cmd/compile: fix cmpstring rewrite rule

We need to ensure that the Select0 lives in the same block as
its argument. Divide up the rule into 2 so that we can put the
parts in the right places.

(This would be simpler if we could use @block syntax mid-rule, but
that feature currently only works at the top level.)

This fixes the ssacheck builder after CL 578835

Change-Id: Id26a01d9fac0684e0b732d35d0f7999f6de07825
Reviewed-on: https://go-review.googlesource.com/c/go/+/580815
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>

net/netip: use the unique package instead of internal/intern

This change replaces net/netip's use of the internal/intern package with
the new unique package.

Below are the benchmark results for the package. Lots of things get
faster, but some things get slower. The single-core performance of
unique.Make is expected to be a tad slower than internal/intern, but is
vastly more scalable. No benchmark in this package currently
demonstrates this, however.

                                    │ before.bench  │             after.bench              │
                                    │    sec/op     │    sec/op      vs base               │
IPNextPrev-48                          85.66n ± ∞ ¹    85.61n ± ∞ ¹        ~ (p=0.690 n=5)
BinaryMarshalRoundTrip/ipv4-48         16.12n ± ∞ ¹    15.84n ± ∞ ¹   -1.74% (p=0.008 n=5)
BinaryMarshalRoundTrip/ipv6-48         33.14n ± ∞ ¹    32.49n ± ∞ ¹   -1.96% (p=0.016 n=5)
BinaryMarshalRoundTrip/ipv6+zone-48    88.84n ± ∞ ¹   101.50n ± ∞ ¹  +14.25% (p=0.008 n=5)
StdIPv4-48                             126.6n ± ∞ ¹    157.2n ± ∞ ¹  +24.17% (p=0.008 n=5)
IPv4-48                                101.2n ± ∞ ¹    101.0n ± ∞ ¹        ~ (p=0.452 n=5)
IPv4_inline-48                         131.7n ± ∞ ¹    101.2n ± ∞ ¹  -23.16% (p=0.008 n=5)
StdIPv6-48                             180.6n ± ∞ ¹    178.7n ± ∞ ¹   -1.05% (p=0.040 n=5)
IPv6-48                                136.7n ± ∞ ¹    137.4n ± ∞ ¹        ~ (p=0.135 n=5)
IPv4Contains-48                        4.650n ± ∞ ¹    5.287n ± ∞ ¹  +13.70% (p=0.008 n=5)
IPv6Contains-48                        5.284n ± ∞ ¹    5.290n ± ∞ ¹        ~ (p=0.540 n=5)
ParseAddr/v4-48                        23.94n ± ∞ ¹    23.62n ± ∞ ¹        ~ (p=0.841 n=5)
ParseAddr/v6-48                        84.95n ± ∞ ¹    80.64n ± ∞ ¹   -5.07% (p=0.008 n=5)
ParseAddr/v6_ellipsis-48               51.54n ± ∞ ¹    51.62n ± ∞ ¹        ~ (p=1.000 n=5)
ParseAddr/v6_v4-48                     55.85n ± ∞ ¹    56.56n ± ∞ ¹   +1.27% (p=0.008 n=5)
ParseAddr/v6_zone-48                   110.0n ± ∞ ¹    124.6n ± ∞ ¹  +13.27% (p=0.008 n=5)
StdParseIP/v4-48                       54.05n ± ∞ ¹    53.74n ± ∞ ¹   -0.57% (p=0.016 n=5)
StdParseIP/v6-48                       111.6n ± ∞ ¹    111.7n ± ∞ ¹        ~ (p=0.651 n=5)
StdParseIP/v6_ellipsis-48              81.81n ± ∞ ¹    82.15n ± ∞ ¹   +0.42% (p=0.008 n=5)
StdParseIP/v6_v4-48                    89.36n ± ∞ ¹    90.72n ± ∞ ¹   +1.52% (p=0.008 n=5)
StdParseIP/v6_zone-48                  138.7n ± ∞ ¹    150.7n ± ∞ ¹   +8.65% (p=0.008 n=5)
AddrString/v4-48                       36.43n ± ∞ ¹    36.33n ± ∞ ¹        ~ (p=0.508 n=5)
AddrString/v6-48                       106.1n ± ∞ ¹    102.5n ± ∞ ¹   -3.39% (p=0.008 n=5)
AddrString/v6_ellipsis-48              93.93n ± ∞ ¹    92.87n ± ∞ ¹   -1.13% (p=0.008 n=5)
AddrString/v6_v4-48                    43.09n ± ∞ ¹    43.33n ± ∞ ¹        ~ (p=0.111 n=5)
AddrString/v6_zone-48                  91.15n ± ∞ ¹    89.22n ± ∞ ¹   -2.12% (p=0.008 n=5)
IPStringExpanded/v4-48                 37.99n ± ∞ ¹    37.77n ± ∞ ¹   -0.58% (p=0.032 n=5)
IPStringExpanded/v6-48                 63.59n ± ∞ ¹    63.58n ± ∞ ¹        ~ (p=0.690 n=5)
IPStringExpanded/v6_ellipsis-48        64.33n ± ∞ ¹    63.50n ± ∞ ¹   -1.29% (p=0.024 n=5)
IPStringExpanded/v6_v4-48              64.79n ± ∞ ¹    63.08n ± ∞ ¹   -2.64% (p=0.008 n=5)
IPStringExpanded/v6_zone-48            111.7n ± ∞ ¹    109.7n ± ∞ ¹   -1.79% (p=0.016 n=5)
AddrMarshalText/v4-48                  34.45n ± ∞ ¹    34.04n ± ∞ ¹        ~ (p=0.151 n=5)
AddrMarshalText/v6-48                  105.2n ± ∞ ¹    102.6n ± ∞ ¹   -2.47% (p=0.008 n=5)
AddrMarshalText/v6_ellipsis-48         99.69n ± ∞ ¹    98.56n ± ∞ ¹        ~ (p=0.151 n=5)
AddrMarshalText/v6_v4-48               48.22n ± ∞ ¹    47.33n ± ∞ ¹        ~ (p=0.151 n=5)
AddrMarshalText/v6_zone-48             98.55n ± ∞ ¹    94.82n ± ∞ ¹   -3.78% (p=0.008 n=5)
AddrPortString/v4-48                   59.03n ± ∞ ¹    50.09n ± ∞ ¹  -15.14% (p=0.008 n=5)
AddrPortString/v6-48                   118.2n ± ∞ ¹    113.3n ± ∞ ¹   -4.15% (p=0.008 n=5)
AddrPortString/v6_ellipsis-48          110.1n ± ∞ ¹    107.3n ± ∞ ¹   -2.54% (p=0.008 n=5)
AddrPortString/v6_v4-48                66.48n ± ∞ ¹    58.80n ± ∞ ¹  -11.55% (p=0.008 n=5)
AddrPortString/v6_zone-48              106.7n ± ∞ ¹    104.9n ± ∞ ¹   -1.69% (p=0.008 n=5)
AddrPortMarshalText/v4-48              56.67n ± ∞ ¹    54.19n ± ∞ ¹   -4.38% (p=0.008 n=5)
AddrPortMarshalText/v6-48              125.1n ± ∞ ¹    122.1n ± ∞ ¹   -2.40% (p=0.008 n=5)
AddrPortMarshalText/v6_ellipsis-48     120.1n ± ∞ ¹    118.0n ± ∞ ¹   -1.75% (p=0.024 n=5)
AddrPortMarshalText/v6_v4-48           71.53n ± ∞ ¹    70.25n ± ∞ ¹        ~ (p=0.548 n=5)
AddrPortMarshalText/v6_zone-48         116.4n ± ∞ ¹    114.0n ± ∞ ¹   -2.06% (p=0.016 n=5)
PrefixMasking/IPv4_/32-48              5.259n ± ∞ ¹    5.289n ± ∞ ¹   +0.57% (p=0.016 n=5)
PrefixMasking/IPv4_/17-48              5.286n ± ∞ ¹    5.284n ± ∞ ¹        ~ (p=0.937 n=5)
PrefixMasking/IPv4_/0-48               5.266n ± ∞ ¹    5.280n ± ∞ ¹        ~ (p=0.151 n=5)
PrefixMasking/IPv6_/128-48             5.271n ± ∞ ¹    5.284n ± ∞ ¹        ~ (p=0.151 n=5)
PrefixMasking/IPv6_/65-48              5.256n ± ∞ ¹    5.282n ± ∞ ¹        ~ (p=0.317 n=5)
PrefixMasking/IPv6_/0-48               5.253n ± ∞ ¹    5.284n ± ∞ ¹        ~ (p=0.095 n=5)
PrefixMasking/IPv6_zone_/128-48        5.274n ± ∞ ¹    5.284n ± ∞ ¹        ~ (p=0.222 n=5)
PrefixMasking/IPv6_zone_/65-48         5.280n ± ∞ ¹    5.291n ± ∞ ¹        ~ (p=0.095 n=5)
PrefixMasking/IPv6_zone_/0-48          5.283n ± ∞ ¹    5.292n ± ∞ ¹        ~ (p=0.095 n=5)
PrefixMarshalText-48                   44.60n ± ∞ ¹    44.81n ± ∞ ¹        ~ (p=1.000 n=5)
ParseAddrPort/v4-48                    46.40n ± ∞ ¹    41.27n ± ∞ ¹  -11.06% (p=0.008 n=5)
ParseAddrPort/v6-48                   106.90n ± ∞ ¹    99.29n ± ∞ ¹   -7.12% (p=0.008 n=5)
ParseAddrPort/v6_ellipsis-48           77.10n ± ∞ ¹    69.98n ± ∞ ¹   -9.23% (p=0.008 n=5)
ParseAddrPort/v6_v4-48                 71.82n ± ∞ ¹    74.97n ± ∞ ¹   +4.39% (p=0.008 n=5)
ParseAddrPort/v6_zone-48               130.0n ± ∞ ¹    148.5n ± ∞ ¹  +14.23% (p=0.008 n=5)
As16-48                                7.446n ± ∞ ¹    7.437n ± ∞ ¹        ~ (p=0.286 n=5)
geomean                                45.52n          45.18n         -0.76%
¹ need >= 6 samples for confidence interval at level 0.95

