| Age | Commit message (Collapse) | Author |
|---|
|
Replace a "fixme" with a more appropriate error. Also invert the condition
so that the error returns early, which is more Go idiomatic.
Change-Id: I03006572c4010fb47037bed3ee1fd7f92bfc20d3
Reviewed-on: https://go-review.googlesource.com/c/go/+/523457
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: M Zhuo <mzh@golangcn.org>
|
|
Change-Id: I68be4110216145ad1fb2e5095e1f2b143f9e69ac
Reviewed-on: https://go-review.googlesource.com/c/go/+/523456
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: M Zhuo <mzh@golangcn.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
Rather than handling shift based scaling in two locations, rework logic
so there is a single exit path.
Change-Id: I832b4932d53183736050059a11019ced08281b3b
Reviewed-on: https://go-review.googlesource.com/c/go/+/523455
Reviewed-by: M Zhuo <mzh@golangcn.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
Unlike RLWNM, the ISA only supports an immediate rotate operand.
Update optab and opirrr to avoid quietly assembling this insn.
Change-Id: I1472a431cb8a870d55d5fff79ab905c4c459f630
Reviewed-on: https://go-review.googlesource.com/c/go/+/449835
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Paul Murphy <murp@ibm.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
This CL is to add assembly instruction mnemonics for the following instructions, mainly used in crypto packages.
* KMA - cipher message with authentication
* KMCTR - cipher message with counter
Fixes #61163
Change-Id: Iff9a69911aeb4fab4bca8755b23a106eaebb2332
Reviewed-on: https://go-review.googlesource.com/c/go/+/515195
Reviewed-by: Carlos Amedee <carlos@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
|
|
Change-Id: I8b3eb51e31139314d222d796b56ab7d42338797f
Reviewed-on: https://go-review.googlesource.com/c/go/+/525315
Run-TryBot: shuang cui <imcusg@gmail.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>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
|
|
Preparatory refactoring before next CL.
Change-Id: I06fb4670b933fddff1a2a70f3cf1eb124cbd86ee
Reviewed-on: https://go-review.googlesource.com/c/go/+/524899
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
The old Go object file format used linker symbols like "gofile..foo"
to record references to the filename "foo". But the current object
file format has a dedicated section for file names, so we don't need
these useless prefixes anymore.
Also, change DWARF generation to pass around the src.Pos directly,
rather than the old file symbols, which it just turned back into a
file index before writing out anyway.
Finally, directly record the FileIndex into src.PosBase, so that we
can skip the map lookups.
Change-Id: Ia4a5ebfa95da271f2522e45befdb9f137c16d373
Reviewed-on: https://go-review.googlesource.com/c/go/+/523378
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
|
|
This adds a modicum of type safety to these APIs, which are otherwise
quite confusing to follow.
Change-Id: I268a9a1a99a47dcfef6dc1e9e5be13673af3fb85
Reviewed-on: https://go-review.googlesource.com/c/go/+/523396
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
|
|
cmd/asm and cmd/compile now always create symbols with the appropriate
package prefixes, so cmd/internal/dwarf and cmd/internal/obj can stop
worrying about qualifying names itself.
Change-Id: I9aee5d759bf0d41a61722c777e7f66fce957e79e
Reviewed-on: https://go-review.googlesource.com/c/go/+/523338
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
This CL changes cmd/internal/obj to also implicitly set
ArgsPointerMaps and ArgInfo for assembly functions that are explicitly
package qualified (e.g., "pkg·name", not just "·name"). This is a
prerequisite for changing cmd/asm to stop emitting `"".`-prefixed
symbol names.
Change-Id: I4e14bc24c87cf4d7114a7aed9beaf0c8d1f9c07f
Reviewed-on: https://go-review.googlesource.com/c/go/+/523335
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
|
|
We have obj.Link.Pkgpath, so we don't need to pass it redundantly in
places where we already have an *obj.Link.
Also, renaming the parser's "compilingRuntime" field to "allowABI", to
match the "AllowAsmABI" name used by objabi.LookupPkgSpecial.
Finally, push the handling of GOEXPERIMENT_* flags up to cmd/asm's
main entry point, by simply appending them to flags.D.
