When trampoline is called, the plt symbol has not been
added. If we add tramp here, plt will not work.
Change-Id: I64e5d2be9e08f78ca5e8a9dcb267620a481d4416
Reviewed-on: https://go-review.googlesource.com/c/go/+/654918
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
This enables the ASAN default behavior of reporting C memory leaks.
It can be disabled with ASAN_OPTIONS=detect_leaks=0.
Fixes#67833
Change-Id: I420da1b5d79cf70d8cf134eaf97bf0a22f61ffd0
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-asan-clang15,gotip-linux-arm64-asan-clang15
Reviewed-on: https://go-review.googlesource.com/c/go/+/651755
Reviewed-by: Cherry Mui <cherryyz@google.com>
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>
Look at the inlining stack of positions for a call site,
if the line/col/file of the call site appears in that
stack, do not inline. This subsumes all the other
recently-added recursive inlining checks, but they are
left in to make this easier+safer to backport.
Fixes#72090
Change-Id: I0f487bb0d4c514015907c649312672b7be464abd
Reviewed-on: https://go-review.googlesource.com/c/go/+/655155
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
The 'used' field on Var and PkgName is fundamentally an aspect of the
type checking pass: it records when objects are used, for the purposes
of reporting errors for unused variables or package names. While
expedient and performant, recording this information in the types.Object
instances themselves increases the memory footprint of type-checked
packages, and (as we saw in golang/go#71817) can lead to data races when
Objects are reused in follow-up type checking, such as is done with the
CheckExpr and Eval APIs.
Fix this by externalizing the 'used' information into two maps (one for
variables and one for packages) on the types.Checker, so that they are
garbage-collected after type checking, and cannot be a source of data
races.
Benchmarks showed essentially no change in performance.
Fixesgolang/go#71817
Change-Id: I40daeabe4ecaca3bcb494e2f1c62a04232098e49
Reviewed-on: https://go-review.googlesource.com/c/go/+/650796
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
This PR adds error handling in net/http toHTTPError to return a 404
instead of a 500 when net/http fs.Dir.Open throws the error http:
invalid or unsafe file path.
Fixes#72091
Change-Id: I7941c8fca5160a4a82732dc1d05b9b95eac84fbf
GitHub-Last-Rev: 04b5019dfb
GitHub-Pull-Request: golang/go#72108
Reviewed-on: https://go-review.googlesource.com/c/go/+/654975
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
cmd/go tests that run builds are generally skipped in short mode. This
change will adds skips for some tests that were running builds.
I found these by sorting tests by elapsed time and removing the top
tests that invoked go build. It's our practice to skip tests that run go
build without the -n flag (which prints but doesn't execute commands).
On my work laptop this reduces test run time from about 20 seconds to
about 16 seconds. On my linux workstation it reduces test run time from
about 10 seconds to about 5 seconds.
Change-Id: I18ffcc231df013cb6ac5f5eb3544bed28dadeda8
Reviewed-on: https://go-review.googlesource.com/c/go/+/653775
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This CL adds an enum type, VarKind, that discriminates among
the various kinds of Var, and adds setter/getter methods
for Var's kind field.
Beware: NewVar has a weaker postcondition: the Var objects it
returns are not completely initialized and require a call to
Var.SetKind. This should only affect importers.
No changes are needed to the export data, since the kind can
always be deduced from the context when decoding.
See CL 645656 for the corresponding x/tools changes.
+ test, relnote, API
Updates golang/go#70250
Change-Id: Icde86ad22a880cde6f50bc12bf38004a5c6a1025
Reviewed-on: https://go-review.googlesource.com/c/go/+/645115
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>
This example illustrates how to overwrite a template function after parsing a template.
This example is intended to clarify the point made in the template.Funcs docstring
that "[i]t is legal to overwrite elements of the map."
