The 1-byte write was silently clearing a byte on the stack.
If there was another function call with more arguments
in the same stack frame, no harm done.
Otherwise, if the variable at that location was already zero,
no harm done.
Otherwise, problems.
Fixes#8139.
LGTM=dsymonds
R=golang-codereviews, dsymonds
CC=golang-codereviews, iant, r
https://golang.org/cl/100940043
We were requiring that the defer stack and the panic stack
be completely processed, thinking that if any were left over
the stack scan and the defer stack/panic stack must be out
of sync. It turns out that the panic stack may well have
leftover entries in some situations, and that's okay.
Fixes#8132.
LGTM=minux, r
R=golang-codereviews, minux, r
CC=golang-codereviews, iant, khr
https://golang.org/cl/100900044
The 'continuation pc' is where the frame will continue
execution, if anywhere. For a frame that stopped execution
due to a CALL instruction, the continuation pc is immediately
after the CALL. But for a frame that stopped execution due to
a fault, the continuation pc is the pc after the most recent CALL
to deferproc in that frame, or else 0. That is where execution
will continue, if anywhere.
The liveness information is only recorded for CALL instructions.
This change makes sure that we never look for liveness information
except for CALL instructions.
Using a valid PC fixes crashes when a garbage collection or
stack copying tries to process a stack frame that has faulted.
Record continuation pc in heapdump (format change).
Fixes#8048.
LGTM=iant, khr
R=khr, iant, dvyukov
CC=golang-codereviews, r
https://golang.org/cl/100870044
This CL forces the optimizer to preserve some memory stores
that would be redundant except that a stack scan due to garbage
collection or stack copying might look at them during a function call.
As such, it forces additional memory writes and therefore slows
down the execution of some programs, especially garbage-heavy
programs that are already limited by memory bandwidth.
The slowdown can be as much as 7% for end-to-end benchmarks.
These numbers are from running go1.test -test.benchtime=5s three times,
taking the best (lowest) ns/op for each benchmark. I am excluding
benchmarks with time/op < 10us to focus on macro effects.
All benchmarks are on amd64.
Comparing tip (a27f34c771cb) against this CL on an Intel Core i5 MacBook Pro:
benchmark old ns/op new ns/op delta
BenchmarkBinaryTree17 3876500413 3856337341 -0.52%
BenchmarkFannkuch11 2965104777 2991182127 +0.88%
BenchmarkGobDecode 8563026 8788340 +2.63%
BenchmarkGobEncode 5050608 5267394 +4.29%
BenchmarkGzip 431191816 434168065 +0.69%
BenchmarkGunzip 107873523 110563792 +2.49%
BenchmarkHTTPClientServer 85036 86131 +1.29%
BenchmarkJSONEncode 22143764 22501647 +1.62%
BenchmarkJSONDecode 79646916 85658808 +7.55%
BenchmarkMandelbrot200 4720421 4700108 -0.43%
BenchmarkGoParse 4651575 4712247 +1.30%
BenchmarkRegexpMatchMedium_1K 71986 73490 +2.09%
BenchmarkRegexpMatchHard_1K 111018 117495 +5.83%
BenchmarkRevcomp 648798723 659352759 +1.63%
BenchmarkTemplate 112673009 112819078 +0.13%
Comparing tip (a27f34c771cb) against this CL on an Intel Xeon E5520:
BenchmarkBinaryTree17 5461110720 5393104469 -1.25%
BenchmarkFannkuch11 4314677151 4327177615 +0.29%
BenchmarkGobDecode 11065853 11235272 +1.53%
BenchmarkGobEncode 6500065 6959837 +7.07%
BenchmarkGzip 647478596 671769097 +3.75%
BenchmarkGunzip 139348579 141096376 +1.25%
BenchmarkHTTPClientServer 69376 73610 +6.10%
BenchmarkJSONEncode 30172320 31796106 +5.38%
BenchmarkJSONDecode 113704905 114239137 +0.47%
BenchmarkMandelbrot200 6032730 6003077 -0.49%
BenchmarkGoParse 6775251 6405995 -5.45%
BenchmarkRegexpMatchMedium_1K 111832 113895 +1.84%
BenchmarkRegexpMatchHard_1K 161112 168420 +4.54%
BenchmarkRevcomp 876363406 892319935 +1.82%
BenchmarkTemplate 146273096 148998339 +1.86%
Just to get a sense of where we are compared to the previous release,
here are the same benchmarks comparing Go 1.2 to this CL.
