Russ Cox [Thu, 24 Feb 2011 21:46:44 +0000 (13:46 -0800)]
runtime: fix signal stack bug
In CL 4188061 I changed malg to allocate the requested
number of bytes n, not n+StackGuard, so that the
allocations would use rounder numbers.
The allocation of the signal stack asks for 32k and
then used g->stackguard as the base, but g->stackguard
is StackGuard bytes above the base. Previously, asking
for 32k meant getting 32k+StackGuard bytes, so using
g->stackguard as the base was safe. Now, the actual base
must be computed, so that the signal handler does not
run StackGuard bytes past the top of the stack.
Was causing flakiness mainly in programs that use the
network, because they sometimes write to closed network
connections, causing SIGPIPEs. Was also causing problems
in the doc/progs test.
Also fix Makefile so that changes to stack.h trigger rebuild.
Russ Cox [Thu, 24 Feb 2011 21:45:45 +0000 (16:45 -0500)]
ld: weak symbols
A reference to the address of weak.foo resolves at link time
to the address of the symbol foo if foo would end up in the
binary anyway, or to zero if foo would not be in the binary.
For example:
int xxx = 1;
int yyy = 2;
int weak·xxx;
int weak·yyy;
Robert Griesemer [Thu, 24 Feb 2011 19:13:20 +0000 (11:13 -0800)]
godoc: fix writeFileAtomically utility function
If the filename was absolute, writeFileAtomically
used the wrong filename for ioutil.TempFile leading
to non-existent directories and the TempFile would
fail.
Russ Cox [Wed, 23 Feb 2011 20:51:20 +0000 (15:51 -0500)]
runtime: always run stackalloc on scheduler stack
Avoids deadlocks like the one below, in which a stack split happened
in order to call lock(&stacks), but then the stack unsplit cannot run
because stacks is now locked.
The only code calling stackalloc that wasn't on a scheduler
stack already was malg, which creates a new goroutine.
Russ Cox [Wed, 23 Feb 2011 20:42:13 +0000 (15:42 -0500)]
runtime: omit breakpoint during terminal panic
A terminal panic (one that prints a stack trace and exits)
has been calling runtime.breakpoint before calling exit,
so that if running under a debugger, the debugger can
take control. When not running under a debugger, though,
this causes an additional SIGTRAP on Unix and pop-up
dialogs on Windows.
Support for debugging Go programs has gotten good
enough that we can rely on the debugger to set its own
breakpoint on runtime.exit if it wants to look around.
Brad Fitzpatrick [Wed, 23 Feb 2011 20:20:50 +0000 (12:20 -0800)]
http: introduce start of Client and ClientTransport
Much yet to come, but this is a safe first step, introducing
an in-the-future configurable Client object (where policy for
cookies, auth, redirects will live) as well as introducing a
ClientTransport interface for sending requests.
The CL intentionally ignores everything around the creation
and configuration of Clients and merely ports/wraps the old
interfaces to/around Client/ClientTransport.
Russ Cox [Wed, 23 Feb 2011 19:47:42 +0000 (14:47 -0500)]
runtime: pass to signal handler value of g at time of signal
The existing code assumed that signals only arrived
while executing on the goroutine stack (g == m->curg),
not while executing on the scheduler stack (g == m->g0).
Most of the signal handling trampolines correctly saved
and restored g already, but the sighandler C code did not
have access to it.
Some rewriting of assembly to make the various
implementations as similar as possible.
Will need to change Windows too but I don't
understand how sigtramp gets called there.
Russ Cox [Wed, 23 Feb 2011 19:47:22 +0000 (14:47 -0500)]
runtime: traceback through active lessstack
With this change, a panic trace due to a signal arriving while
running on the scheduler stack during a lessstack
(a stack unsplit) will trace through the lessstack to show
the state of the goroutine that was unsplitting its stack.
Russ Cox [Wed, 23 Feb 2011 18:21:39 +0000 (13:21 -0500)]
5g: fix optimizer bug
same as in issue below, never fixed on ARM
changeset: 5498:3fa1372ca694
user: Ken Thompson <ken@golang.org>
date: Thu May 20 17:31:28 2010 -0700
description:
fix issue 798
cannot allocate an audomatic temp
while real registers are allocated.
there is a chance that the automatic
will be allocated to one of the
allocated registers. the fix is to
not registerize such variables.
Rob Pike [Wed, 23 Feb 2011 17:49:35 +0000 (09:49 -0800)]
gob: protect against pure recursive types.
There are further changes required for things like
recursive map types. Recursive struct types work
but the mechanism needs generalization. The
case handled in this CL is pathological since it
cannot be represented at all by gob, so it should
be handled separately. (Prior to this CL, encode
would recur forever.)
Gustavo Niemeyer [Wed, 23 Feb 2011 16:48:40 +0000 (11:48 -0500)]
codereview: fix clpatch with empty diffs
Avoid passing the placeholder diff to hgpatch, so that
clpatch-ing an empty diff grabs the metadata and warns
about it being empty, rather than failing with a
hard-to-debug problem ("mkdir: no such file or dir",
no metadata, etc).
Robert Griesemer [Wed, 23 Feb 2011 02:17:03 +0000 (18:17 -0800)]
exp/eval, go/printer: fix build
There are some minor irregularities in the printer
output (some paren's are present that should be
removed), but these are unrelated issues.