                                   │ before.bench │             after.bench             │
                                   │     B/op     │    B/op      vs base                │
StdIPv4-48                            16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4-48                               0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4_inline-48                        0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdIPv6-48                            16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv6-48                               0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4Contains-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv6Contains-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v4-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_ellipsis-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_v4-48                    0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_zone-48                  0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v4-48                      16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6-48                      16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_ellipsis-48             16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_v4-48                   16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_zone-48                 16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v4-48                      16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6-48                      48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_ellipsis-48             24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_v4-48                   24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_zone-48                 24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v4-48                16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6-48                48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_ellipsis-48       48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_v4-48             48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_zone-48           128.0 ± ∞ ¹   128.0 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v4-48                 16.00 ± ∞ ¹   16.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6-48                 48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_ellipsis-48        48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_v4-48              32.00 ± ∞ ¹   32.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_zone-48            48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v4-48                  24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6-48                  48.00 ± ∞ ¹   48.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_ellipsis-48         32.00 ± ∞ ¹   32.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_v4-48               32.00 ± ∞ ¹   32.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_zone-48             32.00 ± ∞ ¹   32.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v4-48             24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6-48             64.00 ± ∞ ¹   64.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_ellipsis-48    64.00 ± ∞ ¹   64.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_v4-48          64.00 ± ∞ ¹   64.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_zone-48        64.00 ± ∞ ¹   64.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/32-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/17-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/0-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/128-48            0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/65-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/0-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/128-48       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/65-48        0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/0-48         0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMarshalText-48                  24.00 ± ∞ ¹   24.00 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v4-48                   0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6-48                   0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_ellipsis-48          0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_v4-48                0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_zone-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
geomean                                         ³                +0.00%               ³
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean

                                   │ before.bench │             after.bench             │
                                   │  allocs/op   │  allocs/op   vs base                │
StdIPv4-48                            1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4-48                               0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4_inline-48                        0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdIPv6-48                            1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv6-48                               0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv4Contains-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPv6Contains-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v4-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6-48                       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_ellipsis-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_v4-48                    0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddr/v6_zone-48                  0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v4-48                      1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6-48                      1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_ellipsis-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_v4-48                   1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
StdParseIP/v6_zone-48                 1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v4-48                      1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6-48                      1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_ellipsis-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_v4-48                   1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrString/v6_zone-48                 1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v4-48                1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6-48                1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_ellipsis-48       1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_v4-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
IPStringExpanded/v6_zone-48           2.000 ± ∞ ¹   2.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v4-48                 1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6-48                 1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_ellipsis-48        1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_v4-48              1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrMarshalText/v6_zone-48            1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v4-48                  1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6-48                  1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_ellipsis-48         1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_v4-48               1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortString/v6_zone-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v4-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6-48             1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_ellipsis-48    1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_v4-48          1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
AddrPortMarshalText/v6_zone-48        1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/32-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/17-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv4_/0-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/128-48            0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/65-48             0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_/0-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/128-48       0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/65-48        0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMasking/IPv6_zone_/0-48         0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
PrefixMarshalText-48                  1.000 ± ∞ ¹   1.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v4-48                   0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6-48                   0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_ellipsis-48          0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_v4-48                0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
ParseAddrPort/v6_zone-48              0.000 ± ∞ ¹   0.000 ± ∞ ¹       ~ (p=1.000 n=5) ²
geomean                                         ³                +0.00%               ³
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean

Change-Id: Ieec7d1f0836f3ee93a33653aee9b0c0fc77f399d
Reviewed-on: https://go-review.googlesource.com/c/go/+/577035
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>

unique: add unique package and implement Make/Handle

This change adds the unique package for canonicalizing values, as
described by the proposal in #62483.

Fixes #62483.

Change-Id: I1dc3d34ec12351cb4dc3838a8ea29a5368d59e99
Reviewed-on: https://go-review.googlesource.com/c/go/+/574355
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ingo Oeser <nightlyone@googlemail.com>
Reviewed-by: David Chase <drchase@google.com>

cmd/go: update comment to not say GO386 does not exist

GO386 was removed by CL 258957, but it was restored by CL 260017.

Change-Id: Iced49ca61512a0f2ef513acbf9700a87ac964c68
Reviewed-on: https://go-review.googlesource.com/c/go/+/580675
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/link/internal/ld: temporarily add skip for new portion of TestElfBindNow

Skip newly added sections in TestElfBindNow from elf_test.go
temporarily pending resolution of failures on loong64 and ppc64le.

Change-Id: I22e0e52c9fbbcad88a4d0088f631ed7be15bf9da
Reviewed-on: https://go-review.googlesource.com/c/go/+/580776
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/compile/internal/ssa: fix the description of auxCCop as auxInt

The auxCCop has been modified from aux to auxInt in CL 252517.

Change-Id: I5c472d684c51fd38fd38f61561d1b1644e20bafe
GitHub-Last-Rev: 78d1cf40f476adbda2f42b929607501b07f0c91c
GitHub-Pull-Request: golang/go#66950
Reviewed-on: https://go-review.googlesource.com/c/go/+/580635
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Joedian Reid <joedian@google.com>

cmd/compile: combine phielim and copyelim into a single pass

Change-Id: Id21145b14169d28bac2144a31f6d3d9729f4be1e
GitHub-Last-Rev: 5413f4753e5acb60db6a93cb3409047bddc8df6d
GitHub-Pull-Request: golang/go#63818
Reviewed-on: https://go-review.googlesource.com/c/go/+/538535
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Keith Randall <khr@google.com>

runtime: always acquire M when acquiring locks by rank

Profiling of runtime-internal locks checks gp.m.locks to see if it's
safe to add a new record to the profile, but direct use of
acquireLockRank can change the list of the M's active lock ranks without
updating gp.m.locks to match. The runtime's internal rwmutex
implementation makes a point of calling acquirem/releasem when
manipulating the lock rank list, but the other user of acquireLockRank
(the GC's Gscan bit) relied on the GC's invariants to avoid deadlocks.

Codify the rwmutex approach by renaming acquireLockRank to
acquireLockRankAndM and having it include a call to aquirem. Do the same
for release.

Fixes #64706
Fixes #66004

Change-Id: Ib76eaa0cc1c45b64861d03345e17e1e843c19713
GitHub-Last-Rev: 160577bdb2bb2a4e869c6fd7e53e3be8fb819182
GitHub-Pull-Request: golang/go#66276
Reviewed-on: https://go-review.googlesource.com/c/go/+/571056
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Rhys Hiltner <rhys.hiltner@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

cmd/link: move .dynamic and .got sections to relro if applicable

This is the second of two CLs to roll forward the changes in CL
473495, which was subsequently reverted.

In this patch we move the .dynamic and .got sections from the writable
data segment to the relro segment if the platform supports relro and
we're producing a PIE binary, and also moves .got.plt into relro if
eager binding is in effect (e.g. -bindnow or -Wl,-z,now).

Updates #45681.

Change-Id: I9f4fba6e825b96d1b5e27fb75844450dd0a650b3
Reviewed-on: https://go-review.googlesource.com/c/go/+/571417
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

internal/reflectlite: remove unused mapType

Change-Id: I715186c053bc9911b912e69904735c9498bf9c13
Reviewed-on: https://go-review.googlesource.com/c/go/+/580376
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>

reflect: remove unnecessary type conversions of untyped PtrSize

Change-Id: I0f20020c2929b58958ce228b9f175d5b4fd77a1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/578855
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Russ Cox <rsc@golang.org>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>

runtime: for fallback hash, get rid of constant xors

There's no need for these. If hashkey[i] is uniform over [0,1<<64), then
hashkey[0]^C has exactly the same distribution, for any constant C.

Change-Id: I4e10c27eff15e4b7a45139654ac41f410b1b12fc
Reviewed-on: https://go-review.googlesource.com/c/go/+/580218
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>

runtime: remove odd-forcing of hash constants

We don't multiply by them any more, so no need to make them odd.

Suggested by me here in 2021:
https://go-review.googlesource.com/c/go/+/280372/comment/1a86b8aa_f64310bf/
Never got around to it until now.

Change-Id: Iddc675e6a18b4a7a785acbf77c96e127003cc3c5
Reviewed-on: https://go-review.googlesource.com/c/go/+/579116
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>

runtime: make it harder to find collisions in the 64-bit fallback hash

Currently the first argument to mix() can be set by an attacker, as it
is just the input bytes xored by some constants. That lets an attacker
set the value being multipled by to 0. That can lead to lots of
collisions. To fix, xor the first argument with the process-wide seed,
so the magic collision-generating value isn't a constant known to the
attacker.  (Maybe there's a timing attack that could figure out the
process-wide seed, but that's a much harder attack.)

