[pull] master from golang:master#45
Merged
Merged
Conversation
While the primary purpose of gp.m.locks may be to count the number of mutex values held by the M, several other parts of the runtime use it as a general-purpose mechanism to disable preemption, independent of the mutex implementation. Adding a sample to the mutex contention profile includes acquiring a mutex, and the easiest way to know that's safe to do is by waiting until the M has released its final mutex. But when calls to lock/unlock are interspersed with acquirem/releasem (or other changes to gp.m.locks), it's hard to tell when the mutex count will reach zero. This can cause TestRuntimeLockMetricsAndProfile to drop samples, attributing more than necessary to _LostContendedRuntimeLock rather than more specific call stacks. Consider the case where an M experiences contention on sched.lock in startm and captures the call stack in its local buffer. The call to unlock is nested within an acquirem/releasem pair, which makes it look like it's not safe to acquire the mutex that protects the contention profile buckets. So the M doesn't flush its local buffer during the unlock call. Then releasem reduces gp.m.locks to 0, but doesn't include any code related to flushing the local contention sample. Later, the same M experiences contention on another mutex. Since it already has a sample in its buffer, it needs to choose which one to keep (adding the other's value to _LostContendedRuntimeLock). This leads to the profile missing known contention events (such as in #70602). Change lock/unlock to use a larger delta. Leave gp.m.locks as an int32, shifting the mutex count by 4 bits (steps of 16), for 2^27-1 before overflow. Leave acquirem (and other ad-hoc gp.m.locks manipulation) with a delta of 1, allowing 15 levels of nesting before interfering with mutex profile sampling. For #70602 For #79409 Change-Id: I7935d9e9f5b221001285fea1131c4d5ed31f0ce2 Reviewed-on: https://go-review.googlesource.com/c/go/+/779160 Auto-Submit: Rhys Hiltner <rhys.hiltner@gmail.com> Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Michael Pratt <mpratt@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The mutex contention profile tests have to create contention between OS threads. Previously, the first thread to obtain the mutex under test would busy-loop until it observed the other thread register its interest in acquiring the mutex. Slow builders (such as gotip-linux-arm and gotip-linux-amd64-noopt) are timing out on this test. Have the leading thread yield to the OS when it hasn't yet seen the contention, to give the other thread a better chance to run. For #70602 For #79409 Change-Id: I3f97e0739fd9cb06b939bfc7c1768ca5e027bd0c Reviewed-on: https://go-review.googlesource.com/c/go/+/779241 Auto-Submit: Rhys Hiltner <rhys.hiltner@gmail.com> Reviewed-by: David Chase <drchase@google.com> TryBot-Bypass: David Chase <drchase@google.com> Reviewed-by: Keith Randall <khr@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Keith Randall <khr@google.com>
While not recognized by cmd/go, this helps with bug reports around the tooling ecosystem. Fixes #75930 Change-Id: I23018d0bb9fcb13df7060d5bf6e11abd6a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/778600 Reviewed-by: qiu laidongfeng <2645477756@qq.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Alan Donovan <adonovan@google.com> Reviewed-by: Alan Donovan <adonovan@google.com> Reviewed-by: Michael Pratt <mpratt@google.com>
p.ip is a value type and cannot ever by nil. Change-Id: I28d54ff1f70d0965fe42cb9f02a7359efd9c9740 Reviewed-on: https://go-review.googlesource.com/c/go/+/779641 LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Michael Pratt <mpratt@google.com> Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com> Reviewed-by: Florian Lehner <lehner.florian86@gmail.com> Reviewed-by: Sean Liao <sean@liao.dev> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Test TestReadUniqueDirectoryEntries requires symbolic links, which are not supported on all platforms. Use testenv.MustHaveSymlink to skip the test on those platforms. Fixes #79516 Change-Id: Iea650022c985c396af6653e3391173cb17bcee20 Reviewed-on: https://go-review.googlesource.com/c/go/+/779642 Reviewed-by: Filippo Valsorda <filippo@golang.org> Reviewed-by: Roland Shoemaker <roland@golang.org> Auto-Submit: Filippo Valsorda <filippo@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Michael Pratt <mpratt@google.com>
v1 is implemented in terms of v2, and as such inherits support for MarshalerTo. This is important to ensure that types implementing MarshalerTo are usable by all clients, not just v2 clients. Update the v1 Marshal docs to add MarshalerTo to the precedence rules. I've reused the nice bulleted formatting used in v2. For #71497. Change-Id: Ifc0e89ce7203aa51016c1056ebaf69ac6a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/775180 Reviewed-by: Damien Neil <dneil@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change-Id: I8a6772cbf6b347d1bd80c5978423046a63a6b743 Reviewed-on: https://go-review.googlesource.com/c/go/+/780320 Auto-Submit: abner chenc <chenguoqi@loongson.cn> Reviewed-by: Michael Pratt <mpratt@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Meidan Li <limeidan@loongson.cn> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
This allows us to properly generate the ops and merge/load rules for various SIMD instructions that can use memory operands. Fixes #78159 Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-simd,gotip-linux-amd64_avx512-simd Change-Id: Idec450c931c41bb903d4cc5b9b9ee8f610ee8796 Reviewed-on: https://go-review.googlesource.com/c/go/+/779521 LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
