which are erroneous because NAME_AUTO implies a base register of
REG_SP, and cmd/internal/obj/x86 makes many assumptions to this
effect. Note that previously we were also producing an extra "ADDQ
SP, CX" instruction, so indexing off of SP was already handled.
The approach taken by this CL to address the problem is to instead
produce addresses like:
Downside to this approach is it requires adding a lot of new
MOV[WLQ]loadidx1 instructions that nearly duplicate functionality of
the existing MOV[WLQ]loadidx[248] instructions, but with a different
Scale.
Fixes #15001.
Change-Id: Iad9a1a41e5e2552f8d22e3ba975e4ea0862dffd2
Reviewed-on: https://go-review.googlesource.com/21245
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Michel Lespinasse [Sat, 19 Mar 2016 00:21:33 +0000 (17:21 -0700)]
cmd/compile: optimize remaining convT2I calls
See #14874
Updates #6853
This change adds a compiler optimization for non pointer shaped convT2I.
Since itab symbols are now emitted by the compiler, the itab address can
be passed directly to convT2I instead of passing the iface type and a
cache pointer argument.
Compilebench results for the 5-commits series ending here:
Michel Lespinasse [Fri, 18 Mar 2016 23:20:20 +0000 (16:20 -0700)]
cmd/compile: optimize convT2I as a two-word copy when T is pointer-shaped
See #14874
This change adds a compiler optimization for pointer shaped convT2I.
Since itab symbols are now emitted by the compiler, the itab address can
be directly moved into the iface structure.
Change-Id: I311483af544519ca682c5f872960717ead772f26
Reviewed-on: https://go-review.googlesource.com/20901 Reviewed-by: Keith Randall <khr@golang.org> Reviewed-by: David Crawshaw <crawshaw@golang.org>
Michel Lespinasse [Thu, 17 Mar 2016 13:18:13 +0000 (06:18 -0700)]
cmd/compile: emit itabs and itablinks
See #14874
This change tells the compiler to emit itab and itablink symbols in
situations where they could be useful; however the compiled code does
not actually make use of the new symbols yet.
Change-Id: I0db3e6ec0cb1f3b7cebd4c60229e4a48372fe586
Reviewed-on: https://go-review.googlesource.com/20888 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Run-TryBot: Michel Lespinasse <walken@google.com>
Richard Miller [Sat, 26 Mar 2016 19:35:21 +0000 (19:35 +0000)]
syscall: fix accidental close of exec status pipe in StartProcess
In syscall.forkAndExecInChild, blocks of code labelled Pass 1
and Pass 2 permute the file descriptors (if necessary) which are
passed to the child process. If Pass 1 begins with fds = {0,2,1},
nextfd = 4 and pipe = 4, then the statement labelled "don't stomp
on pipe" is too late -- the pipe (which will be needed to pass
exec status back to the parent) will have been closed by the
preceding DUP call.
Moving the "don't stomp" test earlier ensures that the pipe is
protected.
Fixes #14979
Change-Id: I890c311527f6aa255be48b3277c1e84e2049ee22
Reviewed-on: https://go-review.googlesource.com/21184
Run-TryBot: David du Colombier <0intro@gmail.com> Reviewed-by: David du Colombier <0intro@gmail.com> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This mostly a mechanical change.
However, the change in assignop (subr.go) is a bug fix.
The code didn’t match the comment,
and the comment was correct.
Nevertheless, this CL passes toolstash -cmp.
The last direct reference to dddBound outside
type.go (in typecheck.go) will go away
in a future CL.
Change-Id: Ifb1691e0a07f906712c18c4a4cd23060807a5da5
Reviewed-on: https://go-review.googlesource.com/21235 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Raul Silvera [Mon, 28 Mar 2016 20:52:28 +0000 (13:52 -0700)]
cmd/pprof/internal: use and accept packed encoding for repeated fields
Packed encoding is the default on the proto3 format. Profiles generated
in the profile.proto format by third parties cannot be decoded by the
Go pprof tool, since its proto decoder does not recognize packed
encoding for repeated fields.
In particular this issue prevents go tool pprof from reading profiles
generated by the version of pprof in github.com/google/pprof
Profiles generated by go tool pprof after this change will use packed
repeating fields, so older versions of pprof will not be able to read
them. pprof will continue to be able to read profiles generated before
this change.