Change-Id: I6ada134522b0cbc90d35bcb145fbe045338fefb7
Reviewed-on: https://go-review.googlesource.com/c/go/+/523297
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
pairs
Implement better classification for load and store pair operations. This in
turn allows us to avoid using pool literals when the offset fits in a 24 bit
unsigned immediate. In this case, the offset can be calculated using two
add immediate instructions, rather than loading the offset from the pool
literal and then adding the offset to the base register. This requires the
same number of instructions, however avoids a load from memory and does
not require the offset to be stored in the literal pool.
Updates #59615
Change-Id: I316ec3d54f1d06ae9d930e98d0c32471775fcb26
Reviewed-on: https://go-review.googlesource.com/c/go/+/515615
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Joedian Reid <joedian@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
Load large constants into vector registers from rodata, instead of placing them
in the literal pool. This treats VMOVQ/VMOVD/VMOVS the same as FMOVD/FMOVS and
makes use of the existing mechanism for storing values in rodata. Two additional
instructions are required for a load, however these instructions are used
infrequently and already have a high latency.
Updates #59615
Change-Id: I54226730267689963d73321e548733ae2d66740e
Reviewed-on: https://go-review.googlesource.com/c/go/+/515617
Reviewed-by: Eric Fang <eric.fang@arm.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
Make it possible to internally link cgo on riscv64, which also adds
support for SDYNIMPORT calls without external linking being required.
This reduces the time of an ./all.bash run on a Sifive Hifive Unleashed by
approximately 20% (~140 minutes down to ~110 minutes).
Change-Id: I43f1348de31672718ae8676cc82f6fdc1dfee054
Reviewed-on: https://go-review.googlesource.com/c/go/+/431104
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Than McIntosh <thanm@google.com>
|
|
Change-Id: I31ba6696e124dccf37d674d090fdf04ba0a049a3
Reviewed-on: https://go-review.googlesource.com/c/go/+/515616
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
KDSA(Compute Digital Signature Authentication) instruction provides
support for the signing and verification of elliptic curves
Change-Id: I19996a307162dd4f476a1cfe4f8d1a74a609e6c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/503215
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
Add support for SEB/SEH instructions, which are introduced in mips32r2.
SEB/SEH can be used to sign-extend byte/halfword in registers directly without passing through memory.
Ref: The MIPS32 Instruction Set, Revision 5.04: https://s3-eu-west-1.amazonaws.com/downloads-mips/documents/MD00086-2B-MIPS32BIS-AFP-05.04.pdf
Updates #60072
Change-Id: I33175ae9d943ead5983ac004bd2a158039046d65
Reviewed-on: https://go-review.googlesource.com/c/go/+/515475
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
Change-Id: I476e2a75f39c876fa9c071cada36573740d546de
GitHub-Last-Rev: dec3fb438f4dd0d8e0aff300356c8d92e8ee6749
GitHub-Pull-Request: golang/go#61783
Reviewed-on: https://go-review.googlesource.com/c/go/+/516395
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
Rather than having register encoding knowledge in each caller of opxrrr,
pass the registers into opxrrr and let it handle the encoding. This reduces
duplication and improves readability.
Change-Id: I202c503465a0169277a0f64340598203c9dcf20c
Reviewed-on: https://go-review.googlesource.com/c/go/+/461140
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
The existing Thread Local Storage (TLS) implementation for riscv64 uses
initial-exec (IE) mode, however a MOV of a TLS symbol currently loads the
thread pointer offset and not the actual address or memory location.
Rework TLS on riscv64 to generate the full instruction sequence needed to
load from or store to a TLS symbol. Additionally, provide support for both
initial-exec (IE) and local-exec (LE) TLS - in many cases we can use LE,
which is slightly more efficient and easier to support in the linker.
Change-Id: I1b43f8888b3b6b10354bbb79d604771e64d92645
Reviewed-on: https://go-review.googlesource.com/c/go/+/431103
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: M Zhuo <mzh@golangcn.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
Add support for WSBH/DSBH/DSHD instructions, which are introduced in mips{32,64}r2.