Comparing Go 1.2 against this CL on an Intel Core i5 MacBook Pro:
BenchmarkBinaryTree17 4370077662 3856337341 -11.76%
BenchmarkFannkuch11 3347052657 2991182127 -10.63%
BenchmarkGobDecode 8791384 8788340 -0.03%
BenchmarkGobEncode 4968759 5267394 +6.01%
BenchmarkGzip 437815669 434168065 -0.83%
BenchmarkGunzip 94604099 110563792 +16.87%
BenchmarkHTTPClientServer 87798 86131 -1.90%
BenchmarkJSONEncode 22818243 22501647 -1.39%
BenchmarkJSONDecode 97182444 85658808 -11.86%
BenchmarkMandelbrot200 4733516 4700108 -0.71%
BenchmarkGoParse 5054384 4712247 -6.77%
BenchmarkRegexpMatchMedium_1K 67612 73490 +8.69%
BenchmarkRegexpMatchHard_1K 107321 117495 +9.48%
BenchmarkRevcomp 733270055 659352759 -10.08%
BenchmarkTemplate 109304977 112819078 +3.21%
Comparing Go 1.2 against this CL on an Intel Xeon E5520:
BenchmarkBinaryTree17 5986953594 5393104469 -9.92%
BenchmarkFannkuch11 4861139174 4327177615 -10.98%
BenchmarkGobDecode 11830997 11235272 -5.04%
BenchmarkGobEncode 6608722 6959837 +5.31%
BenchmarkGzip 661875826 671769097 +1.49%
BenchmarkGunzip 138630019 141096376 +1.78%
BenchmarkHTTPClientServer 71534 73610 +2.90%
BenchmarkJSONEncode 30393609 31796106 +4.61%
BenchmarkJSONDecode 139645860 114239137 -18.19%
BenchmarkMandelbrot200 5988660 6003077 +0.24%
BenchmarkGoParse 6974092 6405995 -8.15%
BenchmarkRegexpMatchMedium_1K 111331 113895 +2.30%
BenchmarkRegexpMatchHard_1K 165961 168420 +1.48%
BenchmarkRevcomp 995049292 892319935 -10.32%
BenchmarkTemplate 145623363 148998339 +2.32%
Fixes#8036.
LGTM=khr
R=golang-codereviews, josharian, khr
CC=golang-codereviews, iant, r
https://golang.org/cl/99660044
CL 51010045 fixed the first one of these:
cmd/gc: return canonical Node* from temp
For historical reasons, temp was returning a copy
of the created Node*, not the original Node*.
This meant that if analysis recorded information in the
returned node (for example, n->addrtaken = 1), the
analysis would not show up on the original Node*, the
one kept in fn->dcl and consulted during liveness
bitmap creation.
Correct this, and watch for it when setting addrtaken.
Fixes#7083.
R=khr, dave, minux.ma
CC=golang-codereviews
https://golang.org/cl/51010045
CL 53200043 fixed the second:
cmd/gc: fix race build
Missed this case in CL 51010045.
TBR=khr
CC=golang-codereviews
https://golang.org/cl/53200043
This CL fixes the third. There are only three nod(OXXX, ...)
calls in sinit.c, so maybe we're done. Embarassing that it
took three CLs to find all three.
Fixes#8028.
LGTM=khr
R=golang-codereviews, khr
CC=golang-codereviews, iant
https://golang.org/cl/100800046
In the first very rough draft of the reordering code
that was introduced in the Go 1.3 cycle, the pre-allocated
temporary for a ... argument was held in n->right.
