Alex Brainman [Tue, 25 Aug 2015 07:01:49 +0000 (17:01 +1000)]
net: return rooted DNS names on windows
This CL also changes windows LookupSRV to return
_xmpp-server._tcp.google.com. as cname instead of google.com
similar to linux. Otherwise TestLookupDots still fails.
Rahul Chaudhry [Thu, 3 Dec 2015 00:42:17 +0000 (16:42 -0800)]
os: skip TestHardLink on Android.
From Android release M (Marshmallow), hard linking files is blocked
and an attempt to call link() on a file will return EACCES.
- https://code.google.com/p/android-developer-preview/issues/detail?id=3150
Rahul Chaudhry [Wed, 2 Dec 2015 21:27:44 +0000 (13:27 -0800)]
cmd/go: allow buildmode=c-shared for android/arm64.
Also, enable test misc/cgo/testcshared for android/arm64.
c/17245 and c/17246 provide the missing pieces for making
this test work.
"androidtest.bash" now passes on a Nexus 9 (volantis)
device running Android build "LMY48T".
Change-Id: Icb9fd2d17d97e0f04cb18d0cd91640c80fbd3fb4
Reviewed-on: https://go-review.googlesource.com/17333 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Michael Hudson-Doyle [Tue, 1 Dec 2015 02:39:41 +0000 (15:39 +1300)]
cmd/internal/obj: fix stack barriers in ppc64le shared libs
runtime.stackBarrier is a strange function: it is only ever "called" by
smashing its address into a LR slot on the stack. Calling it like this
certainly does not adhere to the rule that r12 is set to the global entry point
before calling it and the prologue instrutions that compute r2 from r12 in fact
just corrupt r2, which is bad because the function that stackBarrier returns to
probably uses r2 to access global data.
Fortunately stackBarrier itself does not access any global data and so does not
depend on the value of r2, meaning we can ignore the ABI rules and simply skip
inserting the prologue instructions into this specific function.
Fixes 64bit.go, append.go and fixedbugs/issue13169.go from "cd test; go run
run.go -linkshared".
INADA Naoki [Tue, 3 Mar 2015 12:27:07 +0000 (21:27 +0900)]
database/sql: Add DB.SetConnMaxLifetime
Long lived connections may make some DB operation difficult.
(e.g. retiring load balanced DB server.)
So SetConnMaxLifetime closes long lived connections.
It can be used to limit maximum idle time, too.
Closing idle connections reduces active connections while application is idle
and avoids connections are closed by server side (cause errBadConn while querying).
Alexandre Cesaro [Tue, 20 Oct 2015 15:30:21 +0000 (17:30 +0200)]
net/mail: use base64 encoding when needed in Address.String()
When the name of an Address contains non-ASCII characters,
Address.String() used mime.QEncoding to encode the name.
However certain characters are forbidden when an encoded-word is
in a phrase context (see RFC 2047 section 5.3) and these
characters are not encoded by mime.QEncoding.
In this case we now use mime.BEncoding (base64 encoding) so that
forbidden characters are also encoded.
Joe Tsai [Mon, 30 Nov 2015 22:43:33 +0000 (14:43 -0800)]
compress/bzip2/testdata: make Mark.Twain-Tom.Sawyer.txt free
Commit 7a1fb95d50b7a8302445ebc1296010695151ce7b strips non-free license
from Mark.Twain-Tom.Sawyer.txt, but forgot to remove it from the compressed
version of the file.
Joe Tsai [Wed, 16 Sep 2015 07:58:56 +0000 (00:58 -0700)]
archive/tar: convert Reader.Next to be loop based
Motivation for change:
* Recursive logic is hard to follow, since it tends to apply
things in reverse. On the other hand, the tar formats tend to
describe meta headers as affecting the next entry.
* Recursion also applies changes in the wrong order. Two test
files are attached that use multiple headers. The previous Go
behavior differs from what GNU and BSD tar do.
Brad Fitzpatrick [Tue, 1 Dec 2015 19:19:55 +0000 (19:19 +0000)]
net/http: enable HTTP/2 on all Transports, not just the DefaultTransport
This mirrors the same behavior and API from the server code to the
client side: if TLSNextProto is nil, HTTP/2 is on by default for
both. If it's non-nil, the user was trying to do something fancy and
step out of their way.