As flagged by the modernize linter. Change-Id: Iba6ea62d617f7e3622773c76a47655f1ad8be0e5 Reviewed-on: https://go-review.googlesource.com/c/go/+/779661 Reviewed-by: Michael Pratt <mpratt@google.com> Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Update to a newer version of golang.org/x/arch, in order to bring in changes to the riscv64 disassembler: go get golang.org/x/arch@master go mod vendor go mod tidy Change-Id: I6e3ece09e72e36d50bae8341832d5b8f7e330e72 Reviewed-on: https://go-review.googlesource.com/c/go/+/778961 Reviewed-by: Keith Randall <khr@golang.org> Auto-Submit: Joel Sing <joel@sing.id.au> Reviewed-by: Keith Randall <khr@google.com> Reviewed-by: Cherry Mui <cherryyz@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change floating-point parsing and printing to use fast unrounded scaling, as presented in “Floating-Point Printing and Parsing Can Be Simple And Fast”, https://research.swtch.com/fp This CL deletes almost 900 lines of code while making printing much faster. Parsing is simpler but about the same speed. benchmark \ host local linux-arm64 s7 linux-amd64 s7:GOARCH=386 linux-386 vs base vs base vs base vs base vs base vs base Atof64Decimal ~ -0.47% +1.67% -2.70% +1.21% +2.70% Atof64Float ~ +2.48% ~ ~ ~ ~ Atof64FloatExp ~ +1.15% ~ ~ +1.83% +3.87% Atof64Big ~ +4.74% ~ ~ +4.17% +7.30% Atof64RandomBits -27.95% ~ -10.11% -10.54% ~ ~ Atof64RandomFloats ~ +0.75% ~ ~ ~ ~ Atof64RandomLongFloats -6.24% -4.17% -9.60% -8.55% -2.62% -4.52% Atof32Decimal ~ +3.24% +3.76% +3.89% ~ ~ Atof32Float ~ +2.77% ~ ~ ~ ~ Atof32FloatExp ~ +4.02% ~ ~ +2.56% +6.55% Atof32Random -11.97% -10.38% -14.53% -12.41% -5.65% -3.64% Atof32RandomLong -22.77% -22.45% -20.00% -26.52% -13.36% -13.84% AppendFloat/Decimal +10.82% +2.66% -3.00% ~ ~ ~ AppendFloat/Float -14.26% -2.49% -21.84% -10.52% -4.70% -4.84% AppendFloat/Exp +7.69% +6.01% -4.15% ~ -1.04% ~ AppendFloat/NegExp +8.95% +5.75% -4.73% ~ ~ ~ AppendFloat/LongExp -33.48% -29.01% -31.61% -34.75% -23.30% -21.21% AppendFloat/Big -16.91% -27.85% -32.91% -30.01% -18.67% -8.51% AppendFloat/BinaryExp -19.88% -8.47% -9.40% -18.08% -9.40% -9.33% AppendFloat/32Integer +29.68% -11.91% -12.31% -15.43% +5.49% +14.83% AppendFloat/32ExactFraction -14.62% -7.34% -14.28% -13.78% +6.46% +15.68% AppendFloat/32Point ~ -16.51% -30.84% -22.16% ~ +11.04% AppendFloat/32Exp -7.44% -7.72% -8.47% -12.78% +6.13% +18.65% AppendFloat/32NegExp -6.36% -7.58% -10.05% -13.23% +7.69% +18.18% AppendFloat/32Shortest -17.45% -18.23% -18.03% -19.29% +2.68% +11.41% AppendFloat/32Fixed8Hard -13.57% -14.52% -14.91% -16.55% -20.28% -23.51% AppendFloat/32Fixed9Hard -16.06% -16.69% -11.75% -19.42% -5.12% ~ AppendFloat/64Fixed1 -3.25% -9.70% -8.67% -14.11% -13.08% -14.23% AppendFloat/64Fixed2 -1.77% -9.77% -9.40% -12.25% -12.69% -13.32% AppendFloat/64Fixed2.5 -3.46% -6.21% -12.38% -10.86% -10.47% -11.31% AppendFloat/64Fixed3 ~ -9.39% -11.13% -14.39% -14.50% -11.16% AppendFloat/64Fixed4 ~ -11.91% -20.62% -13.40% -19.78% -22.41% AppendFloat/64Fixed5Hard -6.45% -7.36% -13.88% -12.42% -12.31% -12.92% AppendFloat/64Fixed12 -26.39% -23.15% -29.45% -28.11% -24.63% -27.61% AppendFloat/64Fixed16 ~ -15.85% -21.24% -19.76% -24.24% -26.14% AppendFloat/64Fixed12Hard -16.25% -12.77% -18.74% -19.20% -17.08% -18.95% AppendFloat/64Fixed17Hard -16.81% -9.80% -12.77% -17.19% -2.75% +6.06% AppendFloat/64Fixed18Hard ~ -0.76% ~ ~ ~ -26.49% AppendFloat/64FixedF1 +16.15% -12.93% -18.60% -18.24% +1.57% ~ AppendFloat/64FixedF2 -16.83% -9.77% -12.09% -18.43% -13.44% -15.23% AppendFloat/64FixedF3 ~ -5.68% -9.65% -15.14% -8.87% -11.83% AppendFloat/Slowpath64 -33.56% -28.32% -32.45% -33.42% -22.77% -18.76% AppendFloat/SlowpathDenormal64 -31.53% -25.45% -32.60% -25.27% -13.36% -6.95% AppendFloat/ShorterIntervalCase32 -19.52% -14.41% -13.89% -17.03% ~ +12.21% AppendFloat/ShorterIntervalCase64 +14.00% +14.94% +4.06% ~ +9.43% +8.07% AppendUint -33.66% -13.19% -11.52% -13.39% -13.68% -9.04% AppendUintVarlen/digits=1 ~ -4.96% ~ ~ ~ +13.97% AppendUintVarlen/digits=2 +10.01% +2.45% ~ ~ ~ +11.23% AppendUintVarlen/digits=3 -5.10% +0.53% +2.32% +3.74% +18.05% +61.14% AppendUintVarlen/digits=4 ~ +14.32% ~ +6.86% +22.09% +61.28% AppendUintVarlen/digits=5 -18.17% +0.62% +1.13% ~ +13.94% +48.42% AppendUintVarlen/digits=6 -8.74% +7.58% +2.47% +7.86% +17.45% +50.58% AppendUintVarlen/digits=7 -27.30% -2.17% -1.61% ~ +8.31% +37.41% AppendUintVarlen/digits=8 -19.80% +13.49% ~ ~ +19.81% +55.28% AppendUintVarlen/digits=9 -28.86% +3.29% ~ -7.81% +10.42% +39.20% AppendUintVarlen/digits=10 -33.46% -8.00% -12.57% -19.07% -8.59% +7.48% AppendUintVarlen/digits=11 -37.91% -8.32% -11.85% -16.89% -10.14% ~ AppendUintVarlen/digits=12 -28.93% -7.26% -14.27% -23.26% -12.07% ~ AppendUintVarlen/digits=13 -33.20% -8.87% -13.43% -21.56% -8.39% ~ AppendUintVarlen/digits=14 -33.20% -9.49% -13.50% -21.92% -10.43% ~ AppendUintVarlen/digits=15 -36.90% -9.16% -8.48% -16.95% -10.62% ~ AppendUintVarlen/digits=16 -36.20% -8.06% -13.58% -20.92% -6.67% +3.52% AppendUintVarlen/digits=17 -36.15% -7.47% -14.12% -21.53% -6.00% +3.88% AppendUintVarlen/digits=18 -35.85% -7.56% -14.12% -19.66% -9.16% ~ AppendUintVarlen/digits=19 -43.45% -17.14% -20.38% -28.29% -25.25% -16.47% AppendUintVarlen/digits=20 -40.70% -13.60% -18.66% -24.18% -24.69% -17.33% Change-Id: I4eed57cfbf398b5d5327efd749e13610e17153e9 Reviewed-on: https://go-review.googlesource.com/c/go/+/743860 LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Neal Patel <neal@golang.org> Reviewed-by: Neal Patel <nealpatel@google.com>
Add the testdata binaries and the reproducers for #30250 and #30253 to the seed corpus. Change-Id: I12e4f48a9a4bb91f00f1703cd7b2776eb2bf2e5d Reviewed-on: https://go-review.googlesource.com/c/go/+/780360 Reviewed-by: Michael Pratt <mpratt@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com> Reviewed-by: Quim Muntal <quimmuntal@gmail.com>
Generalize FuncSingleAssignment to FuncAssignments, tracking all
non-zero assignments to func-typed variables. When multiple
callees are known, escape analysis merges their parameter escape
behaviors via teeHole.
