In test files, using testenv.Executable is more reliable than
os.Executable or os.Args[0].
Change-Id: I88e577efeabc20d02ada27bf706ae4523129128e
Reviewed-on: https://go-review.googlesource.com/c/go/+/651955
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
The extension-removing rules for ARM64 were moved to late lower in
CL 568616. This means that the late lower pass can now generate
MOVDreg, however the rules that potentially eliminate MOVDreg only
exist in the earlier pass. Fix this by duplicating the MOVDreg/NOVDnop
rules in late lower, such that we can potentially eliminate conversions.
Removes 400+ instructions from the Go binary on openbsd/arm64.
Change-Id: I14aad06b994c9179f3ecdda566629793ba167511
Reviewed-on: https://go-review.googlesource.com/c/go/+/651819
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
AMD64latelower.rules currently has Windows style line endings,
rather than Unix style line endings. Correct this.
Change-Id: Ie068dc6c64bd51cf2aa5bd192839fca4f28f40b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/651818
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
This patch rolls the main .debug_info DWARF section from version 4 to
version 5, and also introduces machinery in the Go compiler and linker
for taking advantage of the DWARF5 ".debug_addr" section for
subprogram DIE "high" and "low" PC attributes. All functionality is
gated by GOEXPERIMENT=dwarf5.
For the compiler portion of this patch, we add a new DIE attribute
form "DW_FORM_addrx", which accepts as an argument a function (text)
symbol. The dwarf "putattr" function is enhanced to handle this
format by invoking a new dwarf context method "AddIndirectTextRef".
Under the hood, this method invokes the Lsym method WriteDwTxtAddrx,
which emits a new objabi.R_DWTXTADDR_* relocation. The size of the
relocation is dependent on the number of functions in the package; we
pick a size that is just big enough for the largest func index.
In the linker portion of this patch, we now switch over to writing out
a version number of 5 (instead of 4) in the compile unit header (this
is required if we want to use addrx attributes). In the parallel portion
of DWARF gen, within each compilation unit we scan subprogram DIEs to
look for R_DWTXTADDR_* relocations, and when we find such a reloc,
we assign a slot in the .debug_addr section for the func targeted.
After the parallel portion is complete, we then walk through all of the
compilation units to assign a value to their DW_AT_addr_base attribute,
which points to the portion of the single .debug_addr section containing
the text addrs for that compilation unit.
Note that once this patch is in, programs built with GOEXPERIMENT=dwarf5
will have broken/damaged DWARF info; in particular, since we've changed
only the CU and subprogram DIEs and haven't incorported the other
changes mandated by DWARF5 (ex: .debug_ranges => .debug_rnglists)
a lot of the variable location info will be missing/incorrect. This
will obviously change in subsequent patches.
Note also that R_DWTXTADDR_* can't be used effectively for lexical
scope DIE hi/lo PC attrs, since there isn't a viable way to encode
"addrx + constant" in the attribute value (you would need a new entry
for each attr endpoint in .debug_addr, which would defeat the point).
Updates #26379.
Change-Id: I2dfc45c9a8333e7b2a58f8e3b88fc8701fefd006
Reviewed-on: https://go-review.googlesource.com/c/go/+/635337
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Add a new symbol type: SDWARFADDR. This kind of symbol stores content
to be added to the DWARF .debug_addr section (new with DWARF5). At the
moment these symbols are created only in the linker, but it's not hard to
imagine other implementations in which the compiler would create them,
so they are added to both the compiler and linker symbol kind space.
Updates #26379.
Change-Id: I4a82ead0d59fe6028abfd6d6e3fc3df2e28c0ef6
Reviewed-on: https://go-review.googlesource.com/c/go/+/634415
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Add a set of new relocations to be used when the compiler is writing
debug information using DWARF version 5. No changes in compiler or
linker functionality, this patch just adds the relocations themselves
and some helper functions; uses will appear in a later patch. These
relocations are generated by the compiler when writing a DWARF DIE
attribute of form DW_FORM_addrx, or when writing a .debug_addr index
reference in a SDWARFRANGE or SDWARFLOC section. The target symbol of
the relocation is a function (STEXT symbol); the linker resolves the
relocation by replacing the target of the reloc with an index of a
slot in the .debug_addr section (.debug_addr is new with DWARF5).
