Dmitriy Vyukov [Sat, 15 Jun 2013 12:07:06 +0000 (16:07 +0400)]
runtime: fix race condition between GC and setGCPercent
If first GC runs concurrently with setGCPercent,
it can overwrite gcpercent value with default.
Dmitriy Vyukov [Sat, 15 Jun 2013 12:06:28 +0000 (16:06 +0400)]
runtime: improve scheduler fairness
Currently global runqueue is starved if a group of goroutines
constantly respawn each other (local runqueue never becomes empty).
Fixes #5639.
Dmitriy Vyukov [Sat, 15 Jun 2013 12:02:39 +0000 (16:02 +0400)]
runtime: remove unused moreframesize_minalloc field
It was used to request large stack segment for GC
when it was running not on g0.
Now GC is running on g0 with large stack,
and it is not needed anymore.
R=golang-dev, dave
CC=golang-dev
https://golang.org/cl/10242045
Brad Fitzpatrick [Fri, 14 Jun 2013 15:59:43 +0000 (08:59 -0700)]
net: coalesce duplicate in-flight DNS lookups
In Issue 5625, Russ says: "We should at least have a cache of
inflight lookups, so that 100 simultaneous dials of one host
name don't do the work 100x. That's easy and (assume we forget
the answer once they all get it) doesn't pose any consistency
problems. It just merges simultaneous work."
This brings in singleflight (unexported) from Google /
Camlistore, but without its tests. Maybe we should put it
somewhere in the standard library. But not now.
Rémy Oudompheng [Fri, 14 Jun 2013 09:14:45 +0000 (11:14 +0200)]
cmd/gc: instrument arrays properly in race detector.
The previous implementation would only record access to
the address of the array but the memory access to the whole
memory range must be recorded instead.
R=golang-dev, dvyukov, r
CC=golang-dev
https://golang.org/cl/8053044
Rob Pike [Thu, 13 Jun 2013 01:13:34 +0000 (18:13 -0700)]
testing: add -outputdir flag so "go test" controls where the files are written
Obscure misfeature now fixed: When run from "go test", profiles were always
written in the package's source directory. This change puts them in the directory
where "go test" is run.
Also fix a couple of problems causing errors in testing.after to go unreported
unless -v was set.
Dmitriy Vyukov [Wed, 12 Jun 2013 14:46:35 +0000 (18:46 +0400)]
runtime: fix scheduler race condition
In starttheworld() we assume that P's with local work
are situated in the beginning of idle P list.
However, once we start the first M, it can execute all local G's
and steal G's from other P's.
That breaks the assumption above. Thus starttheworld() will fail
to start some P's with local work.
It seems that it can not lead to very bad things, but still
it's wrong and breaks other assumtions
(e.g. we can have a spinning M with local work).
The fix is to collect all P's with local work first,
and only then start them.
The garbage collection routine addframeroots is duplicating
logic in the traceback routine that calls it, sometimes correctly,
sometimes incorrectly, sometimes incompletely.
Pass necessary information to addframeroots instead of
deriving it anew.
Should make addframeroots significantly more robust.
It's certainly smaller.
Also try to standardize on uintptr for saved pc, sp values.
Ian Lance Taylor [Wed, 12 Jun 2013 03:23:21 +0000 (20:23 -0700)]
cmd/gc: save local var list before inlining
This avoids problems with inlining in genwrappers, which
occurs after functions have been compiled. Compiling a
function may cause some unused local vars to be removed from
the list. Since a local var may be unused due to
optimization, it is possible that a removed local var winds up
beingused in the inlined version, in which case hilarity
ensues.
Fixes #5515.
R=golang-dev, khr, dave
CC=golang-dev
https://golang.org/cl/10210043
Russ Cox [Wed, 12 Jun 2013 02:37:07 +0000 (22:37 -0400)]
cmd/gc: turn race detector off for tail-call method wrapper functions
It was off in the old implementation (because there was no high-level
description of the function at all). Maybe some day the race detector
should be fixed to handle the wrapper and then enabled for it, but there's
no reason that has to be today.
Russ Cox [Tue, 11 Jun 2013 20:49:24 +0000 (16:49 -0400)]
runtime: rename _rt0_$GOARCH to _rt0_go
There's no reason to use a different name on each architecture,
and doing so makes it impossible for portable code to refer to
the original Go runtime entry point. Rename it _rt0_go everywhere.
Rob Pike [Tue, 11 Jun 2013 16:35:10 +0000 (09:35 -0700)]
cmd/go: add coverage analysis
This feature is not yet ready for real use. The CL marks a bite-sized
piece that is ready for review. TODOs that remain:
provide control over output
produce output without setting -v
make work on reflect, sync and time packages
(fail now due to link errors caused by inlining)
better documentation
Almost all packages work now, though, if clumsily; try:
go test -v -cover=count encoding/binary
Russ Cox [Tue, 11 Jun 2013 13:41:49 +0000 (09:41 -0400)]
cmd/gc: move genembedtramp into portable code
Requires adding new linker instruction
RET f(SB)
meaning return but then immediately call f.
This is what you'd use to implement a tail call after
fiddling with the arguments, but the compiler only
uses it in genwrapper.
This CL eliminates the copy-and-paste genembedtramp
functions from 5g/8g/6g and makes the code run on ARM
for the first time. It removes a small special case for function
generation, which should help Carl a bit, but at the same time
it does not bother to implement general tail call optimization,
which we do not want anyway.
Gerasimos Dimitriadis [Mon, 10 Jun 2013 22:14:47 +0000 (18:14 -0400)]
asn1: Improved encoding/decoding for OID types
The first identifier in an Object Identifer must be between 0 and 2
inclusive. The range of values that the second one can take depends
on the value of the first one.
The two first identifiers are not necessarily encoded in a single octet,
but in a varint.
Russ Cox [Mon, 10 Jun 2013 20:13:25 +0000 (16:13 -0400)]
cmd/cc: fix lexbody for negative chars
The new code matches the code in cc/lex.c and the #define GETC.
This was causing problems scanning runtime·foo if the leading
· byte was returned by the buffer fill.
Dmitriy Vyukov [Mon, 10 Jun 2013 18:38:58 +0000 (22:38 +0400)]
sync: improve race instrumentation of WaitGroup
Do not synchronize Add(1) with Wait().
Imitate read on first Add(1) and write on Wait(),
it allows to catch common misuses of WaitGroup:
- Add() called in the additional goroutine itself
- incorrect reuse of WaitGroup with multiple waiters
Dmitriy Vyukov [Mon, 10 Jun 2013 05:20:27 +0000 (09:20 +0400)]
runtime: use persistentalloc instead of SysAlloc in FixAlloc
Also reduce FixAlloc allocation granulatiry from 128k to 16k,
small programs do not need that much memory for MCache's and MSpan's.
Dmitriy Vyukov [Thu, 6 Jun 2013 10:56:50 +0000 (14:56 +0400)]
runtime: speedup malloc stats collection
Count only number of frees, everything else is derivable
and does not need to be counted on every malloc.
benchmark old ns/op new ns/op delta
BenchmarkMalloc8 68 66 -3.07%
BenchmarkMalloc16 75 70 -6.48%
BenchmarkMallocTypeInfo8 102 97 -4.80%
BenchmarkMallocTypeInfo16 108 105 -2.78%
Adam Langley [Wed, 5 Jun 2013 00:02:22 +0000 (20:02 -0400)]
crypto/tls: support TLS 1.1.
The significant change between TLS 1.0 and 1.1 is the addition of an explicit IV in the case of CBC encrypted records. Support for TLS 1.1 is needed in order to support TLS 1.2.
Anthony Martin [Tue, 4 Jun 2013 22:02:28 +0000 (15:02 -0700)]
codereview: don't warn about secret commits
Normally the codereview plugin disables adding new commits
when not using the submit command. Unfortunately this makes
it impossible to use the Mercurial Queues extension at the
same time.
A feature called "Phases" was introduced in Mercurial 2.1
that allows marking commits as being secret; this means
they will never be pushed to the upstream repository.
We can take advantage of this feature to allow the use of
Mercurial Queues if the mq.secret option has been set in
hgrc(5) and a recent version of Mercurial is used.
Anthony Martin [Tue, 4 Jun 2013 14:07:22 +0000 (07:07 -0700)]
cmd/ld: fix gcdata and gcbss symbols
These two symbols don't show up in the Go symbol table
since they're defined in dodata which is called sometime
after symtab. They do, however, show up in the ELF symbol
table.
This regression was introduced in changeset 01c40d533367.
Also, remove the corresponding strings from the ELF strtab
section now that they're unused.