WSBH reverse bytes within halfwords for 32-bit word, DSBH reverse bytes within halfwords for 64-bit doubleword, and DSHD reverse halfwords within doublewords. These instructions can be used to optimize byte swaps.
Ref: The MIPS64 Instruction Set, Revision 5.04: https://s3-eu-west-1.amazonaws.com/downloads-mips/documents/MD00087-2B-MIPS64BIS-AFP-05.04.pdf
Updates #60072
Change-Id: I31c043150fe8ac03027f413ef4cb2f3e435775e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/493816
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
This CL add's the following instructions,useful for cipher and
message digest operations:
* KM - cipher message
* KMC - cipher message with chaining
* KLMD - compute last message digest
* KIMD - compute intermediate message digest
Fixes #61163
Change-Id: Ib0636430c3e4888ed61b86c5acae45ee596463ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/509075
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
|
|
Rather than passing registers as uint32, use int16 and cast to uint32 in
the OP_FRRR/OP_SRR implementation. This allows a large number of casts to be
removed and code simplified at call sites.
Change-Id: Ia4bf3bd022f79d65e45fca64db13db12739e5b23
Reviewed-on: https://go-review.googlesource.com/c/go/+/514099
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Rong Zhang <rongrong@oss.cipunited.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
|
|
Rather than passing registers as uint32, use int16 and cast to uint32 in
the OP_IRR implementation. This allows a large number of casts to be removed
and code simplified at call sites. Also be more consistent with op, register
and value ordering.
Change-Id: I510347d97787ce80a338037b25470addf3a2939d
Reviewed-on: https://go-review.googlesource.com/c/go/+/514098
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Rong Zhang <rongrong@oss.cipunited.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
|
|
Rather than passing registers as uint32, use int16 and cast to uint32 in
the OP_RRR implementation. This allows a large number of casts to be removed
and code simplified at call sites.
Change-Id: I2194ea07a6f89861530cd88c4b60c993bb0fa0c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/514097
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Rong Zhang <rongrong@oss.cipunited.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
Instead of comparing register values to 0, use obj.REG_NONE to improve
readability. No functional change.
Change-Id: I848f7ab7e27fe7e5e9fe2caf9cce47e372c6cd6c
Reviewed-on: https://go-review.googlesource.com/c/go/+/514096
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Rong Zhang <rongrong@oss.cipunited.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Junxian Zhu <zhujunxian@oss.cipunited.com>
|
|
The previous implementation would limit itself to 0xfff000 | 0xfff << shift,
while the maximum possible value is 0xfff000 + 0xfff << shift. In practical
terms, this means that an additional ((1 << shift) - 1) * 0x1000 of offset
is reachable for operations that use this splitting format. In the case of
an 8 byte load/store, this is an additional 0x7000 that can be reached
without needing to use the literal pool.
Updates #59615
Change-Id: Ice7023104042d31c115eafb9398c2b999bdd6583
Reviewed-on: https://go-review.googlesource.com/c/go/+/512540
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
In a number of load and store cases, the use of the literal pool can be
entirely avoided by simply adding or subtracting the offset from the
register. This uses the same number of instructions, while avoiding a
load from memory, along with the need for the value to be in the literal
pool. Overall this reduces the size of binaries slightly and should have
lower overhead.
Updates #59615
Change-Id: I9cb6a403dc71e34a46af913f5db87dbf52f8688c
Reviewed-on: https://go-review.googlesource.com/c/go/+/512539
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
Currently, pool literals are added when they are not needed, namely
in the case where the offset is a 24 bit unsigned scaled immediate.
By improving the classification of loads and stores, we can avoid
generating unused pool literals. However, more importantly this
provides a basis for further improvement of the load and store
code generation.
Updates #59615
Change-Id: Ia3bad1709314565a05894a76c434cca2fa4533c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/512538
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
The PCALIGN asm directive was not supported on 386/amd64,
causing a compile-time error when used. The same directive
is currently supported on arm64, loong64 and ppc64 architectures.
