.
R=adonovan
CC=golang-codereviews
https://golang.org/cl/47560043
Also: Provide GcCompatibilityMode for printing types
(intended for testing with gc-generated export data
only).
(TBR adonovan)
R=adonovan
TBR=adonovan
CC=golang-codereviews
https://golang.org/cl/44780043
By using a simple (graph-based) serialization algorithm
and binary encoding, a significantly more compact export
data format is achieved than what the current compilers
use. Furthermore, the exporter and importer are completely
symmetric algorithms that are compact, and much easier to
change/expand.
R=adonovan
CC=golang-dev
https://golang.org/cl/42960043
Obviously in that mode, we can't correctly diagnose such
errors, so we shouldn't attempt it (and emit false positives).
R=gri
CC=golang-dev
https://golang.org/cl/41080043
This test generates a program that declares the constant
elements of an n*n Hilbert matrix, its inverse, and the
constant elements of the explicit product of the two.
The product should be the identity matrix; that check is
also expressed as a constant expression. Type-checking
verifies that the product is indeed the identity matrix
by asserting the result of the identity check (using the
assert built-in which is available for type-check tests).
The test is run for n = 5. Other values can be tested via
the -H flag, say: go test -run=Hilbert -H=100
The generated program can be written to a file for testing
the constant arithmetic of a compiler: go test -out=test.go
Because of the mathematically precise constant arithmetic
of go/types, this test should always succeed and is only
limited by the size of the matrix. It does run successfully
from n = 0 to values larger than 100.
The Hilbert matrix is famous for being ill-conditioned and
thus exposes arithmetic imprecision very quickly. The gc
compiler only produces a correct result for n = 0 (trivially),
and n = 1 at the moment.
R=adonovan, rsc
CC=golang-dev
https://golang.org/cl/35840043
The benchmarks don't permit clear-cut apples-to-apples
comparisons since they depend on the very source code
they are testing. But they do give an indication of
the approximate performance - as a sanity test.
Using 3 different code bases makes it apparent that
there's some difference in performance per code base;
i.e., the lines/s speed varies pretty strongly. This
may be due to setup costs, or other issues. We should
investigate eventually.
R=adonovan
CC=golang-dev
https://golang.org/cl/30650043
Also: Use defer to make sure scopes are always closed
even in case of early exits via bailout (and don't cause
an imbalanced scope error in debug mode).
R=adonovan, gri
CC=golang-dev
https://golang.org/cl/29300043
- consolidate them in (expr|type)string[_test].go
- support for printing all ast.Expr
- fix printing of type: chan (<-chan int) (parentheses)
- more consistent names, comments, and exports
R=adonovan
CC=golang-dev
https://golang.org/cl/28680043
These are variants of Type.String(), Object.String() that
accept a 'from *Package' argument. If provided, package
qualification is omitted when printing named types belonging
to that package.
This is useful for UIs where a package is implied by context
e.g. ssadump disassembly, oracle output.
+ Test.
R=gri, gri, gordon.klaus
CC=golang-dev
https://golang.org/cl/22190048
- fixed a couple of TODOs
- various cleanups along the way
- adjusted clients
Once submitted, clients of go/types that don't explicitly
specify Config.Import will need to add the extra import:
import _ "code.google.com/p/go.tools/go/gcimporter"
to install the default (gc) importer in go/types.
R=adonovan, gri
CC=golang-dev
https://golang.org/cl/26390043
Depending on the context, printing only the package name can
be ambiguous or even incorrect since it is valid only within
the environment of a given file's import specs.
(The standard library packages are mostly unique in their last
segment, but this is not the case for proprietary repos.)
R=gri, gri
CC=golang-dev
https://golang.org/cl/26300043
This check is currently done in go/parser as well but
eventually can be removed from there (after Go 1.2).
R=adonovan
CC=golang-dev
https://golang.org/cl/22240045
Initialization cycles need to reported for cycles
that contain variables, even if they don't end in
a variable.
This fixes the last known issue with the existing
std library tests.
R=adonovan, gri
CC=golang-dev
https://golang.org/cl/22200049
The gc compiler is inconsistent how it handles method
"mentions" with respect to initialization cycle detection
(see issue 6703 for details). Pending a spec clarification,
this CL assumes that for a method to be "mentioned", it
must be mentioned as a method expression rather than a
method value (closer in intent to "syntactic" mention).
