Add sentences to the docs explaining the limit on input
bytes implicit in the choice of litWidth, and the fact that
compress and decompress litWidth must match.
Fixes#11142.
Change-Id: I20cfb4df35739f7bfeb50b92c78249df3d47942c
Reviewed-on: https://go-review.googlesource.com/11063
Reviewed-by: Nigel Tao <nigeltao@golang.org>
Some latency regressions have crept into our system over the past few
weeks. This CL fixes those by having the mark phase more aggressively
blacken objects so that the mark termination phase, a STW phase, has less
work to do. Three approaches were taken when the mark phase believes
it has no more work to do, ie all the work buffers are empty.
If things have gone well the mark phase is correct and there is
in fact little or no work. In that case the following items will
take very little time. If the mark phase is wrong this CL will
ferret that work out and give the mark phase a chance to deal with
it concurrently before mark termination begins.
When the mark phase first appears to be out of work, it does three things:
1) It switches from allocating white to allocating black to reduce the
number of unmarked objects reachable only from stacks.
2) It flushes and disables per-P GC work caches so all work must be in
globally visible work buffers.
3) It rescans the global roots---the BSS and data segments---so there
are fewer objects to blacken during mark termination. We do not rescan
stacks at this point, though that could be done in a later CL.
After these steps, it again drains the global work buffers.
On a lightly loaded machine the garbage benchmark has reduced the
number of GC cycles with latency > 10 ms from 83 out of 4083 cycles
down to 2 out of 3995 cycles. Maximum latency was reduced from
60+ msecs down to 20 ms.
Change-Id: I152285b48a7e56c5083a02e8e4485dd39c990492
Reviewed-on: https://go-review.googlesource.com/10590
Reviewed-by: Austin Clements <austin@google.com>
When a method is called using the Type.Method(receiver, args...) syntax
without the receiver, or enough arguments, provide the more helpful
error message "not enough arguments in call to method expression
Type.Method" instead of the old message "not enough arguments in call
to Type.Method".
Fixes#8385
Change-Id: Id5037eb1ee5fa93687d4a6557b4a8233b29e9df2
Reviewed-on: https://go-review.googlesource.com/2193
Reviewed-by: Russ Cox <rsc@golang.org>
It's easy for someone who wants a time bigger than any
valid time to reach for time.Unix(1<<63-1, 0), so it
makes sense to explicit say such value is not valid.
Fixes#10906 (again).
Change-Id: If71e32472ae40d86c30e629b982406040a73c4c7
Reviewed-on: https://go-review.googlesource.com/10266
Reviewed-by: Russ Cox <rsc@golang.org>
The Slicing-By-8 [1] algorithm has much performance improvements than
current approach. This patch only uses it for IEEE, which is the most
common case in practice.
There is the benchmark on Mac OS X 10.9:
benchmark old MB/s new MB/s speedup
BenchmarkIEEECrc1KB 349.40 353.03 1.01x
BenchmarkIEEECrc4KB 351.55 934.35 2.66x
BenchmarkCastagnoliCrc1KB 7037.58 7392.63 1.05x
This algorithm need 8K lookup table, so it's enabled only for block
larger than 4K.
We can see about 2.6x improvement for IEEE.
Change-Id: I7f786d20f0949245e4aa101d7921669f496ed0f7
Reviewed-on: https://go-review.googlesource.com/1863
Reviewed-by: Russ Cox <rsc@golang.org>
Check that if a version is declared, for example
in '<?xml version="XX" ?>', version must be '1.0'.
Change-Id: I16ba9f78873a5f31977dcf75ac8e671fe6c08280
Reviewed-on: https://go-review.googlesource.com/8961
Reviewed-by: Russ Cox <rsc@golang.org>
When the scanner receives a non-whitespace character in stateEndTop,
it creates an error message and caches it to return on the next
transition. nextValue() uses the scanner to sub-scan for a value
inside a larger JSON structure. Since stateEndTop is triggered
*after* the ending byte, whatever character immediately follows the
sub-value gets pulled into the scanner's state machine as well.
