Change-Id: Ia0a4be56d4e1fbfc73e6ce24f01a658c89a74adb
GitHub-Last-Rev: dd95e50c4b
GitHub-Pull-Request: golang/go#52393
Reviewed-on: https://go-review.googlesource.com/c/go/+/400694
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
This CL exports the existing ir.UintptrKeepAlive via the new directive
//go:uintptrkeepalive. This makes the compiler insert KeepAlives for
pointers converted to uintptr in calls, keeping them alive for the
duration of the call.
//go:uintptrkeepalive requires //go:nosplit, as stack growth can't
handle these arguments (it cannot know which are pointers). We currently
check this on the immediate function, but the actual restriction applies
to all transitive calls.
The existing //go:uintptrescapes is an extension of
//go:uintptrkeepalive which forces pointers to escape to the heap, thus
eliminating the stack growth issue.
This pragma is limited to the standard library.
For #51087
Change-Id: If9a19d484d3561b4219e5539b70c11a3cc09391e
Reviewed-on: https://go-review.googlesource.com/c/go/+/388095
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
CL 388095 will change this file significantly. Move it preemptively to
ensure git tracks the move properly.
For #51087
Change-Id: I1408aecf8675723041b64e54cf44cdec38cc655c
Reviewed-on: https://go-review.googlesource.com/c/go/+/388094
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Fixes#52438.
Change-Id: I5cbf8c448dba037e9e0c5fe8f209401d6bf7d43f
Reviewed-on: https://go-review.googlesource.com/c/go/+/401134
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
The linker performs a global analysis of all nosplit call chains to
check they fit in the stack space ensured by splittable functions.
That analysis has two problems right now:
1. It's inefficient. It performs a top-down analysis, starting with
every nosplit function and the nosplit stack limit and walking *down*
the call graph to compute how much stack remains at every call. As a
result, it visits the same functions over and over, often with
different remaining stack depths. This approach is historical: this
check was originally written in C and this approach avoided the need
for any interesting data structures.
2. If some call chain is over the limit, it only reports a single call
chain. As a result, if the check does fail, you often wind up playing
whack-a-mole by guessing where the problem is in the one chain, trying
to reduce the stack size, and then seeing if the link works or reports
a different path.
This CL completely rewrites the nosplit stack check. It now uses a
bottom-up analysis, computing the maximum stack height required by
every function's call tree. This visits every function exactly once,
making it much more efficient. It uses slightly more heap space for
intermediate storage, but still very little in the scheme of the
overall link. For example, when linking cmd/go, the new algorithm
virtually eliminates the time spent in this pass, and reduces overall
link time:
│ before │ after │
│ sec/op │ sec/op vs base │
Dostkcheck 7.926m ± 4% 1.831m ± 6% -76.90% (p=0.000 n=20)
TotalTime 301.3m ± 1% 296.4m ± 3% -1.62% (p=0.040 n=20)
│ before │ after │
│ B/op │ B/op vs base │
Dostkcheck 40.00Ki ± 0% 212.15Ki ± 0% +430.37% (p=0.000 n=20)
Most of this time is spent analyzing the runtime, so for larger
binaries, the total time saved is roughly the same, and proportionally
less of the overall link.
If the new implementation finds an error, it redoes the analysis,
switching to preferring quality of error reporting over performance.
For error reporting, it computes stack depths top-down (like the old
algorithm), and reports *all* paths that are over the stack limit,
presented as a tree for compactness. For example, this is the output
from a simple test case from test/nosplit with two over-limit paths
from f1:
main.f1: nosplit stack overflow
main.f1
grows 768 bytes, calls main.f2
grows 56 bytes, calls main.f4
grows 48 bytes
80 bytes over limit
grows 768 bytes, calls main.f3
grows 104 bytes
80 bytes over limit
While we're here, we do a few nice cleanups:
- We add a debug output flag, which will be useful for understanding
what our nosplit chains look like and which ones are close to
running over.
- We move the implementation out of the fog of lib.go to its own file.
- The implementation is generally more Go-like and less C-like.
Change-Id: If1ab31197f5215475559b93695c44a01bd16e276
Reviewed-on: https://go-review.googlesource.com/c/go/+/398176
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
The nosplit test was originally written when the stack limit was a
mere 128 bytes. Now it's much larger, but rather than rewriting all of
the tests, we apply a hack to just add the extra space into the stack
frames of the existing tests.