Change-Id: Ibded05974d580c54a24fcc16687fd52ce21133ff
GitHub-Last-Rev: ef19a221ab
GitHub-Pull-Request: golang/go#72094
Reviewed-on: https://go-review.googlesource.com/c/go/+/654416
Reviewed-by: Junyang Shao <shaojunyang@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Fixes#53524
Change-Id: I929ee3c055c3ca564cd6cc374124f493aea2fbf6
Reviewed-on: https://go-review.googlesource.com/c/go/+/421636
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
The original issue62407_test also passes with versions prior to 1.23.
The improvement makes it fail with versions prior to 1.23.
Change-Id: I94bfb9d1ac695c8e07997d7029fc2101535e14f8
GitHub-Last-Rev: 44be2a610a
GitHub-Pull-Request: golang/go#70938
Reviewed-on: https://go-review.googlesource.com/c/go/+/638036
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
This change eliminates sortLines function to avoid strings.Join calls.
It's not a performance problem, this change tries to make the comparison
more straightforward.
Change-Id: I3a7ae877c9fc927833ab9f143205f7e007197f60
GitHub-Last-Rev: a71aa58c58
GitHub-Pull-Request: golang/go#72025
Reviewed-on: https://go-review.googlesource.com/c/go/+/653556
Reviewed-by: Junyang Shao <shaojunyang@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Clone the input slice before adjusting NextProtos
to add or remove "http/1.1" and "h2" entries,
so as not to modify a slice that the caller might be using.
(We clone the tls.Config that contains the slice, but
that's a shallow clone.)
Fixes#72100
Change-Id: I9f228b8fb6f6f2ca5023179ec114929c002dbda9
Reviewed-on: https://go-review.googlesource.com/c/go/+/654875
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Allocates more with -asan after CL 653795.
Change-Id: Ib8cc1de1d649623713b6fc123c1c59a47528857b
Reviewed-on: https://go-review.googlesource.com/c/go/+/654876
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
These methods were previously used by crypto/ecdsa, but now not even
ecdsa_legacy.go uses them. Neither were ever documented.
Inverse was available only on P256() and only on amd64 and arm64, so
hopefully no one used it. CombinedMult was always available on all
curves, so it's possible some application might have used it, but all
the samples on GitHub I can find copied the old crypto/ecdsa package,
which does a conditional interface upgrade with a fallback, so they
won't break.
Change-Id: I6a6a4656ee1ab98438ca0fb20bea53b229cd7e71
Reviewed-on: https://go-review.googlesource.com/c/go/+/640116
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TestReverseProxyWebSocketHalfTCP requires half closed connections,
which aren't supported on Plan 9.
For #35892Fixes#72095
Change-Id: I64b458bc15ac3b8eda43dc871bf67ada32a59708
Reviewed-on: https://go-review.googlesource.com/c/go/+/654636
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>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: David du Colombier <0intro@gmail.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
For instance, this fixes os.ReadFile on plan9's /net/iproute file.
But it's not necessarily plan9-specific; Linux /proc and /sys filesystems
can exhibit the same problems.
Fixes#72080
Change-Id: I60b035913f583a91c6d84df95a6ea7b7ec2b3c92
Reviewed-on: https://go-review.googlesource.com/c/go/+/654315
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
SP and HTAB are allowed after a = before the following CRLF.
RFC 2045 section 6.7 describes the ABNF for the quoted-printable encoding:
qp-line := *(qp-segment transport-padding CRLF)
qp-part transport-padding
qp-segment := qp-section *(SPACE / TAB) "="
transport-padding := *LWSP-char
; Composers MUST NOT generate
; non-zero length transport
; padding, but receivers MUST
; be able to handle padding
; added by message transports.
RFC 822 defines LWSP-char as:
LWSP-char = SPACE / HTAB
Dovecot's imaptest contains such a message in
src/tests/fetch-binary-mime-qp.mbox.
Fixes#70952
Change-Id: Ie05921088d7e4d6c92c4bf79b0f4a13586230753
GitHub-Last-Rev: e6e6eee8eb
GitHub-Pull-Request: golang/go#70951
Reviewed-on: https://go-review.googlesource.com/c/go/+/638276
Reviewed-by: Sean Liao <sean@liao.dev>
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Sean Liao <sean@liao.dev>
This CL propagates closing the write stream from either side of the
reverse proxy and ensures the proxy waits for both copy-to and the
copy-from the backend to complete.
