Assignments of 2-, 3-, and 4-word values were handled
by individual MOV instructions (and for scalars still are).
But if there are pointers involved, those assignments now
go through the write barrier routine. Before this CL, they
went to writebarrierfat, which calls memmove.
Memmove is too much overhead for these small
amounts of data.
Instead, call writebarrierfat{2,3,4}, which are specialized
for the specific amount of data being copied.
Today the write barrier does not care which words are
pointers, so size alone is enough to distinguish the cases.
If we keep these distinctions in Go 1.5 we will need to
expand them for all the pointer-vs-scalar possibilities,
so the current 3 functions will become 3+7+15 = 25,
still not a large burden (we deleted more morestack
functions than that when we dropped segmented stacks).
Adam Langley [Tue, 14 Oct 2014 01:35:53 +0000 (18:35 -0700)]
crypto/x509: continue to recognise MaxPathLen of zero as "no value".
In [1] the behaviour of encoding/asn1 with respect to marshaling
optional integers was changed. Previously, a zero valued integer would
be omitted when marshaling. After the change, if a default value was
set then the integer would only be omitted if it was the default value.
This changed the behaviour of crypto/x509 because
Certificate.MaxPathLen has a default value of -1 and thus zero valued
MaxPathLens would no longer be omitted when marshaling. This is
arguably a bug-fix -- a value of zero for MaxPathLen is valid and
meaningful and now could be expressed. However it broke users
(including Docker) who were not setting MaxPathLen at all.
This change again causes a zero-valued MaxPathLen to be omitted and
introduces a ZeroMathPathLen member that indicates that, yes, one
really does want a zero. This is ugly, but we value not breaking users.
Ian Lance Taylor [Mon, 13 Oct 2014 17:01:34 +0000 (10:01 -0700)]
reflect: generated unrolled GC bitmask directly
The code for a generated type is already generating an
unrolled GC bitmask. Rather than unrolling the the source
type bitmasks and copying them, just generate the required
bitmask directly. Don't mark it as an unrolled GC program,
since there is no need to do so.
Alex Brainman [Sat, 11 Oct 2014 11:01:04 +0000 (22:01 +1100)]
cmd/ld: do not assume that only pe section names start with '.'
Our current pe object reader assumes that every symbol starting with
'.' is section. It appeared to be true, until now gcc 4.9.1 generates
some symbols with '.' at the front. Change that logic to check other
symbol fields in addition to checking for '.'. I am not an expert
here, but it seems reasonable to me.
Added test, but it is only good, if tested with gcc 4.9.1. Otherwise
the test PASSes regardless.
Alex Brainman [Sat, 11 Oct 2014 10:34:10 +0000 (21:34 +1100)]
cmd/ld: correct pe section names if longer then 8 chars
gcc 4.9.1 generates pe sections with names longer then 8 charters.
From IMAGE_SECTION_HEADER definition:
Name
An 8-byte, null-padded UTF-8 string. There is no terminating null character
if the string is exactly eight characters long. For longer names, this
member contains a forward slash (/) followed by an ASCII representation
of a decimal number that is an offset into the string table.
Our current pe object file reader does not read string table when section
names starts with /. Do that, so (issue 8811 example)
c:\go\path\src\isssue8811>go build
# isssue8811
isssue8811/glfw(.text): isssue8811/glfw(/76): not defined
isssue8811/glfw(.text): undefined: isssue8811/glfw(/76)
becomes
c:\go\path\src\isssue8811>go build
# isssue8811
isssue8811/glfw(.text): isssue8811/glfw(.rdata$.refptr._glfwInitialized): not defined
isssue8811/glfw(.text): undefined: isssue8811/glfw(.rdata$.refptr._glfwInitialized)
Adam Langley [Fri, 10 Oct 2014 00:37:40 +0000 (17:37 -0700)]
encoding/asn1: fix explicitly tagged Times.
https://golang.org/cl/153770043/ tried to fix the case where a
implicitly tagged Time, that happened to have the same tag as
GENERALIZEDTIME, shouldn't be parsed as a GENERALIZEDTIME.