Updates #6891
Change-Id: Ia31808b71f336a8d5b44b985591d72113429e1d4
Reviewed-on: https://go-review.googlesource.com/17300 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Joe Tsai [Mon, 28 Sep 2015 20:49:35 +0000 (13:49 -0700)]
archive/tar: move parse/format methods to standalone receiver
Motivations for this change:
* It allows these functions to be used outside of Reader/Writer.
* It allows these functions to be more easily unit tested.
Russ Cox [Wed, 25 Nov 2015 21:15:45 +0000 (16:15 -0500)]
testing: document that T and B are safe for concurrent calls
Fixes #13108.
Change-Id: I474cc2a3b7ced1c9eb978fc815f9c6bae9fb3ecc
Reviewed-on: https://go-review.googlesource.com/17235 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Blake Gentry [Thu, 22 Jan 2015 23:58:25 +0000 (15:58 -0800)]
net/http: retry idempotent HTTP reqs on dead reused conns
If we try to reuse a connection that the server is in the process of
closing, we may end up successfully writing out our request (or a
portion of our request) only to find a connection error when we try to
read from (or finish writing to) the socket. This manifests as an EOF
returned from the Transport's RoundTrip.
The issue, among others, is described in #4677.
This change follows some of the Chromium guidelines for retrying
idempotent requests only when the connection has been already been used
successfully and no header data has yet been received for the response.
As part of this change, an unexported error was defined for
errMissingHost, which was previously defined inline. errMissingHost is
the only non-network error returned from a Request's Write() method.
Additionally, this breaks TestLinuxSendfile because its test server
explicitly triggers the type of scenario this change is meant to retry
on. Because that test server stops accepting conns on the test listener
before the retry, the test would time out. To fix this, the test was
altered to use a non-idempotent test type (POST).
David Benjamin [Wed, 18 Nov 2015 04:24:36 +0000 (23:24 -0500)]
encoding/asn1: Reject invalid INTEGERs.
The empty string is not a valid DER integer. DER also requires that values be
minimally-encoded, so excess padding with leading 0s (0xff for negative
numbers) is forbidden. (These rules also apply to BER, incidentally.)
Joe Tsai [Thu, 1 Oct 2015 08:35:15 +0000 (01:35 -0700)]
archive/tar: fix issues with readGNUSparseMap1x0
Motivations:
* Use of strconv.ParseInt does not properly treat integers as 64bit,
preventing this function from working properly on 32bit machines.
* Use of io.ReadFull does not properly detect truncated streams
when the file suddenly ends on a block boundary.
* The function blindly trusts user input for numEntries and allocates
memory accordingly.
* The function does not validate that numEntries is not negative,
allowing a malicious sparse file to cause a panic during make.
In general, this function was overly complicated for what it was
accomplishing and it was hard to reason that it was free from
bounds errors. Instead, it has been rewritten and relies on
bytes.Buffer.ReadString to do the main work. So long as invariants
about the number of '\n' in the buffer are maintained, it is much
easier to see why this approach is correct.
Joe Tsai [Wed, 4 Nov 2015 02:12:31 +0000 (18:12 -0800)]
archive/tar: properly handle header-only "files" in Reader
Certain special type-flags, specifically 1, 2, 3, 4, 5, 6,
do not have a data section. Thus, regardless of what the size field
says, we should not attempt to read any data for these special types.
The relevant PAX and USTAR specification says:
<<<
If the typeflag field is set to specify a file to be of type 1 (a link)
or 2 (a symbolic link), the size field shall be specified as zero.
If the typeflag field is set to specify a file of type 5 (directory),
the size field shall be interpreted as described under the definition
of that record type. No data logical records are stored for types 1, 2, or 5.
If the typeflag field is set to 3 (character special file),
4 (block special file), or 6 (FIFO), the meaning of the size field is
unspecified by this volume of POSIX.1-2008, and no data logical records shall
be stored on the medium.
Additionally, for type 6, the size field shall be ignored when reading.
