1
0
mirror of https://github.com/golang/go synced 2024-11-19 03:14:42 -07:00
go/oracle/whicherrs.go

327 lines
8.6 KiB
Go
Raw Normal View History

// Copyright 2014 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package oracle
import (
"fmt"
"go/ast"
"go/token"
"sort"
"golang.org/x/tools/go/ast/astutil"
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
"golang.org/x/tools/go/loader"
"golang.org/x/tools/go/ssa"
"golang.org/x/tools/go/ssa/ssautil"
"golang.org/x/tools/go/types"
"golang.org/x/tools/oracle/serial"
)
var builtinErrorType = types.Universe.Lookup("error").Type()
// whicherrs takes an position to an error and tries to find all types, constants
// and global value which a given error can point to and which can be checked from the
// scope where the error lives.
// In short, it returns a list of things that can be checked against in order to handle
// an error properly.
//
// TODO(dmorsing): figure out if fields in errors like *os.PathError.Err
// can be queried recursively somehow.
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
func whicherrs(q *Query) error {
lconf := loader.Config{Build: q.Build}
if err := setPTAScope(&lconf, q.Scope); err != nil {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return err
}
// Load/parse/type-check the program.
lprog, err := lconf.Load()
if err != nil {
return err
}
q.Fset = lprog.Fset
qpos, err := parseQueryPos(lprog, q.Pos, true) // needs exact pos
if err != nil {
return err
}
prog := ssautil.CreateProgram(lprog, ssa.GlobalDebug)
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
ptaConfig, err := setupPTA(prog, lprog, q.PTALog, q.Reflection)
if err != nil {
return err
}
path, action := findInterestingNode(qpos.info, qpos.path)
if action != actionExpr {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return fmt.Errorf("whicherrs wants an expression; got %s",
astutil.NodeDescription(qpos.path[0]))
}
var expr ast.Expr
var obj types.Object
switch n := path[0].(type) {
case *ast.ValueSpec:
// ambiguous ValueSpec containing multiple names
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return fmt.Errorf("multiple value specification")
case *ast.Ident:
obj = qpos.info.ObjectOf(n)
expr = n
case ast.Expr:
expr = n
default:
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return fmt.Errorf("unexpected AST for expr: %T", n)
}
typ := qpos.info.TypeOf(expr)
if !types.Identical(typ, builtinErrorType) {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return fmt.Errorf("selection is not an expression of type 'error'")
}
// Determine the ssa.Value for the expression.
var value ssa.Value
if obj != nil {
// def/ref of func/var object
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
value, _, err = ssaValueForIdent(prog, qpos.info, obj, path)
} else {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
value, _, err = ssaValueForExpr(prog, qpos.info, path)
}
if err != nil {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
return err // e.g. trivially dead code
}
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
// Defer SSA construction till after errors are reported.
prog.BuildAll()
globals := findVisibleErrs(prog, qpos)
constants := findVisibleConsts(prog, qpos)
res := &whicherrsResult{
qpos: qpos,
errpos: expr.Pos(),
}
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
// TODO(adonovan): the following code is heavily duplicated
// w.r.t. "pointsto". Refactor?
// Find the instruction which initialized the
// global error. If more than one instruction has stored to the global
// remove the global from the set of values that we want to query.
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
allFuncs := ssautil.AllFunctions(prog)
for fn := range allFuncs {
for _, b := range fn.Blocks {
for _, instr := range b.Instrs {
store, ok := instr.(*ssa.Store)
if !ok {
continue
}
gval, ok := store.Addr.(*ssa.Global)
if !ok {
continue
}
gbl, ok := globals[gval]
if !ok {
continue
}
// we already found a store to this global
// The normal error define is just one store in the init
// so we just remove this global from the set we want to query
if gbl != nil {
delete(globals, gval)
}
globals[gval] = store.Val
}
}
}
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
ptaConfig.AddQuery(value)
for _, v := range globals {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
ptaConfig.AddQuery(v)
}
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
ptares := ptrAnalysis(ptaConfig)
valueptr := ptares.Queries[value]
for g, v := range globals {
ptr, ok := ptares.Queries[v]
if !ok {
continue
}
if !ptr.MayAlias(valueptr) {
continue
}
res.globals = append(res.globals, g)
}
pts := valueptr.PointsTo()
dedup := make(map[*ssa.NamedConst]bool)
for _, label := range pts.Labels() {
// These values are either MakeInterfaces or reflect
// generated interfaces. For the purposes of this
// analysis, we don't care about reflect generated ones
makeiface, ok := label.Value().(*ssa.MakeInterface)
if !ok {
continue
}
constval, ok := makeiface.X.(*ssa.Const)
if !ok {
continue
}
c := constants[*constval]
if c != nil && !dedup[c] {
dedup[c] = true
res.consts = append(res.consts, c)
}
}
concs := pts.DynamicTypes()
concs.Iterate(func(conc types.Type, _ interface{}) {
// go/types is a bit annoying here.
// We want to find all the types that we can
// typeswitch or assert to. This means finding out
// if the type pointed to can be seen by us.
//
// For the purposes of this analysis, the type is always
// either a Named type or a pointer to one.
// There are cases where error can be implemented
// by unnamed types, but in that case, we can't assert to
// it, so we don't care about it for this analysis.
var name *types.TypeName
switch t := conc.(type) {
case *types.Pointer:
named, ok := t.Elem().(*types.Named)
if !ok {
return
}
name = named.Obj()
case *types.Named:
name = t.Obj()
default:
return
}
if !isAccessibleFrom(name, qpos.info.Pkg) {
return
}
res.types = append(res.types, &errorType{conc, name})
})
sort.Sort(membersByPosAndString(res.globals))
sort.Sort(membersByPosAndString(res.consts))
sort.Sort(sorterrorType(res.types))
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
q.result = res
return nil
}
// findVisibleErrs returns a mapping from each package-level variable of type "error" to nil.