Even though it is not used and doesn't cause an error, it does
cause the state machine to allocate an error that will never be used.
The fix is to probe the state machine with whitespace after
scanEndObject or scanEndArray to see if the next character would
result in a scanEnd state transition. If so, we can return right
away without processing the next character and avoid triggering
an allocation.
benchmark old ns/op new ns/op delta
BenchmarkCodeEncoder 17022194 16611336 -2.41%
BenchmarkCodeMarshal 18443250 18090144 -1.91%
BenchmarkCodeDecoder 61502053 61010936 -0.80%
BenchmarkCodeUnmarshal 61410829 60363605 -1.71%
BenchmarkCodeUnmarshalReuse 59124836 58361772 -1.29%
BenchmarkUnmarshalString 602 603 +0.17%
BenchmarkUnmarshalFloat64 535 537 +0.37%
BenchmarkUnmarshalInt64 482 482 +0.00%
BenchmarkIssue10335 1206 799 -33.75%
BenchmarkSkipValue 17605751 18355391 +4.26%
BenchmarkEncoderEncode 612 604 -1.31%
benchmark old MB/s new MB/s speedup
BenchmarkCodeEncoder 114.00 116.82 1.02x
BenchmarkCodeMarshal 105.21 107.27 1.02x
BenchmarkCodeDecoder 31.55 31.81 1.01x
BenchmarkCodeUnmarshal 31.60 32.15 1.02x
BenchmarkSkipValue 111.63 107.07 0.96x
benchmark old allocs new allocs delta
BenchmarkIssue10335 11 4 -63.64%
BenchmarkEncoderEncode 2 2 +0.00%
benchmark old bytes new bytes delta
BenchmarkIssue10335 376 272 -27.66%
BenchmarkEncoderEncode 40 40 +0.00%
Fixes#10335
Change-Id: I3d4f2b67f7a038adfb33ba48bb6b680f528baf18
Reviewed-on: https://go-review.googlesource.com/9074
Reviewed-by: Russ Cox <rsc@golang.org>
Suggest running 'go help gopath' when the running 'go install .'
and the folder is outside of GOPATH.
Added link to 'https://golang.org/doc/code.html' in gopath help
for more information.
Example output:
% go install .
go install: no install location for directory f:\x\badmessage outside GOPATH
please run 'go help gopath' for more information
% go help gopath
... SNIP ...
See https://golang.org/doc/code.html for an example.
Fixes#8457
Change-Id: I0ef6ee3c65bb12af2168eafeb757258aa3835664
Reviewed-on: https://go-review.googlesource.com/9258
Reviewed-by: Russ Cox <rsc@golang.org>
Check for errors when reading the headers following the pax headers.
Fixes#11169.
Change-Id: Ifec4a949ec8df8b49fa7cb7a67eb826fe2282ad8
Reviewed-on: https://go-review.googlesource.com/11031
Reviewed-by: Russ Cox <rsc@golang.org>
Return a meaningful message when a profile is empty.
Also rename "IO blocking" to "Network blocking",
currently only network blocking is captured.
Fixes#11098
Change-Id: Ib6f1292b8ade4805756fcb6696ba1fca8f9f39a9
Reviewed-on: https://go-review.googlesource.com/11243
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
There were two issues.
1. Delayed EvGoSysExit could have been emitted during TraceStart,
while it had not yet emitted EvGoInSyscall.
2. Delayed EvGoSysExit could have been emitted during next tracing session.
Fixes#10476Fixes#11262
Change-Id: Iab68eb31cf38eb6eb6eee427f49c5ca0865a8c64
Reviewed-on: https://go-review.googlesource.com/9132
Reviewed-by: Russ Cox <rsc@golang.org>
All the other error messages in this package are "lzw: etc".
Change-Id: Ie359a8912d213f92b15f02abc953127380503048
Reviewed-on: https://go-review.googlesource.com/11228
Reviewed-by: David Symonds <dsymonds@golang.org>
In preparation for rename of cgocall_errno into cgocall and
asmcgocall_errno into asmcgocall in the fllowinng CL.
rsc requested CL 9387 to be split into two parts. This is first part.