Unfortunately, we add it in the wrong place. The extra space should be
added just once per chain of nosplit functions, but instead we add it
to every frame that appears first on a line in the test's little
script language. This means that for tests like
start 0 call f1
f1 16 nosplit call f2
f2 16 nosplit call f3
f3 16 nosplit call f4
f4 16 nosplit call f5
f5 16 nosplit call f6
f6 16 nosplit call f7
f7 16 nosplit call f8
f8 16 nosplit call end
end 1000
REJECT
we add 672 bytes to *every* frame, meaning that we wind up way over
the stack limit by the end of the stanza, rather than just a little as
originally intended.
Fix this by instead adding the extra space to the first nosplit
function in a stanza. This isn't perfect either, since we could have a
nosplit -> split -> nosplit chain, but it's the best we can do without
a graph analysis.
Change-Id: Ibf156c68fe3eb1b64a438115f4a17f1a6c7e2bd1
Reviewed-on: https://go-review.googlesource.com/c/go/+/398174
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
So that the inliner knows all the other cases are dead and doesn't
accumulate any cost for them.
The canonical case for this is switching on runtime.GOOS, which occurs
several places in the stdlib.
Fixes#50253
Change-Id: I44823aaebb6c1b03c9b0c12d10086db81954350f
Reviewed-on: https://go-review.googlesource.com/c/go/+/399694
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
runtime.getitab need filled fun[0] to identify whether
implemented the interface.
Fixes#51700Fixes#52228
Change-Id: I0173b98f4e1b45e3a0183a5b60229d289140d1e6
Reviewed-on: https://go-review.googlesource.com/c/go/+/399058
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
gofmt is rewriting +build comments into //go:build anyway, so update
the test script to support both.
Change-Id: Ia6d950cfaa2fca9f184b8b2d3625a551bff88dde
Reviewed-on: https://go-review.googlesource.com/c/go/+/399794
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
After CL 398014 fixed a compiler deadlock on syntax errors,
this CL adds a test case and more details for that.
How it was fixed:
CL 57751 introduced a channel "sem" to limit the number of
simultaneously open files.
Unfortunately, when the number of syntax processing goroutines
exceeds this limit, will easily trigger deadlock.
In the original implementation, "sem" only limited the number
of open files, not the number of concurrent goroutines, which
will cause extra goroutines to block on "sem". When the p.err
of the following iteration happens to be held by the blocking
goroutine, it will fall into a circular wait, which is a deadlock.
CL 398014 fixed the above deadlock, also see issue #52127.
First, move "sem <- struct{}{}" to the outside of the syntax
processing goroutine, so that the number of concurrent goroutines
does not exceed the number of open files, to ensure that all
goroutines in execution can eventually write to p.err.
Second, move the entire syntax processing logic into a separate
goroutine to avoid blocking on the producer side.
Change-Id: I1bb89bfee3d2703784f0c0d4ded82baab2ae867a
Reviewed-on: https://go-review.googlesource.com/c/go/+/399054
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Fixes#52278
Change-Id: Ibf67c7b019feec277d316e04d93b458efea133fb
Reviewed-on: https://go-review.googlesource.com/c/go/+/399574
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
In the load tests, we only want to test the assembly produced by
the load operations. If we use the global variable sink, it will produce
one load operation and one store operation(assign to sink).
For example:
func load_be64(b []byte) uint64 {
sink64 = binary.BigEndian.Uint64(b)
}
If we compile this function with GOAMD64=v3, it may produce MOVBEQload
and MOVQstore or MOVQload and MOVBEQstore, but we only want MOVBEQload.
Discovered when developing CL 395474.
Same for the store tests.
Change-Id: I65c3c742f1eff657c3a0d2dd103f51140ae8079e
Reviewed-on: https://go-review.googlesource.com/c/go/+/397875
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Cherry Mui <cherryyz@google.com>
With this change, the shift checking code matches the corresponding
go/types code, but for the differences in the internal error reporting,
and call of check.overflow.
The change leads to the recording of an untyped int value if the RHS
of a non-constant shift is an untyped integer value. Adjust the type
in the compiler's irgen accordingly. Add test/shift3.go to verify
behavior.