Updates #26379.
Change-Id: I43c587d25d0836972dac487d09c8924d77345f4e
Reviewed-on: https://go-review.googlesource.com/c/go/+/633880
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
This patch rolls out the necessary changes to migrate the DWARF line
table support in the compiler and linker to DWARF version 5, gated by
the "dwarf5" GOEXPERIMENT.
DWARF version 5 includes a number of changes to the line table,
notably a revamped prolog section and a change in the indexing system
used to refer to files and directories within the line table
program. Specifically, prior to DWARF 4 a compilation's directory
table was considered to have an implicit zero entry containing the
compilation directory of the translation unit (package), and the file
table was considered to have an implicit zero entry storing the
"primary source file" (stored in the compilation unit DIE name).
DWARF 5 does away with these implicity entries meaning that files and
dirs are now effectively a 0-based index.
Updates #26379.
Change-Id: I9b4f1be5415aacec1ba57366d60bd48819c56ea5
Reviewed-on: https://go-review.googlesource.com/c/go/+/633879
Reviewed-by: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Add a set of constants for the DWARF version 5 line table content
description values found in the V5 line table prolog, and for the
new DWARF unit type encodings.
Updates #26379.
Change-Id: I8f4989ea6b6cbb303deda1a6a20ad243d73b46b1
Reviewed-on: https://go-review.googlesource.com/c/go/+/633878
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Get rid of the R_DWARFFILEREF relocation type -- we have not used this
relocation for a while now, ever since jfaller's revamp of the DWARF
line table file section in Go 1.15. No change in compiler or linker
functionality; this is purely a dead code cleanup.
Change-Id: I178760c87f3aa79694cfabe7364ca382605c6975
Reviewed-on: https://go-review.googlesource.com/c/go/+/633876
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
For #58508Fixes#71863
Change-Id: Ib1ebaf751bcc6900da6ffd01a9462dd237e2c89a
Reviewed-on: https://go-review.googlesource.com/c/go/+/651295
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
We used to generate register GC maps as an experimental approach
for asynchronous preemption, which later we chose not to take.
Most of the register GC map code are already removed. One
exception is that the ssa.Register type still contains a field
for the register map index. Remove it.
Change-Id: Ib177ebce9548aa5ffbcaedd4b507240ea7df8afe
Reviewed-on: https://go-review.googlesource.com/c/go/+/651076
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Add support for endbr64, which terminates an indirect branch in 64-bit
mode. This is effectively used to mark locations where an indirect branch
is permitted to land, when Indirect Branch Tracking (IBT) is enforced on
Intel CPUs.
Updates #66054
Change-Id: Ib898031711cfaaa6e05c197bfe727ded0bce6f52
Reviewed-on: https://go-review.googlesource.com/c/go/+/649215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
In the few obvious candidates that I found after a bit of grepping.
Change-Id: I36af79c46d29e4422bce1f43bbbac9db7de2001a
Reviewed-on: https://go-review.googlesource.com/c/go/+/650656
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Alan Donovan <adonovan@google.com>
Commit-Queue: Alan Donovan <adonovan@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change-Id: I8d7bbeebbf4a46f2fd8d630b1edbaf79b8ffccc5
Reviewed-on: https://go-review.googlesource.com/c/go/+/420114
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
TryBot-Bypass: Joel Sing <joel@sing.id.au>
This CL is the result of pulling the thread of some unused parameter
diagnostics reported by gopls: remove some variables holding partial
type argument expression information that are no longer needed.
Change-Id: Idfc1d0271252ce9bc5b75feb86e30dd2f6dd550b
Reviewed-on: https://go-review.googlesource.com/c/go/+/650775
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Findley <rfindley@google.com>
ObjectID was a misnaming of OutputID from cacheprog's initial
implementation. It was maintained for compatibility with existing
cacheprog users in 1.24 but can be removed in 1.25.