compilebench results (this CL vs parent):
│ parent.txt │ commit.txt │
│ sec/op │ sec/op vs base │
Template 150.8m ± 12% 149.6m ± 12% ~ (p=0.589 n=6)
Unicode 120.2m ± 11% 117.2m ± 5% ~ (p=0.699 n=6)
GoTypes 890.6m ± 3% 884.8m ± 2% ~ (p=0.310 n=6)
Compiler 161.2m ± 13% 154.4m ± 4% ~ (p=0.180 n=6)
SSA 7.023 ± 1% 7.021 ± 1% ~ (p=0.485 n=6)
Flate 162.6m ± 6% 162.0m ± 6% ~ (p=0.699 n=6)
GoParser 176.9m ± 7% 178.0m ± 6% ~ (p=0.699 n=6)
Reflect 393.2m ± 4% 389.5m ± 5% ~ (p=0.699 n=6)
Tar 171.5m ± 14% 172.9m ± 5% ~ (p=0.937 n=6)
XML 197.1m ± 6% 197.7m ± 2% ~ (p=0.818 n=6)
LinkCompiler 650.5m ± 4% 637.0m ± 2% ~ (p=0.394 n=6)
ExternalLinkCompiler 2.196 ± 1% 2.206 ± 1% ~ (p=0.240 n=6)
LinkWithoutDebugCompiler 421.2m ± 6% 423.4m ± 7% ~ (p=0.699 n=6)
StdCmd 29.21 ± 2% 29.27 ± 1% ~ (p=0.937 n=6)
geomean 525.6m 522.3m -0.63%
│ parent.txt │ commit.txt │
│ user-sec/op │ user-sec/op vs base │
Template 728.1m ± 5% 686.9m ± 11% ~ (p=0.180 n=6)
Unicode 158.9m ± 12% 160.9m ± 8% ~ (p=0.937 n=6)
GoTypes 4.921 ± 4% 4.919 ± 2% ~ (p=0.937 n=6)
Compiler 502.4m ± 9% 505.3m ± 10% ~ (p=0.818 n=6)
SSA 37.77 ± 2% 37.06 ± 4% ~ (p=0.699 n=6)
Flate 751.3m ± 11% 759.7m ± 8% ~ (p=0.937 n=6)
GoParser 677.4m ± 5% 674.4m ± 8% ~ (p=0.394 n=6)
Reflect 1.888 ± 2% 1.901 ± 3% ~ (p=0.589 n=6)
Tar 772.1m ± 14% 789.7m ± 5% ~ (p=0.937 n=6)
XML 911.8m ± 9% 916.0m ± 12% ~ (p=0.937 n=6)
LinkCompiler 1.077 ± 4% 1.059 ± 5% ~ (p=0.310 n=6)
ExternalLinkCompiler 2.539 ± 2% 2.531 ± 3% ~ (p=1.000 n=6)
LinkWithoutDebugCompiler 489.7m ± 4% 485.3m ± 5% ~ (p=1.000 n=6)
geomean 1.192 1.187 -0.36%
│ parent.txt │ commit.txt │
│ text-bytes │ text-bytes vs base │
HelloSize 1.110Mi ± 0% 1.110Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 14.14Mi ± 0% 14.14Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 3.961Mi 3.961Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ data-bytes │ data-bytes vs base │
HelloSize 27.54Ki ± 0% 27.54Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 431.4Ki ± 0% 431.4Ki ± 0% ~ (p=1.000 n=6) ¹
geomean 109.0Ki 109.0Ki +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ bss-bytes │ bss-bytes vs base │
HelloSize 213.9Ki ± 0% 213.9Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 32.27Mi ± 0% 32.27Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 2.597Mi 2.597Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ exe-bytes │ exe-bytes vs base │
HelloSize 1.782Mi ± 0% 1.782Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 21.29Mi ± 0% 21.29Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 6.158Mi 6.158Mi +0.00%
¹ all samples are equal
Fixes #73132
Change-Id: Ice3ed672db28dbfd4bdc788019111d5d3092c5bb
Reviewed-on: https://go-review.googlesource.com/c/go/+/771500
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
The standalone ir.FuncAssignments walked the entire enclosing
function on every call. Move it into ReassignOracle, which collects
the same information in a single walk during Init alongside the
existing singleDef analysis.
compilebench results (this CL vs parent):
│ parent.txt │ commit.txt │
│ sec/op │ sec/op vs base │
Template 153.1m ± 16% 149.5m ± 7% ~ (p=0.699 n=6)
Unicode 115.0m ± 15% 116.1m ± 8% ~ (p=0.310 n=6)
GoTypes 876.2m ± 2% 872.1m ± 2% ~ (p=0.699 n=6)
Compiler 156.4m ± 9% 158.2m ± 8% ~ (p=0.818 n=6)
SSA 7.044 ± 1% 7.044 ± 1% ~ (p=0.937 n=6)
Flate 172.5m ± 4% 165.7m ± 11% ~ (p=0.132 n=6)
GoParser 168.0m ± 3% 169.5m ± 13% ~ (p=0.937 n=6)
Reflect 394.2m ± 3% 391.3m ± 4% ~ (p=0.937 n=6)
Tar 167.1m ± 13% 167.9m ± 15% ~ (p=0.589 n=6)
XML 200.0m ± 7% 197.5m ± 15% ~ (p=1.000 n=6)
LinkCompiler 638.3m ± 3% 631.2m ± 5% ~ (p=0.699 n=6)
ExternalLinkCompiler 2.197 ± 2% 2.192 ± 1% ~ (p=0.699 n=6)
LinkWithoutDebugCompiler 422.2m ± 4% 425.8m ± 3% ~ (p=0.093 n=6)
StdCmd 29.01 ± 2% 28.73 ± 1% ~ (p=0.394 n=6)
geomean 522.0m 519.4m -0.50%
│ parent.txt │ commit.txt │
│ user-sec/op │ user-sec/op vs base │
Template 727.0m ± 15% 692.6m ± 8% ~ (p=1.000 n=6)
Unicode 159.7m ± 6% 161.9m ± 6% ~ (p=0.589 n=6)
GoTypes 4.884 ± 1% 4.907 ± 1% ~ (p=0.937 n=6)
Compiler 470.5m ± 9% 476.9m ± 11% ~ (p=0.699 n=6)
SSA 37.17 ± 2% 37.40 ± 2% ~ (p=0.132 n=6)
Flate 776.9m ± 9% 749.1m ± 16% ~ (p=0.699 n=6)
GoParser 688.5m ± 5% 676.7m ± 6% ~ (p=0.937 n=6)
Reflect 1.861 ± 2% 1.910 ± 3% ~ (p=0.065 n=6)
Tar 750.6m ± 11% 762.4m ± 23% ~ (p=0.937 n=6)
XML 947.8m ± 13% 926.4m ± 12% ~ (p=0.485 n=6)
LinkCompiler 1.035 ± 6% 1.037 ± 5% ~ (p=0.818 n=6)
ExternalLinkCompiler 2.505 ± 2% 2.506 ± 2% ~ (p=0.699 n=6)
LinkWithoutDebugCompiler 479.0m ± 6% 490.4m ± 3% ~ (p=0.394 n=6)
geomean 1.181 1.180 -0.15%
│ parent.txt │ commit.txt │
│ text-bytes │ text-bytes vs base │
HelloSize 1.110Mi ± 0% 1.110Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 14.14Mi ± 0% 14.14Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 3.961Mi 3.961Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ data-bytes │ data-bytes vs base │
HelloSize 27.54Ki ± 0% 27.54Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 431.4Ki ± 0% 431.4Ki ± 0% ~ (p=1.000 n=6) ¹
geomean 109.0Ki 109.0Ki +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ bss-bytes │ bss-bytes vs base │
HelloSize 213.9Ki ± 0% 213.9Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 32.27Mi ± 0% 32.27Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 2.597Mi 2.597Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ exe-bytes │ exe-bytes vs base │
HelloSize 1.782Mi ± 0% 1.782Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 21.29Mi ± 0% 21.29Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 6.158Mi 6.158Mi +0.00%
¹ all samples are equal
Change-Id: Iba7c593a31946fb17450106d1ee0ba9e28eedba3
Reviewed-on: https://go-review.googlesource.com/c/go/+/771161
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
…ysis
Split out of CL 771163; this solves the TODO and saves a double-walk,
without requiring extra fixes needed to use it more broadly.