Fixes #66841

Change-Id: I33e073c78355d1cee08660de52074e6ccc38b426
Reviewed-on: https://go-review.googlesource.com/c/go/+/579115
Reviewed-by: M Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

runtime: track frame pointer while in syscall

Currently the runtime only tracks the PC and SP upon entering a syscall,
but not the FP (BP). This is mainly for historical reasons, and because
the tracer (which uses the frame pointer unwinder) does not need it.

Until it did, of course, in CL 567076, where the tracer tries to take a
stack trace of a goroutine that's in a syscall from afar. It tries to
use gp.sched.bp and lots of things go wrong. It *really* should be using
the equivalent of gp.syscallbp, which doesn't exist before this CL.

This change introduces gp.syscallbp and tracks it. It also introduces
getcallerfp which is nice for simplifying some code. Because we now have
gp.syscallbp, we can also delete the frame skip count computation in
traceLocker.GoSysCall, because it's now the same regardless of whether
frame pointer unwinding is used.

Fixes #66889.

Change-Id: Ib6d761c9566055e0a037134138cb0f81be73ecf7
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-nocgo
Reviewed-on: https://go-review.googlesource.com/c/go/+/580255
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/trace/v2: tolerate traces with broken tails

This change modifies cmd/trace/v2 to tolerate traces with
incomplete/broken generations at the tail. These broken tails can be
created if a program crashes while a trace is being produced. Although
the runtime tries to flush the trace on some panics, it may still
produce some extra trace data that is incomplete.

This change modifies cmd/trace/v2 to still work on any complete
generations, even if there are incomplete/broken generations at the tail
end of the trace. Basically, the tool now just tracks when the last good
generation ended (via Sync events) and truncates the trace to that point
when it encounters an error.

This change also revamps the text output of the tool to emit regular
progress notifications as well as warnings as to how much of the trace
data was lost.

Fixes #65316.

Change-Id: I877d39993bc02a81eebe647db9c2be17635bcec8
Reviewed-on: https://go-review.googlesource.com/c/go/+/580135
Reviewed-by: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>

internal/concurrent: handle boundary case for hash bits in HashTrieMap

Currently the HashTrieMap has a panic for running out of hash bits, but
it turns out we can end up in these paths in valid cases, like inserting
or deleting an element that requires *all* the hash bits to finds its
position in the tree. There's basically an off-by-one error here where
the panic fires erroneously.

This wasn't caught before the original CL landed because it's very
unlikely on 64-bit platforms, with a 64-bit hash, but much more likely
on 32-bit platforms, where using all 32 bits of a 32-bit hash is much
more likely.

This CL makes the condition for panicking much more explicit, which
avoids the off-by-one error.

After this CL, I can't get the tests to fail on 32-bit under stress
testing.

Change-Id: I855e301e3b3893e2b6b017f6dd9f3d83a94a558d
Reviewed-on: https://go-review.googlesource.com/c/go/+/580138
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: David Chase <drchase@google.com>

reflect: omit anonymous field name from StructOf type string

This makes the reflect package match the compiler for StructOf
with an embedded field.

Fixes #24781

Change-Id: Ice64b167cbe0b9d30a953c5d8e2a86f3ad1158bf
Reviewed-on: https://go-review.googlesource.com/c/go/+/567897
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

crypto/tls: don't cache marshal'd bytes

Only cache the wire representation for clientHelloMsg and serverHelloMsg
during unmarshal, which are the only places we actually need to hold
onto them. For everything else, remove the raw field.

This appears to have zero performance impact:

name                                               old time/op   new time/op   delta
CertCache/0-10                                       177µs ± 2%    189µs ±11%   ~     (p=0.700 n=3+3)
CertCache/1-10                                       184µs ± 3%    182µs ± 6%   ~     (p=1.000 n=3+3)
CertCache/2-10                                       187µs ±12%    187µs ± 2%   ~     (p=1.000 n=3+3)
CertCache/3-10                                       204µs ±21%    187µs ± 1%   ~     (p=0.700 n=3+3)
HandshakeServer/RSA-10                               410µs ± 2%    410µs ± 3%   ~     (p=1.000 n=3+3)
HandshakeServer/ECDHE-P256-RSA/TLSv13-10             473µs ± 3%    460µs ± 2%   ~     (p=0.200 n=3+3)
HandshakeServer/ECDHE-P256-RSA/TLSv12-10             498µs ± 3%    489µs ± 2%   ~     (p=0.700 n=3+3)
HandshakeServer/ECDHE-P256-ECDSA-P256/TLSv13-10      140µs ± 5%    138µs ± 5%   ~     (p=1.000 n=3+3)
HandshakeServer/ECDHE-P256-ECDSA-P256/TLSv12-10      132µs ± 1%    133µs ± 2%   ~     (p=0.400 n=3+3)
HandshakeServer/ECDHE-X25519-ECDSA-P256/TLSv13-10    168µs ± 1%    171µs ± 4%   ~     (p=1.000 n=3+3)
HandshakeServer/ECDHE-X25519-ECDSA-P256/TLSv12-10    166µs ± 3%    163µs ± 0%   ~     (p=0.700 n=3+3)
HandshakeServer/ECDHE-P521-ECDSA-P521/TLSv13-10     1.87ms ± 2%   1.81ms ± 0%   ~     (p=0.100 n=3+3)
HandshakeServer/ECDHE-P521-ECDSA-P521/TLSv12-10     1.86ms ± 0%   1.86ms ± 1%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/1MB/TLSv12-10                  6.79ms ± 3%   6.73ms ± 0%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/1MB/TLSv13-10                  6.73ms ± 1%   6.75ms ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/2MB/TLSv12-10                  12.8ms ± 2%   12.7ms ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/2MB/TLSv13-10                  13.1ms ± 3%   12.8ms ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/4MB/TLSv12-10                  24.9ms ± 2%   24.7ms ± 1%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/4MB/TLSv13-10                  26.0ms ± 4%   24.9ms ± 1%   ~     (p=0.100 n=3+3)
Throughput/MaxPacket/8MB/TLSv12-10                  50.0ms ± 3%   48.9ms ± 0%   ~     (p=0.200 n=3+3)
Throughput/MaxPacket/8MB/TLSv13-10                  49.8ms ± 2%   49.3ms ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/16MB/TLSv12-10                 97.3ms ± 1%   97.4ms ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/16MB/TLSv13-10                 97.9ms ± 0%   97.9ms ± 1%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/32MB/TLSv12-10                  195ms ± 0%    194ms ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/32MB/TLSv13-10                  196ms ± 0%    196ms ± 1%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/64MB/TLSv12-10                  405ms ± 3%    385ms ± 0%   ~     (p=0.100 n=3+3)
Throughput/MaxPacket/64MB/TLSv13-10                  391ms ± 1%    388ms ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/1MB/TLSv12-10              6.75ms ± 0%   6.75ms ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/1MB/TLSv13-10              6.84ms ± 1%   6.77ms ± 0%   ~     (p=0.100 n=3+3)
Throughput/DynamicPacket/2MB/TLSv12-10              12.8ms ± 1%   12.8ms ± 1%   ~     (p=0.400 n=3+3)
Throughput/DynamicPacket/2MB/TLSv13-10              12.8ms ± 1%   13.0ms ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/4MB/TLSv12-10              24.8ms ± 1%   24.8ms ± 0%   ~     (p=1.000 n=3+3)
Throughput/DynamicPacket/4MB/TLSv13-10              25.1ms ± 2%   25.1ms ± 1%   ~     (p=1.000 n=3+3)
Throughput/DynamicPacket/8MB/TLSv12-10              49.2ms ± 2%   48.9ms ± 0%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/8MB/TLSv13-10              49.3ms ± 1%   49.4ms ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/16MB/TLSv12-10             97.1ms ± 0%   98.0ms ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/16MB/TLSv13-10             98.8ms ± 1%   98.4ms ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/32MB/TLSv12-10              192ms ± 0%    198ms ± 5%   ~     (p=0.100 n=3+3)
Throughput/DynamicPacket/32MB/TLSv13-10              194ms ± 0%    196ms ± 1%   ~     (p=0.400 n=3+3)
Throughput/DynamicPacket/64MB/TLSv12-10              385ms ± 1%    384ms ± 0%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/64MB/TLSv13-10              387ms ± 0%    388ms ± 0%   ~     (p=0.400 n=3+3)
Latency/MaxPacket/200kbps/TLSv12-10                  694ms ± 0%    694ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/200kbps/TLSv13-10                  699ms ± 0%    699ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/500kbps/TLSv12-10                  278ms ± 0%    278ms ± 0%   ~     (p=0.400 n=3+3)
Latency/MaxPacket/500kbps/TLSv13-10                  280ms ± 0%    280ms ± 0%   ~     (p=1.000 n=3+3)
Latency/MaxPacket/1000kbps/TLSv12-10                 140ms ± 1%    140ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/1000kbps/TLSv13-10                 141ms ± 0%    141ms ± 0%   ~     (p=1.000 n=3+3)
Latency/MaxPacket/2000kbps/TLSv12-10                70.5ms ± 0%   70.4ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/2000kbps/TLSv13-10                70.7ms ± 0%   70.7ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/5000kbps/TLSv12-10                28.8ms ± 0%   28.8ms ± 0%   ~     (p=0.700 n=3+3)
Latency/MaxPacket/5000kbps/TLSv13-10                28.9ms ± 0%   28.9ms ± 0%   ~     (p=0.700 n=3+3)
Latency/DynamicPacket/200kbps/TLSv12-10              134ms ± 0%    134ms ± 0%   ~     (p=0.700 n=3+3)
Latency/DynamicPacket/200kbps/TLSv13-10              138ms ± 0%    138ms ± 0%   ~     (p=1.000 n=3+3)
Latency/DynamicPacket/500kbps/TLSv12-10             54.1ms ± 0%   54.1ms ± 0%   ~     (p=1.000 n=3+3)
Latency/DynamicPacket/500kbps/TLSv13-10             55.7ms ± 0%   55.7ms ± 0%   ~     (p=0.100 n=3+3)
Latency/DynamicPacket/1000kbps/TLSv12-10            27.6ms ± 0%   27.6ms ± 0%   ~     (p=0.200 n=3+3)
Latency/DynamicPacket/1000kbps/TLSv13-10            28.4ms ± 0%   28.4ms ± 0%   ~     (p=0.200 n=3+3)
Latency/DynamicPacket/2000kbps/TLSv12-10            14.4ms ± 0%   14.4ms ± 0%   ~     (p=1.000 n=3+3)
Latency/DynamicPacket/2000kbps/TLSv13-10            14.6ms ± 0%   14.6ms ± 0%   ~     (p=1.000 n=3+3)
Latency/DynamicPacket/5000kbps/TLSv12-10            6.44ms ± 0%   6.45ms ± 0%   ~     (p=0.100 n=3+3)
Latency/DynamicPacket/5000kbps/TLSv13-10            6.49ms ± 0%   6.49ms ± 0%   ~     (p=0.700 n=3+3)