Change-Id: I20386fcb1d5c48becffdc2203081fb70c08b282d
Reviewed-on: https://go-review.googlesource.com/c/go/+/398236
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
The SHRX/SHLX instruction can take any general register as the shift count operand, and can read source from memory. This CL introduces some operators to combine load and shift to one instruction.
For #47120
Change-Id: I13b48f53c7d30067a72eb2c8382242045dead36a
Reviewed-on: https://go-review.googlesource.com/c/go/+/385174
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Cherry Mui <cherryyz@google.com>
LZCNT is similar to BSR, but BSR(x) is undefined when x == 0, so using
LZCNT can avoid a special case for zero input. Except that case,
LZCNTQ(x) == 63-BSRQ(x) and LZCNTL(x) == 31-BSRL(x).
And according to https://www.agner.org/optimize/instruction_tables.pdf,
LZCNT instructions are much faster than BSR on AMD CPU.
name old time/op new time/op delta
LeadingZeros-8 0.91ns ± 1% 0.80ns ± 7% -11.68% (p=0.000 n=9+9)
LeadingZeros8-8 0.98ns ±15% 0.91ns ± 1% -7.34% (p=0.000 n=9+9)
LeadingZeros16-8 0.94ns ± 3% 0.92ns ± 2% -2.36% (p=0.001 n=10+10)
LeadingZeros32-8 0.89ns ± 1% 0.78ns ± 2% -12.49% (p=0.000 n=10+10)
LeadingZeros64-8 0.92ns ± 1% 0.78ns ± 1% -14.48% (p=0.000 n=10+10)
Change-Id: I125147fe3d6994a4cfe558432780408e9a27557a
Reviewed-on: https://go-review.googlesource.com/c/go/+/396794
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Emmanuel Odeke <emmanuel@orijtech.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
This CL add MOVBE support for 16-bit version, but MOVBEWload is
excluded because it does not satisfy zero extented.
For #51724
Change-Id: I3fadf20bcbb9b423f6355e6a1e340107e8e621ac
Reviewed-on: https://go-review.googlesource.com/c/go/+/396617
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Trust: Emmanuel Odeke <emmanuel@orijtech.com>
The -G compiler option doesn't exist anymore. Update some variable
names and comments to reflect the new reality.
Change-Id: I227e9c59a01615c3a40c3869102e8045cb012980
Reviewed-on: https://go-review.googlesource.com/c/go/+/397254
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
For syntax errors in various (syntactic) lists, instead of reporting
a set of "expected" tokens (which may be incomplete), provide context
and mention "possibly missing" tokens. The result is a friendlier and
more accurate error message.
Fixes#49205.
Change-Id: I38ae7bf62febfe790075e62deb33ec8c17d64476
Reviewed-on: https://go-review.googlesource.com/c/go/+/396914
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
When parsing method declarations in an interface, the parser has
for historic reasons gracefully handled a list of method names with
a single (common) signature, and then reported an error. For example
interface {
m1, m2, m3 (x int)
}
This code originally came from the very first parser for Go which
initially permitted such declarations (or at least assumed that
people would write such declarations). Nobody is doing this at this
point, so there's no need for being extra careful here. Remove the
respective code and adjust the corresponding test.
Change-Id: If6f9b398bbc9e425dcd4328a80d8bf77c37fe8b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/396654
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
CL 367755 added soleComponent for handling 1-byte type interface conversion.
This implementation must be kept in sync with Type.SoleComponent, but it
does not. When seeing a blank field in struct, we must continue looking
at the field type to find sole component, if any. The current code just
terminate immediately, which causes wrong sole component type returned.
Fixes#52020
Change-Id: I4f506fe094fa7c5532de23467a4f9139476bb0a4
Reviewed-on: https://go-review.googlesource.com/c/go/+/396614
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Use the 1.17 compiler error message, sans "array" prefix.
Change-Id: I0e70781c5ff02dca30a2004ab4d0ea82b0849eae
Reviewed-on: https://go-review.googlesource.com/c/go/+/396296
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
CL 394074 broke the noopt builder. Something about time.After's inlining
depends on the build flags to make.bash, not the build flags that run.go
passes.
Change-Id: Ib284c66ea2008a4d32829c055d57c54a34ec3fb4
Reviewed-on: https://go-review.googlesource.com/c/go/+/396037
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Historically, we sometimes recorded imports based on either package
path ("net/http") or object file path ("net/http.a"). But modern Go
build systems always use package path, and the extra ".a" suffix
doesn't mean anything anyway.