R=adonovan
CC=golang-dev
https://golang.org/cl/22050044
- Info.InitOrder now provides list of Initializers
(vars + init expr), handling n:1 decls and blank
identifiers
- added respective API test
- cycles detected through function "mentions"
Missing: cycles through method "mentions" and via
closures
R=adonovan
CC=golang-dev
https://golang.org/cl/21810043
Missing:
- dependencies via functions (incl. closures and methods)
- more tests (incl. API test)
R=adonovan
CC=golang-dev
https://golang.org/cl/20510043
- better error messages
- in contrast to a long-standing TODO, comparisons
between interface and non-interface types always
worked correctly
R=adonovan
CC=golang-dev
https://golang.org/cl/17310043
The spec does not exclude blank _ method names in interfaces
from the uniqueness criteria; i.e., at most one blank method
may appear in an interface type.
Arguably the spec is vague (and possibly incorrect) here.
gccgo handles it the same way. gc crashes with an internal
compiler error.
R=adonovan
CC=golang-dev
https://golang.org/cl/16380043
Motivation:
Previously, we assumed that the set of types for which a
complete method set (containing all synthesized wrapper
functions) is required at runtime was the set of types
used as operands to some *ssa.MakeInterface instruction.
In fact, this is an underapproximation because types can
be derived from other ones via reflection, and some of
these may need methods. The reflect.Type API allows *T to
be derived from T, and these may have different method
sets. Reflection also allows almost any subcomponent of a
type to be accessed (with one exception: given T, defined
'type T struct{S}', you can reach S but not struct{S}).
As a result, the pointer analysis was unable to generate
all necessary constraints before running the solver,
causing a crash when reflection derives types whose
methods are unavailable. (A similar problem would afflict
an ahead-of-time compiler based on ssa. The ssa/interp
interpreter was immune only because it does not require
all wrapper methods to be created before execution
begins.)
Description:
This change causes the SSA builder to record, for each
package, the set of all types with non-empty method sets that
are referenced within that package. This set is accessed via
Packages.TypesWithMethodSets(). Program.TypesWithMethodSets()
returns its union across all packages.
The set of references that matter are:
- types of operands to some MakeInterface instruction (as before)
- types of all exported package members
- all subcomponents of the above, recursively.
This is a conservative approximation to the set of types
whose methods may be called dynamically.
We define the owning package of a type as follows:
- the owner of a named type is the package in which it is defined;
- the owner of a pointer-to-named type is the owner of that named type;
- the owner of all other types is nil.
A package must include the method sets for all types that it
owns, and all subcomponents of that type that are not owned by
another package, recursively. Types with an owner appear in
exactly one package; types with no owner (such as struct{T})
may appear within multiple packages.
(A typical Go compiler would emit multiple copies of these
methods as weak symbols; a typical linker would eliminate
duplicates.)
Also:
- go/types/typemap: implement hash function for *Tuple.
- pointer: generate nodes/constraints for all of
ssa.Program.TypesWithMethodSets().
Add rtti.go regression test.
- Add API test of Package.TypesWithMethodSets().
- Set Function.Pkg to nil (again) for wrapper functions,
since these may be shared by many packages.
- Remove a redundant logging statement.
- Document that ssa CREATE phase is in fact sequential.
Fixesgolang/go#6605
R=gri
CC=golang-dev
https://golang.org/cl/14920056
This will make it a bit easier to create commonly used "custom" sizes for types.
With this CL, interfaces are now by default 2*WordSize (= 16) instead of 1*WordSize
as before.
Also: minor unrelated cleanups.
R=adonovan
CC=golang-dev
https://golang.org/cl/14719043
Given a built-in call f(args), Info.Types now maps f to the call-site
specific type of f (by looking at the argument types) if the built-in
call is not producing a constant (at typecheck time) result. If the
result is constant, the recorded type is invalid (a back-end won't
need it).
R=adonovan
CC=golang-dev
https://golang.org/cl/14598045
Alan Donovan
Robert Griesemer
Gordon Klaus
Alex Brainman
Peter Collingbourne
Richard Musiol