If the typeflag field is set to any other value, the number of logical
records written following the header shall be (size+511)/512, ignoring
any fraction in the result of the division.
>>>
Contrary to the specification, we do not assert that the size field
is zero for type 1 and 2 since we liberally accept non-conforming formats.
Colin Cross [Thu, 5 Nov 2015 23:47:20 +0000 (15:47 -0800)]
archive/zip: enable overriding (de)compressors per file
Implement setting the compression level for a zip archive by registering
a per-Writer compressor through Writer.RegisterCompressor. If no
compressors are registered, fall back to the ones registered at the
package level. Also implements per-Reader decompressors.
Fixes #8359
Change-Id: I93b27c81947b0f817b42e0067aa610ff267fdb21
Reviewed-on: https://go-review.googlesource.com/16669 Reviewed-by: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Joe Tsai <joetsai@digital-static.net> Reviewed-by: Klaus Post <klauspost@gmail.com> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Brad Fitzpatrick [Mon, 30 Nov 2015 19:33:49 +0000 (19:33 +0000)]
mime: let FormatMediaType format slash-less media types, to mirror ParseMediaType.
A Content-Type always has a slash (type/subtype)
A Content-Disposition does not (e.g. "attachment" or "line").
A "media type" is either one of those, plus optional parameters afterwards.
Our ParseMediaType and FormatMediaType weren't consistent in whether
they permitted Content-Dispositions. Now they both do.
Brad Fitzpatrick [Mon, 30 Nov 2015 20:28:19 +0000 (20:28 +0000)]
mime: don't accept single-quoted strings in media type parameter values
Fix an old bug where media type parameter values could be escaped by
either double quotes (per the spec) or single quotes (due to my bug).
The original bug was introduced by me in git rev 90e4ece3
(https://golang.org/cl/4430049) in April 2011 when adding more tests
from http://greenbytes.de/tech/tc2231/ and misinterpreting the
expected value of test "attwithfntokensq" and not apparently thinking
about it enough.
No known spec or existing software produces or expects single quotes
around values. In fact, it would have be a parsing ambiguity if it
were allowed: the string `a=', b='` could parse as two keys "a" and
"b" both with value "'", or it could be parse as a single key "a" with
value "', b=".
Michael Hudson-Doyle [Tue, 1 Dec 2015 01:47:22 +0000 (14:47 +1300)]
runtime: set r12 to sigpanic before jumping to it in sighandler
The ppc64le shared library ABI demands that r12 is set to a function's global
entrypoint before jumping to the global entrypoint. Not doing so means that
handling signals that usually panic actually crashes (and so, e.g. can't be
recovered). Fixes several failures of "cd test; go run run.go -linkshared".
Change-Id: Ia4d0da4c13efda68340d38c045a52b37c2f90796
Reviewed-on: https://go-review.googlesource.com/17280 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Ian Lance Taylor [Fri, 20 Nov 2015 01:03:01 +0000 (17:03 -0800)]
cmd/cgo: make the char * pointer in GoString const
This makes it more convenient for C code to use GoString with string
constants. Since Go string values are immutable, the const qualifier is
appropriate in C.
Change-Id: I5fb3cdce2ce5079f1f0467a1544bb3a1eb27b811
Reviewed-on: https://go-review.googlesource.com/17067
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Russ Cox <rsc@golang.org>
Mikio Hara [Fri, 27 Nov 2015 03:09:14 +0000 (12:09 +0900)]
net: force LookupAddr results to be rooted DNS paths even in the case of local source
The builtin name resolver using various resolution techniques is a bit
complicated and we sometimes fotget to take care of all the go and cgo
code paths and exchanging information to local and remote sources. This
change makes LookupAddr return absolute domain names even in the case of
local source.
Robert Griesemer [Sat, 28 Nov 2015 00:11:05 +0000 (16:11 -0800)]
cmd/compile: remove gratuituous type conversions
Follow-up cleanup for https://go-review.googlesource.com/17248:
Use properly typed local variable op now that that variable use
is not overloaded anymore.
Also: Remove unnecessary if stmt from common lexical path.
Adam Langley [Fri, 27 Nov 2015 21:50:36 +0000 (13:50 -0800)]
crypto/x509: permit serial numbers to be negative.