Change-Id: Ife0b353a535ae1e495515b9bcec588dd967e171b
Reviewed-on: https://go-review.googlesource.com/21240 Reviewed-by: David Symonds <dsymonds@golang.org>
Run-TryBot: David Symonds <dsymonds@golang.org>
Robert Griesemer [Mon, 28 Mar 2016 21:12:10 +0000 (14:12 -0700)]
cmd/compile: avoid allocation in Nodes.Set in common case
When building make.bash, calling Nodes.Set(s) where len(s) == 0 occurs 4738678 times vs 1465415 calls where len(s) > 0; i.e., it is over 3x
more common to set Nodes.slice to nil rather than to s.
Make a copy of slice (header) and take address of that copy instead
to avoid allocating the argument slice on the heap always even when
not needed.
Saves 4738678 slice header allocations and slice header value copies.
Change-Id: I88e8e919ea9868ceb2df46173d187af4109bd947
Reviewed-on: https://go-review.googlesource.com/21241 Reviewed-by: Alan Donovan <adonovan@google.com>
Michael Munday [Sat, 19 Mar 2016 04:44:17 +0000 (00:44 -0400)]
cmd/asm: add s390x support
s390x doesn't introduce any new assembly syntax. There are a few
instructions which require the operands to be reordered, notably
the storage-storage instructions that put the length into From3 so
that the memory operands can be put into From and To.
The assembly test currently covers a subset of instructions but
tries to hit edge cases as much as possible. Unlike the other ports
it can be linked as an executable to make disassembling it easy.
It would be nice to autogenerate it at some point in the future.
Change-Id: I7615ac6ecf239e3f347fad9ae1f8eede91742859
Reviewed-on: https://go-review.googlesource.com/20934
Run-TryBot: Rob Pike <r@golang.org> Reviewed-by: Rob Pike <r@golang.org>
Shinji Tanaka [Sat, 26 Mar 2016 11:14:15 +0000 (07:14 -0400)]
runtime: use set_thread_area instead of modify_ldt on linux/386
linux/386 depends on modify_ldt system call, but recent Linux kernels
can disable this system call. Any Go programs built as linux/386
crash with the message 'Trace/breakpoint trap'.
The kernel config CONFIG_MODIFY_LDT_SYSCALL, which control
enable/disable modify_ldt, is disabled on Amazon Linux 2016.03.
This fixes this problem by using set_thread_area instead of modify_ldt
on linux/386.
David Chase [Fri, 11 Mar 2016 05:10:52 +0000 (00:10 -0500)]
cmd/compile: added some intrinsics to SSA back end
One intrinsic was needed to help get the very best
performance out of a future GC; as long as that one was
being added, I also added Bswap since that is sometimes
a handy thing to have. I had intended to fill out the
bit-scan intrinsic family, but the mismatch between the
"scan forward" instruction and "count leading zeroes"
was large enough to cause me to leave it out -- it poses
a dilemma that I'd rather dodge right now.
These intrinsics are not exposed for general use.
That's a separate issue requiring an API proposal change
( https://github.com/golang/proposal )
All intrinsics are tested, both that they are substituted
on the appropriate architecture, and that they produce the
expected result.
Change-Id: I5848037cfd97de4f75bdc33bdd89bba00af4a8ee
Reviewed-on: https://go-review.googlesource.com/20564 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Shahar Kohanim [Sun, 27 Mar 2016 07:06:12 +0000 (10:06 +0300)]
cmd/link: refactor symbol lookup
Calling the read only Linkrlookup will now not cause the name
string to escape. So a lookup can be performed on a []byte
casted to a string without allocating. This will help a followup
cl and it is also much simpler and cleaner.
Performance not impacted by this.
name old s/op new s/op delta
LinkCmdGo 0.51 ± 6% 0.51 ± 5% ~ (p=0.192 n=98+98)
Change-Id: I7846ba3160eb845a3a29cbf0be703c47369ece16
Reviewed-on: https://go-review.googlesource.com/21187 Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org> Reviewed-by: David Crawshaw <crawshaw@golang.org>
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Mon, 28 Mar 2016 02:44:06 +0000 (19:44 -0700)]
cmd/compile: clear OTFUNC info when converting to OTYPE
I want to get rid of OTFUNC, which serves no useful purpose. However,
it turns out that the escape analysis pass looks at the node slices set
up for OTFUNC, even though by the time escape analysis runs the OTFUNC
has been converted to OTYPE. This CL converts the escape analysis code
to look at the function decls instead, and clears the OTFUNC info when
converting to OTYPE to ensure that nothing else looks at it.