name                                               old speed     new speed     delta
Throughput/MaxPacket/1MB/TLSv12-10                 155MB/s ± 3%  156MB/s ± 0%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/1MB/TLSv13-10                 156MB/s ± 1%  155MB/s ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/2MB/TLSv12-10                 163MB/s ± 2%  165MB/s ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/2MB/TLSv13-10                 160MB/s ± 3%  164MB/s ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/4MB/TLSv12-10                 168MB/s ± 2%  170MB/s ± 1%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/4MB/TLSv13-10                 162MB/s ± 4%  168MB/s ± 1%   ~     (p=0.100 n=3+3)
Throughput/MaxPacket/8MB/TLSv12-10                 168MB/s ± 3%  172MB/s ± 0%   ~     (p=0.200 n=3+3)
Throughput/MaxPacket/8MB/TLSv13-10                 168MB/s ± 2%  170MB/s ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/16MB/TLSv12-10                172MB/s ± 1%  172MB/s ± 0%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/16MB/TLSv13-10                171MB/s ± 0%  171MB/s ± 1%   ~     (p=1.000 n=3+3)
Throughput/MaxPacket/32MB/TLSv12-10                172MB/s ± 0%  173MB/s ± 1%   ~     (p=0.400 n=3+3)
Throughput/MaxPacket/32MB/TLSv13-10                171MB/s ± 0%  172MB/s ± 1%   ~     (p=0.700 n=3+3)
Throughput/MaxPacket/64MB/TLSv12-10                166MB/s ± 3%  174MB/s ± 0%   ~     (p=0.100 n=3+3)
Throughput/MaxPacket/64MB/TLSv13-10                171MB/s ± 1%  173MB/s ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/1MB/TLSv12-10             155MB/s ± 0%  155MB/s ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/1MB/TLSv13-10             153MB/s ± 1%  155MB/s ± 0%   ~     (p=0.100 n=3+3)
Throughput/DynamicPacket/2MB/TLSv12-10             164MB/s ± 1%  164MB/s ± 1%   ~     (p=0.400 n=3+3)
Throughput/DynamicPacket/2MB/TLSv13-10             163MB/s ± 1%  162MB/s ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/4MB/TLSv12-10             169MB/s ± 1%  169MB/s ± 0%   ~     (p=1.000 n=3+3)
Throughput/DynamicPacket/4MB/TLSv13-10             167MB/s ± 1%  167MB/s ± 1%   ~     (p=1.000 n=3+3)
Throughput/DynamicPacket/8MB/TLSv12-10             170MB/s ± 2%  171MB/s ± 0%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/8MB/TLSv13-10             170MB/s ± 1%  170MB/s ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/16MB/TLSv12-10            173MB/s ± 0%  171MB/s ± 1%   ~     (p=0.200 n=3+3)
Throughput/DynamicPacket/16MB/TLSv13-10            170MB/s ± 1%  170MB/s ± 1%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/32MB/TLSv12-10            175MB/s ± 0%  170MB/s ± 5%   ~     (p=0.100 n=3+3)
Throughput/DynamicPacket/32MB/TLSv13-10            173MB/s ± 0%  171MB/s ± 1%   ~     (p=0.300 n=3+3)
Throughput/DynamicPacket/64MB/TLSv12-10            174MB/s ± 1%  175MB/s ± 0%   ~     (p=0.700 n=3+3)
Throughput/DynamicPacket/64MB/TLSv13-10            174MB/s ± 0%  173MB/s ± 0%   ~     (p=0.400 n=3+3)

Change-Id: Ifa79cce002011850ed8b2835edd34f60e014eea8
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest,gotip-linux-arm64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/580215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Auto-Submit: Roland Shoemaker <roland@golang.org>

runtime: use bootstrapRand to initialize hashkey

The seed for rand is not initialized until after alginit. Before
initialization, rand returns a deterministic sequence, making hashkey
deterministic across processes.

Switch to bootstrapRand, like other early rand calls, such as
initialization of aeskeysched.

Fixes #66885.

Change-Id: I5023a9161232b49fda2ebd1d5f9338bbdd17b1fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/580136
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>

cmd/compile: remove redundant calls to cmpstring

The results of cmpstring are reuseable if the second call has the
same arguments and memory.

Note that this gets rid of cmpstring, but we still generate a
redundant </<= test and branch afterwards, because the compiler
doesn't know that cmpstring only ever returns -1,0,1.

Update #61725

Change-Id: I93a0d1ccca50d90b1e1a888240ffb75a3b10b59b
Reviewed-on: https://go-review.googlesource.com/c/go/+/578835
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>

internal/goexperiment: remove unused range experiment

This experiment was added mistakenly in https://go.dev/cl/537980,
probably as the result of a bad rebase on https://go.dev/cl/539277.

Change-Id: Ia3fc75725535ac70a2e6dd655440a2f515405c48
Reviewed-on: https://go-review.googlesource.com/c/go/+/580375
Auto-Submit: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

internal/syscall/unix: implement Eaccess on darwin

Like on other Unix-like platforms, use faccessat(AT_FDCWD, path, mode,
AT_EACCESS).

Change-Id: I6e04b4244f8e90bd3b35b1b8a9ca38fe845b34ae
Reviewed-on: https://go-review.googlesource.com/c/go/+/579976
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

syscall/js: allocate arg slices on stack for small numbers of args

The existing implementation causes unnecessary heap allocations for
javascript syscalls: Call, Invoke, and New. The new change seeks to
hint the Go compiler to allocate arg slices with length <=16 to the
stack.

Original Work: CL 367045
- Calling a JavaScript function with 16 arguments or fewer will not
induce two additional heap allocations, at least with the current Go
compiler.
- Using syscall/js features with slices and strings of
statically-known length will not cause them to be escaped to the heap,
at least with the current Go compiler.
- The reduction in allocations has the additional benefit that the
garbage collector runs less often, blocking WebAssembly's one and only
thread less often.

Fixes #39740

Change-Id: I815047e1d4f8ada796318e2064d38d3e63f73098
GitHub-Last-Rev: 36df1b33a4506e216767d8a73395f2fafdd80eba
GitHub-Pull-Request: golang/go#66684
Reviewed-on: https://go-review.googlesource.com/c/go/+/576575
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

doc: make symbol links simpler for relnotes

This is a follow-up to CL 578195.

Change-Id: I6466ff8753f05a9424a8098cd88490aab6b236c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/580277
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Jonathan Amsterdam <jba@google.com>

net/http: add field Cookie.Quoted bool

The current implementation of the http package strips double quotes
from the cookie-value during parsing, resulting in the serialized
cookie not including them. This patch addresses this limitation by
introducing a new field to track whether the original value was
enclosed in quotes.

Additionally, the internal representation of a cookie in the cookiejar
package has been adjusted to align with the new representation.

The syntax of cookies is outlined in RFC 6265 Section 4.1.1:
https://datatracker.ietf.org/doc/html/rfc6265\#section-4.1.1

Fixes #46443

Change-Id: Iac12a56397d77a6060a75757ab0daeacc60457f3
GitHub-Last-Rev: a76440e741440cddaa05944b6828a14a32b5a44a
GitHub-Pull-Request: golang/go#66752
Reviewed-on: https://go-review.googlesource.com/c/go/+/577755
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

go/types: add Func.Signature method

Unfortunately we can't enforce the repr invariant
that Func.typ != nil without thinking about the
object color invariants. For now, return a trivial
Signature if typ == nil, which should never happen
in bug-free client code.

Fixes golang/go#65772

Change-Id: I7f89c6d8fdc8dcd4b8880572e54bb0ed9b6136eb
Reviewed-on: https://go-review.googlesource.com/c/go/+/565375
Commit-Queue: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

internal/weak: add package implementing weak pointers

This change adds the internal/weak package, which exposes GC-supported
weak pointers to the standard library. This is for the upcoming weak
package, but may be useful for other future constructs.

For #62483.

Change-Id: I4aa8fa9400110ad5ea022a43c094051699ccab9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/576297
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

internal/concurrent: add HashTrieMap

This change adds a concurrent hash-trie map implementation to the
standard library in the new internal/concurrent package, intended to
hold concurrent data structures. (The name comes from how Java names
their concurrent data structure library in the standard library.)