This has potential for noticeable performance improvements on
amd64 across multiple packages, I did a quick test aligning a hot
loop on bytes.IndexByte:
```
IndexByte/10-16 3.477n ± ∞ ¹ 3.462n ± ∞ ¹ ~ (p=0.198 n=5)
IndexByte/32-16 4.675n ± ∞ ¹ 4.834n ± ∞ ¹ +3.40% (p=0.008 n=5)
IndexByte/4K-16 67.47n ± ∞ ¹ 44.44n ± ∞ ¹ -34.13% (p=0.008 n=5)
IndexByte/4M-16 61.98µ ± ∞ ¹ 45.07µ ± ∞ ¹ -27.28% (p=0.008 n=5)
IndexByte/64M-16 1206.6µ ± ∞ ¹ 940.9µ ± ∞ ¹ -22.02% (p=0.008 n=5)
IndexBytePortable/10-16 4.064n ± ∞ ¹ 4.044n ± ∞ ¹ ~ (p=0.325 n=5)
IndexBytePortable/32-16 9.999n ± ∞ ¹ 9.934n ± ∞ ¹ ~ (p=0.151 n=5)
IndexBytePortable/4K-16 975.8n ± ∞ ¹ 965.5n ± ∞ ¹ ~ (p=0.151 n=5)
IndexBytePortable/4M-16 973.3µ ± ∞ ¹ 972.3µ ± ∞ ¹ ~ (p=0.222 n=5)
IndexBytePortable/64M-16 15.68m ± ∞ ¹ 15.89m ± ∞ ¹ ~ (p=0.310 n=5)
geomean 1.478µ 1.342µ -9.20%
IndexByte/10-16 2.678Gi ± ∞ ¹ 2.690Gi ± ∞ ¹ ~ (p=0.151 n=5)
IndexByte/32-16 6.375Gi ± ∞ ¹ 6.165Gi ± ∞ ¹ -3.30% (p=0.008 n=5)
IndexByte/4K-16 56.54Gi ± ∞ ¹ 85.85Gi ± ∞ ¹ +51.83% (p=0.008 n=5)
IndexByte/4M-16 63.03Gi ± ∞ ¹ 86.68Gi ± ∞ ¹ +37.52% (p=0.008 n=5)
IndexByte/64M-16 51.80Gi ± ∞ ¹ 66.42Gi ± ∞ ¹ +28.23% (p=0.008 n=5)
IndexBytePortable/10-16 2.291Gi ± ∞ ¹ 2.303Gi ± ∞ ¹ ~ (p=0.421 n=5)
IndexBytePortable/32-16 2.980Gi ± ∞ ¹ 3.000Gi ± ∞ ¹ ~ (p=0.151 n=5)
IndexBytePortable/4K-16 3.909Gi ± ∞ ¹ 3.951Gi ± ∞ ¹ ~ (p=0.151 n=5)
IndexBytePortable/4M-16 4.013Gi ± ∞ ¹ 4.017Gi ± ∞ ¹ ~ (p=0.222 n=5)
IndexBytePortable/64M-16 3.987Gi ± ∞ ¹ 3.933Gi ± ∞ ¹ ~ (p=0.310 n=5)
geomean 8.183Gi 9.013Gi +10.14%
```
Fixes #56474
Change-Id: Idea022b1a16e6d4b8dd778723adb862c46602c4f
GitHub-Last-Rev: 2eb7e31dc378a02fd83faa7d41239df0f2859677
GitHub-Pull-Request: golang/go#61516
Reviewed-on: https://go-review.googlesource.com/c/go/+/511662
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
|
|
Currently the Optab structure contains four arguments of an instruction,
excludes the fifth argument p.RegTo2. It does not participate in
instruction matching and is usually handled separately.
Instructions with five operands are common in the newer arm instruction
set, so this CL adds the fifth argument to Optab, so that instruction
matching is easier. This caused the oplook function also needs to be
updated synchronously, this CL also made some cleaning and modifications
to this function.
Change-Id: I1d95ad99e72a44dfad1e00db182cfc369a0e55c6
Reviewed-on: https://go-review.googlesource.com/c/go/+/505975
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>
|
|
immediates
Rather than duplicating this code, factor it out into a function and
add test coverage.
Change-Id: I37ce568ded4659d98a4ff1361520c5fb2207e947
Reviewed-on: https://go-review.googlesource.com/c/go/+/512537
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
This code stems from the original 7l C code, where one way to determine
the end of a table is to put a sentinel entry, then scan for it. This is
now Go code and the length of an array is readily available.