compilebench results (this CL vs parent):
│ parent.txt │ commit.txt │
│ sec/op │ sec/op vs base │
Template 153.9m ± 9% 157.1m ± 5% ~ (p=0.818 n=6)
Unicode 116.7m ± 7% 115.8m ± 10% ~ (p=0.699 n=6)
GoTypes 879.9m ± 3% 876.5m ± 2% ~ (p=0.937 n=6)
Compiler 165.3m ± 8% 160.5m ± 8% ~ (p=0.699 n=6)
SSA 6.998 ± 0% 7.097 ± 1% +1.42% (p=0.002 n=6)
Flate 166.1m ± 10% 163.3m ± 6% ~ (p=0.310 n=6)
GoParser 170.3m ± 6% 169.3m ± 5% ~ (p=0.937 n=6)
Reflect 388.3m ± 5% 390.2m ± 2% ~ (p=0.937 n=6)
Tar 173.7m ± 3% 173.6m ± 5% ~ (p=0.818 n=6)
XML 197.6m ± 9% 194.6m ± 8% ~ (p=0.485 n=6)
LinkCompiler 651.5m ± 7% 637.9m ± 3% ~ (p=0.394 n=6)
ExternalLinkCompiler 2.194 ± 1% 2.193 ± 2% ~ (p=0.937 n=6)
LinkWithoutDebugCompiler 419.0m ± 6% 412.4m ± 3% ~ (p=0.589 n=6)
StdCmd 29.02 ± 2% 29.36 ± 4% ~ (p=0.937 n=6)
geomean 524.6m 522.2m -0.47%
│ parent.txt │ commit.txt │
│ user-sec/op │ user-sec/op vs base │
Template 727.8m ± 12% 736.0m ± 8% ~ (p=0.589 n=6)
Unicode 156.5m ± 8% 162.3m ± 8% ~ (p=0.310 n=6)
GoTypes 4.889 ± 2% 4.898 ± 2% ~ (p=0.937 n=6)
Compiler 505.5m ± 6% 491.1m ± 4% ~ (p=0.394 n=6)
SSA 36.98 ± 1% 37.35 ± 3% ~ (p=0.093 n=6)
Flate 797.3m ± 8% 761.3m ± 13% ~ (p=0.180 n=6)
GoParser 677.9m ± 5% 695.3m ± 7% ~ (p=0.394 n=6)
Reflect 1.871 ± 8% 1.875 ± 2% ~ (p=1.000 n=6)
Tar 763.1m ± 10% 825.0m ± 15% ~ (p=0.485 n=6)
XML 915.9m ± 15% 877.2m ± 14% ~ (p=0.065 n=6)
LinkCompiler 1.035 ± 5% 1.059 ± 5% ~ (p=0.394 n=6)
ExternalLinkCompiler 2.488 ± 3% 2.509 ± 3% ~ (p=0.485 n=6)
LinkWithoutDebugCompiler 477.8m ± 2% 478.1m ± 6% ~ (p=0.937 n=6)
geomean 1.185 1.192 +0.60%
│ parent.txt │ commit.txt │
│ text-bytes │ text-bytes vs base │
HelloSize 1.110Mi ± 0% 1.110Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 14.14Mi ± 0% 14.14Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 3.961Mi 3.961Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ data-bytes │ data-bytes vs base │
HelloSize 27.54Ki ± 0% 27.54Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 431.4Ki ± 0% 431.4Ki ± 0% ~ (p=1.000 n=6) ¹
geomean 109.0Ki 109.0Ki +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ bss-bytes │ bss-bytes vs base │
HelloSize 213.9Ki ± 0% 213.9Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 32.27Mi ± 0% 32.27Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 2.597Mi 2.597Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ exe-bytes │ exe-bytes vs base │
HelloSize 1.782Mi ± 0% 1.782Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 21.29Mi ± 0% 21.29Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 6.158Mi 6.158Mi +0.00%
¹ all samples are equal
Change-Id: If4d51835045e27e5f67bac901f9edd1dc4da9b97
Reviewed-on: https://go-review.googlesource.com/c/go/+/778300
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Teach ReassignOracle.StaticValue to peel OPAREN nodes, matching the
package-level ir.StaticValue helper.
This is a small parity step needed before more callers can switch from
ir.StaticValue to ReassignOracle.StaticValue.
Passes toolstash -cmp.
compilebench results (this CL vs parent):
│ parent.txt │ commit.txt │
│ sec/op │ sec/op vs base │
Template 154.2m ± 5% 153.6m ± 5% ~ (p=0.485 n=6)
Unicode 114.7m ± 2% 117.2m ± 4% ~ (p=0.310 n=6)
GoTypes 897.8m ± 2% 878.8m ± 5% ~ (p=0.310 n=6)
Compiler 160.2m ± 22% 159.0m ± 6% ~ (p=0.310 n=6)
SSA 7.171 ± 4% 7.136 ± 1% ~ (p=0.485 n=6)
Flate 171.2m ± 11% 161.8m ± 5% ~ (p=0.065 n=6)
GoParser 172.5m ± 6% 172.9m ± 18% ~ (p=0.589 n=6)
Reflect 398.9m ± 11% 391.8m ± 13% ~ (p=0.394 n=6)
Tar 173.2m ± 4% 168.8m ± 3% ~ (p=0.093 n=6)
XML 201.6m ± 23% 194.2m ± 3% ~ (p=0.240 n=6)
LinkCompiler 639.2m ± 3% 646.7m ± 4% ~ (p=0.937 n=6)
ExternalLinkCompiler 2.204 ± 2% 2.196 ± 2% ~ (p=0.699 n=6)
LinkWithoutDebugCompiler 423.5m ± 4% 415.7m ± 13% ~ (p=0.937 n=6)
StdCmd 29.73 ± 3% 29.41 ± 2% ~ (p=0.589 n=6)
geomean 528.6m 522.2m -1.22%
│ parent.txt │ commit.txt │
│ user-sec/op │ user-sec/op vs base │
Template 685.3m ± 8% 746.5m ± 16% ~ (p=0.132 n=6)
Unicode 158.6m ± 5% 156.5m ± 9% ~ (p=0.589 n=6)
GoTypes 5.001 ± 2% 4.900 ± 1% -2.01% (p=0.009 n=6)
Compiler 511.1m ± 4% 500.2m ± 9% ~ (p=0.699 n=6)
SSA 37.50 ± 3% 37.34 ± 1% ~ (p=0.394 n=6)
Flate 793.4m ± 10% 735.2m ± 8% ~ (p=0.180 n=6)
GoParser 698.0m ± 4% 696.3m ± 8% ~ (p=0.818 n=6)
Reflect 1.895 ± 5% 1.897 ± 2% ~ (p=0.937 n=6)
Tar 794.7m ± 13% 764.9m ± 10% ~ (p=0.394 n=6)
XML 936.5m ± 19% 865.7m ± 5% -7.57% (p=0.009 n=6)
LinkCompiler 1.054 ± 7% 1.077 ± 8% ~ (p=0.485 n=6)
ExternalLinkCompiler 2.498 ± 2% 2.515 ± 3% ~ (p=0.937 n=6)
LinkWithoutDebugCompiler 485.8m ± 5% 482.1m ± 14% ~ (p=1.000 n=6)
geomean 1.197 1.184 -1.13%
│ parent.txt │ commit.txt │
│ text-bytes │ text-bytes vs base │
HelloSize 1.110Mi ± 0% 1.110Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 14.14Mi ± 0% 14.14Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 3.961Mi 3.961Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ data-bytes │ data-bytes vs base │
HelloSize 27.49Ki ± 0% 27.49Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 431.4Ki ± 0% 431.4Ki ± 0% ~ (p=1.000 n=6) ¹
geomean 108.9Ki 108.9Ki +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ bss-bytes │ bss-bytes vs base │
HelloSize 213.9Ki ± 0% 213.9Ki ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 32.27Mi ± 0% 32.27Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 2.597Mi 2.597Mi +0.00%
¹ all samples are equal
│ parent.txt │ commit.txt │
│ exe-bytes │ exe-bytes vs base │
HelloSize 1.777Mi ± 0% 1.777Mi ± 0% ~ (p=1.000 n=6) ¹
CmdGoSize 21.29Mi ± 0% 21.29Mi ± 0% ~ (p=1.000 n=6) ¹
geomean 6.150Mi 6.150Mi +0.00%
¹ all samples are equal
Change-Id: I31c77871939a057415d2b99b051b28624f88e9d2
Reviewed-on: https://go-review.googlesource.com/c/go/+/778800
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
In the no scan reuse path, goto the postMallocgc call in the end rather
than duplicating it. This only has an effect in the no scan slow path
function, but adds some noise to the generated files: we put
nextFree/nextFreeFast into a block so that there are no new variables we
jump over. Also the label has a goto before it so the compiler doesn't
complain if we don't use the label, and a nop statement after it so that
the call to postMallocgc doesn't appear in a labeledstatement, but is
instead in a block, where mkmalloc expects it.