This data structure is created specially for the upcoming unique
package. It is built specifically around frequent successful lookups and
comparatively rare insertions and deletions.

A valid question is whether this is worth it over a simple locked map.
Some microbenchmarks in this new package show that yes, this extra
complexity appears to be worth it.

Single-threaded performance for LoadOrStore is comparable to a locked
map for a map with 128k small string elements. The map scales perfectly
up to 24 cores for Loads, which is the maximum available parallelism
on my machine. LoadOrStore operations scale less well. Small maps will
have a high degree of contention, but for the unique library, small maps
are very unlikely to stay small if there are a lot of inserts, since
they have a full GC cycle to grow.

For #62483.

Change-Id: I38e5ac958d19ebdd0c8c02e36720bb3338fe2e35
Reviewed-on: https://go-review.googlesource.com/c/go/+/573956
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go: only attempt to match subset of $GOCACHE in test

This test attempted to be clever by looking for the entirety of $GOCACHE
in the compile command line to ensure that the profile was coming from
cache.

Unfortunately, on Windows $GOCACHE contains \, which needs extra
escaping in a regexp. As an approximate alternative, just look for the
"gocache" component specified when defining GOCACHE.

This fixes the Windows longtest builders.

Cq-Include-Trybots: luci.golang.try:gotip-windows-amd64-longtest
Change-Id: If6c77cf066d8612431e0720405254e1fdf528e9b
Reviewed-on: https://go-review.googlesource.com/c/go/+/580137
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Bypass: Michael Pratt <mpratt@google.com>

go/types, types2: use correct predicate when asserting comma-ok types

While at it and unrelated, up-date testdata/manual.go sample file so
we can just copy its contents into a test file after debugging, without
fixing the date.

Fixes #66878.

Change-Id: Ie49a341b78d99bdc0f1a0ba1ca42fa2d3a807bd6
Reviewed-on: https://go-review.googlesource.com/c/go/+/580075
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

doc: simplify symbol links for relnotes

This is a follow-up to CL 578195.

Change-Id: Iab38ee008df8b06c300b0027019cc0464bdadca1
Reviewed-on: https://go-review.googlesource.com/c/go/+/579796
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>

net/http: add case for TestParseSetCookie

Updates #66008

Change-Id: Idd36a7f0b4128adfd0a3d7fe76eb6c2cea4306a3
Reviewed-on: https://go-review.googlesource.com/c/go/+/579795
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Damien Neil <dneil@google.com>

cmd/compile/internal: stack slot merging region formation enhancements

This patch revises the algorithm/strategy used for overlapping the
stack slots of disjointly accessed local variables. The main change
here is to allow merging the stack slot of B into the slot for A if
B's size is less then A (prior to this they had to be identical), and
to also allow merging a non-pointer variables into pointer-variable
slots.

The new algorithm sorts the candidate list first by pointerness
(pointer variables first), then by alignment, then by size, and
finally by name. We no longer check that two variables have the same
GC shape before merging: since it should never be the case that we
have two vars X and Y both live across a given callsite where X and Y
share a stack slot, their gc shape doesn't matter.

Doing things this new way increases the total number of bytes saved
(across all functions) from 91256 to 124336 for the sweet benchmarks.

Updates #62737.
Updates #65532.
Updates #65495.

Change-Id: I1daaac1b1240aa47a6975e98ccd24e03304ab602
Reviewed-on: https://go-review.googlesource.com/c/go/+/577615
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/dist: allow arbitrary package renames for bootstrap imports

Import declaration matching currently has a list of specific cases. It
allows bare imports, dot imports, and renamed imports named "exec" and
"rtabi".

Keeping a specific allowlist of renamed imports is unnecessary and
causes annoyance for developers adding such imports, as the bootstrap
build errors do not make it clear that this is where the issue lies.

We can simplify this to be much more general. The body of the condition
will still only rewrite imports in cmd/ or in bootstrapDirs.

I believe the only downside of this change is that it makes it a bit
more likely to match and replace within comments. That said, there
should be no harm in replacements within comments.

This change results in no change to the resulting bootstrap source tree:

$ diff -u -r /tmp/bootstrap.before/src /tmp/bootstrap.after/src
diff -u -r /tmp/bootstrap.before/src/bootstrap/internal/buildcfg/zbootstrap.go /tmp/bootstrap.after/src/bootstrap/internal/buildcfg/zbootstrap.go
--- /tmp/bootstrap.before/src/bootstrap/internal/buildcfg/zbootstrap.go 2024-03-27 12:29:27.439540946 -0400
+++ /tmp/bootstrap.after/src/bootstrap/internal/buildcfg/zbootstrap.go  2024-03-27 12:28:08.516211238 -0400
@@ -20,6 +20,6 @@
 const defaultGOEXPERIMENT = ``
 const defaultGO_EXTLINK_ENABLED = ``
 const defaultGO_LDSO = ``
-const version = `devel go1.23-38087c80ae Wed Mar 27 12:09:16 2024 -0400`
+const version = `devel go1.23-fa64f04409 Wed Mar 27 12:22:52 2024 -0400`
 const defaultGOOS = runtime.GOOS
 const defaultGOARCH = runtime.GOARCH

Change-Id: Ia933c6373f366f2e607b28d900227c24cb214674
Reviewed-on: https://go-review.googlesource.com/c/go/+/574735
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Pratt <mpratt@google.com>

net/http: add comments that servemux121.go should remain frozen

Change-Id: I6d1e02ffd07dbb722669abb2e72784dcb2bfffd0
GitHub-Last-Rev: 55a82f724a32237be059ae0696cf023db414f1d0
GitHub-Pull-Request: golang/go#66848
Reviewed-on: https://go-review.googlesource.com/c/go/+/579118
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Cherry Mui <cherryyz@google.com>
Reviewed-by: Damien Neil <dneil@google.com>

cmd/go: use cache for PGO preprocessing

This is the final CL in the series adding PGO preprocessing support to
cmd/go. Now that the tool is hooked up, we integrate with the build
cache to cache the result.

This is fairly straightforward. One difference is that the compile and
link do caching through updateBuildID. However, preprocessed PGO files
don't have a build ID, so it doesn't make much sense to hack our way
through that function when it is simple to just add to the cache
ourselves.

As as aside, we could add a build ID to the preproccessed file format,
though it is not clear if it is worthwhile. The one place a build ID
could be used is in buildActionID, which currently compute the file hash
of the preprocessed profile. With a build ID it could simply read the
build ID. This would save one complete read of the file per build
(cmd/go caches the hash), but each compile process also reads the entire
file, so this is a small change overall.

Fixes #58102.

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Change-Id: I86e2999a08ccd264230fbb1c983192259b7288e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/569425
Auto-Submit: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go: preprocess PGO profiles

Following the previous CL, now actually run the preprofile tool to create the
preprocessed output.

There is still no build cache integration, so the tool will run on every
build even if nothing has changed.

For #58102.

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Change-Id: I0414377a956889f457e50898737fcaa8a698658d
Reviewed-on: https://go-review.googlesource.com/c/go/+/569424
Auto-Submit: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/go: inital plumbing for PGO profiles preprocessing

The new go tool preprofile preprocesses a PGO pprof profile into an
intermediate representation that is more efficient for the compiler to
consume. Performing preprocessing avoids having every single compile
process from duplicating the same processing.

This CL prepares the initial plumbing to support automatic preprocessing
by cmd/go.

Each compile action takes a new dependency on a new "preprocess PGO
profile" action. The same action instance is shared by all compile
actions (assuming they have the same input profile), so the action only
executes once.

Builder.build retrieves the file to pass to -pgofile from the output of
the preprocessing action, rather than directly from
p.Internal.PGOProfile.

Builder.buildActionID also uses the preprocess output as the PGO
component of the cache key, rather than the original source. This
doesn't matter for normal toolchain releases, as the two files are
semantically equivalent, but it is useful for correct cache invalidation
in development. For example, if _only_ go tool preprofile changes
(potentially changing the output), then we must regenerate the output
and then rebuild all packages.

This CL does not actually invoke go tool preprocess. That will come in
the next CL. For now, it just copies the input pprof profile.

This CL shouldn't be submitted on its own, only with the children. Since
the new action doesn't yet use the build cache, every build (even fully
cached builds) unconditionally run the PGO action.

For #58102.

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Change-Id: I594417cfb0164cd39439a03977c904e4c0c83b8b
Reviewed-on: https://go-review.googlesource.com/c/go/+/569423
Auto-Submit: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

go/types: track gotypesalias non-default behavior

Fixes #66216.

Change-Id: I04d7389e5712b35db078844ce424e10f5b96156c
Reviewed-on: https://go-review.googlesource.com/c/go/+/579936
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>

go/types, types2: use types2.Config flag to control Alias node creation

Move Checker.enableAlias to Config.EnableAlias (for types2) and
Config._EnableAlias (for go/types), and adjust all uses.

Use Config.EnableAlias to control Alias creation for types2 and
with that remove dependencies on the gotypesalias GODEBUG setting
and problems during bootstrap. The only client is the compiler and
there we simply use the desired configuration; it is undesirable
for the compiler to be dependent on gotypesalias.

Use the gotypesalias GODEBUG setting to control Config._EnableAlias
for go/types (similar to before).

Adjust some related code. We plan to remove gotypesalias eventually
which will remove some of the new discrepancies between types2 and
go/types again.

Fixes #66874.