Updates #64876
Change-Id: I8ff53bc581c16b7739e1cfbaa8bd35d285d3231d
Reviewed-on: https://go-review.googlesource.com/c/go/+/649435
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
While looking at the SSA of following code, i noticed
that these rules do not work properly, and the types
are loaded indirectly through an itab, instead of statically.
type M interface{ M() }
type A interface{ A() }
type Impl struct{}
func (*Impl) M() {}
func (*Impl) A() {}
func main() {
var a M = &Impl{}
a.(A).A()
}
Change-Id: Ia275993f81a2e7302102d4ff87ac28586023d13c
GitHub-Last-Rev: 4bfc901917
GitHub-Pull-Request: golang/go#71784
Reviewed-on: https://go-review.googlesource.com/c/go/+/649500
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
This change improves escape analysis by attempting to
deduce static values for the len and cap parameters,
allowing allocations to be made on the stack.
Change-Id: I1161019aed9f60cf2c2fe4d405da94ad415231ac
GitHub-Last-Rev: d78c1b4ca5
GitHub-Pull-Request: golang/go#71693
Reviewed-on: https://go-review.googlesource.com/c/go/+/649035
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
The go command assumes that GOMODCACHE is immutable. As an example of
one place the assumption is made, the modindex won't stat the files in
GOMODCACHE when getting the cache key for the index entry and just uses
the path of the module in the modcache (basically the module's name and
version). Explicitly reject overlays affecting GOMODCACHE to avoid
surprising and incorrect behavior.
For #71783
For #71075
Change-Id: I21dd5d39d71037de473b09ac8482a1867864e11f
Reviewed-on: https://go-review.googlesource.com/c/go/+/650475
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
The existing code for recover from deferrangefunc was broken in
several ways.
1. the code following a deferrangefunc call did not check the return
value for an out-of-band value indicating "return now" (i.e., recover
was called)
2. the returned value was delivered using a bespoke ABI that happened
to match on register-ABI platforms, but not on older stack-based
ABI.
3. the returned value was the wrong width (1 word versus 2) and
type/value(integer 1, not a pointer to anything) for deferrangefunc's
any-typed return value (in practice, the OOB value check could catch
this, but still, it's sketchy).
This -- using the deferreturn lookup method already in place for
open-coded defers -- turned out to be a much-less-ugly way of
obtaining the desired transfer of control for recover().
TODO: we also could do this for regular defer, and delete some code.
Fixes#71675
Change-Id: If7d7ea789ad4320821aab3b443759a7d71647ff0
Reviewed-on: https://go-review.googlesource.com/c/go/+/650476
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
And use it to unify all codes that need parent/closure checking.
Change-Id: I0b0aa1b007598668dff2c4bee31e21f0fb3830ce
Reviewed-on: https://go-review.googlesource.com/c/go/+/650315
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
CL 630696 changes budget for once-called closures, making them more
inlinable. However, when recursive inlining involve both the closure and
its parent, the inliner goes into an infinite loop:
parent (a closure) -> closure -> parent -> ...
The problem here dues to the closure name mangling, causing the inlined
checking condition failed, since the closure name affects how the
linker symbol generated.
To fix this, just prevent the closure from inlining its parent into
itself, avoid the infinite inlining loop.
Fixes#71680
Change-Id: Ib27626d70f95e5f1c24a3eb1c8e6c3443b7d90c8
Reviewed-on: https://go-review.googlesource.com/c/go/+/649656
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
By moving the bit field, we can reduce the miniExpr size by
8 bytes, reducing the sizes of Exprs embedding this type.
Hopefully we get a few types to a lower memory size class.
Change-Id: I4b1d4471cf905f998b26d235980e40ca91446f45
GitHub-Last-Rev: 6dea0bd27c
GitHub-Pull-Request: golang/go#71823
Reviewed-on: https://go-review.googlesource.com/c/go/+/650435
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Change-Id: I27e86c221da7f541c4823f501801e02942c9a829
Reviewed-on: https://go-review.googlesource.com/c/go/+/649935
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Currently, exported Go functions with no parameters generate C functions
with an empty parameter list. In C, a function with an empty parameter
list can accept any number of arguments, whereas a function with a single
void parameter explicitly declares that it takes no arguments.