Change-Id: I3f2f5997ea8ea7a127a858e94b20aabfab84a5bf
Reviewed-on: https://go-review.googlesource.com/21202
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Dominik Honnef [Mon, 28 Mar 2016 00:44:43 +0000 (02:44 +0200)]
cmd/go: remove code specific to Google Code
Remove all special handling of Google Code, which has shut down.
Commit 4ec2fd3e6ac4f869d39348bf48016687b731d910 suggested that maybe the
shutdown warning should remain. However, it has been missing from Go 1.6
already, and by Go 1.7 people will most likely have realised that Google
Code has shut down.
Martin Möhrmann [Tue, 15 Mar 2016 16:14:03 +0000 (17:14 +0100)]
fmt: unify array and slice formatting for bytes and other types
Make verbs b,c,o and U work for any array and slice of integer
type including byte and uint8.
Fix a bug that triggers badverb for []uint8 and []byte type
on the slice/array level instead of on each element like for
any other slice or array type.
Add tests that make sure we do not accidentally alter the
behavior of printing []byte for []byte and []uint8 type
if they are used at the top level when formatting with %#v.
name old time/op new time/op delta
SprintfHexBytes-2 177ns ± 2% 176ns ± 2% ~ (p=0.066 n=48+49)
SprintfBytes-2 330ns ± 1% 329ns ± 1% ~ (p=0.118 n=45+47)
Fixes #13478
Change-Id: I99328a184973ae219bcc0f69c3978cb1ff462888
Reviewed-on: https://go-review.googlesource.com/20686
Run-TryBot: Rob Pike <r@golang.org> Reviewed-by: Rob Pike <r@golang.org>
Martin Möhrmann [Sat, 19 Mar 2016 07:54:07 +0000 (08:54 +0100)]
fmt: cleanup reflect value handling
Merge printReflectValue into printValue. Determine if handleMethods
was already called in printArg by checking if depth is 0. Do not
call handleMethods on depth 0 again in printValue to not introduce
a performance regression. handleMethods is called already in printArg
to not introduce a performance penalty for top-level Stringer,
GoStringer, Errors and Formatters by using reflect.ValueOf on them
just to retrieve them again as interface{} values in printValue.
Clear p.arg in printValue after handleMethods to print the type
of the value inside the reflect.Value when a bad verb is encountered
on the top level instead of printing "reflect.Value=" as the type of
the argument. This also fixes a bug that incorrectly prints the
whole map instead of just the value for a key if the returned value
by the map for the key is an invalid reflect value.
Also updates the pattern matching algorithm in
https://mimesniff.spec.whatwg.org/#matching-a-mime-type-pattern
by implementing clause 4 that dictates that the number of bytes in
the pattern must match the number of bytes in the mask.
Christopher Nelson [Thu, 24 Mar 2016 21:50:21 +0000 (17:50 -0400)]
misc/cgo/testcarchive: split main.c into unix and non-unix
This is in support of https://golang.org/cl/18057 which adds
support for c-archive to the Windows platform.
The signal handling tests do not compile on Windows. This splits
them out into a separate main_unix.c file, and conditionally
includes them for non-Windows platforms.
Change-Id: Ic79ce83da7656d6703505e514554748a482b81a1
Reviewed-on: https://go-review.googlesource.com/21086 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Robert Griesemer [Fri, 25 Mar 2016 22:34:55 +0000 (15:34 -0700)]
cmd/compile: reduce slice header allocation when parsing := assignments
The colas function allocates 2 slice headers in each call (via Nodes.Set)
only to throw away those slice headers in the common case where both the
lhs and rhs in "lhs := rhs" have length 1.
Avoid the Nodes.Set calls in those cases. For make.bash, this eliminates
~63,000 slice header allocations.
Shenghou Ma [Thu, 24 Mar 2016 23:35:10 +0000 (19:35 -0400)]
runtime, runtime/cgo: save callee-saved FP registers on arm64
For #14876.