Change-Id: I6060ef8bafa324168710d152a353f4d8db062133
Reviewed-on: https://go-review.googlesource.com/c/go/+/395254
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Add a new rewrite rule to merge ANDconst and UBFX into
UBFX.
Add test cases.
Change-Id: I24d6442d0c956d7ce092c3a3858d4a3a41771670
Reviewed-on: https://go-review.googlesource.com/c/go/+/377054
Trust: Fannie Zhang <Fannie.Zhang@arm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
This CL updates test/run.go to compile xxx.dir/x.go with a package
path of "test/x" instead of just "x". This prevents collisions with
standard library packages.
It also requires updating a handful of tests to account for the
updated package paths.
Fixes#25693.
Change-Id: I49208c56ab3cb229ed667d547cd6e004d2175fcf
Reviewed-on: https://go-review.googlesource.com/c/go/+/395258
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
CL 187617 removed oldescape_linkname.go, but forgot to remove this
directory too.
Change-Id: I6d208c4d96d636b3df93adec1ee22fe1d4f5f61d
Reviewed-on: https://go-review.googlesource.com/c/go/+/395259
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
bug302 compiles p.go with -p=p, and then manually creates a pp.a
archive, and imports it as both "p" and "pp". This is a misuse of
cmd/compile's -p flag, and it isn't representative of how any actual
Go build systems work anyway.
This test made sense back when cmd/compile still wrote out bare object
files, which was then split into separate __.PKGDEF and _go_.o archive
entries when added to a pack archive. But since CL 102236, cmd/compile
always writes out pack files.
Updates #51734.
Change-Id: I4b5de22d348ecc0a72c98b512351c2d267c77736
Reviewed-on: https://go-review.googlesource.com/c/go/+/393896
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Currently, run.go's *dir tests allow "x.go" to be imported
interchangeably as either "x" or "./x". This is generally fine, but
can cause problems when "x" is the name of a standard library
package (e.g., "fixedbugs/bug345.dir/io.go").
This CL is an automated rewrite to change all `import "x"` directives
to use `import "./x"` instead. It has no effect today, but will allow
subsequent CLs to update test/run.go to resolve "./x" to "test/x" to
avoid stdlib collisions.
Change-Id: Ic76cd7140e83b47e764f8a499e59936be2b3c876
Reviewed-on: https://go-review.googlesource.com/c/go/+/395116
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The importer type param index used package name type parameter key,
causing type parameters to be reused/overwritten if two packages in the
import graph had the same combination of (name, declaration name, type
parameter name).
Fix this by instead using the *Package in the key.
Fixes#51836
Change-Id: I881ceaf3cf7c1ab4e0835962350feb552e79b233
Reviewed-on: https://go-review.googlesource.com/c/go/+/394219
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
First law of cmd/compile frontend development: thou shalt not rely on
types.Sym.
This CL replaces Type.OrigSym with Type.OrigType, which semantically
matches what all of the uses within the frontend actually care about,
and avoids using types.Sym, which invariably leads to mistakes because
symbol scoping in the frontend doesn't work how anyone intuitively
expects it to.
Fixes#51765.
Change-Id: I4affe6ee0718103ce5006ab68aa7e1bb0cac6881
Reviewed-on: https://go-review.googlesource.com/c/go/+/394274
Trust: Matthew Dempsky <mdempsky@google.com>
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: Keith Randall <khr@golang.org>
CL 342350 fixed deadcode panic with dead hidden closures. However, a
closure may contains nested dead hidden closures, so we need to mark
them dead as well.
Fixes#51839
Change-Id: Ib54581adfc1bdea60e74d733cd30fd8e783da983
Reviewed-on: https://go-review.googlesource.com/c/go/+/394079
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
They should not share a shape with regular pointers. We could coalesce
multiple pointer-to-not-in-heap types, but doesn't seem worth it - just
make them fully stenciled.
Fixes#51733
Change-Id: Ie8158177226fbc46a798e71c51897a82f15153df
Reviewed-on: https://go-review.googlesource.com/c/go/+/393895
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
CL 338129 added getDictionaryType to get the dictionary type from the
specified dict param, but still using the one in info.dictParam, which
is wrong.