Remove the sentinel and sentinel scan, then adjust the remaining code to
work accordingly.
Change-Id: I8964c787f5149f3548fa78bf8923aa7a93f9482e
Reviewed-on: https://go-review.googlesource.com/c/go/+/512536
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
|
|
The following instruction is wrongly encoded on arm64:
MOVD (R2)(R3<<0), R1
It's incorrectly encoded as
MOVD (R2)(R3<<3), R1
The reason for the error is that we hard-coded the shift encoding to 6,
which is correct for the MOVB and MOVBU instructions because it only
allows a shift amount of 0, but it is wrong for the MOVD instruction
because it also allows other shift values.
For instructions MOVB, MOVBU and FMOVB, the extension amount must be 0,
encoded in "S" as 0 if omitted, or as 1 if present. But in Go, we don't
distinguish between Rn.<EXT> and Rn.<EXT><<0, so we encode it as that
does not present. This makes no difference to the function of the
instruction.
Change-Id: I2afe3498392cc9b2ecd524c7744f28b9d6d107b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/510995
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
The compiler/assembler's -S output prints relocation type
numerically, which is hard to understand. Every time I need to
count the relocation type constants to figure out which relocation
it actually is. Print the symbolic name instead.
Change-Id: I4866873bbae8b3dc0ee212609cb00280f9164243
Reviewed-on: https://go-review.googlesource.com/c/go/+/501856
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
|
|
Go's memory model closely follows the approach C++ concurrency memory
model (https://go.dev/ref/mem) and Go atomic "has the same semantics as C++'s
sequentially consistent atomics".
Meanwhile according to RISCV manual A.6 "Mappings from C/C++ primitives to RISC-V primitives".
C/C++ atomic operations (memory_order_acq_rel) should be map to "amo<op>.{w|d}.aqrl"
LR/SC (memory_order_acq_rel) should map to "lr.{w|d}.aq; <op>; sc.{w|d}.rl"
goos: linux
goarch: riscv64
pkg: runtime/internal/atomic
│ atomic.old.bench │ atomic.new.bench │
│ sec/op │ sec/op vs base │
AtomicLoad64-4 4.216n ± 1% 4.202n ± 0% ~ (p=0.127 n=10)
AtomicStore64-4 5.040n ± 0% 6.718n ± 0% +33.30% (p=0.000 n=10)
AtomicLoad-4 4.217n ± 0% 4.213n ± 0% ~ (p=0.145 n=10)
AtomicStore-4 5.040n ± 0% 6.718n ± 0% +33.30% (p=0.000 n=10)
And8-4 9.237n ± 0% 9.240n ± 0% ~ (p=0.582 n=10)
And-4 5.878n ± 0% 6.719n ± 0% +14.31% (p=0.000 n=10)
And8Parallel-4 28.44n ± 0% 28.46n ± 0% +0.07% (p=0.000 n=10)
AndParallel-4 28.40n ± 0% 28.43n ± 0% +0.11% (p=0.000 n=10)
Or8-4 8.399n ± 0% 8.398n ± 0% ~ (p=0.357 n=10)
Or-4 5.879n ± 0% 6.718n ± 0% +14.27% (p=0.000 n=10)
Or8Parallel-4 28.43n ± 0% 28.45n ± 0% +0.09% (p=0.000 n=10)
OrParallel-4 28.40n ± 0% 28.43n ± 0% +0.11% (p=0.000 n=10)
Xadd-4 30.05n ± 0% 30.10n ± 0% +0.18% (p=0.000 n=10)
Xadd64-4 30.05n ± 0% 30.09n ± 0% +0.12% (p=0.000 n=10)
Cas-4 60.48n ± 0% 61.13n ± 0% +1.08% (p=0.000 n=10)
Cas64-4 62.28n ± 0% 62.34n ± 0% ~ (p=0.810 n=10)
Xchg-4 30.05n ± 0% 30.09n ± 0% +0.15% (p=0.000 n=10)
Xchg64-4 30.05n ± 0% 30.09n ± 0% +0.13% (p=0.000 n=10)
geomean 15.42n 16.17n +4.89%
Fixes #61295
Change-Id: I97b5325db50467eeec36fb079bded7b09a32330f
Reviewed-on: https://go-review.googlesource.com/c/go/+/508715
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Run-TryBot: M Zhuo <mzh@golangcn.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
Change-Id: I510b0a4bf3472d937393800dd57472c30beef329
GitHub-Last-Rev: 8d289b73a37bd86080936423d981d21e152aaa33
GitHub-Pull-Request: golang/go#60960
Reviewed-on: https://go-review.googlesource.com/c/go/+/505398
Auto-Submit: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
|
|
The encoding of instructions "FMOVD F1, ZR" and "FMOVS F1, ZR" is wrong,
the assembler encodes them as "FMOVD F1, F31" and "FMOVS F1, F31". This
CL fixes the bug.