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
func findVisibleErrs(prog *ssa.Program, qpos *queryPos) map[*ssa.Global]ssa.Value {
globals := make(map[*ssa.Global]ssa.Value)
for _, pkg := range prog.AllPackages() {
for _, mem := range pkg.Members {
gbl, ok := mem.(*ssa.Global)
if !ok {
continue
}
gbltype := gbl.Type()
// globals are always pointers
if !types.Identical(deref(gbltype), builtinErrorType) {
continue
}
if !isAccessibleFrom(gbl.Object(), qpos.info.Pkg) {
continue
}
globals[gbl] = nil
}
}
return globals
}
// findVisibleConsts returns a mapping from each package-level constant assignable to type "error", to nil.
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
func findVisibleConsts(prog *ssa.Program, qpos *queryPos) map[ssa.Const]*ssa.NamedConst {
constants := make(map[ssa.Const]*ssa.NamedConst)
for _, pkg := range prog.AllPackages() {
for _, mem := range pkg.Members {
obj, ok := mem.(*ssa.NamedConst)
if !ok {
continue
}
consttype := obj.Type()
if !types.AssignableTo(consttype, builtinErrorType) {
continue
}
if !isAccessibleFrom(obj.Object(), qpos.info.Pkg) {
continue
}
constants[*obj.Value] = obj
}
}
return constants
}
type membersByPosAndString []ssa.Member
func (a membersByPosAndString) Len() int { return len(a) }
func (a membersByPosAndString) Less(i, j int) bool {
cmp := a[i].Pos() - a[j].Pos()
return cmp < 0 || cmp == 0 && a[i].String() < a[j].String()
}
func (a membersByPosAndString) Swap(i, j int) { a[i], a[j] = a[j], a[i] }
type sorterrorType []*errorType
func (a sorterrorType) Len() int { return len(a) }
func (a sorterrorType) Less(i, j int) bool {
cmp := a[i].obj.Pos() - a[j].obj.Pos()
return cmp < 0 || cmp == 0 && a[i].typ.String() < a[j].typ.String()
}
func (a sorterrorType) Swap(i, j int) { a[i], a[j] = a[j], a[i] }
type errorType struct {
typ types.Type // concrete type N or *N that implements error
obj *types.TypeName // the named type N
}
type whicherrsResult struct {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
qpos *queryPos
errpos token.Pos
globals []ssa.Member
consts []ssa.Member
types []*errorType
}
func (r *whicherrsResult) display(printf printfFunc) {
if len(r.globals) > 0 {
printf(r.qpos, "this error may point to these globals:")
for _, g := range r.globals {
printf(g.Pos(), "\t%s", g.RelString(r.qpos.info.Pkg))
}
}
if len(r.consts) > 0 {
printf(r.qpos, "this error may contain these constants:")
for _, c := range r.consts {
printf(c.Pos(), "\t%s", c.RelString(r.qpos.info.Pkg))
}
}
if len(r.types) > 0 {
printf(r.qpos, "this error may contain these dynamic types:")
for _, t := range r.types {
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
printf(t.obj.Pos(), "\t%s", r.qpos.typeString(t.typ))
}
}
}
func (r *whicherrsResult) toSerial(res *serial.Result, fset *token.FileSet) {
we := &serial.WhichErrs{}
we.ErrPos = fset.Position(r.errpos).String()
for _, g := range r.globals {
we.Globals = append(we.Globals, fset.Position(g.Pos()).String())
}
for _, c := range r.consts {
we.Constants = append(we.Constants, fset.Position(c.Pos()).String())
}
for _, t := range r.types {
var et serial.WhichErrsType
oracle: several major improvements Features: More robust: silently ignore type errors in modes that don't need SSA form: describe, referrers, implements, freevars, description. This makes the tool much more robust for everyday queries. Less configuration: don't require a scope argument for all queries. Only queries that do pointer analysis need it. For the rest, the initial position is enough for importQueryPackage to deduce the scope. It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles. (It no longer works for ad-hoc main packages like $GOROOT/src/net/http/triv.go) More complete: "referrers" computes the scope automatically by scanning the import graph of the entire workspace, using gorename's refactor/importgraph package. This requires two passes at loading. Faster: simplified start-up logic avoids unnecessary package loading and SSA construction (a consequence of bad abstraction) in many cases. "callgraph": remove it. Unlike all the other commands it isn't related to the current selection, and we have golang.org/x/tools/cmdcallgraph now. Internals: Drop support for long-running clients (i.e., Pythia), since godoc -analysis supports all the same features except "pointsto", and precomputes all the results so latency is much lower. Get rid of various unhelpful abstractions introduced to support long-running clients. Expand out the set-up logic for each subcommand. This is simpler, easier to read, and gives us more control, at a small cost in duplication---the familiar story of abstractions. Discard PTA warnings. We weren't showing them (nor should we). Split tests into separate directories (so that importgraph works). Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237 Reviewed-on: https://go-review.googlesource.com/8243 Reviewed-by: David Crawshaw <crawshaw@golang.org>
2015-03-30 09:21:48 -06:00
et.Type = r.qpos.typeString(t.typ)
et.Position = fset.Position(t.obj.Pos()).String()
we.Types = append(we.Types, et)
}
res.WhichErrs = we
}