Use NEGW to produce a negated and sign extended word, rather than doing
the same via two instructions:
neg t0, t0
sext.w a0, t0
Becomes:
negw t0, t0
Change-Id: I824ab25001bd3304bdbd435e7b244fcc036ef212
Reviewed-on: https://go-review.googlesource.com/c/go/+/652319
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
On riscv64, subtraction from a constant is typically implemented as an
ADDI with the negative constant, followed by a negation. However this can
lead to multiple NEG/ADDI/NEG sequences that can be optimised out.
For example, runtime.(*_panic).nextDefer currently contains:
lbu t0, 0(t0)
addi t0, t0, -8
neg t0, t0
addi t0, t0, -7
neg t0, t0
Which is now optimised to:
lbu t0, 0(t0)
addi t0, t0, -1
Change-Id: Idf5815e6db2e3705cc4a4811ca9130a064ae3d80
Reviewed-on: https://go-review.googlesource.com/c/go/+/652318
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: David Chase <drchase@google.com>
Nil checks need to stay in their original blocks. They cannot
be moved to a following conditionally-executed block.
Fixes#72860
Change-Id: Ic2d66cdf030357d91f8a716a004152ba4c016f77
Reviewed-on: https://go-review.googlesource.com/c/go/+/657715
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
In CL 475375 the Go command started to generate the "preferlinkext"
token file for "strange/dangerous" compiler flags. This serves as a hint
to the Go linker whether to call the external linker or not.
Permit compiler flag used by the hermetic_cc_toolchain bzlmod.
As a side effect, it also allows these flags to appear
in #cgo directives in source code. We don't know of any cases
where that is actually useful, but it appears to be harmless
and simplifies the implementation of the internal linking change.
Fixes#72842
Change-Id: Ic6de29b535a4e2c0720f383567ea6b3c7ca4f541
Reviewed-on: https://go-review.googlesource.com/c/go/+/657575
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
As suggested by Michael in CL 655515.
Change-Id: Idf0b879287bd777d03443aebc7351fcb0d724885
GitHub-Last-Rev: 58eda020f5
GitHub-Pull-Request: golang/go#72806
Reviewed-on: https://go-review.googlesource.com/c/go/+/656856
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This patch extends the change in CL 657175 to apply the same abbrev
selection strategy to single-range lexical scopes that we're now using
for inlined routine bodies, when DWARF 5 is in effect. Ranges are more
compact and use fewer relocation than explicit hi/lo PC values, so we
might as well always use them.
Updates #26379.
Change-Id: Ieeaddf50e82acc4866010e29af32bcd1fb3b4f02
Reviewed-on: https://go-review.googlesource.com/c/go/+/657177
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Currently we check the size difference between non-PIE and PIE binaries
without specifying a linkmode (and that is presumed to be internal).
However, on some platforms (like openbsd/arm64), the use of
-buildmode=pie results in external linking. Ensure that we only test
internally linked non-PIE against internally linked PIE and externally
linked non-PIE against externally linked PIE, avoiding unexpected
differences.
Fixes#72818
Change-Id: I7e1da0976a4b5de387a59d0d6c04f58498a8eca0
Reviewed-on: https://go-review.googlesource.com/c/go/+/657035
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@golang.org>
The Go command had a behavior of writing its own toolchain name when
updating the go line in a go.mod (for example when a user runs go get
go@version). This behavior was often undesirable and the toolchain line
was often removed by users before checking in go.mod files (including in
the x/ repos). It also led to user confusion.
This change removes that behavior. A toolchain line will not be added if
one wasn't present before. The toolchain line can still be removed
though: the toolchain line must be at least the go version, so if the go
version is increased above the toolchain version, the toolchain version
will be bumped up to that go version. Then the toolchain line will then
be dropped because go <version> implies toolchain <version>.
Making this change slightly hurts reproducability because future go
commands run on the go.mod file may be run with a different toolchain
than the one that used it, but that doesn't seem to be worth the
confusion the behavior resulted in.