Change-Id: I2d31520b58f9950ce2534a04f4a3275bf103a673
Reviewed-on: https://go-review.googlesource.com/c/go/+/503135
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
Fix spelling errors discovered using https://github.com/codespell-project/codespell. Errors in data files and vendored packages are ignored.
Change-Id: I83c7818222f2eea69afbd270c15b7897678131dc
GitHub-Last-Rev: 3491615b1b82832cc0064f535786546e89aa6184
GitHub-Pull-Request: golang/go#60758
Reviewed-on: https://go-review.googlesource.com/c/go/+/502576
Auto-Submit: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
|
|
sort.Slice was being used to sort some newly added entries by name
to make the ctxt.Data slice reproducible, but some existing entries
have the same name, and the effect was to take the non-determinism
of the tail entries and scatter it into the earlier, deterministic section
when multiple entries had the same name.
The specific entries with the same name are type SDWARFVAR, which
all have an empty name but different relocations. If they are shuffled,
then the relocation symbols are visited in a different order, which
enters them into the string table in a different order, which results in
different object files, different object file hashes, and different build IDs
for the final executables.
Use sort.SliceStable to avoid reordering entries we don't mean to reorder.
Also add a simple test for scheduling-related non-determinism.
I debugged this originally using 'go install -race cmd/compile',
but that was slow and turned out not to be terribly reliable.
Using a few different GOMAXPROCS settings turns out to be a much more
effective (and faster) way to scramble scheduling decisions.
Fixes #60759.
Change-Id: Ia966b02b9fdaefa971f998a09319ca375bdf8604
Reviewed-on: https://go-review.googlesource.com/c/go/+/502755
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>
TryBot-Bypass: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
HAUTO should be handled the same way as other stack offsets for
adding to constant pool. Add the missing cases.
Fixes #57955.
Change-Id: If7fc82cafb2bbf0a6121e73e353b8825cb36b5bc
Reviewed-on: https://go-review.googlesource.com/c/go/+/463138
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
|
|
Before this CL, we use GetFrom3&SetFrom3 to get or set a source operand
which not fit into Prog.Reg. Those APIs operate the first element in
Prog.RestArgs without checking the type so they're fragile to break if
we have more than one different type of operands in the slice, which
will be a common case in Arm64.
This CL deprecates & renames some APIs related to Prog.RestArgs to make
those APIs more reasonable and robust than before.
Change-Id: I70d56edc1f23ccfffbcd6df34844e2cef2288432
Reviewed-on: https://go-review.googlesource.com/c/go/+/493355
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Eric Fang <eric.fang@arm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>
|
|
In Go 1.17, cmd/compile gained the ability to inline calls to
functions that contain function literals (aka "closures"). This was
implemented by duplicating the function literal body and emitting a
second LSym, because in general it might be optimized better than the
original function literal.
However, the second LSym was named simply as any other function
literal appearing literally in the enclosing function would be named.
E.g., if f has a closure "f.funcX", and f is inlined into g, we would
create "g.funcY" (N.B., X and Y need not be the same.). Users then
have no idea this function originally came from f.
With this CL, the inlined call stack is incorporated into the clone
LSym's name: instead of "g.funcY", it's named "g.f.funcY".