Some software that produces certificates doesn't encode integers
correctly and, about half the time, ends up producing certificates with
serial numbers that are actually negative.
This buggy software, sadly, appears to be common enough that we should
let these errors pass. This change allows a Certificate.SerialNumber to
be negative.
Shenghou Ma [Thu, 26 Nov 2015 22:45:09 +0000 (17:45 -0500)]
cmd/cover: allow part selection to be retained across page refreshes
Usually, you are primarily interested to see the coverage of a particular
file (e.g. when you're changing tests that affects a given source file),
it is very valuable if you can just refresh the page and immediately see
changes to the part you're already looking at (without selecting from the
selector again.)
Change-Id: I615207c9be6713f436e444771134fceaf4600ff3
Reviewed-on: https://go-review.googlesource.com/17238
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Andrew Gerrand <adg@golang.org>
Tamir Duberstein [Mon, 9 Nov 2015 15:34:02 +0000 (10:34 -0500)]
cmd/yacc: memory allocation improvements
Places a fixed size initial stack and the lval inside the parser
struct so that they are allocated together. Places $$char inside the
parser struct to avoid allocating the closure used in Lookahead().
Ralph Corderoy [Sat, 7 Nov 2015 12:18:36 +0000 (12:18 +0000)]
crypto/tls: Server can specify an unadvertised cipher suite
During the TLS handshake, check the cipher suite the server selects is
one of those offered in the ClientHello. The code was checking it was
in the larger list that was sometimes whittled down for the ClientHello.
Fixes #13174
Change-Id: Iad8eebbcfa5027f30403b9700c43cfa949e135bb
Reviewed-on: https://go-review.googlesource.com/16698 Reviewed-by: Adam Langley <agl@golang.org>
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Russ Cox <rsc@golang.org>
Make sure spaces are tolerated in ${SRCDIR} when this variable
is expanded. This applies to ${SRCDIR} only. Shell safety
checks are still done in the same exact way for anything else.
Matthew Dempsky [Wed, 25 Nov 2015 23:08:00 +0000 (15:08 -0800)]
doc: update go1.6.txt for cmd/cgo's C.complexfloat and C.complexdouble fix
Updates #13402.
Change-Id: Ia7b729d81fb78206d214444911f2e6573b88717a
Reviewed-on: https://go-review.googlesource.com/17240 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Matthew Dempsky [Wed, 25 Nov 2015 21:09:14 +0000 (13:09 -0800)]
cmd/cgo: fix C.complexfloat and C.complexdouble
This also fixes an unintended behavior where C's "complex float" and
"complex double" types were interchangeable with Go's "complex64" and
"complex128" types.
Fixes #13402.
Change-Id: I73f96d9a4772088d495073783c6982e9634430e8
Reviewed-on: https://go-review.googlesource.com/17208 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Matthew Dempsky [Wed, 25 Nov 2015 20:40:43 +0000 (12:40 -0800)]
cmd/cgo: handle another Clang DWARF quirk
Without the fix:
$ CC=clang-3.5 ./test.bash
misc/cgo/errors/test.bash: BUG: expected error output to contain "C.ushort" but saw:
# command-line-arguments
./issue13129.go:13: cannot use int(0) (type int) as type C.unsignedshort in assignment
Fixes #13129.
Change-Id: I2c019d2d000f5bfa3e33c477e533aff97031a84f
Reviewed-on: https://go-review.googlesource.com/17207
Run-TryBot: Matthew Dempsky <mdempsky@google.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Caleb Spare [Tue, 24 Nov 2015 00:16:42 +0000 (16:16 -0800)]
regexp: fix one-pass compilation
The one-pass transformation is structured as a search over the input
machine for conditions that violate the one-pass requisites. At each
iteration, we should fully explore all non-input paths that proceed from
the current instruction; this is implemented via recursive check calls.
But when we reach instructions that demand input (InstRune*), these
should be put onto the search queue.
Instead of searching this way, the routine previously (effectively)
proceeded through the machine one instruction at a time until finding an
Inst{Match,Fail,Rune*}, calling check on each instruction. This caused
bug #11905, where the transformation stopped before rewriting all
InstAlts as InstAltMatches.