We expect this change will not have negative consequences, but it could
be possible, and we would like to hear from any users that depended on
the previous behavior in case we need to roll it back before the
release.
Fixes#65847
Change-Id: Id795b7f762e4f90ba0fa8c7935d03f32dfc8590e
Reviewed-on: https://go-review.googlesource.com/c/go/+/656835
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This patch changes the strategy we use in the compiler for handling
range information for inlined subroutine bodies, fixing a bug in how
this was handled for DWARF 5. The high and lo PC values being emitted
for DW_TAG_inlined_subroutine DIEs were incorrect, pointing to the
start of functions instead of the proper location. The fix in this
patch is to move to unconditionally using DW_AT_ranges for inlined
subroutines, even those with only a single range.
Background: prior to this point, if a given inlined function body had
a single contiguous range, we'd pick an abbrev entry for it with
explicit DW_AT_low_pc and DW_AT_high_pc attributes. If the extent of
the code for the inlined body was not contiguous (which can happen),
we'd select an abbrev that used a DW_AT_ranges attribute instead. This
strategy (preferring explicit hi/lo PC attrs for a single-range func)
made sense for DWARF 4, since in DWARF 4 the representation used in
the .debug_ranges section was especially heavyweight (lots of space,
lots of relocations), so having explicit hi/lo PC attrs was less
expensive.
With DWARF 5 range info is written to the .debug_rnglists section, and
the representation here is much more compact. Specifically, a single
hi/lo range can be represented using a base address in addrx format
(max of 4 bytes, but more likely 2 or 3) followed by start and
endpoints of the range in ULEB128 format. This combination is more
compact spacewise than the explicit hi/lo values, and has fewer
relocations (0 as opposed to 2).
Note: we should at some point consider applying this same strategy to
lexical scopes, since we can probably reap some of the same benefits
there as well.
Updates #26379.
Fixes#72821.
Change-Id: Ifb65ecc6221601bad2ca3939f9b69964c1fafc7c
Reviewed-on: https://go-review.googlesource.com/c/go/+/657175
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Rather than using a specific intrinsic for math/bits.Len, use a pair of
aliases instead. This requires less code and automatically adapts when
platforms have a math/bits.Len32 or math/bits.Len64 intrinsic.
Change-Id: I28b300172daaee26ef82a7530d9e96123663f541
Reviewed-on: https://go-review.googlesource.com/c/go/+/656995
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Jorropo <jorropo.pgm@gmail.com>
Reviewed-by: Jorropo <jorropo.pgm@gmail.com>
noDuffDevice was for Plan 9, but Plan 9 doesn't need it anymore.
It was also being set in s390x, mips, mipsle, and wasm, but
on those systems it had no effect since the SSA rules for those
architectures don't refer to it at all.
Change-Id: Ib85c0832674c714f3ad5091f0a022eb7cd3ebcdf
Reviewed-on: https://go-review.googlesource.com/c/go/+/655878
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Russ Cox <rsc@golang.org>
Every OS uses FMA now.
Change-Id: Ia7ffa77c52c45aefca611ddc54e9dfffb27a48da
Reviewed-on: https://go-review.googlesource.com/c/go/+/655877
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Every OS uses SSE now.
Change-Id: I4df7e2fbc8e5ccb1fc84a884d4c922b7a2a628e4
Reviewed-on: https://go-review.googlesource.com/c/go/+/655876
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Fixes#72800
Change-Id: Idde7eae13d1c0098e5314935cf8ca823cbc7a7cc
Reviewed-on: https://go-review.googlesource.com/c/go/+/656855
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
test/decoratemappingszero.go is intended to test that
//go:debug decoratemappings=0 disables annonations.
Unfortunately, //go:debug processing is handled by cmd/go, but
cmd/internal/testdir (which runs tests from test/) generally invokes the
compiler directly, thus it does not set default GODEBUGs.