Change-Id: Id7cc4805e7ea0697e0d023c7f510867e59a24871
Reviewed-on: https://go-review.googlesource.com/c/go/+/579935
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>

net/http: don't cancel Dials when requests are canceled

Currently, when a Transport creates a new connection for a request,
it uses the request's Context to make the Dial. If a request
times out or is canceled before a Dial completes, the Dial is
canceled.

Change this so that the lifetime of a Dial call is not bound
by the request that originated it.

This change avoids a scenario where a Transport can start and
then cancel many Dial calls in rapid succession:

  - Request starts a Dial.
  - A previous request completes, making its connection available.
  - The new request uses the now-idle connection, and completes.
  - The request Context is canceled, and the Dial is aborted.

Fixes #59017

Change-Id: I996ffabc56d3b1b43129cbfd9b3e9ea7d53d263c
Reviewed-on: https://go-review.googlesource.com/c/go/+/576555
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

internal/abi: define EmptyInterface, TypeOf, and NoEscape

This change defines two commonly-defined functions and a
commonly-defined type in internal/abi to try and deduplicate some
definitions. This is motivated by a follow-up CL which will want access
to TypeOf in yet another package.

There still exist duplicate definitions of all three of these things in
the runtime, and this CL doesn't try to handle that yet. There are far
too many uses in the runtime to handle manually in a way that feels
comfortable; automated refactoring will help.

For #62483.

Change-Id: I02fc64a28f11af618f6071f94d27f45c135fa8ac
Reviewed-on: https://go-review.googlesource.com/c/go/+/573955
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>

crypto/tls: add a bogo shim

Run the BoGo test suite.

For now a number of tests are disabled, so that we can land the shim.
Once the shim is in the tree I'll work on fixing tests, and aligning
the TLS stack with the boringssl stack.

Eventually we should also remove the --loose-errors flag.

Fixes #51434

Change-Id: Ic8339fc34552936b798acf834011a129e375750e
Reviewed-on: https://go-review.googlesource.com/c/go/+/486495
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Roland Shoemaker <roland@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>

cmd/compile/internal/ssa: delay rewrite cycle detection for huge funcs

The SSA rewrite pass has some logic that looks to see whether a
suspiciously large number of rewrites is happening, and if so, turns
on logic to try to detect rewrite cycles. The cycle detection logic is
quite expensive (hashes the entire function), meaning that for very
large functions we might get a successful compilation in a minute or
two with no cycle detection, but take a couple of hours once cycle
detection kicks in.

This patch moves from a fixed limit of 1000 iterations to a limit set
partially based on the size of the function (meaning that we'll wait
longer before turning cycle detection for a large func).

Fixes #66773.

Change-Id: I72f8524d706f15b3f0150baf6abeab2a5d3e15c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/578215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

net/http: add ParseCookie, ParseSetCookie

Fixes #66008

Change-Id: I64acb7da47a03bdef955f394682004906245a18b
Reviewed-on: https://go-review.googlesource.com/c/go/+/578275
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/covdata: close cpu and mem profile

Change-Id: Iaf14989eb2981f724c4091f992ed99687ce3a60e
GitHub-Last-Rev: 6e6c82bb8fba5c32c24eafb50eec6751a93d09f9
GitHub-Pull-Request: golang/go#66852
Reviewed-on: https://go-review.googlesource.com/c/go/+/579255
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>

all: update vendored golang.org/x/crypto

Pull in CL 578715:

5defcc19 sha3: fix Sum results for SHAKE functions on s390x

Fixes #66804.

Change-Id: I72bb7862778c6e10a40b1aaeeafea49e1a0d80f5
Reviewed-on: https://go-review.googlesource.com/c/go/+/579455
Reviewed-by: Michael Munday <mike.munday@lowrisc.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>

Revert "go/types, types2: track gotypesalias non-default behavior"

This reverts commit c51f6c62576ab3088a981b9cef90e413ac99e333.

Reason for revert: This breaks toolchain bootstrapping in Google. Root cause investigation is pending.

Error message is:

```
<unknown line number>: internal compiler error: panic: godebug: Value of name not listed in godebugs.All: gotypesalias
```

Change-Id: Ie3dff566a29b3b0846ebc8fe0a371c656a043a4c
Reviewed-on: https://go-review.googlesource.com/c/go/+/579575
Reviewed-by: Chressie Himpel <chressie@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Lasse Folger <lassefolger@google.com>

go/types, types2: track gotypesalias non-default behavior

Fixes #66216.

Change-Id: I2750a744d0dcf636a00d388299e1f2f993e5ac26
Reviewed-on: https://go-review.googlesource.com/c/go/+/572575
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>

types2: flip the default value of GODEBUG=gotypesalias=1

This CL changes the interpretation of the unset value
of gotypesalias to not equal "0".

This is a port of CL 577715 from go/types to types2,
with adjustments to go/types to keep the source code
in sync. Specifically:

- Re-introduce testing of both modes (gotypesalias=0,
  gotypesalias=1) in go/types.
- Re-introduce setting of gotypesalias in some of the
  tests for explicit documentation in go/types.

The compiler still uses the (now) non-default setting
due to a panic with the default setting that needs to
be debugged.

Also, the type checkers still don't call IncNonDefault
when the non-default setting of gotypesalias is used.

Change-Id: I1feed3eb334c202950ac5aadf49a74adcce0d8c3
Reviewed-on: https://go-review.googlesource.com/c/go/+/579076
TryBot-Bypass: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>

go/internal/gcimporter: suppress 3 test cases when gotypesalias=1

CL 577715 caused these test cases to fail, but this was not
detected by CI because they are "long" tests.

Updates #66859

Change-Id: I84320d9091772540df2ab15f57d93031596bb89b
Reviewed-on: https://go-review.googlesource.com/c/go/+/579415
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Alan Donovan <adonovan@google.com>

runtime/pprof: test for Darwin flake in TestVMInfo

If it contains
"No process corpse slots currently available, waiting to get one"
skip the test in short mode, so that run.bash works reliably
on developer laptops, but the flake is still recorded on builders.

The problem also seems to get better after a laptop reboot?

Updates #62352.

Change-Id: I12e8f594f0b830bacda5d8bfa594782345764c4a
Reviewed-on: https://go-review.googlesource.com/c/go/+/579295
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

math/big: improve use of addze in mulAddVWW on ppc64x

Improve the use of addze to avoid unnecessary register
moves on ppc64x.

goos: linux
goarch: ppc64le
pkg: math/big
cpu: POWER10
                 │   old.out    │               new.out               │
                 │    sec/op    │    sec/op     vs base               │
MulAddVWW/1         4.524n ± 3%   4.248n ±  0%   -6.10% (p=0.002 n=6)
MulAddVWW/2         5.634n ± 0%   5.283n ±  0%   -6.24% (p=0.002 n=6)
MulAddVWW/3         6.406n ± 0%   5.918n ±  0%   -7.63% (p=0.002 n=6)
MulAddVWW/4         6.484n ± 0%   5.859n ±  0%   -9.64% (p=0.002 n=6)
MulAddVWW/5         7.363n ± 0%   6.766n ±  0%   -8.11% (p=0.002 n=6)
MulAddVWW/10       10.920n ± 0%   9.856n ±  0%   -9.75% (p=0.002 n=6)
MulAddVWW/100       83.46n ± 0%   66.95n ±  0%  -19.78% (p=0.002 n=6)
MulAddVWW/1000      856.0n ± 0%   681.6n ±  0%  -20.38% (p=0.002 n=6)
MulAddVWW/10000     8.589µ ± 1%   6.774µ ±  0%  -21.14% (p=0.002 n=6)
MulAddVWW/100000    86.22µ ± 0%   67.71µ ± 43%  -21.48% (p=0.065 n=6)
geomean             73.34n        63.62n        -13.26%

Change-Id: I95d6ac49ff6b64aa678e6896f57af9d85c923aad
Reviewed-on: https://go-review.googlesource.com/c/go/+/579235
Reviewed-by: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Paul Murphy <murp@ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd,crypto/elliptic: fix typos in comments

Replace these incorrect fullwidth brackets with halfwidth brackets.

Change-Id: Ie17561c18334f9c07eedbff79e5f64ed4fc281bd
GitHub-Last-Rev: 94214be6cea75b4d1fa7b536cac523f061614d38
GitHub-Pull-Request: golang/go#66846
Reviewed-on: https://go-review.googlesource.com/c/go/+/579117
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

os: remove unused issueNo field

It's no longer set since CL 31118.

Change-Id: Ibe77b1454b5e7fd02eaed432f04cf993f53791fc
Reviewed-on: https://go-review.googlesource.com/c/go/+/579135
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>

doc: fix tense in godebug (s/revert/reverted)

Change-Id: Ida6ed22102a6da36739c7581aeab297fdd7bc9f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/561715
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

runtime: remove no-op slice operation in Caller

rpc was an array prior to CL 152537, so it was necessary to slice
it since callers accepts a slice. Now that rpc is already a slice,
slicing it is no longer required.

Change-Id: Ie646ef5e494323c9fb58f3a24f942e3b1ff639ea
Reviewed-on: https://go-review.googlesource.com/c/go/+/579016
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

debug/elf: define non-standard but well-known symbol types

Fixes #66836

Change-Id: I603faca2acd2bcffabbcaca8b8670d46387d2a5b
Reviewed-on: https://go-review.googlesource.com/c/go/+/578995
Commit-Queue: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/cgo: create -objdir if not exist

Currently the directory is created only if -objdir is omited.
Creating the directory here is useful to avoid doing this in each build system.
And also this is consistent with similar flags of other tools like `-o`.