Change-Id: I0992859264cbaf9c9b691fad53345bbb01b4cf3b
Reviewed-on: https://go-review.googlesource.com/21085 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Shenghou Ma [Thu, 24 Mar 2016 23:12:44 +0000 (19:12 -0400)]
runtime/cgo: save callee-saved xmm registers on windows/amd64
For #14876.
Change-Id: I33947f74e8058437a784862f1f064974afc99250
Reviewed-on: https://go-review.googlesource.com/21084 Reviewed-by: Alex Brainman <alex.brainman@gmail.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
David Benjamin [Tue, 5 Jan 2016 00:11:02 +0000 (16:11 -0800)]
encoding/asn1: tags should be encoded in minimal form.
High tag number form may not be used for tag numbers that fit in low tag number
form.
Change-Id: I93edde0e1f86087047e0b3f2e55d6180b01e78bf
Reviewed-on: https://go-review.googlesource.com/18224 Reviewed-by: Adam Langley <agl@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Joe Sylve [Fri, 25 Mar 2016 18:51:56 +0000 (13:51 -0500)]
runtime: fix signal handling on Solaris
This fixes the problems with signal handling that were inadvertently
introduced in https://go-review.googlesource.com/21006.
Fixes #14899
Change-Id: Ia746914dcb3146a52413d32c57b089af763f0810
Reviewed-on: https://go-review.googlesource.com/21145 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Mohit Agarwal [Thu, 24 Mar 2016 17:16:10 +0000 (22:46 +0530)]
cmd/go: stat the archive file only when executing the commands
Fixes #14944
Change-Id: I73e0997cb6ebaeced1045b0ddadac893319bd78f
Reviewed-on: https://go-review.googlesource.com/21065 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
David Crawshaw [Fri, 25 Mar 2016 15:03:47 +0000 (11:03 -0400)]
reflect: name interface type that pins method info
I recently added TestUnexportedMethods which uses an interface type
to pin type information for an unexported method. But as written,
the interface type is not accessible to the reflect package.
You can imagine a future compiler optimization realizing that and
removing the type information for f. In fact, cl/20901 happens to
do that.
Change-Id: I1ddb67f50cb9b5737253b58f10545f3de652c29d
Reviewed-on: https://go-review.googlesource.com/21112 Reviewed-by: Michel Lespinasse <walken@google.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Change-Id: Ice5fa648058fe9d7fa89dee97ca359966f671128
Reviewed-on: https://go-review.googlesource.com/21101 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Elias Naur [Fri, 25 Mar 2016 14:40:44 +0000 (15:40 +0100)]
misc/ios: fix exec wrapper locking
The exec wrapper lock file was opened, locked and then never used
again, assuming it would close and unlock at process exit.
However, the garbage collector could collect and run the *os.File
finalizer that closes the file prematurely, rendering the lock
ineffective.
Make the lock global so that the lock is live during the entire
execution.
(Hopefully) fix the iOS builders.
Change-Id: I62429e92042a0a49c4f1ea553fdb32b6ea53a43e
Reviewed-on: https://go-review.googlesource.com/21137 Reviewed-by: David Crawshaw <crawshaw@golang.org>
David Crawshaw [Fri, 25 Mar 2016 02:32:38 +0000 (22:32 -0400)]
cmd/compile: avoid pointers in go.string.* symbols
When creating binaries for dynamic linking, the linker moves
read-only data symbols that contain pointers into relro sections.
It is not setup for handling a go.string symbol moving to relro.
Instead of teaching it how (because go.string symbols with pointers
are unusual anyhow), put the data in a type.. section.
Richard Miller [Fri, 25 Mar 2016 12:50:35 +0000 (12:50 +0000)]
runtime: avoid fork/exit race in plan9
There's a race between runtime.goexitsall killing all OS processes
of a go program in order to exit, and runtime.newosproc forking a
new one. If the new process has been created but not yet stored
its pid in m.procid, it will not be killed by goexitsall and
deadlock results.
This CL prevents the race by making the newly forked process
check whether the program is exiting. It also prevents a
potential "shoot-out" if multiple goroutines call Exit at
the same time, which could possibly lead to two processes
killing each other and leaving the rest deadlocked.