Further, the check function unnecessarily recurred on InstRune*
instructions. (I believe this helps to mask the above bug.)
This change also deletes some unused functions and duplicate test cases.
Caleb Spare [Thu, 22 Oct 2015 06:03:54 +0000 (23:03 -0700)]
regexp: fix LiteralPrefix for certain onepass progs
The prefix computation for onepass was incorrectly returning
complete=true when it encountered a beginning-of-text empty width match
(^) in the middle of an expression.
Fix by returning complete only when the prefix is followed by $ and then
an accepting state.
Caleb Spare [Tue, 20 Oct 2015 08:37:06 +0000 (01:37 -0700)]
regexp: add Copy method to Regexp
This helps users who wish to use separate Regexps in each goroutine to
avoid lock contention. Previously they had to parse the expression
multiple times to achieve this.
I used variants of the included benchmark to evaluate this change. I
used the arguments -benchtime 20s -cpu 1,2,4,8,16 on a machine with 16
hardware cores.
Comparing a single shared Regexp vs. copied Regexps, we can see that
lock contention causes huge slowdowns at higher levels of parallelism.
The copied version shows the expected linear speedup.
I also compared a modified version of Regexp that has no mutex and a
single machine (the "RegexpForSingleGoroutine" rsc mentioned in
https://github.com/golang/go/issues/8232#issuecomment-66096128).
In this next test, I compared using N copied Regexps vs. N separate
RegexpForSingleGoroutines. This shows that, even for this relatively
simple regex, avoiding the lock entirely would only buy about 10-12%
further improvement.
Joe Tsai [Fri, 20 Nov 2015 01:41:57 +0000 (17:41 -0800)]
hash/crc32: add noescape tags to assembly functions
CRC-32 computation is stateless and the p slice does not get stored
anywhere. Thus, we mark the assembly functions as noescape so that
it doesn't believe that p leaks in:
func Update(crc uint32, tab *Table, p []byte) uint32
Brad Fitzpatrick [Tue, 24 Nov 2015 16:55:43 +0000 (08:55 -0800)]
net/http: more HTTP/2 tests and fixes
This compares the behavior of server handlers and the net/http
Transport in both HTTP/1 and HTTP/2 mode and verifies they're the
same.
This also moves some client<->server tests into clientserver_test.go.
Many of them were in serve_test.go or transport_test.go but were
basically testing both.
h2_bundle.go is an update of the golang.org/x/net/http2 code
from https://golang.org/cl/17204 (x/net git rev c745c36eab10)
Fixes #13315
Fixes #13316
Fixes #13317
Fixes other stuff found in the process too
Updates #6891 (http2 support in general)
Shenghou Ma [Thu, 3 Sep 2015 06:44:26 +0000 (02:44 -0400)]
runtime: check that masks and shifts are correct aligned
We need a runtime check because the original issue is encountered
when running cross compiled windows program from linux. It's better
to give a meaningful crash message earlier than to segfault later.
The added test should not impose any measurable overhead to Go
programs.
For #12415.
Change-Id: Ib4a24ef560c09c0585b351d62eefd157b6b7f04c
Reviewed-on: https://go-review.googlesource.com/14207 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Robert Griesemer [Tue, 24 Nov 2015 22:49:10 +0000 (14:49 -0800)]
cmd/compile: remove references to go.y and update documentation
This is a comment/documentation change only but for a minor
code change in the file and package_ methods (move recognition
of semi to match grammar better).
Per request from r.
Change-Id: I81ec985cc5831074d9eb5e8ffbf7e59466284819
Reviewed-on: https://go-review.googlesource.com/17202 Reviewed-by: Rob Pike <r@golang.org>
Jakub Čajka [Tue, 24 Nov 2015 14:36:32 +0000 (15:36 +0100)]
compress: make Mark.Twain-Tom.Sawyer.txt licensed under non-free license free again
This change strips non-free license from Mark.Twain-Tom.Sawyer.txt along with all reference to Project Gutenberg in the file and the whole source tree. Making the file public domain again.