Change-Id: Ic39d6eb3e003bc4884089f80f790e30df4a54b01
Reviewed-on: https://go-review.googlesource.com/c/go/+/576815
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

os: make File.Readdir et al concurrency-safe

Before, all methods of File (including Close) were
safe for concurrent use (I checked), except the three
variants of ReadDir.

This change makes the ReadDir operations
atomic too, and documents explicitly that all methods
of File have this property, which was already implied
by the package documentation.

Fixes #66498

Change-Id: I05c88b4e60b44c702062e99ed8f4a32e7945927a
Reviewed-on: https://go-review.googlesource.com/c/go/+/578322
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

go/types, types2: simplify TestUnaliasTooSoonInCycle (cleanup)

Follow-up on CL 576975 and CL 579015.

Updates #66704
Updates #65294

Change-Id: Ied95386a346be38ccda86d332d09b2089a68c5e7
Reviewed-on: https://go-review.googlesource.com/c/go/+/579075
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>

go/types: flip the default value of GODEBUG=gotypesalias=1

This CL changes the interpretation of the unset value
of gotypesalias to equal "1". The actual deletion of
all the transitional logic will happen in a follow-up.

Note that the compiler still interprets unset as "0".
More work appears to be required within the compiler
before it is safe to flip its default.

Change-Id: I854ab1fd856c7c361a757676b0670e2f23402816
Reviewed-on: https://go-review.googlesource.com/c/go/+/577715
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Alan Donovan <adonovan@google.com>
Commit-Queue: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile/internal/types2: port CL 576975 to types2

This CL ports to types2 the (passing) test from CL 576975,
which fixed a bug in go/types.

Updates #66704
Updates #65294

Change-Id: Icdf77e39ed177d9f9ecc435d5125f02f2ee4dd0f
Reviewed-on: https://go-review.googlesource.com/c/go/+/579015
Auto-Submit: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>

compress/bzip2: simplify Huffman tree construction

This change simplifies the construction of the Huffman tree in the
bzip2 package by replacing custom sort logic with the more concise and
idiomatic use of "slices" and "cmp" packages.

Change-Id: I2a8aef146b54b9433038b133d2cc8856ba077c72
GitHub-Last-Rev: c031bb56635552205c12b87add5f4337a7def74f
GitHub-Pull-Request: golang/go#66817
Reviewed-on: https://go-review.googlesource.com/c/go/+/578438
Reviewed-by: Ian Lance Taylor <iant@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: Cherry Mui <cherryyz@google.com>

cmd/internal/obj/loong64: recheck jump offset boundary after auto-aligning loop heads

After the alignment of the loop header is performed, the offset of the checked
conditional branch instruction may overflow, so it needs to be checked again.

When checking whether the offset of the branch jump instruction overflows, it
can be classified and processed according to the range of the immediate field
of the specific instruction, which can reduce the introduction of unnecessary
jump instructions.

Fixes #61819

Change-Id: I772a5b5b8b8de21c78d7566be30be8ff65fdbce8
Reviewed-on: https://go-review.googlesource.com/c/go/+/519915
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Run-TryBot: qiu laidongfeng2 <2645477756@qq.com>
Reviewed-by: WANG Xuerui <git@xen0n.name>

internal/abi, internal/buildcfg: always enable register ABI on loong64

Change-Id: Ia3a31556c8d355eb6bdb42d3b3c8bc6b37311ed9
Reviewed-on: https://go-review.googlesource.com/c/go/+/525575
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

io: fix typo inside copyBuffer

The "rt" seems to be caused after copy-pasting the previous "wt" block
which make sense as WriterTo, but for ReaderFrom it makes more sense
thinking of rf instead of rt.

Change-Id: I873699c27211bea6cdba3e199f36eb3c38188d70
GitHub-Last-Rev: 1795600a9b29946d824ba645c137da216bdf6302
GitHub-Pull-Request: golang/go#66811
Reviewed-on: https://go-review.googlesource.com/c/go/+/578635
Reviewed-by: Ian Lance Taylor <iant@google.com>
Commit-Queue: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>

internal/goexperiment: remove the ExecTracer2 experiment

This change removes the ExecTracer2 experiment flag. This flag was
created as part of the tracer overhaul described in #60773.

Updates #66703
For #60773

Change-Id: Ib95b9dd8bb68b48e7543c4d28d853c6a518438f6
Reviewed-on: https://go-review.googlesource.com/c/go/+/576258
Auto-Submit: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

runtime: rename v2 execution tracer files

This change renames the v2 execution tracer files created as part of

Updates #66703
For #60773

Change-Id: I91bfdc08fec4ec68ff3a6e8b5c86f6f8bcae6e6d
Reviewed-on: https://go-review.googlesource.com/c/go/+/576257
Auto-Submit: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

runtime, cmd/trace: remove code paths that include v1 tracer

This change makes the new execution tracer described in #60773, the
default tracer. This change attempts to make the smallest amount of
changes for a single CL.

Updates #66703
For #60773

Change-Id: I3742f3419c54f07d7c020ae5e1c18d29d8bcae6d
Reviewed-on: https://go-review.googlesource.com/c/go/+/576256
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/pprof: close resp body after the profile.Parse is called

Change-Id: I5bccef3c46072e388bfe9985e70745853f673a42
GitHub-Last-Rev: adee7b9f7fa21c3ee5561b8233bbfe94168e45ed
GitHub-Pull-Request: golang/go#66829
Reviewed-on: https://go-review.googlesource.com/c/go/+/578875
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/cover: fix typo in comment

Change-Id: I7507e6cff00d027fd7840e0661499efc63353f6e
GitHub-Last-Rev: 81348ed39d055e445d943eedfe4b4db3a1fd73d8
GitHub-Pull-Request: golang/go#66820
Reviewed-on: https://go-review.googlesource.com/c/go/+/578441
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net: implement TCP_KEEPIDLE, TCP_KEEPINTVL, and TCP_KEEPCNT on Solaris 11.4

Also simulate TCP_KEEPIDLE, TCP_KEEPINTVL, and TCP_KEEPCNT with
TCP_KEEPALIVE_THRESHOLD + TCP_KEEPALIVE_ABORT_THRESHOLD for
Solaris prior to 11.4

Fixes #9614
Fixes #64251

Change-Id: Ia0777076a7952630bc52761cddd0b06b0d81c6a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/577195
Commit-Queue: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

cmd/go/internal/modfetch: fix typos in comment

Change-Id: I3eb41562bf6478b288d9f41915fd7d027399a6ba
GitHub-Last-Rev: 218b9cb0672f0be18834733c0907a244a8cfcfd9
GitHub-Pull-Request: golang/go#66551
Reviewed-on: https://go-review.googlesource.com/c/go/+/574575
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net: add Unwrap to *DNSError

Fixes #63116

Change-Id: Iab8c415555ab85097be6d2d133b3349c5219a23b
GitHub-Last-Rev: 8a8177b9af5509ebbaa701b06c79126aae7510a8
GitHub-Pull-Request: golang/go#63348
Reviewed-on: https://go-review.googlesource.com/c/go/+/532217
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net: separate the Solaris fast/slow path of setting SOCK_* from others

Along with the removal of the slow path from Linux and *BSD.

For #59359

Change-Id: I6c79594252e5e5f1c1c57c11e09458fcae3793d2
Reviewed-on: https://go-review.googlesource.com/c/go/+/577175
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Andy Pan <panjf2000@gmail.com>

net: bifurcate the TCP Keep-Alive mechanism into Solaris and illumos

Fixes #65812

Change-Id: I63facb32eeddbe9b6e0279f1c039779ba2e2ab7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/575015
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Andy Pan <panjf2000@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Commit-Queue: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: utilize EVFILT_USER to wake up kevent for kqueue

Fixes #66760

Change-Id: I6ba5bc5b00506b66cb8dc3984a61f32a6358d9bc
Reviewed-on: https://go-review.googlesource.com/c/go/+/577895
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Run-TryBot: Andy Pan <panjf2000@gmail.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

doc: make use of simpler symbol links for relnotes

This is a practical use of CL 577915, follow-up to CL 577835.

Change-Id: Ibe7e2fa11b444afa1898dc6f6aba1512fe98f1fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/578195
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>

go/types: update package doc

It wasn't DocLink compatible in a number of ways.

Change-Id: Ib7ab9a908ef47561ac70cdc0c157d49dcfd03a02
Reviewed-on: https://go-review.googlesource.com/c/go/+/577375
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

os: document file mode of temporary files

Fixes #66784

Change-Id: Ifd17e0830e04e7028d8a876c6c12c496f5167887
Reviewed-on: https://go-review.googlesource.com/c/go/+/578395
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net/http: fix debugRoundTrip log typo

Change-Id: I449fcb008fa4c1f51de0bef9f016907d8b99da67
GitHub-Last-Rev: b2e84f1a8a96199e1f9e4ae548225273e2ec94c0
GitHub-Pull-Request: golang/go#66772
Reviewed-on: https://go-review.googlesource.com/c/go/+/578135
Reviewed-by: Carlos Amedee <carlos@golang.org>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/trace/v2: make the -pprof actually useful

In both the v1 and v2 cmd/trace, pprofMatchingGoroutines will generate
no output at all if the filter name passed to it is the empty string.

This is rather pointless because there are at least two places where we
don't pass a name to filter. Modify pprofMatchingGoroutines to include
*all* goroutines in the trace if the name to filter by is not specified.

For #66782.