Move this test to the cmd/go script tests, alongside the similar test
for language version.
Fixes#72772.
Cq-Include-Trybots: luci.golang.try:gotip-linux-ppc64le_power10
Change-Id: I6a6a636c9d380ef984f760be5689fdc7f5cb2aeb
Reviewed-on: https://go-review.googlesource.com/c/go/+/656795
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Pratt <mpratt@google.com>
Change-Id: Id5258a72b0727bf7c66d558e30486eac2c6c8c36
Reviewed-on: https://go-review.googlesource.com/c/go/+/655875
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David du Colombier <0intro@gmail.com>
We've been slowly moving packages from runtime/internal to
internal/runtime. For now, runtime/internal only has test packages.
It's a good chance to clean up the references to runtime/internal
in the toolchain.
For #65355.
Change-Id: Ie6f9091a44511d0db9946ea6de7a78d3afe9f063
GitHub-Last-Rev: fad32e2e81
GitHub-Pull-Request: golang/go#72137
Reviewed-on: https://go-review.googlesource.com/c/go/+/655515
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
We previously disallowed all non-regular files being embedded. This CL
relaxes the restriction a little: if the GODEBUG embedfollowsymlinks=1
is set, we allow the leaf files being embedded (not the directories
containing them) to be symlinks. The files pointed to by the symlinks
must still be regular files.
This will be used when a Bazel build action executing the Go command is
running in a symlink-based sandbox. It's not something we want to enable
in general for now, so it's behind a GODEBUG.
Fixes#59924
Change-Id: I895be14c12de55b7d1b663d81bdda1df37d54804
Reviewed-on: https://go-review.googlesource.com/c/go/+/643215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Change-Id: I4593aeb4cad1e2c3f4705ed5249ac0bad910162f
Reviewed-on: https://go-review.googlesource.com/c/go/+/655518
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Explicitly compute the common underlying type and while doing
so report better slice-expression relevant error messages.
Streamline message format for index and slice errors.
This removes the last uses of the coreString and match functions.
Delete them.
Change-Id: I4b50dda1ef7e2ab5e296021458f7f0b6f6e229cd
Reviewed-on: https://go-review.googlesource.com/c/go/+/655935
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Also, add additional test cases for NewSignatureType
to check expected panic behavior.
Change-Id: If26cd81a2af384bf2084dd09119483c0584715c8
Reviewed-on: https://go-review.googlesource.com/c/go/+/655695
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Rather than relying on coreString, use the new commonUnder function
to determine the argument slice element types.
Factor out this functionality, which is shared for append and copy,
into a new helper function sliceElem (similar to chanElem).
Use sliceElem for both the append and copy implementation.
As a result, the error messages for invalid copy calls are
now more detailed.
While at it, handle the special cases for append and copy first
because they don't need the slice element computation.
Finally, share the same type recording code for the special and
general cases.
As an aside, in commonUnder, be clearer in the code that the
result is either a nil type and an error, or a non-nil type
and a nil error. This matches in style what we do in sliceElem.
Change-Id: I318bafc0d2d31df04f33b1b464ad50d581918671
Reviewed-on: https://go-review.googlesource.com/c/go/+/655675
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Closing the file after mmap will reduce the number of files associated
with the process. This will not likely help with #71698 but it doesn't
hurt to close the files and should simplify lsof output.
For #71698
Change-Id: I06a1bf91914afc7703783fe1a38d8bc5a6fb3d9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/653055
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Require GCC 11 or greater to turn on the location checking portion of
the asan tests in this directory; the copy of libasan.so.6 shipped
with GCC 10 doesn't seem to properly digest the new DWARF 5 being
generated by the Go compiler+linker.
Updates #72752.
Change-Id: I92718c112df844d9333c4c798cddaae95665feb2
Reviewed-on: https://go-review.googlesource.com/c/go/+/656175
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This adds a new godebug to control whether the runtime applies the
anonymous memory mapping annotations added in https://go.dev/cl/646095.