It did so, mistakenly, by testing whether params.tag != nil. But
explicitly tagged values also have a non-nil tag and there the inner
tag actually does encode the type of the value.
This change instead tests whether the tag class is UNIVERSAL before
assuming that the tag contains type information.
That was complete failure - builders are broken,
but original cl worked fine on my system.
I will need access to builders
to test this change properly.
««« original CL description
runtime: handle all windows exception
Keith Randall [Wed, 8 Oct 2014 22:57:20 +0000 (15:57 -0700)]
runtime: delay freeing of shrunk stacks until gc is done.
This change prevents confusion in the garbage collector.
The collector wants to make sure that every pointer it finds
isn't junk. Its criteria for junk is (among others) points
to a "free" span.
Because the stack shrinker modifies pointers in the heap,
there is a race condition between the GC scanner and the
shrinker. The GC scanner can see old pointers (pointers to
freed stacks). In particular this happens with SudoG.elem
pointers.
Normally this is not a problem, as pointers into stack spans
are ok. But if the freed stack is the last one in its span,
the span is marked as "free" instead of "contains stacks".
This change makes sure that even if the GC scanner sees
an old pointer, the span into which it points is still
marked as "contains stacks", and thus the GC doesn't
complain about it.
This change will make the GC pause a tiny bit slower, as
the stack freeing now happens in serial with the mark pause.
We could delay the freeing until the mutators start back up,
but this is the simplest change for now.
Ian Lance Taylor [Wed, 8 Oct 2014 22:48:46 +0000 (15:48 -0700)]
reflect: add tests for variadic method calls
These tests fail when using gccgo. In gccgo using Interface
on the value of a method function is implemented using a
variant of MakeFunc. That approach did not correctly handle
variadic functions.
LGTM=r
R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/151280043
Dmitriy Vyukov [Wed, 8 Oct 2014 09:51:12 +0000 (13:51 +0400)]
runtime: faster GC scan
The change contains 3 spot optimizations to scan loop:
1. Don't use byte vars, use uintptr's instead.
This seems to alleviate some codegen issue,
and alone accounts to a half of speedup.
2. Remove bitmap cache. Currently we cache only 1 byte,
so caching is not particularly effective anyway.
Removal of the cache simplifies code and positively affects regalloc.
3. Replace BitsMultiword switch with if and
do debug checks only in Debug mode.
I've benchmarked changes separately and ensured that
each of them provides speedup on top of the previous one.
This change as a whole fixes the unintentional regressions
of scan loop that were introduced during development cycle.
Fixes #8625.
Fixes #8565.
Russ Cox [Wed, 8 Oct 2014 03:17:31 +0000 (23:17 -0400)]
runtime: clear Defer.panic before removing from G.defer list
Another dangling stack pointer in a cached structure.
Same as SudoG.elem and SudoG.selectdone.
Definitely a fix, and the new test in freedefer makes the
crash reproducible, but probably not a complete fix.
I have seen one dangling pointer in a Defer.panic even
after this fix; I cannot see where it could be coming from.
I think this will fix the solaris build.
I do not think this will fix the occasional failure on the darwin build.
Keith Randall [Tue, 7 Oct 2014 22:21:00 +0000 (15:21 -0700)]
runtime: zero pointer-looking scalararg values
I have a CL which at every gc looks through data and bss
sections for nonpointer data (according to gc maps) that
looks like a pointer. These are potential missing roots.
The only thing it finds are begnign, storing stack pointers
into m0.scalararg[1] and never cleaning them up. Let's
clean them up now so the test CL passes all.bash cleanly.
The test CL can't be checked in because we might store
pointer-looking things in nonpointer data by accident.