The new unit test checks communication through the reverse proxy when
the backend or frontend closes either the read or write streams.
That closing the write stream is propagated through the proxy from
either the backend or the frontend. That closing the read stream is
not propagated through the proxy.
Fixes#35892
Change-Id: I83ce377df66a0f17b9ba2b53caf9e4991a95f6a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/637939
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Sean Liao <sean@liao.dev>
Auto-Submit: Sean Liao <sean@liao.dev>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Matej Kramny <matejkramny@gmail.com>
The root cause of issue #65802 is a small race condition that occurs between
two events:
1. During the HTTP server shutdown, a connection in an idle state is identified
and closed.
2. The connection, although idle, has just finished reading a complete request
before being closed and hasn't yet updated its state to active.
In this scenario, despite the connection being closed, the request continues to
be processed. This not only wastes server resources but also prevents the
client request from being retried.
Fixes#65802
Change-Id: Ic22abb4497be04f6c84dff059df00f2c319d8652
GitHub-Last-Rev: 426099a3e7
GitHub-Pull-Request: golang/go#65805
Reviewed-on: https://go-review.googlesource.com/c/go/+/565277
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Sean Liao <sean@liao.dev>
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Sean Liao <sean@liao.dev>
This patch updates the compiler to generate DWARF5-style location
lists (e.g. entries that feed into .debug_loclists) as opposed to
DWARF4-style location lists (which wind up in .debug_loc). The DWARF5
format is much more compact, and can make indirect references to text
addresses via the .debug_addr section for further space savings.
Updates #26379.
Change-Id: If2e6fce1136d9cba5125ea51a71419596d1d1691
Reviewed-on: https://go-review.googlesource.com/c/go/+/635836
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
This patch updates the compiler to generate DWARF5-style range lists
(e.g. entries that feed into .debug_rnglists) as opposed to
DWARF4-style range lists (which wind up in .debug_ranges). The DWARF5
format is much more compact, and can make indirect references to text
address via the .debug_addr section for further space savings.
Updates #26379.
Change-Id: I273a6283484b7fe33d79d5412e31c5155b22a7c0
Reviewed-on: https://go-review.googlesource.com/c/go/+/635345
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
CL 629195 strongly favor closure inlining, allowing closures to be
inlined more aggressively.
However, if the closure body contains a call to a function, which itself
is one of the call arguments, it causes the infinite inlining.
Fixing this by prevent this kind of functions from being inlinable.
Fixes#72063
Change-Id: I5fb5723a819b1e2c5aadb57c1023ec84ca9fa53c
Reviewed-on: https://go-review.googlesource.com/c/go/+/654195
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
avo v0.4.0 x/tools dependency crashes while parsing with Go 1.25.
Change-Id: Ic951066b0b39b477887ad0e32be44f4d88d4c2f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/653175
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Fixes#68590
Change-Id: Ie7cf1fe8379182f86317d5ebb7f45a404ecd70e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/601555
Auto-Submit: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
The implementatiom still calls coreType in places and refers to
"core types" in comments, but user-visible error messages don't
know about core types anymore.
This brings the user-visible part of the implementation in sync with
the spec which doesn't have the notion of core types anymore.
For #70128.
Change-Id: I14bc6767a83e8f54b10ebe99a7df0b98cd9fca87
Reviewed-on: https://go-review.googlesource.com/c/go/+/654395
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
There is no need to call GetConsoleMode if we know that the file
type is not FILE_TYPE_CHAR. This is a tiny performance optimization,
as I sometimes see this call in profiles.
Change-Id: I9e9237908585d0ec8360930a0406b26f52699b92
Reviewed-on: https://go-review.googlesource.com/c/go/+/654155
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
For now, use commonUnder (formerly called sharedUnder) and update
error messages and comments. We can provide better error messages
in individual cases eventually.