goos: linux
goarch: amd64
pkg: runtime
cpu: Intel(R) Xeon(R) Platinum 8481C CPU @ 2.70GHz
│ baseline │ experiment │
│ sec/op │ sec/op vs base │
Mallocgc/scan=noscan/size=1/kind=new-88 4.701n ± 0% 4.695n ± 0% -0.14% (p=0.049 n=10)
Mallocgc/scan=noscan/size=1/kind=mallocgc-88 6.976n ± 0% 6.953n ± 0% -0.33% (p=0.000 n=10)
Mallocgc/scan=noscan/size=2/kind=new-88 5.113n ± 0% 5.074n ± 0% -0.75% (p=0.000 n=10)
Mallocgc/scan=noscan/size=2/kind=mallocgc-88 6.646n ± 0% 6.611n ± 0% -0.54% (p=0.000 n=10)
Mallocgc/scan=noscan/size=3/kind=new-88 6.304n ± 8% 6.271n ± 1% -0.53% (p=0.000 n=10)
Mallocgc/scan=noscan/size=3/kind=mallocgc-88 7.528n ± 0% 7.479n ± 0% -0.65% (p=0.000 n=10)
Mallocgc/scan=noscan/size=4/kind=new-88 6.556n ± 0% 6.496n ± 0% -0.92% (p=0.000 n=10)
Mallocgc/scan=noscan/size=4/kind=mallocgc-88 7.901n ± 0% 7.841n ± 0% -0.76% (p=0.000 n=10)
Mallocgc/scan=noscan/size=5/kind=new-88 7.778n ± 1% 7.707n ± 1% -0.92% (p=0.008 n=10)
Mallocgc/scan=noscan/size=5/kind=mallocgc-88 9.588n ± 2% 9.046n ± 0% -5.65% (p=0.000 n=10)
Mallocgc/scan=noscan/size=6/kind=new-88 9.366n ± 0% 9.280n ± 0% -0.91% (p=0.000 n=10)
Mallocgc/scan=noscan/size=6/kind=mallocgc-88 10.80n ± 0% 10.70n ± 0% -0.97% (p=0.000 n=10)
Mallocgc/scan=noscan/size=7/kind=new-88 9.737n ± 0% 9.642n ± 0% -0.98% (p=0.000 n=10)
Mallocgc/scan=noscan/size=7/kind=mallocgc-88 11.01n ± 0% 10.92n ± 0% -0.82% (p=0.000 n=10)
Mallocgc/scan=noscan/size=8/kind=new-88 9.149n ± 0% 9.081n ± 0% -0.74% (p=0.000 n=10)
Mallocgc/scan=noscan/size=8/kind=mallocgc-88 10.51n ± 0% 10.45n ± 1% -0.57% (p=0.000 n=10)
Mallocgc/scan=noscan/size=9/kind=new-88 15.12n ± 1% 14.38n ± 1% -4.89% (p=0.000 n=10)
Mallocgc/scan=noscan/size=9/kind=mallocgc-88 16.14n ± 4% 15.90n ± 0% -1.52% (p=0.000 n=10)
Mallocgc/scan=noscan/size=10/kind=new-88 14.34n ± 1% 14.13n ± 1% -1.46% (p=0.000 n=10)
Mallocgc/scan=noscan/size=10/kind=mallocgc-88 15.56n ± 1% 15.38n ± 0% -1.16% (p=0.000 n=10)
Mallocgc/scan=noscan/size=11/kind=new-88 14.53n ± 0% 14.29n ± 0% -1.62% (p=0.000 n=10)
Mallocgc/scan=noscan/size=11/kind=mallocgc-88 15.97n ± 0% 15.84n ± 0% -0.81% (p=0.000 n=10)
Mallocgc/scan=noscan/size=12/kind=new-88 14.31n ± 1% 14.05n ± 1% -1.82% (p=0.000 n=10)
Mallocgc/scan=noscan/size=12/kind=mallocgc-88 15.55n ± 0% 15.32n ± 0% -1.51% (p=0.000 n=10)
Mallocgc/scan=noscan/size=13/kind=new-88 14.59n ± 0% 14.33n ± 0% -1.78% (p=0.000 n=10)
Mallocgc/scan=noscan/size=13/kind=mallocgc-88 16.09n ± 0% 15.80n ± 1% -1.83% (p=0.000 n=10)
Mallocgc/scan=noscan/size=14/kind=new-88 14.59n ± 0% 14.11n ± 1% -3.26% (p=0.000 n=10)
Mallocgc/scan=noscan/size=14/kind=mallocgc-88 15.74n ± 1% 15.30n ± 1% -2.86% (p=0.000 n=10)
Mallocgc/scan=noscan/size=15/kind=new-88 14.59n ± 1% 14.44n ± 1% -0.99% (p=0.000 n=10)
Mallocgc/scan=noscan/size=15/kind=mallocgc-88 16.03n ± 1% 15.85n ± 0% -1.12% (p=0.000 n=10)
Mallocgc/scan=noscan/size=16/kind=new-88 13.35n ± 0% 13.15n ± 0% -1.50% (p=0.000 n=10)
Mallocgc/scan=noscan/size=16/kind=mallocgc-88 14.52n ± 0% 14.28n ± 0% -1.62% (p=0.000 n=10)
Mallocgc/scan=noscan/size=24/kind=new-88 16.06n ± 0% 15.86n ± 1% -1.25% (p=0.000 n=10)
Mallocgc/scan=noscan/size=24/kind=mallocgc-88 17.36n ± 1% 17.11n ± 1% -1.41% (p=0.000 n=10)
Mallocgc/scan=noscan/size=32/kind=new-88 18.87n ± 1% 18.38n ± 1% -2.60% (p=0.000 n=10)
Mallocgc/scan=noscan/size=32/kind=mallocgc-88 20.04n ± 0% 19.76n ± 1% -1.40% (p=0.000 n=10)
Mallocgc/scan=noscan/size=48/kind=new-88 24.79n ± 1% 24.33n ± 1% -1.86% (p=0.000 n=10)
Mallocgc/scan=noscan/size=48/kind=mallocgc-88 25.49n ± 1% 25.07n ± 1% -1.61% (p=0.000 n=10)
Mallocgc/scan=noscan/size=64/kind=new-88 31.02n ± 1% 30.30n ± 1% -2.34% (p=0.000 n=10)
Mallocgc/scan=noscan/size=64/kind=mallocgc-88 31.75n ± 1% 31.09n ± 1% -2.09% (p=0.000 n=10)
Mallocgc/scan=noscan/size=80/kind=new-88 37.26n ± 1% 36.43n ± 0% -2.23% (p=0.000 n=10)
Mallocgc/scan=noscan/size=80/kind=mallocgc-88 37.57n ± 1% 36.49n ± 1% -2.87% (p=0.000 n=10)
Mallocgc/scan=noscan/size=96/kind=new-88 43.11n ± 1% 42.17n ± 1% -2.18% (p=0.000 n=10)
Mallocgc/scan=noscan/size=96/kind=mallocgc-88 44.12n ± 1% 42.95n ± 1% -2.65% (p=0.000 n=10)
Mallocgc/scan=noscan/size=112/kind=new-88 50.54n ± 1% 48.53n ± 1% -3.99% (p=0.000 n=10)
Mallocgc/scan=noscan/size=112/kind=mallocgc-88 50.53n ± 2% 48.51n ± 1% -4.01% (p=0.000 n=10)
Mallocgc/scan=noscan/size=128/kind=new-88 56.10n ± 1% 54.55n ± 1% -2.76% (p=0.000 n=10)
Mallocgc/scan=noscan/size=128/kind=mallocgc-88 56.66n ± 1% 54.72n ± 1% -3.42% (p=0.000 n=10)
Mallocgc/scan=noscan/size=144/kind=new-88 64.96n ± 1% 63.57n ± 1% -2.15% (p=0.000 n=10)
Mallocgc/scan=noscan/size=144/kind=mallocgc-88 64.61n ± 1% 63.43n ± 1% -1.83% (p=0.000 n=10)
Mallocgc/scan=noscan/size=160/kind=new-88 71.78n ± 1% 70.39n ± 1% -1.94% (p=0.000 n=10)
Mallocgc/scan=noscan/size=160/kind=mallocgc-88 71.64n ± 1% 69.60n ± 1% -2.85% (p=0.000 n=10)