It moved to n->alloc but the code avoiding n->right
was left behind in order.c. In copy(x, <-c), the receive
is in n->right and must be processed. Delete the special
case code, removing the bug.
Fixes#8039.
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/100820044
The code cannot have worked before, because it was
trying to use the old value in a range check for the new
type, which might have a different representation
(hence the 'internal compiler error').
Fixes#8073.
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/98630045
I don't know when the bug was fixed, but empirically it was.
Make sure it stays fixed by adding a test.
Fixes#7884.
LGTM=adg
R=golang-codereviews, adg
CC=golang-codereviews
https://golang.org/cl/93500043
The temporary-introducing pass was not recursing
into the argumnt of a receive operation.
Fixes#8011.
LGTM=r
R=golang-codereviews, r
CC=golang-codereviews, iant, khr
https://golang.org/cl/91540043
The introduction of temporaries in order.c was not
quite right for two corner cases:
1) The rewrite that pushed new variables on the lhs of
a receive into the body of the case was dropping the
declaration of the variables. If the variables escape,
the declaration is what allocates them.
Caught by escape analysis sanity check.
In fact the declarations should move into the body
always, so that we only allocate if the corresponding
case is selected. Do that. (This is an optimization that
was already present in Go 1.2. The new order code just
made it stop working.)
Fixes#7997.
2) The optimization to turn a single-recv select into
an ordinary receive assumed it could take the address
of the destination; not so if the destination is _.
Fixes#7998.
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/100480043
The GC program describing a data structure sometimes trusts the
pointer base type and other times does not (if not, the garbage collector
must fall back on per-allocation type information stored in the heap).
Make the scanning of a pointer in an interface do the same.
This fixes a crash in a particular use of reflect.SliceHeader.
Fixes#8004.
LGTM=khr
R=golang-codereviews, khr
CC=0xe2.0x9a.0x9b, golang-codereviews, iant, r
https://golang.org/cl/100470045
Globals, function arguments, and results are special cases in
registerization.
Globals must be flushed aggressively, because nearly any
operation can cause a panic, and the recovery code must see
the latest values. Globals also must be loaded aggressively,
because nearly any store through a pointer might be updating a
global: the compiler cannot see all the "address of"
operations on globals, especially exported globals. To
accomplish this, mark all globals as having their address
taken, which effectively disables registerization.
If a function contains a defer statement, the function results
must be flushed aggressively, because nearly any operation can
cause a panic, and the deferred code may call recover, causing
the original function to return the current values of its
function results. To accomplish this, mark all function
results as having their address taken if the function contains
any defer statements. This causes not just aggressive flushing
but also aggressive loading. The aggressive loading is
overkill but the best we can do in the current code.
Function arguments must be considered live at all safe points
in a function, because garbage collection always preserves
them: they must be up-to-date in order to be preserved
correctly. Accomplish this by marking them live at all call
sites. An earlier attempt at this marked function arguments as
having their address taken, which disabled registerization
completely, making programs slower. This CL's solution allows
registerization while preserving safety. The benchmark speedup
is caused by being able to registerize again (the earlier CL
lost the same amount).
benchmark old ns/op new ns/op delta
BenchmarkEqualPort32 61.4 56.0 -8.79%
benchmark old MB/s new MB/s speedup
BenchmarkEqualPort32 521.56 570.97 1.09x
Fixes#1304. (again)
Fixes#7944. (again)
Fixes#7984.
Fixes#7995.
LGTM=khr
R=golang-codereviews, khr
CC=golang-codereviews, iant, r
https://golang.org/cl/97500044
The inputs to a function are marked live at all times in the
liveness bitmaps, so that the garbage collector will not free
the things they point at and reuse the pointers, so that the
pointers shown in stack traces are guaranteed not to have
been recycled.