It is enabled by default.
This has several effects:
* The feature is only enabled by default when the main go.mod has go >=
1.25.
* This feature can be disabled with GODEBUG=decoratemappings=0, or the
equivalents in go.mod or package main. See https://go.dev/doc/godebug.
* As an opaque setting, this option will not appear in runtime/metrics.
* This setting is non-atomic, so it cannot be changed after startup.
I am not 100% sure about my decision for the last two points.
I've made this an opaque setting because it affects every memory mapping
the runtime performs. Thus every mapping would report "non-default
behavior", which doesn't seem useful.
This setting could trivially be atomic and allow changes at run time,
but those changes would only affect future mappings. That seems
confusing and not helpful. On the other hand, going back to annotate or
unannotate every previous mapping when the setting changes is
unwarranted complexity.
For #71546.
Change-Id: I6a6a636c5ad551d76691cba2a6f668d5cff0e352
Reviewed-on: https://go-review.googlesource.com/c/go/+/655895
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Pratt <mpratt@google.com>
As described in the associated comment, we need to reallocate usedVars
and usedPkgNames in initFiles, as they are nilled out at the end of
Checker.Files, which may be called multiple times.
Fixes#72122
Change-Id: I9f6eb86e072d9d43a8720f6a5e86d827de6006a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/655437
Auto-Submit: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
For go/types, generate its range.go file from the corresponding types2 file.
Change-Id: Iaff3ecbf1c536143c92f7b50e2461140469f9280
Reviewed-on: https://go-review.googlesource.com/c/go/+/655536
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Change-Id: Ib8a63cdaa12dacb5223318a7166fe3dfdac71a45
Reviewed-on: https://go-review.googlesource.com/c/go/+/654655
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
If the error cause is not further specified (empty string),
avoid allocating a new errorCause. This makes using errorCauses
as boolean signals efficient.
While at it, fix an error message for incomparable arrays:
report the array type rather than its underlying type.
Change-Id: I844b18a76695330ca726932ee760aa89635f6a38
Reviewed-on: https://go-review.googlesource.com/c/go/+/654575
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Rather than reporting "non-function" for an invalid type parameter,
report which type in the type parameter's type set is not a function.
Change-Id: I8beec25cc337bae8e03d23e62d97aa82db46bab4
Reviewed-on: https://go-review.googlesource.com/c/go/+/654475
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Combine commonUnder and commonUnderOrChan:
- Provide an optional cond(ition) function argument to commonUnder
to establish additional type set conditions.
- Instead of a *Checker and *string argument for error reporting,
return an error cause that is only allocated in the presence of
an error.
- Streamline some error messages.
Replace all calls to coreType with calls to commonUnder.
Change-Id: I81ac86d0d532cddc09164309acced61d90718b44
Reviewed-on: https://go-review.googlesource.com/c/go/+/654455
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
We currently test the leak detector by running "go build -asan",
which will pass -fsanitize=address to the C compiler.
So use that when testing whether the option works.
Fixes#72128
Change-Id: I4efc0b689bfda04c80dbac30a5c757215f297d2f
Reviewed-on: https://go-review.googlesource.com/c/go/+/655535
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The 'work' package pattern will resolve to the set of packages in the
work (formerly called main) modules. It's essentially 'all', but without
the dependencies. And the implementation is similar to that of 'all',
except that we don't expand to the dependencies.
Fixes#71294
Change-Id: I3d02beb74fa4e5c6de2290e24eedc51745d13080
Reviewed-on: https://go-review.googlesource.com/c/go/+/643235
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Xiaolin Zhao
Joel Sing
Keith Randall
Holger Hans Peter Freyther
Jes Cok
Than McIntosh
Michael Matloob
Russ Cox
Jorropo
Ian Lance Taylor
Michael Pratt
Alexander Musman
Robert Griesemer
Guoqi Chen
Filippo Valsorda
Rob Findley