Mallocgc/scan=noscan/size=176/kind=new-88 79.03n ± 1% 77.52n ± 1% -1.92% (p=0.000 n=10)
Mallocgc/scan=noscan/size=176/kind=mallocgc-88 78.59n ± 1% 77.16n ± 1% -1.83% (p=0.000 n=10)
Mallocgc/scan=noscan/size=192/kind=new-88 83.71n ± 1% 82.55n ± 1% -1.39% (p=0.005 n=10)
Mallocgc/scan=noscan/size=192/kind=mallocgc-88 83.21n ± 1% 82.48n ± 1% -0.88% (p=0.002 n=10)
Mallocgc/scan=noscan/size=208/kind=new-88 92.03n ± 1% 90.84n ± 1% -1.29% (p=0.002 n=10)
Mallocgc/scan=noscan/size=208/kind=mallocgc-88 91.67n ± 1% 90.78n ± 2% -0.97% (p=0.015 n=10)
Mallocgc/scan=noscan/size=224/kind=new-88 96.05n ± 1% 96.22n ± 1% ~ (p=0.529 n=10)
Mallocgc/scan=noscan/size=224/kind=mallocgc-88 96.02n ± 1% 94.05n ± 2% -2.05% (p=0.001 n=10)
Mallocgc/scan=noscan/size=240/kind=new-88 103.9n ± 1% 102.9n ± 2% ~ (p=0.060 n=10)
Mallocgc/scan=noscan/size=240/kind=mallocgc-88 104.0n ± 0% 102.9n ± 1% -1.15% (p=0.000 n=10)
Mallocgc/scan=noscan/size=256/kind=new-88 111.2n ± 1% 109.0n ± 1% -1.98% (p=0.000 n=10)
Mallocgc/scan=noscan/size=256/kind=mallocgc-88 110.2n ± 1% 109.1n ± 1% -1.00% (p=0.002 n=10)
Mallocgc/scan=noscan/size=288/kind=new-88 122.2n ± 1% 120.7n ± 2% -1.19% (p=0.000 n=10)
Mallocgc/scan=noscan/size=288/kind=mallocgc-88 121.0n ± 1% 119.4n ± 1% -1.40% (p=0.001 n=10)
Mallocgc/scan=noscan/size=320/kind=new-88 135.2n ± 1% 132.8n ± 1% -1.74% (p=0.000 n=10)
Mallocgc/scan=noscan/size=320/kind=mallocgc-88 133.8n ± 1% 132.9n ± 1% -0.64% (p=0.011 n=10)
Mallocgc/scan=noscan/size=352/kind=new-88 151.9n ± 1% 151.1n ± 1% -0.56% (p=0.008 n=10)
Mallocgc/scan=noscan/size=352/kind=mallocgc-88 151.1n ± 1% 151.3n ± 1% ~ (p=0.424 n=10)
Mallocgc/scan=noscan/size=384/kind=new-88 159.4n ± 1% 157.5n ± 1% -1.22% (p=0.000 n=10)
Mallocgc/scan=noscan/size=384/kind=mallocgc-88 158.3n ± 1% 157.8n ± 2% ~ (p=0.148 n=10)
Mallocgc/scan=noscan/size=416/kind=new-88 174.0n ± 1% 172.2n ± 2% ~ (p=0.305 n=10)
Mallocgc/scan=noscan/size=416/kind=mallocgc-88 174.7n ± 1% 173.3n ± 1% -0.80% (p=0.022 n=10)
Mallocgc/scan=noscan/size=448/kind=new-88 183.0n ± 1% 182.3n ± 1% ~ (p=0.110 n=10)
Mallocgc/scan=noscan/size=448/kind=mallocgc-88 182.7n ± 1% 180.4n ± 1% -1.23% (p=0.007 n=10)
Mallocgc/scan=noscan/size=480/kind=new-88 205.5n ± 1% 204.6n ± 1% ~ (p=0.101 n=10)
Mallocgc/scan=noscan/size=480/kind=mallocgc-88 204.4n ± 1% 205.0n ± 1% ~ (p=0.077 n=10)
Mallocgc/scan=noscan/size=512/kind=new-88 218.4n ± 1% 217.6n ± 1% ~ (p=0.643 n=10)
Mallocgc/scan=noscan/size=512/kind=mallocgc-88 217.6n ± 1% 215.5n ± 1% -0.97% (p=0.004 n=10)
Mallocgc/scan=scan/size=8/kind=new-88 10.93n ± 0% 10.88n ± 0% -0.41% (p=0.006 n=10)
Mallocgc/scan=scan/size=8/kind=mallocgc-88 11.94n ± 0% 11.85n ± 0% -0.75% (p=0.000 n=10)
Mallocgc/scan=scan/size=16/kind=new-88 14.81n ± 1% 15.03n ± 1% +1.49% (p=0.000 n=10)
Mallocgc/scan=scan/size=16/kind=mallocgc-88 16.12n ± 0% 16.05n ± 1% -0.37% (p=0.003 n=10)
Mallocgc/scan=scan/size=24/kind=new-88 18.23n ± 1% 18.23n ± 1% ~ (p=0.954 n=10)
Mallocgc/scan=scan/size=24/kind=mallocgc-88 19.46n ± 0% 19.30n ± 1% -0.80% (p=0.014 n=10)
Mallocgc/scan=scan/size=32/kind=new-88 20.48n ± 0% 20.22n ± 0% -1.25% (p=0.000 n=10)
Mallocgc/scan=scan/size=32/kind=mallocgc-88 21.72n ± 1% 21.47n ± 1% -1.13% (p=0.000 n=10)
Mallocgc/scan=scan/size=48/kind=new-88 26.56n ± 1% 26.03n ± 0% -2.01% (p=0.000 n=10)
Mallocgc/scan=scan/size=48/kind=mallocgc-88 27.96n ± 1% 27.61n ± 1% -1.25% (p=0.000 n=10)
Mallocgc/scan=scan/size=64/kind=new-88 31.95n ± 1% 31.37n ± 1% -1.83% (p=0.001 n=10)
Mallocgc/scan=scan/size=64/kind=mallocgc-88 32.48n ± 2% 32.16n ± 1% -1.00% (p=0.005 n=10)
Mallocgc/scan=scan/size=80/kind=new-88 39.54n ± 2% 38.49n ± 1% -2.64% (p=0.000 n=10)
Mallocgc/scan=scan/size=80/kind=mallocgc-88 40.04n ± 2% 39.25n ± 1% -1.99% (p=0.000 n=10)
Mallocgc/scan=scan/size=96/kind=new-88 44.99n ± 1% 44.51n ± 1% -1.07% (p=0.025 n=10)
Mallocgc/scan=scan/size=96/kind=mallocgc-88 45.75n ± 1% 45.28n ± 2% -1.05% (p=0.003 n=10)
Mallocgc/scan=scan/size=112/kind=new-88 52.31n ± 2% 50.68n ± 1% -3.12% (p=0.000 n=10)
Mallocgc/scan=scan/size=112/kind=mallocgc-88 51.62n ± 1% 50.62n ± 1% -1.93% (p=0.000 n=10)
Mallocgc/scan=scan/size=128/kind=new-88 56.66n ± 1% 56.33n ± 1% ~ (p=0.052 n=10)
Mallocgc/scan=scan/size=128/kind=mallocgc-88 57.20n ± 1% 56.30n ± 1% -1.57% (p=0.000 n=10)
Mallocgc/scan=scan/size=144/kind=new-88 68.48n ± 2% 67.41n ± 2% -1.57% (p=0.003 n=10)
Mallocgc/scan=scan/size=144/kind=mallocgc-88 67.55n ± 1% 66.06n ± 1% -2.21% (p=0.000 n=10)
Mallocgc/scan=scan/size=160/kind=new-88 75.42n ± 1% 73.86n ± 2% -2.07% (p=0.000 n=10)
Mallocgc/scan=scan/size=160/kind=mallocgc-88 74.39n ± 1% 73.36n ± 1% -1.38% (p=0.003 n=10)
Mallocgc/scan=scan/size=176/kind=new-88 80.22n ± 1% 79.15n ± 1% -1.34% (p=0.008 n=10)
Mallocgc/scan=scan/size=176/kind=mallocgc-88 79.47n ± 1% 79.03n ± 2% ~ (p=0.110 n=10)
Mallocgc/scan=scan/size=192/kind=new-88 86.95n ± 1% 85.90n ± 1% -1.21% (p=0.034 n=10)