To align the generated C functions with their Go prototypes, update the
code generation to explicitly include a void parameter for functions
with no parameters.
Fixes#68411
Change-Id: Iab9456aa0236200bf21d1181a2e18e82869df63f
GitHub-Last-Rev: 6ff21a98df
GitHub-Pull-Request: golang/go#70981
Reviewed-on: https://go-review.googlesource.com/c/go/+/638635
Reviewed-by: Michael Knyszek <mknyszek@google.com>
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>
In particular, we apply it only to functions where it is always
a code improvement to inline the call.
We also apply it to some constants.
In a few cases this may introduce a panic statement at the
caller, which is debatable, but making the potential for panic
evident is the purpose of the deprecation.
The gofix analyzer in gopls v0.18 will show a diagnostic for calls
to the annotated functions, and will offer to inline the call.
The new //go:fix annotation needs a special exemption in the
pragma check in the compiler.
Updates #32816
Change-Id: I43bf15648ac12251734109eb7102394f8a76d55e
Reviewed-on: https://go-review.googlesource.com/c/go/+/648995
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Commit-Queue: Alan Donovan <adonovan@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Currently, we only support loading of values from memory (or other
registers). Add floating point constant support to MOVD. This is
implemented by storing the floating point constant to a symbol,
which is then loaded into the floating point register.
Change-Id: I6db242d27f606f0d5d084a3ab93538698d3a4f8c
Reviewed-on: https://go-review.googlesource.com/c/go/+/631876
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The package "path" is already imported under the name pathpkg and used
many times in these files. It's not worth it to also make it available
under the name path, so keep using pathpkg.
Change-Id: I7f8fa7a11de338b9a7c0f58a48d5af68b5639cfe
Reviewed-on: https://go-review.googlesource.com/c/go/+/630475
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Make ssa.disjoint call ssa.disjointTypes to disambiguate Values based on
their types. Only one type-based rule is employed: a Type can't alias
with a pointer (https://pkg.go.dev/unsafe#Pointer).
Fixes#70488
Change-Id: I5a7e75292c2b6b5a01fb9048e3e2360e31dbcdd9
Reviewed-on: https://go-review.googlesource.com/c/go/+/632176
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Replace the usage of runtime.SetFinalizer with runtime.AddCleanup.
This changes a test and how when the Go command panics when a file is
left locked.
Updates #70907
Change-Id: I8d8c56d16486728f9bd4b910b81796ae506bda74
Reviewed-on: https://go-review.googlesource.com/c/go/+/640736
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Implement vector configuration setting instructions (VSETVLI,
VSETIVLI, VSETL). These allow the vector length (vl) and vector
type (vtype) CSRs to be configured via a single instruction.
Unfortunately each instruction has its own dedicated encoding.
In the case of VSETVLI/VSETIVLI, the vector type is specified via
a series of special operands, which specify the selected element
width (E8, E16, E32, E64), the vector register group multiplier
(M1, M2, M4, M8, MF2, MF4, MF8), the vector tail policy (TU, TA)
and vector mask policy (MU, MA). Note that the order of these
special operands matches non-Go assemblers.
Partially based on work by Pengcheng Wang <wangpengcheng.pp@bytedance.com>.
Cq-Include-Trybots: luci.golang.try:gotip-linux-riscv64
Change-Id: I431f59c1e048a3e84754f0643a963da473a741fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/631936
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Change-Id: Ifcab176faa2ac55e60576cf6acd96a18d0e860ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/648859
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
This CL adds new relocation type for riscv64: R_GOT_PCREL_ITYPE_RELOC
which generate an AUIPC + I-type pair with relocation type of GOT_HI20
and PCREL_LO12_I.
According to RISCV elf psabi doc, medium position independent code
model, the GNU as example is:
```
# Calculate address of non-local symbol
.Ltmp3: aupipc a0, %got_pcrel_hi(symbol)
ld a0, %pcrel_lo(.Ltmp3)(a0)
```
Change-Id: I719dd05e009ca2d9291f0689b346c059f9c56918
Reviewed-on: https://go-review.googlesource.com/c/go/+/612635
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
It currently isn't because it does load/store/load/store/...