Change-Id: I3170b4a62d2461f6b029b3d6aad70373714ed53e
Reviewed-on: https://go-review.googlesource.com/21135
Run-TryBot: David du Colombier <0intro@gmail.com> Reviewed-by: Marvin Stenger <marvin.stenger94@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David du Colombier <0intro@gmail.com>
Dmitry Vyukov [Fri, 18 Mar 2016 11:52:52 +0000 (12:52 +0100)]
runtime: improve randomized stealing logic
During random stealing we steal 4*GOMAXPROCS times from random procs.
One would expect that most of the time we check all procs this way,
but due to low quality PRNG we actually miss procs with frightening
probability. Below are modelling experiment results for 1e6 tries:
GOMAXPROCS = 2 : missed 1 procs 7944 times
GOMAXPROCS = 3 : missed 1 procs 101620 times
GOMAXPROCS = 3 : missed 2 procs 3571 times
GOMAXPROCS = 4 : missed 1 procs 63916 times
GOMAXPROCS = 4 : missed 2 procs 61 times
GOMAXPROCS = 4 : missed 3 procs 16 times
GOMAXPROCS = 5 : missed 1 procs 133136 times
GOMAXPROCS = 5 : missed 2 procs 1025 times
GOMAXPROCS = 5 : missed 3 procs 101 times
GOMAXPROCS = 5 : missed 4 procs 15 times
GOMAXPROCS = 8 : missed 1 procs 151765 times
GOMAXPROCS = 8 : missed 2 procs 5057 times
GOMAXPROCS = 8 : missed 3 procs 1726 times
GOMAXPROCS = 8 : missed 4 procs 68 times
GOMAXPROCS = 12 : missed 1 procs 199081 times
GOMAXPROCS = 12 : missed 2 procs 27489 times
GOMAXPROCS = 12 : missed 3 procs 3113 times
GOMAXPROCS = 12 : missed 4 procs 233 times
GOMAXPROCS = 12 : missed 5 procs 9 times
GOMAXPROCS = 16 : missed 1 procs 237477 times
GOMAXPROCS = 16 : missed 2 procs 30037 times
GOMAXPROCS = 16 : missed 3 procs 9466 times
GOMAXPROCS = 16 : missed 4 procs 1334 times
GOMAXPROCS = 16 : missed 5 procs 192 times
GOMAXPROCS = 16 : missed 6 procs 5 times
GOMAXPROCS = 16 : missed 7 procs 1 times
GOMAXPROCS = 16 : missed 8 procs 1 times
A missed proc won't lead to underutilization because we check all procs
again after dropping P. But it can lead to an unpleasant situation
when we miss a proc, drop P, check all procs, discover work, acquire P,
miss the proc again, repeat.
Improve stealing logic to cover all procs.
Also don't enter spinning mode and try to steal when there is nobody around.
Marcel van Lohuizen [Wed, 23 Mar 2016 20:24:22 +0000 (21:24 +0100)]
testing: probe with N=1
Change control flow to probe with N=1. This calls benchFunc
the same number of times as the old implementation in the
absence of subbenchmarks.
To be compatible with existing tools, benchmarking only
prints a line for "leaf" benchmarks. This means, though, that
the name of a benchmark can only be printed after the first
iteration.
Dominik Honnef [Sun, 20 Mar 2016 23:12:18 +0000 (00:12 +0100)]
all: delete dead non-test code
This change removes a lot of dead code. Some of the code has never been
used, not even when it was first commited. The rest shouldn't have
survived refactors.
This change doesn't remove unused routines helpful for debugging, nor
does it remove code that's used in commented out blocks of code that are
only unused temporarily. Furthermore, unused constants weren't removed
when they were part of a set of constants from specifications.
One noteworthy omission from this CL are about 1000 lines of unused code
in cmd/fix, 700 lines of which are the typechecker, which hasn't been
used ever since the pre-Go 1 fixes have been removed. I wasn't sure if
this code should stick around for future uses of cmd/fix or be culled as
well.
Dmitry Chestnykh [Tue, 22 Mar 2016 15:41:11 +0000 (16:41 +0100)]
crypto/hmac: simplify implementation
Store already padded keys instead of storing key and padding it during
Reset and Sum. This simplifies code and makes Reset-Write-Sum sequences
faster, which helps /x/crypto/pbkdf2.