Mallocgc/scan=scan/size=192/kind=mallocgc-88 85.85n ± 1% 85.34n ± 1% -0.59% (p=0.004 n=10)
Mallocgc/scan=scan/size=208/kind=new-88 92.53n ± 1% 92.22n ± 1% ~ (p=0.105 n=10)
Mallocgc/scan=scan/size=208/kind=mallocgc-88 92.27n ± 1% 92.41n ± 2% ~ (p=0.912 n=10)
Mallocgc/scan=scan/size=224/kind=new-88 101.0n ± 1% 102.4n ± 1% +1.34% (p=0.006 n=10)
Mallocgc/scan=scan/size=224/kind=mallocgc-88 99.61n ± 1% 99.37n ± 1% ~ (p=0.393 n=10)
Mallocgc/scan=scan/size=240/kind=new-88 104.7n ± 1% 105.5n ± 1% ~ (p=0.100 n=10)
Mallocgc/scan=scan/size=240/kind=mallocgc-88 104.4n ± 1% 105.0n ± 1% ~ (p=0.403 n=10)
Mallocgc/scan=scan/size=256/kind=new-88 110.2n ± 1% 110.6n ± 1% ~ (p=0.362 n=10)
Mallocgc/scan=scan/size=256/kind=mallocgc-88 110.2n ± 1% 110.5n ± 1% ~ (p=0.592 n=10)
Mallocgc/scan=scan/size=288/kind=new-88 127.4n ± 2% 127.0n ± 1% ~ (p=0.492 n=10)
Mallocgc/scan=scan/size=288/kind=mallocgc-88 126.4n ± 1% 126.3n ± 1% ~ (p=0.755 n=10)
Mallocgc/scan=scan/size=320/kind=new-88 140.4n ± 1% 139.8n ± 1% ~ (p=0.361 n=10)
Mallocgc/scan=scan/size=320/kind=mallocgc-88 141.1n ± 1% 140.9n ± 1% ~ (p=0.927 n=10)
Mallocgc/scan=scan/size=352/kind=new-88 152.6n ± 1% 152.6n ± 1% ~ (p=0.927 n=10)
Mallocgc/scan=scan/size=352/kind=mallocgc-88 152.9n ± 1% 151.8n ± 1% ~ (p=0.066 n=10)
Mallocgc/scan=scan/size=384/kind=new-88 168.2n ± 1% 165.6n ± 1% -1.58% (p=0.002 n=10)
Mallocgc/scan=scan/size=384/kind=mallocgc-88 166.8n ± 1% 165.3n ± 1% -0.93% (p=0.018 n=10)
Mallocgc/scan=scan/size=416/kind=new-88 174.6n ± 1% 173.8n ± 1% ~ (p=0.101 n=10)
Mallocgc/scan=scan/size=416/kind=mallocgc-88 174.4n ± 1% 173.7n ± 1% ~ (p=0.424 n=10)
Mallocgc/scan=scan/size=448/kind=new-88 194.5n ± 1% 195.7n ± 1% +0.59% (p=0.012 n=10)
Mallocgc/scan=scan/size=448/kind=mallocgc-88 194.0n ± 1% 194.6n ± 1% ~ (p=0.286 n=10)
Mallocgc/scan=scan/size=480/kind=new-88 204.3n ± 1% 204.0n ± 1% ~ (p=0.810 n=10)
Mallocgc/scan=scan/size=480/kind=mallocgc-88 204.1n ± 1% 203.8n ± 0% ~ (p=0.616 n=10)
Mallocgc/scan=scan/size=512/kind=new-88 217.8n ± 1% 217.0n ± 1% ~ (p=0.210 n=10)
Mallocgc/scan=scan/size=512/kind=mallocgc-88 218.7n ± 1% 215.4n ± 1% -1.51% (p=0.000 n=10)
Mallocgc/scan=scanslice/size=8-88 13.33n ± 0% 13.20n ± 0% -0.98% (p=0.000 n=10)
Mallocgc/scan=scanslice/size=16-88 22.67n ± 0% 22.55n ± 0% -0.53% (p=0.000 n=10)
Mallocgc/scan=scanslice/size=24-88 25.52n ± 0% 25.37n ± 0% -0.61% (p=0.000 n=10)
Mallocgc/scan=scanslice/size=32-88 27.77n ± 0% 27.83n ± 1% ~ (p=0.382 n=10)
Mallocgc/scan=scanslice/size=48-88 33.97n ± 0% 33.80n ± 1% ~ (p=0.085 n=10)
Mallocgc/scan=scanslice/size=64-88 38.23n ± 1% 37.95n ± 1% -0.73% (p=0.050 n=10)
Mallocgc/scan=scanslice/size=80-88 45.38n ± 0% 45.17n ± 1% -0.47% (p=0.004 n=10)
Mallocgc/scan=scanslice/size=96-88 50.79n ± 0% 50.56n ± 0% ~ (p=0.063 n=10)
Mallocgc/scan=scanslice/size=112-88 55.70n ± 1% 55.92n ± 1% ~ (p=0.424 n=10)
Mallocgc/scan=scanslice/size=128-88 60.19n ± 1% 59.73n ± 1% ~ (p=0.060 n=10)
Mallocgc/scan=scanslice/size=144-88 69.30n ± 1% 67.95n ± 1% -1.94% (p=0.000 n=10)
Mallocgc/scan=scanslice/size=160-88 74.91n ± 1% 74.58n ± 1% ~ (p=0.052 n=10)
Mallocgc/scan=scanslice/size=176-88 80.61n ± 1% 79.55n ± 1% -1.31% (p=0.004 n=10)
Mallocgc/scan=scanslice/size=192-88 87.58n ± 1% 86.90n ± 1% ~ (p=0.105 n=10)
Mallocgc/scan=scanslice/size=208-88 93.00n ± 1% 92.48n ± 1% -0.55% (p=0.022 n=10)
Mallocgc/scan=scanslice/size=224-88 101.15n ± 1% 99.08n ± 1% -2.05% (p=0.001 n=10)
Mallocgc/scan=scanslice/size=240-88 105.3n ± 1% 104.0n ± 1% -1.19% (p=0.002 n=10)
Mallocgc/scan=scanslice/size=256-88 110.2n ± 1% 110.2n ± 1% ~ (p=0.810 n=10)
Mallocgc/scan=scanslice/size=288-88 127.4n ± 1% 127.0n ± 1% ~ (p=0.446 n=10)
Mallocgc/scan=scanslice/size=320-88 142.6n ± 2% 141.2n ± 2% ~ (p=0.072 n=10)
Mallocgc/scan=scanslice/size=352-88 153.8n ± 1% 152.3n ± 1% -0.98% (p=0.005 n=10)
Mallocgc/scan=scanslice/size=384-88 168.0n ± 1% 166.4n ± 1% ~ (p=0.060 n=10)
Mallocgc/scan=scanslice/size=416-88 175.1n ± 2% 174.3n ± 1% ~ (p=0.127 n=10)
Mallocgc/scan=scanslice/size=448-88 196.5n ± 1% 196.4n ± 1% ~ (p=0.754 n=10)
Mallocgc/scan=scanslice/size=480-88 204.0n ± 2% 204.3n ± 1% ~ (p=0.446 n=10)
Mallocgc/scan=scanslice/size=512-88 218.2n ± 1% 217.4n ± 1% ~ (p=0.423 n=10)
geomean 50.01n 49.45n -1.11%
For #79286
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64_c2s16-perf_vs_parent,gotip-linux-amd64_c3h88-perf_vs_parent,gotip-linux-arm64_c4ah72-perf_vs_parent,gotip-linux-arm64_c4as16-perf_vs_parent
Change-Id: Ie5fe8007733ae94556792146d1cd7c666a6a6964
Reviewed-on: https://go-review.googlesource.com/c/go/+/779220
Reviewed-by: Michael Matloob <matloob@google.com>
TryBot-Bypass: Michael Matloob <matloob@google.com>