Fixes#51413
Change-Id: Ie13460c1e5751c4c5fc44479a44f6eed8b3b06e4
Reviewed-on: https://go-review.googlesource.com/c/go/+/391994
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Trust: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
For hidden closure built during stenciling to implement a function
instantiation, the function may come from other package, not local
package, which causes the ICE for code that re-export the hidden closure
after inlining.
To fix it, use the closure package for export writer when writing out
the closure itself.
Fixes#51423
Change-Id: I23b067ba14e2d602a0fc3b2e99bd9317afbe53ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/391574
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Both the thing we're switching on, as well as the cases we're switching for.
Convert anything containing a type parameter to interface{} before the
comparison happens.
Fixes#51522
Change-Id: I97ba9429ed332cb7d4240cb60f46d42226dcfa5f
Reviewed-on: https://go-review.googlesource.com/c/go/+/391594
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
At this point in stenciling, we have shape types, not raw type parameters.
The code was correct in the other part of this function.
Update #51522
Change-Id: Ife495160a2be5f6af5400363c3efb68dda518b5f
Reviewed-on: https://go-review.googlesource.com/c/go/+/391475
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The -p flag specifies the import path of the package being compiled.
This CL makes it required when invoking the compiler and
adjusts tests that invoke the compiler directly to conform to this
new requirement. The go command already passes the flag, so it
is unmodified in this CL. It is expected that any other Go build systems
also already pass -p, or else they will need to arrange to do so before
updating to Go 1.19. Of particular note, Bazel already does for rules
with an importpath= attribute, which includes all Gazelle-generated rules.
There is more cleanup possible now in cmd/compile, cmd/link,
and other consumers of Go object files, but that is left to future CLs.
Additional historical background follows but can be ignored.
Long ago, before the go command, or modules, or any kind of
versioning, symbols in Go archive files were named using just the
package name, so that for example func F in math/rand and func F in
crypto/rand would both be the object file symbol 'rand.F'. This led to
collisions even in small source trees, which made certain packages
unusable in the presence of other packages and generally was a problem
for Go's goal of scaling to very large source trees.
Fixing this problem required changing from package names to import
paths in symbol names, which was mostly straightforward. One wrinkle,
though, is that the compiler did not know the import path of the
package being compiled; it only knew the package name. At the time,
there was no go command, just Makefiles that people had invoking 6g
(now “go tool compile”) and then copying the resulting object file to
an importable location. That is, everyone had a custom build setup for
Go, because there was no standard one. So it was not particularly
attractive to change how the compiler was invoked, since that would
break approximately every Go user at the time. Instead, we arranged
for the compiler to emit, and other tools reading object files to
recognize, a special import path (the empty string, it turned out)
denoting “the import path of this object file”. This worked well
enough at the time and maintained complete command-line compatibility
with existing Go usage.
The changes implementing this transition can be found by searching
the Git history for “package global name space”, which is what they
eliminated. In particular, CL 190076 (a6736fa4), CL 186263 (758f2bc5),
CL 193080 (1cecac81), CL 194053 (19126320), and CL 194071 (531e6b77)
did the bulk of this transformation in January 2010.
Later, in September 2011, we added the -p flag to the compiler for
diagnostic purposes. The problem was that it was easy to create import
cycles, especially in tests, and these could not be diagnosed until
link time. You'd really want the compiler to diagnose these, for
example if the compilation of package sort noticed it was importing a
package that itself imported "sort". But the compilation of package
sort didn't know its own import path, and so it could not tell whether
it had found itself as a transitive dependency. Adding the -p flag
solved this problem, and its use was optional, since the linker would
still diagnose the import cycle in builds that had not updated to
start passing -p. This was CL 4972057 (1e480cd1).
There was still no go command at this point, but when we introduced
the go command we made it pass -p, which it has for many years at this
point.
Over time, parts of the compiler began to depend on the presence of
the -p flag for various reasonable purposes. For example:
In CL 6497074 (041fc8bf; Oct 2012), the race detector used -p to
detect packages that should not have race annotations, such as
runtime/race and sync/atomic.
In CL 13367052 (7276c02b; Sep 2013), a bug fix used -p to detect the
compilation of package reflect.
In CL 30539 (8aadcc55; Oct 2016), the compiler started using -p to
identify package math, to be able to intrinsify calls to Sqrt inside
that package.