Change-Id: I7434f0e4b44dd37017540695834bfcb1eebf0b2f
Reviewed-on: https://go-review.googlesource.com/11166
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Work-around issue #11265 and re-enable tests for Plan9.
Change-Id: I3aabb674a149b8eb936f948dd4cda5fd81454646
Reviewed-on: https://go-review.googlesource.com/11194
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This change will brake the build. The immediately following change
contains the necessary adjustments to make it work again. We're
doing this in two steps to expose the manual changes applied.
Change-Id: I225947da23e190b12e12cbd0c5e6e91628de7f53
Reviewed-on: https://go-review.googlesource.com/11151
Reviewed-by: Alan Donovan <adonovan@google.com>
This fixes a hang during runtime.TestTraceStress.
It also fixes double-scan of stacks, which leads to
stack barrier installation failures.
Both of these have shown up as flaky failures on the dashboard.
Fixes#10941.
Change-Id: Ia2a5991ce2c9f43ba06ae1c7032f7c898dc990e0
Reviewed-on: https://go-review.googlesource.com/11089
Reviewed-by: Austin Clements <austin@google.com>
Adjust timestamps in TestABIChecking to make sure that the library and
executable are rebuilt when expected.
Change-Id: I3288c254ba8201b5b4255347b0cb056fa0908657
Reviewed-on: https://go-review.googlesource.com/11128
Reviewed-by: Michael Hudson-Doyle <michael.hudson@canonical.com>
Reviewed-by: Yves Junqueira <yves.junqueira@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
//go:systemstack means that the function must run on the system stack.
Add one use in runtime as a demonstration.
Fixes#9174.
Change-Id: I8d4a509cb313541426157da703f1c022e964ace4
Reviewed-on: https://go-review.googlesource.com/10840
Reviewed-by: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
I updated some references to 6g, 6l and friends that I came across, as those
programs don't exist anymore. I also fixed some echos in make.rc to match other make.* scripts while I was there.
Change-Id: Ib84532cd4688cf65174dd9869e5d42af98a20a48
Reviewed-on: https://go-review.googlesource.com/11162
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Corrected several issues:
* RFC1951 section 3.2.7 dictates that it is okay for the HDist tree to have a
single code of zero bits. Furthermore, the behavior of the C zlib library
permits empty trees even when there are more than one codes.
* RFC1951 section 3.2.5 shows that HLit codes 286 and 287 are invalid. Thus,
Go's implementation should choke on inputs using these codes.
* RFC1951 section 3.2.5 and 3.2.7 are ambiguous about whether the number of
HDist codes can be greater than 30. The C zlib library (which is the canonical
reference implementation) performs this check here:
62d6112a79/inflate.c (L906)
In addition, a number of test cases were added to the unit tests that exercises
these edge cases. The test cases listed in TestStreams will either fail or
succeed in a manner matching the behaviour of the C zlib version. Given that the
C zlib implementation is the reference for the world, Go's implementation should
match C zlib behaviour.
Fixes#11030
Change-Id: Ic24e4e40ce5832c7e1930249246e86d34bfedaa6
Reviewed-on: https://go-review.googlesource.com/11000
Reviewed-by: Nigel Tao <nigeltao@golang.org>
Also modified test/run.go to ignore messages prefixed <autogenerated>
because those cannot be described with "// ERROR ...", and backed out
patch from issue #9537 because it is no longer necessary. The reasons
described in the 9537 discussion for why escape analysis cannot run
late no longer hold, happily.
Fixes#11053.
Change-Id: Icb14eccdf2e8cde3d0f8fb8a216b765400a96385
Reviewed-on: https://go-review.googlesource.com/11088
Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: David Chase <drchase@google.com>
This change reintroduces CL 8523. CL 8523 was reverted because
it broke darwin and netbsd builds. Now that this test is part
of "go tool dist test" command we could skip OSes that fail.
Updates #10360
Change-Id: Iaaeb5b800126492f36415a439c333a218fe4ab67
Reviewed-on: https://go-review.googlesource.com/11119
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
So the tests don't interfere with each other on windows.