Rework to do overwrite processing in pairs so it is instead
load/load/store/store/...
Change-Id: If7be629bc4048da5f2386dafb8f05759b79e9e2b
Reviewed-on: https://go-review.googlesource.com/c/go/+/631495
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Particularly with 2-word load instructions, this becomes important.
Classic example is:
func f(p *string) string {
return *p
}
We want the two loads to put the return values directly into
the two ABI return registers.
At this point in the stack, cmd/go is 1.1% smaller.
Change-Id: I51fd1710238e81d15aab2bfb816d73c8e7c207b1
Reviewed-on: https://go-review.googlesource.com/c/go/+/631137
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Look for possible paired load/store operations on arm64.
I don't expect this would be a lot faster, but it will save
binary space, and indirectly through the icache at least a bit
of time.
Change-Id: I4dd73b0e6329c4659b7453998f9b75320fcf380b
Reviewed-on: https://go-review.googlesource.com/c/go/+/629256
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
These will be used in a subsequent CL.
Change-Id: I96562668da502e5cb41096c9831c59292644be72
Reviewed-on: https://go-review.googlesource.com/c/go/+/629255
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
If we call slicebytetostring immediately (with no intervening writes)
before calling map access or delete functions with the resulting
string as the key, then we can just use the ptr/len of the
slicebytetostring argument as the key. This avoids an allocation.
Fixes#44898
Update #71132
There's old code in cmd/compile/internal/walk/order.go that handles
some of these cases.
1. m[string(b)]
2. s := string(b); m[s]
3. m[[2]string{string(b1),string(b2)}]
The old code handled cases 1&3. The new code handles cases 1&2.
We'll leave the old code around to keep 3 working, although it seems
not terribly common.
Case 2 happens particularly after inlining, so it is pretty common.
Change-Id: I8913226ca79d2c65f4e2bd69a38ac8c976a57e43
Reviewed-on: https://go-review.googlesource.com/c/go/+/640656
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Do not use New16, New20, Sum16, Sum20 anymore.
As of CL 641096, these are just wrappers around New32 and Sum32.
Change call sites to use them directly.
Change-Id: Icea91a77449f6839b903894997057ba404bd04e0
Reviewed-on: https://go-review.googlesource.com/c/go/+/641076
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
In addition to unsigned loads which already exist.
This helps code that does switches on strings to constant-fold
the switch away when the string being switched on is constant.
Fixes#71699
Change-Id: If3051af0f7255d2a573da6f96b153a987a7f159d
Reviewed-on: https://go-review.googlesource.com/c/go/+/649295
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@google.com>
The testing package already does this. go test should do the same thing.
Fixes: #69181
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest,gotip-windows-amd64-longtest
Change-Id: I942bd09c5832b48d498a2eb1f1500e1d294d0a2c
Reviewed-on: https://go-review.googlesource.com/c/go/+/648236
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
GOFIPS140=latest turns on the GODEBUG by default, and it's otherwise
untested.
Change-Id: I6a6a4656ff7ad313ce2c61ee4144ad2858bd148c
Reviewed-on: https://go-review.googlesource.com/c/go/+/648819
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Pull in the latest published version of github.com/google/pprof
as part of the continuous process of keeping Go's dependencies
up to date.
For #36905.
[git-generate]
cd src/cmd
go get github.com/google/pprof@v0.0.0-20250208200701-d0013a598941
go mod tidy
go mod vendor
Change-Id: I87e5621286d3db85f358fb86875aaf65bd7811a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/648916
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
qmuntal
Joel Sing
Than McIntosh
Ian Lance Taylor
Cherry Mui
Jorropo
limeidan
Daniel Martí
Meng Zhuo
Rob Findley
Dan Peterson
Mateusz Poliwczak
Michael Matloob
David Chase
Cuong Manh Le
cuishuang
Mauri de Souza Meneguzzo
Jakub Ciolek
Alan Donovan
Keith Randall
Andrey Bokhanko
Carlos Amedee
qiulaidongfeng
Russ Cox
Sam Thanawalla
Filippo Valsorda
Dmitri Shuralyov