In CL 61019 (9daee931; Sep 2017), CL 71430 (2c1d2e06; Oct 2017), and
later related CLs, the compiler started using the -p value when
creating various DWARF debugging information.
In CL 174657 (cc5eaf93; May 2019), the compiler started writing
symbols without the magic empty string whenever -p was used, to reduce
the amount of work required in the linker.
In CL 179861 (dde7c770; Jun 2019), the compiler made the second
argument to //go:linkname optional when -p is used, because in that
case the compiler can derive an appropriate default.
There are more examples. Today it is impossible to compile the Go
standard library without using -p, and DWARF debug information is
incomplete without using -p.
All known Go build systems pass -p. In particular, the go command
does, which is what nearly all Go developers invoke to build Go code.
And Bazel does, for go_library rules that set the importpath
attribute, which is all rules generated by Gazelle.
Gccgo has an equivalent of -p and has required its use in order to
disambiguate packages with the same name but different import paths
since 2010.
On top of all this, various parts of code generation for generics
are made more complicated by needing to cope with the case where -p
is not specified, even though it's essentially always specified.
In summary, the current state is:
- Use of the -p flag with cmd/compile is required for building
the standard library, and for complete DWARF information,
and to enable certain linker speedups.
- The go command and Bazel, which we expect account for just
about 100% of Go builds, both invoke cmd/compile with -p.
- The code in cmd/compile to support builds without -p is
complex and has become more complex with generics, but it is
almost always dead code and therefore not worth maintaining.
- Gccgo already requires its equivalent of -p in any build
where two packages have the same name.
All this supports the change in this CL, which makes -p required
and adjusts tests that invoke cmd/compile to add -p appropriately.
Future CLs will be able to remove all the code dealing with the
possibility of -p not having been specified.
Change-Id: I6b95b9d4cffe59c7bac82eb273ef6c4a67bb0e43
Reviewed-on: https://go-review.googlesource.com/c/go/+/391014
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
This is a feature that is not understood well enough and may have
subtle repercussions impacting future changes. Disable for Go 1.18.
The actual change is trivial: disable a branch through a flag.
The remaining changes are adjustments to tests.
Fixes#51576.
Change-Id: Ib77b038b846711a808315a8889b3904e72367bce
Reviewed-on: https://go-review.googlesource.com/c/go/+/391135
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
If you attempt to instantiate a generic type or func and run 'go build'
with a language version < 1.18 in the 'go' directive inside the go.mod
file, cmd/compile emits a friendly message that includes the suggestion
to 'check go.mod':
type instantiation requires go1.18 or later (-lang was set to go1.17; check go.mod)
However, if the code instead only declares a generic type or func
without instantiating, cmd/compile currently emits a less friendly
message:
type parameters require go1.18 or later
With this CL, the error in that situation becomes:
type parameter requires go1.18 or later (-lang was set to go1.17; check go.mod)
Within cmd/compile/internal/types2, it already calls check.versionErrorf
in a dozen or so places, including three existing calls to
check.versionErrorf within typeset.go (e.g., for embedding a constraint
interface).
This CL adds two more calls to check.versionErrorf, replacing calls to
check.softErrorf. Both check.versionErrorf and check.softErrorf call
check.err(at, <string>, true) after massaging the string message.
Fixes#51531
Change-Id: If54e179f5952b97701d1dfde4abb08101de07811
GitHub-Last-Rev: b0b7c1346f
GitHub-Pull-Request: golang/go#51536
Reviewed-on: https://go-review.googlesource.com/c/go/+/390578
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Robert Findley <rfindley@google.com>
Inference for type instances has dependencies on type-checking order
that can lead to subtle bugs. As explained in #51527, disable it for
1.18.
Fixes#51527
Change-Id: I42795bad30ce53abecfc5a4914599ae5a2041a9e
Reviewed-on: https://go-review.googlesource.com/c/go/+/387934
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Now that we always use types2 to validate user source code, we can
remove the constSet logic from typecheck for detecting duplicate
expression switch cases and duplicate map literal keys. This logic is
redundant with types2, and currently causes unified IR to report
inappropriate duplicate constant errors that only appear after type
substitution.
Updates #42758.
Change-Id: I51ee2c5106eec9abf40eba2480dc52603c68ba21
Reviewed-on: https://go-review.googlesource.com/c/go/+/390474
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>