Unfortunately, no one told the register optimizer that the
inputs need to be preserved at all call sites. If a function
is done with a particular input value, the optimizer will stop
preserving it across calls. For single-word values this just
means that the value recorded might be stale. For multi-word
values like slices, the value recorded could be only partially stale:
it can happen that, say, the cap was updated but not the len,
or that the len was updated but not the base pointer.
Either of these possibilities (and others) would make the
garbage collector misinterpret memory, leading to memory
corruption.
This came up in a real program, in which the garbage collector's
'slice len ≤ slice cap' check caught the inconsistency.
Fixes#7944.
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews, khr
https://golang.org/cl/100370045
This is joint work with Daniel Morsing.
In order for the register allocator to alias two variables, they must have the same width, stack offset, and etype. Code generation was altering a variable's etype in a few places. This prevented the variable from being moved to a register, which in turn prevented peephole optimization. This failure to alias was very common, with almost 23,000 instances just running make.bash.
This phenomenon was not visible in the register allocation debug output because the variables that failed to alias had the same name. The debugging-only change to bits.c fixes this by printing the variable number with its name.
This CL fixes the source of all etype mismatches for 6g, all but one case for 8g, and depressingly few cases for 5g. (I believe that extending CL 6819083 to 5g is a prerequisite.) Fixing the remaining cases in 8g and 5g is work for the future.
The etype mismatch fixes are:
* [gc] Slicing changed the type of the base pointer into a uintptr in order to perform arithmetic on it. Instead, support addition directly on pointers.
* [*g] OSPTR was giving type uintptr to slice base pointers; undo that. This arose, for example, while compiling copy(dst, src).
* [8g] 64 bit float conversion was assigning int64 type during codegen, overwriting the existing uint64 type.
Note that some etype mismatches are appropriate, such as a struct with a single field or an array with a single element.
With these fixes, the number of registerizations that occur while running make.bash for 6g increases ~10%. Hello world binary size shrinks ~1.5%. Running all benchmarks in the standard library show performance improvements ranging from nominal to substantive (>10%); a full comparison using 6g on my laptop is available at https://gist.github.com/josharian/8f9b5beb46667c272064. The microbenchmarks must be taken with a grain of salt; see issue 7920. The few benchmarks that show real regressions are likely due to issue 7920. I manually examined the generated code for the top few regressions and none had any assembly output changes. The few benchmarks that show extraordinary improvements are likely also due to issue 7920.
Performance results from 8g appear similar to 6g.
5g shows no performance improvements. This is not surprising, given the discussion above.
Update #7316
LGTM=rsc
R=rsc, daniel.morsing, bradfitz
CC=dave, golang-codereviews
https://golang.org/cl/91850043
Before we used line 1 of the first source file.
This should be clearer.
Fixes#4388.
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/92250044
The register allocator decides which variables should be placed into registers by charging for each load/store and crediting for each use, and then selecting an allocation with minimal cost. NOPs will be eliminated, however, so using a variable in a NOP should not generate credit.
Issue 7867 arises from attempted registerization of multi-word variables because they are used in NOPs. By not crediting for that use, they will no longer be considered for registerization.
This fix could theoretically lead to better register allocation, but NOPs are rare relative to other instructions.
Fixes#7867.
LGTM=rsc
R=rsc
CC=golang-codereviews
https://golang.org/cl/94810044
Variables declared with 'var' have no sym->def.
Fixes#7794.
LGTM=rsc
R=golang-codereviews, bradfitz, rsc
CC=golang-codereviews
https://golang.org/cl/88360043
1. On entry to a function, only zero the ambiguously live stack variables.
Before, we were zeroing all stack variables containing pointers.
The zeroing is pretty inefficient right now (issue 7624), but there are also
too many stack variables detected as ambiguously live (issue 7345),
and that must be addressed before deciding how to improve the zeroing code.
(Changes in 5g/ggen.c, 6g/ggen.c, 8g/ggen.c, gc/pgen.c)
Fixes#7647.