Fixes#11217
Change-Id: I4b3936bc64c95c7274298d6f137b24a28876b625
Reviewed-on: https://go-review.googlesource.com/11138
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This change allows LookupAddr to use getnameinfo through cgo for working
together with various name services other than DNS.
Fixes#7855.
Change-Id: I5b3b4aefe3d1b904541c3350865734d8cbb1c1c4
Reviewed-on: https://go-review.googlesource.com/3420
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This flag is not needed in the std repo because we don't have
tests requiring it. Remove it before it's frozen into the API.
Change-Id: I18b861eea146ad67e7a3c26ee8be681d8065ef12
Reviewed-on: https://go-review.googlesource.com/11150
Reviewed-by: Alan Donovan <adonovan@google.com>
This makes the behaviour match what happens when duplicate symbols are read
from regular object files and fixes errors about cgoAlwaysFalse when linking
an executable that uses cgo against a shared library.
Change-Id: Ibb8cd8fe3f7813cde504b7483f1e857868d7e063
Reviewed-on: https://go-review.googlesource.com/11117
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TestHostname was re-enabled in CL 10753.
However, on Plan 9 the hostname is not obtained
by executing a "hostname" command, but by reading
the #c/sysname file.
Change-Id: I80c0e303f4983fe39ceb300ad64e2c4a8392b695
Reviewed-on: https://go-review.googlesource.com/11033
Run-TryBot: David du Colombier <0intro@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Currently, when shrinkstack computes whether the halved stack
allocation will have enough room for the stack, it accounts for the
stack space that's actively in use but fails to leave extra room for
the stack guard space. As a result, *if* the minimum stack size is
small enough or the guard large enough, it may shrink the stack and
leave less than enough room to run nosplit functions. If the next
function called after the stack shrink is a nosplit function, it may
overflow the stack without noticing and overwrite non-stack memory.
We don't think this is happening under normal conditions right now.
The minimum stack allocation is 2K and the guard is 640 bytes. The
"worst case" stack shrink is from 4K (4048 bytes after stack barrier
array reservation) to 2K (2016 bytes after stack barrier array
reservation), which means the largest "used" size that will qualify
for shrinking is 4048/4 - 8 = 1004 bytes. After copying, that leaves
2016 - 1004 = 1012 bytes of available stack, which is significantly
more than the guard space.
If we were to reduce the minimum stack size to 1K or raise the guard
space above 1012 bytes, the logic in shrinkstack would no longer leave
enough space.
It's also possible to trigger this problem by setting
firstStackBarrierOffset to 0, which puts stack barriers in a debug
mode that steals away *half* of the stack for the stack barrier array
reservation. Then, the largest "used" size that qualifies for
shrinking is (4096/2)/4 - 8 = 504 bytes. After copying, that leaves
(2096/2) - 504 = 8 bytes of available stack; much less than the
required guard space. This causes failures like those in issue #11027
because func gc() shrinks its own stack and then immediately calls
casgstatus (a nosplit function), which overflows the stack and
overwrites a free list pointer in the neighboring span. However, since
this seems to require the special debug mode, we don't think it's
responsible for issue #11027.
To forestall all of these subtle issues, this commit modifies
shrinkstack to correctly account for the guard space when considering
whether to halve the stack allocation.
Change-Id: I7312584addc63b5bfe55cc384a1012f6181f1b9d
Reviewed-on: https://go-review.googlesource.com/10714
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Issues #10240, #10541, #10941, #11023, #11027 and possibly others are
indicating memory corruption in the runtime. One of the easiest places
to both get corruption and detect it is in the allocator's free lists
since they appear throughout memory and follow strict invariants. This
commit adds a check when sweeping a span that its free list is sane
and, if not, it prints the corrupted free list and panics. Hopefully
this will help us collect more information on these failures.
Change-Id: I6d417bcaeedf654943a5e068bd76b58bb02d4a64
Reviewed-on: https://go-review.googlesource.com/10713
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: Austin Clements <austin@google.com>