In the future, it seems desirable to arrange for the clone's name to
appear exactly as the original name, so stack traces remain the same
as when -l or -d=inlfuncswithclosures are used. But it's unclear
whether the linker supports that today, or whether any downstream
tooling would be confused by this.
Updates #60324.
Change-Id: Ifad0ccef7e959e72005beeecdfffd872f63982f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/497137
Reviewed-by: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
|
|
Before this CL, instructions such as SHA1SU0, AESD and AESE are encoded
in case 1 together with FMOV/ADD, and some error checking is missing,
for example:
SHA1SU0 V1.B16, V2.B16, V3.B16 // wrong data arrangement
SHA1SU0 V1.4S, V2.S4, V3.S4 // correct
Both will be accepted by the assembler, but the first one is totally
incorrect.
This CL fixes these potential encoding issues by moving them into
separate cases, adds some error tests, and also fixes a wrong encoding
operand for ASHA1C.
Change-Id: Ic778321a567735d48bc34a1247ee005c4ed9e11f
Reviewed-on: https://go-review.googlesource.com/c/go/+/493195
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
An explicit TOC pointer is not needed when building with GOPPC64=power10
on linux/PPC64 as all addressing is PC relative.
Apply changes to the compiler, assembler, and linker to remove R2/TOC
maintenance in these build configurations.
This results in noticeably smaller PIC binaries. For example the size
(in bytes) of the k8s binaries before and with this patch:
GOFLAGS="-buildmode=pie" \
FORCE_HOST_GO=y \
GOPPC64=power10 \
CGO_CFLAGS="-mcpu=power10 -O2 -g" \
make all
apiextensions-apiserver 66060288 64487424 -1572864 -2.38%
e2e_node.test 163520856 159850760 -3670096 -2.24%
e2e.test 178167304 174890432 -3276872 -1.83%
ginkgo 11010048 10747904 -262144 -2.38%
go-runner 2162688 2162688 0 0%
k8s-tests 170182216 166970880 -3211336 -1.88%
kubeadm 52625408 51314688 -1310720 -2.49%
kube-aggregator 62849024 61341696 -1507328 -2.39%
kube-apiserver 147783680 144375808 -3407872 -2.30%
kube-controller-manager 131137536 127991808 -3145728 -2.39%
kubectl 53608448 52363264 -1245184 -2.32%
kubectl-convert 52625408 51314688 -1310720 -2.49%
kubelet 120913920 118095872 -2818048 -2.33%
kube-log-runner 1900544 1835008 -65536 -3.44%
kubemark 119078912 116326400 -2752512 -2.31%
kube-proxy 58392576 56885248 -1507328 -2.58%
kube-scheduler 60751872 59244544 -1507328 -2.48%
mounter 1835008 1769472 -65536 -3.57%
watch-termination 38076416 37158912 -917504 -2.40%
And text size changes:
apiextensions-apiserver 30243288 28654116 -1589172 -5.25%
e2e_node.test 71132064 67520288 -3611776 -5.07%
e2e.test 61843984 58635088 -3208896 -5.18%
ginkgo 4975916 4769304 -206612 -4.15%
go-runner 896532 858400 -38132 -4.25%
k8s-tests 60925792 57752032 -3173760 -5.20%
kubeadm 24643240 23404100 -1239140 -5.02%
kube-aggregator 28688060 27160976 -1527084 -5.32%
kube-apiserver 65627332 62259460 -3367872 -5.13%
kube-controller-manager 56773844 53706532 -3067312 -5.40%
kubectl 24344276 23080640 -1263636 -5.19%
kubectl-convert 23733764 22521768 -1211996 -5.10%
kubelet 52494580 49720340 -2774240 -5.28%
kube-log-runner 787128 753232 -33896 -4.30%
kubemark 51576580 48837380 -2739200 -5.31%
kube-proxy 26541092 25124080 -1417012 -5.33%
kube-scheduler 27448512 25976172 -1472340 -5.36%
mounter 744100 712628 -31472 -4.22%
watch-termination 18047276 17139912 -907364 -5.02%
Change-Id: Ib4872823b06f93861e46a00679b5d4e5e30b538a
Reviewed-on: https://go-review.googlesource.com/c/go/+/495416
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Paul Murphy <murp@ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
|
|
Retrying the original CL with a small modification. The original CL
did not handle the case of reading an itab out of a dictionary
correctly. When we read an itab out of a dictionary, we must treat
the type inside that itab as maybe being put in an interface.