Keith Randall [Tue, 7 Oct 2014 20:36:16 +0000 (13:36 -0700)]
runtime: update heap dump format for 1.4
We no longer have full type information in the heap, so
we can't dump that any more. Instead we dump ptr/noptr
maps so at least we can compute graph connectivity.
In addition, we still dump Iface/Eface types so together
with dwarf type info we might be able to reconstruct
types of most things in the heap.
Russ Cox [Tue, 7 Oct 2014 15:07:18 +0000 (11:07 -0400)]
runtime: crash if we see an invalid pointer into GC arena
This will help find bugs during the release freeze.
It's not clear it should be kept for the release itself.
That's issue 8861.
The most likely thing that would trigger this is stale
pointers that previously were ignored or caused memory
leaks. These were allowed due to the use of conservative
collection. Now that everything is precise, we should not
see them anymore.
The small number check reinforces what the stack copier
is already doing, catching the storage of integers in pointers.
It caught issue 8864.
The check is disabled if _cgo_allocate is linked into the binary,
which is to say if the binary is using SWIG to allocate untyped
Go memory. In that case, there are invalid pointers and there's
nothing we can do about it.
LGTM=rlh
R=golang-codereviews, dvyukov, rlh
CC=golang-codereviews, iant, khr, r
https://golang.org/cl/148470043
Russ Cox [Tue, 7 Oct 2014 15:06:51 +0000 (11:06 -0400)]
runtime: remove type-punning for Type.gc[0], gc[1]
Depending on flags&KindGCProg,
gc[0] and gc[1] are either pointers or inlined bitmap bits.
That's not compatible with a precise garbage collector:
it needs to be always pointers or never pointers.
Change the inlined bitmap case to store a pointer to an
out-of-line bitmap in gc[0]. The out-of-line bitmaps are
dedup'ed, so that for example all pointer types share the
same out-of-line bitmap.
Fixes #8864.
LGTM=r
R=golang-codereviews, dvyukov, r
CC=golang-codereviews, iant, khr, rlh
https://golang.org/cl/155820043
Jens Frederich [Tue, 7 Oct 2014 14:13:42 +0000 (07:13 -0700)]
net/http: fix authentication info leakage in Referer header (potential security risk)
http.Client calls URL.String() to fill in the Referer header, which may
contain authentication info. This patch removes authentication info from
the Referer header without introducing any API changes.
A new test for net/http is also provided.
This is the polished version of Alberto García Hierro's
https://golang.org/cl/9766046/
Brad Fitzpatrick [Mon, 6 Oct 2014 22:10:51 +0000 (15:10 -0700)]
strings: use fast path for IndexRune
Noticed while reviewing https://golang.org/cl/147690043/
I'd never seen anybody use IndexRune before, and
unsurprisingly it doesn't use the other fast paths in the
strings/bytes packages. IndexByte uses assembly.
Rob Pike [Mon, 6 Oct 2014 21:50:58 +0000 (14:50 -0700)]
go/build: do not consider "android.go" to be android-specific
A file name must have a non-empty underscore-separated
prefix before its suffix matches GOOS. This is what the
documentation already said but is not what the code did.
Fixes #8838.
This needs to be called out in the release notes.
The he single affected file
code.google.com/p/go.text/collate/tools/colcmp/darwin.go
could use a renaming but works because it has a build tag inside.
Russ Cox [Mon, 6 Oct 2014 18:18:09 +0000 (14:18 -0400)]
runtime: update docs, code for SetFinalizer
At last minute before 1.3 we relaxed SetFinalizer to avoid
crashes when you pass the result of a global alloc to it.
This avoids the crash but makes SetFinalizer a bit too relaxed.
Document that the finalizer of a global allocation may not run.
Tighten the SetFinalizer check to ignore a global allocation but
not ignore everything else.
Fixes #7656.