2. Make the regopt word-based liveness analysis preserve the
whole-variable liveness property expected by the garbage collection
bitmap liveness analysis. That is, if the regopt liveness decides that
one word in a struct needs to be preserved, make sure it preserves
the entire struct. This is particularly important for multiword values
such as strings, slices, and interfaces, in which all the words need
to be present in order to understand the meaning.
(Changes in 5g/reg.c, 6g/reg.c, 8g/reg.c.)
Fixes#7591.
3. Make the regopt word-based liveness analysis treat a variable
as having its address taken - which makes it preserved across
all future calls - whenever n->addrtaken is set, for consistency
with the gc bitmap liveness analysis, even if there is no machine
instruction actually taking the address. In this case n->addrtaken
is incorrect (a nicer way to put it is overconservative), and ideally
there would be no such cases, but they can happen and the two
analyses need to agree.
(Changes in 5g/reg.c, 6g/reg.c, 8g/reg.c; test in bug484.go.)
Fixes crashes found by turning off "zero everything" in step 1.
4. Remove spurious VARDEF annotations. As the comment in
gc/pgen.c explains, the VARDEF must immediately precede
the initialization. It cannot be too early, and it cannot be too late.
In particular, if a function call sits between the VARDEF and the
actual machine instructions doing the initialization, the variable
will be treated as live during that function call even though it is
uninitialized, leading to problems.
(Changes in gc/gen.c; test in live.go.)
Fixes crashes found by turning off "zero everything" in step 1.
5. Do not treat loading the address of a wide value as a signal
that the value must be initialized. Instead depend on the existence
of a VARDEF or the first actual read/write of a word in the value.
If the load is in order to pass the address to a function that does
the actual initialization, treating the load as an implicit VARDEF
causes the same problems as described in step 4.
The alternative is to arrange to zero every such value before
passing it to the real initialization function, but this is a much
easier and more efficient change.
(Changes in gc/plive.c.)
Fixes crashes found by turning off "zero everything" in step 1.
6. Treat wide input parameters with their address taken as
initialized on entry to the function. Otherwise they look
"ambiguously live" and we will try to emit code to zero them.
(Changes in gc/plive.c.)
Fixes crashes found by turning off "zero everything" in step 1.
7. An array of length 0 has no pointers, even if the element type does.
Without this change, the zeroing code complains when asked to
clear a 0-length array.
(Changes in gc/reflect.c.)
LGTM=khr
R=khr
CC=golang-codereviews
https://golang.org/cl/80160044
Revision 3ae4607a43ff introduced CONVNOP layers
to fix type checking issues arising from comparisons.
The added complexity made 8g run out of registers
when compiling an equality function in go.net/ipv6.
A similar issue occurred in test/sizeof.go on
amd64p32 with 6g.
Fixes#7405.
LGTM=khr
R=rsc, dave, iant, khr
CC=golang-codereviews
https://golang.org/cl/78100044
A too large float constant is an error.
A too small float constant is rounded to zero.
Fixes#7419
Update #6902
LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/76730046
The lowering to runtime calls introduces hidden pointers to the
arguments of select clauses. When implicit conversions were
involved it could end up with incompatible pointers. Since the
pointed-to types have the same representation, we can introduce a
forced conversion.
Fixes#6847.
LGTM=rsc
R=rsc, iant, khr
CC=golang-codereviews
https://golang.org/cl/72380043
The byte that r is or'd into is already 0x7, so the failure to zero r only
impacts the generated machine code if the register is > 7.
Fixes#7044.
LGTM=dave, minux.ma, rsc
R=dave, minux.ma, bradfitz, rsc
CC=golang-codereviews
https://golang.org/cl/73730043
Revision c0e0467635ec (cmd/gc: return canonical Node* from temp)
exposed original nodes of temporaries, allowing callers to mutate
their types.
In walkcompare a temporary could be typed as ideal because of
this. Additionnally, assignment of a comparison result to
a custom boolean type was broken.
Fixes#7366.
LGTM=rsc
R=rsc, iant, khr
CC=golang-codereviews
https://golang.org/cl/66930044
The error message was previously off by one in all cases.
Fixes#7150.