Original CL: 486895
Revert CL: 490156
Change-Id: Id2dc1699d184cd8c63dac83986a70b60b4e6cbd7
Reviewed-on: https://go-review.googlesource.com/c/go/+/491495
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
ARM64 doesn't have MOVNP/MOVNPW and STLP/STLPW instructions, which are
currently useless instructions as well. This CL deletes them. At the
same time this CL sorts the opcodes by name, which looks cleaner.
Change-Id: I25cfb636b23356ba0a50cba527a8c85b3f7e2ee4
Reviewed-on: https://go-review.googlesource.com/c/go/+/495695
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
|
|
This CL updates riscv instructions by
https://github.com/riscv/riscv-opcodes
which adds instruction:
APAUSE
And removes the following unused instructions:
AFENCEI
AFMVQX
AFMVXQ
AHFENCEGVMA
AHFENCEVVMA
ASLLIRV32
ASRAIRV32
ASRLIRV32
AURET
Change-Id: I314570c643af3e6bbc9d2cd471b6b39985bcbdff
Reviewed-on: https://go-review.googlesource.com/c/go/+/409415
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Benny Siegert <bsiegert@gmail.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: M Zhuo <mzh@golangcn.org>
|
|
The call sites that cared all reversed inner-to-outer to outer-to-inner already.
The ones that didn't care left it alone. No one explicitly wanted inner-to-outer.
Also change to a callback-based interface, so that call sites aren't required
to accumulate the results in a slice (the main reason for that before was to
reverse the slice!).
There were three places where these lists were printed:
1. -d=ssa/genssa/dump, explicitly reversing to outer-to-inner
2. node dumps like -W, leaving the default inner-to-outer
3. file positions for HashDebugs, explicitly reversing to outer-to-inner
It makes no sense that (1) and (2) would differ. The reason they do is that
the code for (2) was too lazy to bother to fix it to be the right way.
Consider this program:
package p
func f() {
g()
}
func g() {
println()
}
Both before and after this change, the ssa dump for f looks like:
# x.go:3
00000 (3) TEXT <unlinkable>.f(SB), ABIInternal
00001 (3) FUNCDATA $0, gclocals·g2BeySu+wFnoycgXfElmcg==(SB)
00002 (3) FUNCDATA $1, gclocals·g2BeySu+wFnoycgXfElmcg==(SB)
v4 00003 (-4) XCHGL AX, AX
# x.go:4
# x.go:8
v5 00004 (+8) PCDATA $1, $0
v5 00005 (+8) CALL runtime.printlock(SB)
v7 00006 (-8) CALL runtime.printnl(SB)
v9 00007 (-8) CALL runtime.printunlock(SB)
# x.go:5
b2 00008 (5) RET
00009 (?) END
Note # x.go:4 (f) then # x.go:8 (g, called from f) between v4 and v5.
The -W node dumps used the opposite order:
before walk f
. AS2 Def tc(1) # x.go:4:3
. INLMARK # +x.go:4:3
. PRINTN tc(1) # x.go:8:9,x.go:4:3
. LABEL p..i0 # x.go:4:3
Now they match the ssa dump order, and they use spaces as separators,
to avoid potential problems with commas in some editors.
before walk f
. AS2 Def tc(1) # x.go:4:3
. INLMARK # +x.go:4:3
. PRINTN tc(1) # x.go:4:3 x.go:8:9
. LABEL p..i0 # x.go:4:3
I'm unaware of any argument for the old order other than it was easier
to compute without allocation. The new code uses recursion to reverse
the order without allocation.
Now that the callers get the results outer-to-inner, most don't need
any slices at all.
This change is particularly important for HashDebug, which had been
using a locked temporary slice to walk the inline stack without allocation.
Now the temporary slice is gone.
Change-Id: I5cb6d76b2f950db67b248acc928e47a0460569f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/493735
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Russ Cox <rsc@golang.org>
|