HMAC benchmark:
benchmark old ns/op new ns/op delta
BenchmarkHMACSHA256_1K-4 7669 7613 -0.73%
BenchmarkHMACSHA256_32-4 1880 1737 -7.61%
benchmark old MB/s new MB/s speedup
BenchmarkHMACSHA256_1K-4 133.52 134.50 1.01x
BenchmarkHMACSHA256_32-4 17.02 18.41 1.08x
PBKDF2 benchmark:
benchmark old ns/op new ns/op delta
BenchmarkPBKDF2HMACSHA256-4 19431961807699 -6.97%
Elias Naur [Thu, 24 Mar 2016 19:45:54 +0000 (20:45 +0100)]
cmd/go: remove limits on parallel tasks in the Go command for iOS
Ther darwin/arm{,64} exec wrapper now limits the number of concurrent
executions to 1, so remove the higher level parallel task limit from
the Go command.
Change-Id: Id84f65c3908305bde0452b3c8db6df8c5a8881bb
Reviewed-on: https://go-review.googlesource.com/21100 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Elias Naur [Fri, 25 Mar 2016 00:27:34 +0000 (01:27 +0100)]
misc/cgo/testcarchive: fix build
I failed to rebase (and re-test) CL 21102 before submit, which meant
that two extra tests sneaked into testcarchive that still referenced
runtime.GOOS and runtime.GOARCH.
Convert the new tests.
While we're here, make sure pending tasks are flushed before running
the host tests. If not, the "##### misc/cgo/testcarchive" banner
and "PASS" won't show up in the all.bash output.
David Crawshaw [Mon, 21 Mar 2016 17:21:55 +0000 (13:21 -0400)]
cmd/compile, runtime: new static name encoding
Create a byte encoding designed for static Go names.
It is intended to be a compact representation of a name
and optional tag data that can be turned into a Go string
without allocating, and describes whether or not it is
exported without unicode table.
The encoding is described in reflect/type.go:
// The first byte is a bit field containing:
//
// 1<<0 the name is exported
// 1<<1 tag data follows the name
// 1<<2 pkgPath *string follow the name and tag
//
// The next two bytes are the data length:
//
// l := uint16(data[1])<<8 | uint16(data[2])
//
// Bytes [3:3+l] are the string data.
//
// If tag data follows then bytes 3+l and 3+l+1 are the tag length,
// with the data following.
//
// If the import path follows, then ptrSize bytes at the end of
// the data form a *string. The import path is only set for concrete
// methods that are defined in a different package than their type.
Shrinks binary sizes:
cmd/go: 164KB (1.6%)
jujud: 1.0MB (1.5%)
For #6853.
Change-Id: I46b6591015b17936a443c9efb5009de8dfe8b609
Reviewed-on: https://go-review.googlesource.com/20968
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Elias Naur [Thu, 24 Mar 2016 20:47:02 +0000 (21:47 +0100)]
misc/cgo/testcarchive: re-enable c-archive test on iOS and Android
The c-archive test were recently converted from shell script to Go.
Unfortunately, it also lost the ability to target iOS and Android
that lack C compilers and require exec wrappers.
Compile the c-archive test for the host and run it with the target
GOOS/GOARCH environment. Change the test to rely on go env GOOS
and go env GOARCH instead of runtime.GOOS and runtime.GOARCH.
Fixes #8345
Change-Id: I290ace2f7e96b87c55d99492feb7d660140dcb32
Reviewed-on: https://go-review.googlesource.com/21102 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Alexandru Moșoi [Thu, 24 Mar 2016 12:40:11 +0000 (13:40 +0100)]
cmd/compile: simplify shifts when the counter fits 6 bits.
In f the extra & 63 is redundant because SHRQ already
looks at the bottom 6 bits only. This is a trick on AMD64
to get rid of CMPQ/SBBQ/ANDQ if one knows that the shift
counter is small.
func f(x uint64, s uint) uint64 {
return x >> (s & 63)
}
Change-Id: I4861c902168dabec9a6a14a85750246dde94fc08
Reviewed-on: https://go-review.googlesource.com/21073 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Alexandru Moșoi <alexandru@mosoi.ro>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Joe Sylve [Wed, 23 Mar 2016 03:11:42 +0000 (22:11 -0500)]
runtime: improve last ditch signal forwarding for Unix libraries
The current runtime attempts to forward signals generated by non-Go
code to the original signal handler. If it can't call the original
handler directly, it currently attempts to re-raise the signal after
resetting the handler. In this case, the original context is lost.