Fixes #13216
Change-Id: I2f41b0de225f627dde152efe93c006a4c24be668
Reviewed-on: https://go-review.googlesource.com/17196 Reviewed-by: Andrew Gerrand <adg@golang.org>
Austin Clements [Thu, 19 Nov 2015 16:25:55 +0000 (11:25 -0500)]
runtime: make gcFlushBgCredit go:nowritebarrierrec
Write barriers in gcFlushBgCredit lead to very subtle bugs because it
executes after the getfull barrier. I tracked some bugs of this form
down before go:nowritebarrierrec was implemented. Ensure that they
don't reappear by making gcFlushBgCredit go:nowritebarrierrec.
Rob Pike [Tue, 24 Nov 2015 18:32:37 +0000 (10:32 -0800)]
encoding/gob: document that zero elements of arrays and slices are sent
Fixes #13378
Change-Id: Ia78624ca1aa36ee906cef15416ea5554fa8229f2
Reviewed-on: https://go-review.googlesource.com/17201 Reviewed-by: Ian Lance Taylor <iant@golang.org>
David du Colombier [Tue, 24 Nov 2015 17:54:58 +0000 (18:54 +0100)]
syscall: don't check result of close(fd) in forkAndExecInChild on Plan9
On multiprocessor machines, a file descriptor could be
closed twice in forkAndExecInChild. Consequently, the close
syscall returns the "fd out of range or not open" error
and forkAndExecInChild fails.
This changes forkAndExecInChild to ignore the error
returned by close(fd), as on other operating systems.
Fixes #12851.
Change-Id: I96a8463ce6599bfd1362353283e0329a00f738da
Reviewed-on: https://go-review.googlesource.com/17188 Reviewed-by: Rob Pike <r@golang.org>
Elias Naur [Tue, 17 Nov 2015 10:41:06 +0000 (11:41 +0100)]
runtime: use a proper type, sigset, for m.sigmask
Replace the cross platform but unsafe [4]uintptr type with a OS
specific type, sigset. Most OSes already define sigset, and this
change defines a suitable sigset for the OSes that don't (darwin,
openbsd). The OSes that don't use m.sigmask (windows, plan9, nacl)
now defines sigset as the empty type, struct{}.
The gain is strongly typed access to m.sigmask, saving a dynamic
size sanity check and unsafe.Pointer casting. Also, some storage is
saved for each M, since [4]uinptr was conservative for most OSes.
The cost is that OSes that don't need m.sigmask has to define sigset.
completes ./all.bash with GOOS linux, on amd64
completes ./make.bash with GOOSes openbsd, android, plan9, windows,
darwin, solaris, netbsd, freebsd, dragonfly, all amd64.
With GOOS=nacl ./make.bash failed with a seemingly unrelated error.
[Replay of CL 16942 by Elias Naur.]
Change-Id: I98f144d626033ae5318576115ed635415ac71b2c
Reviewed-on: https://go-review.googlesource.com/17033 Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Caleb Spare [Tue, 17 Nov 2015 03:34:54 +0000 (19:34 -0800)]
testing: pause the test timer while waiting in T.Parallel
Before, we reset the timer at the end of T.Parallel, which is okay
assuming that T.Parallel is the first thing in the test.
Snapshot the elapsed time at the beginning of Parallel and include it in
the total duration so that any time spent in the test before calling
Parallel is reported in the test duration as well.
Dave Cheney [Fri, 11 Sep 2015 00:15:17 +0000 (10:15 +1000)]
runtime: skip CallbackGC test in short mode on linux/arm
Fixes #11959
Fixes #12035
Skip the CallbackGC test on linux/arm. This test takes between 30 and 60
seconds to run by itself, and is run 4 times over the course of ./run.bash
(once during the runtime test, three times more later in the build).
Alex Brainman [Tue, 24 Nov 2015 03:29:06 +0000 (14:29 +1100)]
runtime: fix handling VirtualAlloc failure in sysUsed
Original code is mistakenly panics on VirtualAlloc failure - we want
it to go looking for smaller memory region that VirtualAlloc will
succeed to allocate. Also return immediately if VirtualAlloc succeeds.
See rsc comment on issue #12587 for details.
I still don't have a test for this. So I can only hope that this