LGTM=r, iant
R=golang-codereviews, iant, r
CC=dvyukov, golang-codereviews, khr, rlh
https://golang.org/cl/145930043
Alex Brainman [Sun, 5 Oct 2014 02:15:13 +0000 (13:15 +1100)]
syscall: another attempt to keep windows syscall pointers live
This approach was suggested in
https://golang.org/cl/138250043/#msg15.
Unlike current version of mksyscall_windows.go,
new code could be used in go.sys and other external
repos without help from asm.
LGTM=iant
R=golang-codereviews, iant, r
CC=golang-codereviews
https://golang.org/cl/143160046
Rob Pike [Sat, 4 Oct 2014 03:27:08 +0000 (20:27 -0700)]
fmt: print &map like &slice and &struct
It was inconsistent.
Also test these better.
Also document the default format for types.
This wasn't written down.
Rob Pike [Fri, 3 Oct 2014 20:23:35 +0000 (13:23 -0700)]
fmt: part 2 of the great flag rebuild: make %+v work in formatters
Apply a similar transformation to %+v that we did to %#v, making it
a top-level setting separate from the + flag itself. This fixes the
appearance of flags in Formatters and cleans up the code too,
probably making it a little faster.
Russ Cox [Fri, 3 Oct 2014 16:44:20 +0000 (12:44 -0400)]
cmd/cc, runtime: disallow structs without tags
Structs without tags have no unique name to use in the
Go definitions generated from the C types.
This caused issue 8812, fixed by CL 149260043.
Avoid future problems by requiring struct tags.
Update runtime as needed.
(There is no other C code in the tree.)
LGTM=bradfitz, iant
R=golang-codereviews, bradfitz, dave, iant
CC=golang-codereviews, khr, r
https://golang.org/cl/150360043
Rob Pike [Thu, 2 Oct 2014 21:16:58 +0000 (14:16 -0700)]
fmt: make the %#v verb a special flag
The %#v verb is special: it says all values below need to print as %#v.
However, for some situations the # flag has other meanings and this
causes some issues, particularly in how Formatters work. Since %#v
dominates all formatting, translate it into actual state of the formatter
and decouple it from the # flag itself within the calculations (although
it must be restored when methods are doing the work.)
The result is cleaner code and correct handling of # for Formatters.
TODO: Apply the same thinking to the + flag in a followup CL.
Also, the wasString return value in handleMethods is always false,
so eliminate it.
Robert Griesemer [Thu, 2 Oct 2014 20:02:25 +0000 (13:02 -0700)]
math/big: math.Exp should return result >= 0 for |m| > 0
The documentation states that Exp(x, y, m)
computes x**y mod |m| for m != nil && m > 0.
In math.big, Mod is the Euclidean modulus,
which is always >= 0.
Rob Pike [Wed, 1 Oct 2014 22:25:56 +0000 (15:25 -0700)]
doc/go_faq.html: explain the policy about unused imports a little better
This new text won't stop the whining but it might focus the whining a little more.
Rob Pike [Wed, 1 Oct 2014 21:35:12 +0000 (21:35 +0000)]
fmt: fix internal unknownType function
This thing should never be called, but before 151960044 it was being called, incorrectly.
This is now just a precaution but let's pretend it
Fixes #8843
even though that was fixed by 151960044.
The test case was already there and ran, another mystery.
Rob Pike [Wed, 1 Oct 2014 20:18:44 +0000 (13:18 -0700)]
net/rpc: shut down connection if gob has error
The nicest solution would be to buffer the message and only write
it if it encodes correctly, but that adds considerable memory and
CPU overhead for a very rare condition. Instead, we just shut
down the connection if this happens.
Fixes #7689.
Paul van Brouwershaven [Tue, 30 Sep 2014 20:38:48 +0000 (13:38 -0700)]
x509: Fixed ASN.1 encoding in CRL Distribution Points extension
The ASN.1 encoding of the CRL Distribution Points extension showed an invalid false 'IsCompound' which caused a display problem in the Windows certificate viewer.