This fix prevents that problem by simply returning from the go signal
handler after resetting the original handler. It only does this when
the original handler is the system default handler, which in all cases
is known to not recover. The signal is not reset, so it is retriggered
and the original handler takes over with the proper context.
Fixes #14899
Change-Id: Ib1c19dfa4b50d9732d7a453de3784c8141e1cbb3
Reviewed-on: https://go-review.googlesource.com/21006 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Elias Naur [Thu, 24 Mar 2016 18:55:40 +0000 (19:55 +0100)]
os: skip TestStatStdin on Android
Android doesn't (generally) have /bin/sh.
Change-Id: I343817c342e3473d09c85155761682b5ddb043e4
Reviewed-on: https://go-review.googlesource.com/21075 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Marvin Stenger [Thu, 24 Mar 2016 00:55:42 +0000 (01:55 +0100)]
runtime: fix inconsistency in slice.go
Fixes #14938.
Additionally some simplifications along the way.
Change-Id: I2c5fb7e32dcc6fab68fff36a49cb72e715756abe
Reviewed-on: https://go-review.googlesource.com/21046 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Elias Naur [Thu, 24 Mar 2016 15:03:07 +0000 (16:03 +0100)]
misc/ios: serialize iOS execution
The iOS exec wrapper use complicated machinery to run a iOS binary
on a device.
Running several binaries concurrently doesn't work (reliably), which
can break tests running concurrently. For my setup, the
runtime:cpu124 and sync_cpu tests can't run reliably without one of them
crashing.
Add a file lock to the exec wrapper to serialize execution.
Fixes #14318 (for me)
Change-Id: I023610e014b327f8d66f1d2fd2e54dd0e56f2be0
Reviewed-on: https://go-review.googlesource.com/21074 Reviewed-by: David Crawshaw <crawshaw@golang.org>
cmd/compile: avoid allocating a Nodes for readonly method receivers
We were allocating a Nodes for common method calls
that did not modify the Nodes.
Though there is no clear wall time impact,
this significantly reduces the number of allocations,
so it seems worth doing.
Elias Naur [Thu, 24 Mar 2016 09:16:46 +0000 (10:16 +0100)]
cmd/dist: skip testcarchive test in Android and iOS
CL 20892 converted the misc/cgo/testcarchive test to Go.
Unfortunately, dist does not (yet) support tests running off the host
so the testcarchive is disabled for now.
For #14318
Change-Id: Iab3d0a7b5309187a603b48f22a7fa736f089f89d
Reviewed-on: https://go-review.googlesource.com/21070 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Elias Naur [Wed, 23 Mar 2016 15:17:44 +0000 (16:17 +0100)]
misc/ios: deflake tests on darwin/arm{,arm64}
A retry mechanism is in place to combat the inherent flakiness of
launching iOS test binaries. Before it covered just the starting of
lldb; expand it to cover the setup steps as well. Note that the
running of the binary itself is (still) not retried, to avoid
covering over genuine bugs.
On my test device (iPhone 5S, iOS 9.3) starting lldb can take longer
than 10 seconds, so increase the timeout for that.
Furthermore, some basic steps such as setting breakpoints in lldb
can take longer than the 1 second timeout. Increase that timeout
as well, to 2 seconds.
Finally, improve the error message for when ios-deploy is not
installed.
For #14318
Change-Id: Iba41d1bd9d023575b9454cb577b08f8cae081c2a
Reviewed-on: https://go-review.googlesource.com/21072 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Elias Naur [Thu, 24 Mar 2016 07:39:54 +0000 (08:39 +0100)]
runtime/cgo: block signals to the iOS mach exception handler
For darwin/arm{,64} a non-Go thread is created to convert
EXC_BAD_ACCESS to panics. However, the Go signal handler refuse to
handle signals that would otherwise be ignored if they arrive at
non-Go threads.
Block all (posix) signals to that thread, making sure that
no unexpected signals arrive to it. At least one test, TestStop in
os/signal, depends on signals not arriving on any non-Go threads.