For #70128.
Change-Id: I906ba9a0c768f6499c1683dc9be3ad27da8007a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/653156
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
For now, use commonUnder (formerly called sharedUnder) and update
error messages and comments. We can provide better error messages
in individual cases eventually.
Kepp using coreType for make built-in for now because it must accept
different channel types with non-conflicting directions and identical
element types. Added extra test cases.
While at it, rename sharedUnder, sharedUnderOrChan to commonUnder
and commonUnderOrChan, respectively (per suggestion from rfindley).
For #70128.
Change-Id: I11f3d5ce858746574f4302271d8cb763c2cdcf98
Reviewed-on: https://go-review.googlesource.com/c/go/+/653139
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Div is way faster. We could actually test a lot more primes and still
gain performance despite the diminishing returns, but necessarily it
would have marginal impact overall.
fips140: off
goos: linux
goarch: amd64
pkg: crypto/rsa
cpu: AMD Ryzen 7 PRO 8700GE w/ Radeon 780M Graphics
│ e325b41ad1 │ 0f611af2e1 │
│ sec/op │ sec/op vs base │
GenerateKey/2048-16 124.19m ± 0% 39.93m ± 0% -67.85% (p=0.000 n=20)
Surprisingly, the performance gain is similar on ARM64, which doesn't
have intrinsified math.Div.
fips140: off
goos: darwin
goarch: arm64
pkg: crypto/rsa
cpu: Apple M2
│ e325b41ad1 │ 6276161a7f │
│ sec/op │ sec/op vs base │
GenerateKey/2048-8 136.49m ± 0% 47.97m ± 1% -64.86% (p=0.000 n=20)
Change-Id: I6a6a46560331198312bd09c1cbe4d2b3c370c552
Reviewed-on: https://go-review.googlesource.com/c/go/+/639955
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
Writing explicit code for this case turned out to be simpler
and easier to reason about then relying on a helper functions
(except for typeset).
While at it, make append error messages more consistent.
For #70128.
Change-Id: I3dc79774249929de5061b4301ab2506d4b3da0d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/653095
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
CL 652276 reduced the m struct by 8 bytes, which has changed the
allocation class on 64 bit OpenBSD platforms. This results in build
failures due to:
M structure uses sizeclass 1792/0x700 bytes; incompatible with mutex flag mask 0x3ff
Add 128 bytes of padding when spinbitmutex is enabled on 64 bit
architectures, moving the size to the half point between the
1792 and 2048 allocation size.
Change-Id: I71623a1f75714543c302217e619d20cf0e717aeb
Reviewed-on: https://go-review.googlesource.com/c/go/+/653335
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
The map is automatically generated by running the latest version of
parse.py from github.com/riscv/riscv-opcodes.
Change-Id: I05e00ab27ec583750752c25e1835c2578b339fbf
Reviewed-on: https://go-review.googlesource.com/c/go/+/630518
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Pengcheng Wang <wangpengcheng.pp@bytedance.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: Michael Pratt <mpratt@google.com>
The tool IDs can be calculated once and reused across multiple
threads. This is a small optimization that helps optimize system
resources.
On a normal Windows machine with 12 virtual CPUs, the time to build
a hello world program is reduced from over 1 second, with spikes of 2
seconds, to a consistent 0.7 seconds.
Updates #71981.
Change-Id: I85f4a19f8ad4230afa32213780c761b7eb22fa29
Reviewed-on: https://go-review.googlesource.com/c/go/+/653715
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
GetPackage is sometimes called with a modroot or pkgdir that ends with a
path separator, and sometimes without. Clean them before passing them to
openIndexPackage to deduplicate calls to mcache.Do and action entry
files.
This shouldn't affect #71698 but was discovered while debugging that
issue.
Change-Id: I6a7fa4de93f45801504abea11bd97f6c6577f296
Reviewed-on: https://go-review.googlesource.com/c/go/+/652435
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>