Will review in a 2nd step.
Robert Griesemer [Wed, 23 Feb 2011 01:25:10 +0000 (17:25 -0800)]
go/ast, parser: condition in if statement is mandatory
As a result, parsing a "control clause" is now sufficiently
different for if, switch, and for statements that the code
is not factored out anymore. The code is a bit longer but
clearer in each individual case.
Russ Cox [Tue, 22 Feb 2011 22:40:40 +0000 (17:40 -0500)]
ld: detect stack overflow due to NOSPLIT
Fix problems found.
On amd64, various library routines had bigger
stack frames than expected, because large function
calls had been added.
runtime.assertI2T: nosplit stack overflow
120 assumed on entry to runtime.assertI2T
8 after runtime.assertI2T uses 112
0 on entry to runtime.newTypeAssertionError
-8 on entry to runtime.morestack01
runtime.assertE2E: nosplit stack overflow
120 assumed on entry to runtime.assertE2E
16 after runtime.assertE2E uses 104
8 on entry to runtime.panic
0 on entry to runtime.morestack16
-8 after runtime.morestack16 uses 8
runtime.assertE2T: nosplit stack overflow
120 assumed on entry to runtime.assertE2T
16 after runtime.assertE2T uses 104
8 on entry to runtime.panic
0 on entry to runtime.morestack16
-8 after runtime.morestack16 uses 8
runtime.newselect: nosplit stack overflow
120 assumed on entry to runtime.newselect
56 after runtime.newselect uses 64
48 on entry to runtime.printf
8 after runtime.printf uses 40
0 on entry to vprintf
-8 on entry to runtime.morestack16
runtime.selectdefault: nosplit stack overflow
120 assumed on entry to runtime.selectdefault
56 after runtime.selectdefault uses 64
48 on entry to runtime.printf
8 after runtime.printf uses 40
0 on entry to vprintf
-8 on entry to runtime.morestack16
runtime.selectgo: nosplit stack overflow
120 assumed on entry to runtime.selectgo
0 after runtime.selectgo uses 120
-8 on entry to runtime.gosched
On arm, 5c was tagging functions NOSPLIT that should
not have been, like the recursive function printpanics:
printpanics: nosplit stack overflow
124 assumed on entry to printpanics
112 after printpanics uses 12
108 on entry to printpanics
96 after printpanics uses 12
92 on entry to printpanics
80 after printpanics uses 12
76 on entry to printpanics
64 after printpanics uses 12
60 on entry to printpanics
48 after printpanics uses 12
44 on entry to printpanics
32 after printpanics uses 12
28 on entry to printpanics
16 after printpanics uses 12
12 on entry to printpanics
0 after printpanics uses 12
-4 on entry to printpanics
Rob Pike [Tue, 22 Feb 2011 20:31:57 +0000 (12:31 -0800)]
gob: compute information about a user's type once.
Other than maybe cleaning the code up a bit, this has
little practical effect for now, but lays the foundation
for remembering the method set of a type, which can
be expensive.
Dave Cheney [Fri, 18 Feb 2011 23:49:46 +0000 (10:49 +1100)]
build: reduce the use of subshells in recursive make
Using make -C $* rather than (cd $* ; make) results in a small,
but measurable improvement in build times where compilation is
not the major component. eg.
before - ~/go/src/pkg$ time make
real 0m1.176s
user 0m0.639s
sys 0m0.399s
after - ~/go/src/pkg$ time make
real 0m0.916s
user 0m0.571s
sys 0m0.243s
There are other places in the distribution src/make.common for example
that could also benefit from this change.
Adam Langley [Fri, 18 Feb 2011 16:31:10 +0000 (11:31 -0500)]
crypto/rsa: left-pad OAEP results when needed.
PKCS#1 v2.1 section 7.1.1 says that the result of an OAEP encryption
is "an octet string of length $k$". Since we didn't left-pad the
result it was previously possible for the result to be smaller when
the most-significant byte was zero.
Nigel Tao [Wed, 16 Feb 2011 23:45:30 +0000 (10:45 +1100)]
html: tokenize HTML comments.
I'm not sure if it's 100% correct wrt the HTML5 specification,
but the test suite has plenty of HTML comment test cases, and
we'll shake out any tokenization bugs as the parser improves its
coverage.
Dave Cheney [Wed, 16 Feb 2011 20:07:13 +0000 (15:07 -0500)]
net: add IPv4 multicast to UDPConn
notes:
Darwin is very particular about joining a multicast group if the
listneing socket is not created in "udp4" mode, the other supported
OS's are more flexible.
A simple example sets up a socket to listen on the mdns/bonjour
group 224.0.0.251:5353
// ensure the sock is udp4, and the IP is a 4 byte IPv4
socket, err := net.ListenUDP("udp4", &net.UDPAddr {
IP: net.IPv4zero,
// currently darwin will not allow you to bind to
// a port if it is already bound to another process
Port: 5353,
})
if err != nil {
log.Exitf("listen %s", err)
}
defer socket.Close()
err = socket.JoinGroup(net.IPv4(224, 0, 0, 251))
if err != nil {
log.Exitf("join group %s", err)
}