For #14318
Change-Id: I901467fb53bdadb0d03b0f1a537116c7f4754423
Reviewed-on: https://go-review.googlesource.com/21047 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Tilman Dilo [Wed, 23 Mar 2016 21:38:52 +0000 (22:38 +0100)]
image/png: ignore trailing IDAT chunks
Ignore superfluous trailing IDAT chunks which were not consumed when decoding
the image. This change fixes decoding of valid images in which a zero-length
IDAT chunk appears after the actual image data. It also prevents decoding of
trailing garbage IDAT chunks or maliciously embedded additional images.
Fixes #14936
Change-Id: I8c76cfa9a03496d9576f72bed2db109271f97c5e
Reviewed-on: https://go-review.googlesource.com/21045 Reviewed-by: Nigel Tao <nigeltao@golang.org>
Remove almost all the remaining Nod(OXXX, ... ) uses. The performance
change is due entirely to the changes to func temp(*Type). The other
cleanups have no effect, as expected.
I'll address the remaining Nod(OXXX, ...) uses in a followup CL as they
are very sensitive to change.
Brad Fitzpatrick [Sun, 20 Mar 2016 20:20:01 +0000 (07:20 +1100)]
cmd/compile: reduce some SSA garbage
It's pretty hard to get reliable CPU numbers, even with 50 runs on an
otherwise-idle physical Linux machine, but the garbage reduction
numbers are nice. To get useful time/op numbers, I modified
compilebench to report user CPU time instead of wall time:
Matthew Dempsky [Wed, 23 Mar 2016 18:07:20 +0000 (11:07 -0700)]
cmd/compile: small Mpint method simplifications
Get rid of (*Mpint).Add's "quiet" parameter: it's always set to 0.
Inline (*Mpint).shift into (*Mpint).Lsh and (*Mpint).Rsh. There's no
need for a common shift method that can handle both left or right
shifts based on sign when the higher level abstractions only ever do
one or the other.
Change-Id: Icd3b082413f9193961b6835279e0bd4b6a6a6621
Reviewed-on: https://go-review.googlesource.com/21050
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
Alexandru Moșoi [Wed, 23 Mar 2016 10:11:34 +0000 (11:11 +0100)]
cmd/compile: don't simplify nilchecks in loops
khr: Lifting the nil check out of the loop altogether is an admirable
goal, and this rewrite is one step on the way. But without lifting it
out of the loop, the rewrite is just hurting us.
Fixes #14917
Change-Id: Idb917f37d89f50f8e046d5ebd7c092b1e0eb0633
Reviewed-on: https://go-review.googlesource.com/21040 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Alexandru Moșoi <alexandru@mosoi.ro>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Lynn Boger [Mon, 7 Mar 2016 17:59:16 +0000 (11:59 -0600)]
bytes: Equal perf improvements on ppc64le/ppc64
The existing implementation for Equal and similar
functions in the bytes package operate on one byte at
at time. This performs poorly on ppc64/ppc64le especially
when the byte buffers are large. This change improves
those functions by loading and comparing double words where
possible. The common code has been moved to a function
that can be shared by the other functions in this
file which perform the same type of comparison.
Further optimizations are done for the case where
>= 32 bytes are being compared. The new function
memeqbody is used by memeq_varlen, Equal, and eqstring.
When running the bytes test with -test.bench=Equal
Klaus Post [Mon, 21 Mar 2016 09:21:55 +0000 (10:21 +0100)]
compress/flate: rework matching algorithm
This changes how matching is done in deflate algorithm.
The major change is that we do not look for matches that are only
3 bytes in length, matches must be 4 bytes at least.
Contrary to what you would expect this actually improves the
compresion ratio, since 3 literal bytes will often be shorter
than a match after huffman encoding.
This varies a bit by source, but is most often the case when the
source is "easy" to compress.
Second of all, a "stronger" hash is used. The hash is similar to
the hashing function used by Snappy.
Overall, the speed impact is biggest on higher compression levels.
I intend to replace the "speed" compression level, which can be
seen in CL 21021.
The built-in benchmark using "digits" is slower at level 1.
I see this as an exception, since "digits" is a special type
of data, where you have low entropy (numbers 0->9), but no
significant matches. Again, CL 20021 fixes that case.
NewWriterDict is also made considerably faster, by not running data
through the entire encoder. This is not reflected by the benchmark.
Overall, the speed impact is biggest on higher compression levels.
I intend to replace the "speed" compression level.