Change-Id: I6b72298d676bc93892b075a7426e6e56bc6656c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/578356
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Auto-Submit: Michael Knyszek <mknyszek@google.com>

cmd/trace/v2: handle the -pprof flag

Turns out we ported all the profile generation, but forgot to actually
support the command line flags for them! This change fixes the issue by
handling the different kinds of profiles and writing them out to stdout.

Fixes #66782.

Change-Id: I7756fb4636ce8daaf11ed471be79c86ce3d463cc
Reviewed-on: https://go-review.googlesource.com/c/go/+/578318
Reviewed-by: Carlos Amedee <carlos@golang.org>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net: check SkipAdditional error result

This will avoid a potential endless loop for a corrupt DNS packet.

For #66754

Change-Id: I46591b3f7695bcc88d2312833e45955f8c129d2b
Reviewed-on: https://go-review.googlesource.com/c/go/+/578375
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

internal/trace/v2: make TestGCStress less random

Currently TestGCStress' main loop allocates a randomly-sized byte slice
in a loop. On the windows-386 builder, it looks like the following is
happening.

In such heavily-allocating scenarios, the test seems to be able to
outrun the GC. This is a known issue on all platforms, but it looks to
me like there may be a real issue with mark termination. (Fixing that is
outside the scope of this change, but relevant here.)

Furthermore, while the test is ramping up, the pacer is taking time to
acclimate to the high allocation rate. This is probably made worse due
to the coarse time granularity on Windows, since the pacer relies on
accurate time measurements.

Because the pacer is ramping up, it isn't starting early enough, causing
a lot of memory to get allocated black and inflate the live heap size.
This happens for more than one cycle.

Last but not least, because the core allocating loop of this test
allocates randomly-sized byte slices, we could just get unlucky and
inflate the live heap by much more sometimes. Furthermore, the
randomness creates chaos for the pacer that is totally unnecessary for
this test.

Although I couldn't reproduce the issue we're seeing on the trybots in a
gomote, I *could* reproduce memory spikes in general. These memory
spikes always occurred before the pacer had a chance to "warm up," in
the first two cycles after the heavy allocating begins.

I believe the flakiness we're seeing is all of these factors lining up,
because if I just make the size of the allocated byte slices smaller and
non-random, I can no longer reproduce the memory spikes. This change
implements this as a fix in the hope that it'll resolve the flakiness.

Fixes #66624.

Change-Id: I478d45e7c600e5aee4b21dbe831e1f287284f5e3
Reviewed-on: https://go-review.googlesource.com/c/go/+/578319
Reviewed-by: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

all: consistently use "IEEE 754" over "IEEE-754"

There is no hyphen between the organization and the number.

For example, https://standards.ieee.org/ieee/754/6210/
shows the string "IEEE 754-2019" and not "IEEE-754-2019".

This assists in searching for "IEEE 754" in documentation
and not missing those using "IEEE-754".

Change-Id: I9a50ede807984ff1e2f17390bc1039f6a5d162e5
Reviewed-on: https://go-review.googlesource.com/c/go/+/575438
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
TryBot-Result: Gopher Robot <gobot@golang.org>
TryBot-Bypass: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>

misc/wasm: drop wasmtime < 14 support

For Go 1.23, we decided to no longer support the old CLI interface
exposed by wasmtime. This removes the extra logic included to support
both the new and the old CLI interface. Now only versions of wasmtime
14 and newer are supported.

Fixes #63718

Change-Id: Iea31388dc41bc8d73caa923c7e4acae2228bf515
Reviewed-on: https://go-review.googlesource.com/c/go/+/577135
Reviewed-by: Randy Reddig <randy.reddig@fastly.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>

encoding/json: use slices to simplify the code

Use "slices.Equal" instead of "reflect.DeepEqual".

Replace unnecessary helper type "byIndex" with "slices.SortFunc".

No effect on benchmarks.

Change-Id: I1fb2768ea6d9db7f487408fa109343be3f1741d5
GitHub-Last-Rev: 8429bc145272ae620fcd001b1de393bf3c0b6108
GitHub-Pull-Request: golang/go#66646
Reviewed-on: https://go-review.googlesource.com/c/go/+/575715
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>

internal/syscall/unix: implement Eaccess on openbsd

Like on other BSDs, use faccessat(AT_FDCWD, path, mode, AT_EACCESS)

Change-Id: I80f8d327dd152576165b9206e32dfb749b41d187
Reviewed-on: https://go-review.googlesource.com/c/go/+/538836
Reviewed-by: Carlos Amedee <carlos@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>

image/gif: revert incorrect usage of clear

CL 570555 replaced a loop which added empty
color.RGBA elements with a call to clear.

color.Palette is a slice of interfaces, so using
clear results in a slice of nil elements, rather
than what we previously had which was empty
color.RGBA elements. This could cause a panic when
attempting to re-encode a GIF which had an
extended color palette because of the weird
transparency hack.

This was discovered by OSS-Fuzz. I've added a test
case using their reproducer in order to prevent
future regressions.

Change-Id: I00a89257d90b6cca68672173eecdaa0a24f18d9c
Reviewed-on: https://go-review.googlesource.com/c/go/+/577555
Reviewed-by: Nigel Tao <nigeltao@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Roland Shoemaker <roland@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>

internal/poll: better panic for invalid write return value

For #61060

Change-Id: I13cd73b4062cb7bd248d2a4afae06dfa29ac0203
Reviewed-on: https://go-review.googlesource.com/c/go/+/577955
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd/link: close cpu profile

Change-Id: Ieaf1c165307a98d58c40d4d970eecfd6d74b2411
GitHub-Last-Rev: bd7271097896a4003b08c28e8ea11c6ab74077d9
GitHub-Pull-Request: golang/go#66736
Reviewed-on: https://go-review.googlesource.com/c/go/+/577416
Auto-Submit: Than McIntosh <thanm@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

database/sql: avoid clobbering driver-owned memory in RawBytes

Depending on the query, a RawBytes can contain memory owned by the
driver or by database/sql:

If the driver provides the column as a []byte,
RawBytes aliases that []byte.

If the driver provides the column as any other type,
RawBytes contains memory allocated by database/sql.
Prior to this CL, Rows.Scan will reuse existing capacity in a
RawBytes to permit a single allocation to be reused across rows.

When a RawBytes is reused across queries, this can result
in database/sql writing to driver-owned memory.

Add a buffer to Rows to store RawBytes data, and reuse this
buffer across calls to Rows.Scan.

Fixes #65201

Change-Id: Iac640174c7afa97eeb39496f47dec202501b2483
Reviewed-on: https://go-review.googlesource.com/c/go/+/557917
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

doc/README.md: note simpler symbol links

Document that links to symbols in the standard library can be
written as "[foo]", without the actual link.

For #64169.

Change-Id: I9d8a33e85df70037320a169d55a2bb4a8a981ebf
Reviewed-on: https://go-review.googlesource.com/c/go/+/577915
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

internal/trace/v2: halve the memory footprint of TestGCStress

This test has been OOMing on 32-bit platforms for a bit. I suspect the
very high allocation rate is causing the program to outrun the GC in
some corner-case scenarios, especially on 32-bit Windows.

I don't have a strong grasp of what's going on yet, but lowering the
memory footprint should help with the flakiness. This shouldn't
represent a loss in test coverage, since we're still allocating and
assisting plenty (tracing the latter is a strong reason this test
exists).

For #66624.

Change-Id: Idd832cfc5cde04701386919df4490f201c71130a
Reviewed-on: https://go-review.googlesource.com/c/go/+/577475
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>

runtime: rewrite traceMap to scale better

The existing implementation of traceMap is a hash map with a fixed
bucket table size which scales poorly with the number of elements added
to the map. After a few thousands elements are in the map, it tends to
fall over.

Furthermore, cleaning up the trace map is currently non-preemptible,
without very good reason.

This change replaces the traceMap implementation with a simple
append-only concurrent hash-trie. The data structure is incredibly
simple and does not suffer at all from the same scaling issues.

Because the traceMap no longer has a lock, and the traceRegionAlloc it
embeds is not thread-safe, we have to push that lock down. While we're
here, this change also makes the fast path for the traceRegionAlloc
lock-free. This may not be inherently faster due to contention on the
atomic add, but it creates an easy path to sharding the main allocation
buffer to reduce contention in the future. (We might want to also
consider a fully thread-local allocator that covers both string and
stack tables. The only reason a thread-local allocator isn't feasible
right now is because each of these has their own region, but we could
certainly group all them together.)

Change-Id: I8c06d42825c326061a1b8569e322afc4bc2a513a
Reviewed-on: https://go-review.googlesource.com/c/go/+/570035
Reviewed-by: Carlos Amedee <carlos@golang.org>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
TryBot-Bypass: Michael Knyszek <mknyszek@google.com>
Reviewed-by: David Chase <drchase@google.com>

runtime: push down systemstack requirement for tracer where possible

Currently lots of functions require systemstack because the trace buffer
might get flushed, but that will already switch to the systemstack for
the most critical bits (grabbing trace.lock). That means a lot of this
code is non-preemptible when it doesn't need to be. We've seen this
cause problems at scale, when dumping very large numbers of stacks at
once, for example.

This is a re-land of CL 572095 which was reverted in CL 577376. This
re-land includes a fix of the test that broke on the longtest builders.

Change-Id: Ia8d7cbe3aaa8398cf4a1818bac66c3415a399348
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/577377
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>

doc: update relnote for sync.Map.Clear

Change-Id: I28baf375502ccf3b0dfcc8250650e914ac7bae78
Reviewed-on: https://go-review.googlesource.com/c/go/+/577835
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Jes Cok <xigua67damn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>