Commit-Queue: Michael Matloob <matloob@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Profile.Write defers gzip.Writer.Close without checking its return value. Close flushes buffered data and writes the gzip footer (CRC32 checksum and size). If the flush or footer write fails, the caller receives a nil error despite the output being truncated or corrupt. This is called by runtime/pprof and net/http/pprof to serialize profiles. A write error during gzip finalization (e.g., broken pipe, full disk) produces corrupt output reported as success. Remove the defer and call Close explicitly after Write, returning its error. On Write failure, Close is still called to release resources, but the Write error is returned. Fixes #79423 Change-Id: I1d93c20158010f45c72d57caf5d5b25b72ca7511 GitHub-Last-Rev: cd79f2a GitHub-Pull-Request: #79424 Reviewed-on: https://go-review.googlesource.com/c/go/+/778280 Reviewed-by: Michael Pratt <mpratt@google.com> Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The 0xffff directorySize was a typo, and File.zip64 was unused. The ErrFormat was intentional, but it's actually possible to have a valid zip file with compressed size 2³²-1, for example by storing uncompressed a file of size 2³²-1 using Info-ZIP. The following CL introduces a couple such files (infozip-store-4g-minus-1 and infozip-offset-eq-4g). Fixes #31692 Fixes #56249 Updates #14185 Updates #13367 Updates #13166 Change-Id: I503805cace50316a665633d43dcc7fa46a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/779180 Reviewed-by: Russ Cox <rsc@golang.org> Reviewed-by: David Chase <drchase@google.com> Auto-Submit: Filippo Valsorda <filippo@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change-Id: I98c8eb513131942c885457912a2101316a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/768460 LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com> Reviewed-by: Michael Matloob <matloob@golang.org>
Print only the removals that would actually happen, and print them for all removals. Fixes #33573 Change-Id: Ic73e5e73329dd5dcc111696e9732e7e36a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/768461 Reviewed-by: Michael Matloob <matloob@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com> Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
…rce directory The behaviour of the compiler and go/scanner is unified: Relative //line directives are now relative to the directory of the scanned file. Fixes #70478 Change-Id: I804edbb5f9540d386e0fb1a4e64b10b717125d1b Reviewed-on: https://go-review.googlesource.com/c/go/+/706795 Reviewed-by: Alan Donovan <adonovan@google.com> Reviewed-by: Robert Griesemer <gri@google.com> Auto-Submit: Robert Griesemer <gri@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
I ran into this because the broken Writer caused mysterious and very
hard to debug failures uploading archive/zip-generated files to the
Internet Archive. (Only zip files bigger than 4GiB *and* smaller than
around 6.5GiB failed. I still don't have an explanation for the latter
part, maybe the parser has different logic for when the count of records
crosses 65535 and the Zip64 EOCD is used.)
Reproducing testdata/zip64/*.zsparse:
Inputs (sparse zero files via `truncate -s N NAME`, sizes in bytes):
big5g.bin 5<<30 big4g.bin 4<<30
big4g-1.bin (4<<30) - 1 big4g-2.bin (4<<30) - 2
under4g.bin (4<<30) - 59 first (4<<30) - 36
small.bin 42 (use `dd` for the non-sparse 42-byte file)
Cases (case → entries, M=0 Store, M=9 Deflate):
store-5g big5g.bin/0
deflate-zeros-5g big5g.bin/9
store-4g-minus-1 big4g-1.bin/0
store-4g-minus-2 big4g-2.bin/0
store-just-under-4g under4g.bin/0
store-exact-4g big4g.bin/0
offset-past-4g big5g.bin/0, small.bin/0
offset-eq-4g first/0, small.bin/0
Producers:
infozip-* Info-ZIP 3.0:
zip -q -X -M OUT.zip <entries>
libarchive-* bsdtar (libarchive):
bsdtar -cf OUT.zip --format zip \
--options zip:compression={store|deflate} <entries>
go126-* archive/zip from Go 1.26. Build with GOTOOLCHAIN=go1.26.0
from a tempdir whose go.mod declares `go 1.26.0`.
For each entry:
zip.FileHeader{Name, Method: zip.Store|zip.Deflate},
CreateHeader, io.CopyN(fw, zeros, size), w.Close().
Convert each OUT.zip to ${producer}-${case}.zsparse using the format
defined in archive/zip/zip64_sparse_test.go (scanSparse / readSparse):
walk the zip in 4 KiB chunks, drop chunks that are entirely zero,
coalesce adjacent non-zero chunks into spans, and serialize the result
as gzip of:
uint64 LE totalSize
uint32 LE numSpans
numSpans times:
uint64 LE offset
uint32 LE dataLen
dataLen bytes
Updates #22520
Fixes #23572
Fixes #33116
Fixes #69415
Change-Id: I6e24e7170094346af494da153c63e6b56a6a6964
Reviewed-on: https://go-review.googlesource.com/c/go/+/725161
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
SLSA-Policy-Verified: SLSA Policy Verification Service <devtools-gerritcodereview-exitgate@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
14 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )