Joel Sing [Tue, 15 May 2012 15:52:20 +0000 (01:52 +1000)]
runtime: fix netbsd runtime defs
Fix and regenerate runtime defs for NetBSD.
Whilst the mcontext struct can be handled across architectures,
the registers are provided as defines that index an array, rather
than as members of the struct. Since these are architecture
dependent, include them via a defs_netbsd_<arch>.go file.
Dmitriy Vyukov [Tue, 15 May 2012 15:10:16 +0000 (19:10 +0400)]
runtime: refactor helpgc functionality in preparation for parallel GC
Parallel GC needs to know in advance how many helper threads will be there.
Hopefully it's the last patch before I can tackle parallel sweep phase.
The benchmarks are unaffected.
Joel Sing [Tue, 15 May 2012 01:26:05 +0000 (11:26 +1000)]
runtime: fix TLS handling for netbsd
Set the TLS base using the _lwp_setprivate() syscall, instead of via
sysarch(). NetBSD tracks the pointer passed to _lwp_setprivate() and
restores this value when restoring mcontext. If sysarch() is used
directly, restoring an mcontext trashes the FS/GS value, resulting
in a segfault when we next try to access the TLS.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6206062
Brad Fitzpatrick [Tue, 15 May 2012 01:16:47 +0000 (18:16 -0700)]
mime/multipart: fix handling of empty parts without CRLF before next part
Empty parts can be either of the form:
a) "--separator\r\n", header (w/ trailing 2xCRLF), \r\n "--separator"...
or
b) "--separator\r\n", header (w/ trailing 2xCRLF), "--separator"...
We never handled case b). In fact the RFC seems kinda vague about
it, but browsers seem to do a), and App Engine's synthetic POST
bodies after blob uploads is of form b).
So handle them both, and add a bunch of tests.
(I can't promise these are the last fixes to multipart, especially
considering its history, but I'm growing increasingly confident at
least, and I've never submitted a multipart CL with known bugs
outstanding, including this time.)
Dmitriy Vyukov [Mon, 14 May 2012 15:27:29 +0000 (19:27 +0400)]
sync: use atomic.Store in Once.Do
No perf/semantic changes, merely improves code health.
There were several questions as to why Once.Do uses
atomic.CompareAndSwap to do a store.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6208057
Dmitriy Vyukov [Fri, 11 May 2012 09:30:34 +0000 (13:30 +0400)]
runtime: revert MaxGcproc from 16 to 4
The change accidentally come in with this revision:
https://code.google.com/p/go/source/detail?spec=svn345cbca96c5550f2e89bc727703301933802923c&r=14c38c23c819a17021b1808cf4a34ef3a1a17db5
Marcel van Lohuizen [Wed, 9 May 2012 10:03:55 +0000 (12:03 +0200)]
exp/locale/collate: Add maketables tool and generated tables.
Also set maxContractLen automatically.
Note that the table size is much bigger than it needs to be.
Optimization is best done, though, when the language specific
tables are added.
Nigel Tao [Wed, 9 May 2012 00:43:15 +0000 (10:43 +1000)]
go: fix the import path "./..." not matching ".".
Tested manually.
Fixes #3554.
Before:
$ cd $GOROOT/src/pkg
$ go list io
io
$ go list io/...
io
io/ioutil
$ cd $GOROOT/src/pkg/io
$ go list .
io
$ go list ./...
io/ioutil
After:
$ cd $GOROOT/src/pkg
$ go list io
io
$ go list io/...
io
io/ioutil
$ cd $GOROOT/src/pkg/io
$ go list .
io
$ go list ./...
io
io/ioutil
$ go list ././...
io
io/ioutil
$ go list ././.././io/...
io
io/ioutil
$ go list ../image
image
$ go list ../image/...
image
image/color
image/draw
image/gif
image/jpeg
image/png
$ go list ../.../template
html/template
text/template
$ cd $GOROOT/src/pkg
$ go list ./io
io
$ go list ./io/...
io
io/ioutil
$ go list ./.../pprof
net/http/pprof
runtime/pprof
$ go list ./compress
can't load package: package compress: no Go source files in /home/nigeltao/go/src/pkg/compress
$ go list ./compress/...
compress/bzip2
compress/flate
compress/gzip
compress/lzw
compress/zlib
$ cd $GOROOT/src/pkg/code.google.com
$ go list ./p/leveldb-go/...
code.google.com/p/leveldb-go/leveldb
code.google.com/p/leveldb-go/leveldb/crc
code.google.com/p/leveldb-go/leveldb/db
code.google.com/p/leveldb-go/leveldb/memdb
code.google.com/p/leveldb-go/leveldb/memfs
code.google.com/p/leveldb-go/leveldb/record
code.google.com/p/leveldb-go/leveldb/table
code.google.com/p/leveldb-go/manualtest/filelock
$ go list ./p/.../truetype
code.google.com/p/freetype-go/example/truetype
code.google.com/p/freetype-go/freetype/truetype
$ go list ./p/.../example
warning: "./p/.../example" matched no packages
$ go list ./p/.../example/...
code.google.com/p/freetype-go/example/freetype
code.google.com/p/freetype-go/example/gamma
code.google.com/p/freetype-go/example/raster
code.google.com/p/freetype-go/example/round
code.google.com/p/freetype-go/example/truetype
code.google.com/p/x-go-binding/example/imgview
code.google.com/p/x-go-binding/example/xgb
Nigel Tao [Wed, 9 May 2012 00:02:28 +0000 (10:02 +1000)]
compress/testdata: change {e,pi}.txt from 10k to 100k digits.
These files change from exactly 10003 bytes long to 100003: a digit,
a '.', 100k digits, and a '\n'.
The magic constants in compress/flate/deflate_test.go change since
deflateInflateStringTests checks that the compressed form of e.txt
is not 'too large'. I'm not exactly sure how these numbers were
originally calculated (they were introduced in codereview 5554066
"make lazy matching work"); perhaps krasin@golang.org can comment.
My change was to increase the first one (no compression) to a tight
bound, and multiply all the others by 10.
Benchcmp numbers for compress/flate and compress/lzw below. LZW's
window size of 4096 is less than 10k, so shows no significant change.
Flate's window size is 32768, between 10k and 100k, and so the .*1e5
and .*1e6 benchmarks show a dramatic drop, since the compressed forms
are no longer a trivial forward copy of 10k digits repeated over and
over, but should now be more representative of real world usage.
Nigel Tao [Tue, 8 May 2012 22:57:54 +0000 (08:57 +1000)]
compress/flate: benchmark some English text, not just the digits of e.
The testdata/e.txt input is repeated on the longer benchmarks, but the
length of that data is less than flate's window size, so the numbers are
essentially measuring the performance of a trivial compression. A follow-up
CL will add more data to testdata/e.txt.
Nigel Tao [Tue, 8 May 2012 00:41:45 +0000 (10:41 +1000)]
compress/testdata: remove Byte Order Mark from the Tom Sawyer data.
I'm not sure where the BOM came from, originally.
http://www.gutenberg.org/files/74/74.txt doesn't have it, although
a fresh download of that URL gives me "\r\n"s instead of plain "\n"s,
and the extra line "Character set encoding: ASCII". Maybe Project
Gutenberg has changed their server configuration since we added that
file to the Go repo.
Anyway, this change is just manually excising the BOM from the start
of the file, leaving pure ASCII.
Marcel van Lohuizen [Mon, 7 May 2012 09:51:40 +0000 (11:51 +0200)]
exp/locale/collate: from the regression test we derive that the spec
dictates a CJK rune is only part of a certain specified range if it
is explicitly defined in the Unicode Codepoint Database.
Fixed the code and some of the tests accordingly.
Shenghou Ma [Fri, 4 May 2012 18:02:36 +0000 (02:02 +0800)]
sync/atomic: fix 64-bit atomic cas for Linux/ARM
This is a follow-up to CL 5978051.
Use kernel cas64 helper if we can, fallback to LDREXD/STREXD if
we are on ARMv6 or higher, and to lock-emulated cas64 if on ARMv5.
A future CL will fix {Add,Load,Store}{Int,Uint}64 and issue 3331.
Shenghou Ma [Fri, 4 May 2012 17:59:14 +0000 (01:59 +0800)]
runtime: get per-process random number from auxv for hash table
Decode AT_RANDOM, AT_HWCAP, and AT_PLATFORM.
This CL only make use of AT_RANDOM, but future CLs will make use of the others.
Shenghou Ma [Fri, 4 May 2012 17:54:16 +0000 (01:54 +0800)]
cmd/cgo, cmd/cc, cmd/ld: detect dynamic linker automatically
Some newer Linux distributions (Ubuntu ARM at least) use a new multiarch
directory organization, where dynamic linker is no longer in the hardcoded
path in our linker.
For example, Ubuntu 12.04 ARM hardfloat places its dynamic linker at
/lib/arm-linux-gnueabihf/ld-linux.so.3
Ref: http://lackof.org/taggart/hacking/multiarch/
Also, to support Debian GNU/kFreeBSD as a FreeBSD variant, we need this capability, so it's part of issue 3533.
This CL add a new pragma (#pragma dynlinker "path") to cc.
Shenghou Ma [Fri, 4 May 2012 17:35:13 +0000 (01:35 +0800)]
runtime, misc/cgo/test: fix build for Linux/ARM
1. In CL 5989057, I made a mistake in the last minute change.
"MOVW.W R4, -4(SP)" should really be "MOVW.W R4, -4(R13)",
as 5l will rewrite offset for SP.
2. misc/cgo/test/issue1560.go tests for parallel sleep of 1s,
but on ARM, the deadline is frequently missed, so change sleep
time to 2s on ARM.
Akshat Kumar [Fri, 4 May 2012 10:48:34 +0000 (03:48 -0700)]
pkg/runtime: Plan 9 signal handling in Go
This adds proper note handling for Plan 9,
and fixes the issue of properly killing go procs.
Without this change, the first go proc that dies
(using runtime·exit()) would kill all the running
go procs. Proper signal handling is needed.
Russ Cox [Wed, 2 May 2012 20:22:56 +0000 (16:22 -0400)]
cmd/cc: add PREFETCH built-in (like SET, USED)
This makes it possible to inline the prefetch of upcoming
memory addresses during garbage collection, instead of
needing to flush registers, make a function call, and
reload registers. On garbage collection-heavy workloads,
this results in a 5% speedup.
Fixes #3493.
R=dvyukov, ken, r, dave
CC=golang-dev
https://golang.org/cl/5990066
Marcel van Lohuizen [Wed, 2 May 2012 15:01:41 +0000 (17:01 +0200)]
exp/locale/collate: fixed two bugs uncovered by regression tests.
The first bug was that tertiary ignorables had the same colElem as
implicit colElems, yielding unexpected results. The current encoding
ensures that a non-implicit colElem is never 0. This fix uncovered
another bug of the trie that indexed incorrectly into the null block.
This was caused by an unfinished optimization that would avoid the
need to max out the most-significant bits of continuation bytes.
This bug was also present in the trie used in exp/norm and has been
fixed there as well. The appearence of the bug was rare, as the lower
blocks happened to be nearly nil.
David Symonds [Tue, 1 May 2012 01:33:25 +0000 (11:33 +1000)]
misc/dashboard/codereview: send mail immediately, and fill in time.Time fields.
If we delay the mail sending, we can't send as the current user.
If we don't fill in the time.Time fields, datastore.Put will fail
because the zero time.Time value is out of its range.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6136053
Nigel Tao [Tue, 1 May 2012 00:51:34 +0000 (10:51 +1000)]
compress/flate: optimize history-copy decoding.
The forwardCopy function could be re-written in asm, and the copyHuff
method could probably be rolled into huffmanBlock and copyHist, but
I'm leaving those changes for future CLs.
Wall time for Denis Cheremisov's PNG-decoding program given in
https://groups.google.com/group/golang-nuts/browse_thread/thread/22aa8a05040fdd49
Before: 3.07s
After: 2.32s
Delta: -24%
Shenghou Ma [Mon, 30 Apr 2012 19:55:07 +0000 (15:55 -0400)]
runtime: give proper messages when user defined DYLD_INSERT_LIBRARIES on Darwin
also gives clear error when bsdthread_register fails on Darwin
Fixes #2992.
Dave Cheney [Sun, 29 Apr 2012 10:41:13 +0000 (20:41 +1000)]
compress/flate: add Encoder/Decoder benchmarks
In CL 6127051, nigeltao suggested that further gains
were possible by improving the performance of flate.
This CL adds a set of benchmarks (based on compress/lzw)
that can be used to judge any future improvements.
Dave Cheney [Fri, 27 Apr 2012 12:17:08 +0000 (22:17 +1000)]
net: consolidate common socket functions
In resolving 3507, the fix had to be applied individually to
the four *Conn types, tcp, udp, rawip and unix, due to the
duplicate code in each Conn type.
This CL consolidates the common net.Conn methods that all four
*Conn types implement into a base conn type.
Pros:
* The fix for 3507 would have only needed to be applied to one
method. Further improvements, such as possibly removing the
c.fd != nil check in c.ok(), would benefit from this CL.
* Nearly 300 lines removed from the net package.
* The public interface and documentation are not changed.
* I think this is an excellent example of the power of embedding.
Cons:
* The net package is already distributed over many files, this
CL adds another place to look.
* The fix for 3507 was a total of 16 lines changed, this follow
up CL could be considered to be an overreaction as new Conn types
are unlikely to be added in the near future.
The new logic is "use go1 if it's there, otherwise no tag."
Nothing needs to say "I require go1.0.1", and I want to
preserve some flexibility in defining what tags mean.
Right now (before go1.0.1) there is only one possible tag,
"go1", and I'd like to keep it that way.
Anthony Martin [Thu, 26 Apr 2012 09:59:13 +0000 (02:59 -0700)]
syscall: fix a number of exec bugs on Plan 9
1. Readdirnames was erroneously returning an
empty slice on every invocation.
2. The logic for determining which files to
close before exec was incorrect. If the
set of files to be kept open (provided by
the caller) did not include the files
opened at startup, those files would be
accidentally closed.
I also cleaned up readdupdevice while I was
in the vicinity.