LGTM=r
R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/65850043
Not recording the address being taken was causing
the liveness analysis not to preserve x in the absence
of direct references to x, which in turn was making the
net test fail with GOGC=0.
In addition to the test, this fixes a bug wherein
GOGC=0 go test -short net
crashed if liveness analysis was in use (like at tip, not like Go 1.2).
TBR=ken2
CC=golang-codereviews
https://golang.org/cl/64470043
The existing tests issue4463.go and issue4654.go had failures at
typechecking and did not test walking the AST.
Fixes#7272.
LGTM=khr
R=khr, rsc, iant
CC=golang-codereviews
https://golang.org/cl/60550044
For historical reasons, temp was returning a copy
of the created Node*, not the original Node*.
This meant that if analysis recorded information in the
returned node (for example, n->addrtaken = 1), the
analysis would not show up on the original Node*, the
one kept in fn->dcl and consulted during liveness
bitmap creation.
Correct this, and watch for it when setting addrtaken.
Fixes#7083.
R=khr, dave, minux.ma
CC=golang-codereviews
https://golang.org/cl/51010045
Nodes of goto statements were corrupted when written
to export data.
Fixes#7023.
R=rsc, dave, minux.ma
CC=golang-codereviews
https://golang.org/cl/46190043
Gccgo doesn't have the same equivalent of file name and
package as the gc compiler, so the error messages are
necessarily different.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/40510048
fixedbugs/issue4510.dir/f2.go:7: error: 'fmt' defined as both imported name and global name
f1.go:7: note: 'fmt' imported here
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/41530044
const1.go:33: error: integer constant overflow
<similar errors omitted>
const1.go:64: error: division by zero
const1.go:67: error: floating point constant overflow
const1.go:68: error: floating point constant overflow
const1.go:69: error: floating point constant overflow
const1.go:70: error: division by zero
const1.go:71: error: expected integer type
const1.go:77: error: argument 1 has incompatible type (cannot use type int8 as type int)
const1.go:78: error: argument 1 has incompatible type (cannot use type int8 as type int)
const1.go:79: error: argument 1 has incompatible type (cannot use type uint8 as type int)
const1.go:81: error: argument 1 has incompatible type (cannot use type float32 as type int)
const1.go:82: error: argument 1 has incompatible type (cannot use type float64 as type int)
const1.go:83: error: floating point constant truncated to integer
const1.go:85: error: argument 1 has incompatible type (cannot use type float64 as type int)
const1.go:86: error: argument 1 has incompatible type (cannot use type string as type int)
const1.go:87: error: argument 1 has incompatible type (cannot use type bool as type int)
const1.go:90: error: const initializer cannot be nil
const1.go:91: error: expression is not constant
const1.go:92: error: expression is not constant
const1.go:93: error: invalid constant type
const1.go:94: error: invalid constant type
fixedbugs/bug462.go:17: error: unknown field 'os.File' in 'T'
fixedbugs/issue3705.go:9: error: cannot declare init - must be func
fixedbugs/issue4251.go:12: error: inverted slice range
fixedbugs/issue4251.go:16: error: inverted slice range
fixedbugs/issue4251.go:20: error: inverted slice range
fixedbugs/issue4405.go:11: error: invalid character 0x7 in identifier
fixedbugs/issue4405.go:12: error: invalid character 0x8 in identifier
fixedbugs/issue4405.go:13: error: invalid character 0xb in identifier
fixedbugs/issue4405.go:14: error: invalid character 0xc in identifier
fixedbugs/issue4429.go:15: error: expected pointer
fixedbugs/issue4517d.go:9: error: cannot import package as init
fixedbugs/issue4545.go:17: error: invalid context-determined non-integer type for left operand of shift
fixedbugs/issue4545.go:16: error: incompatible types in binary expression
fixedbugs/issue4610.go:15: error: expected ';' or '}' or newline
fixedbugs/issue4610.go:16: error: expected declaration
fixedbugs/issue4654.go:15: error: value computed is not used
<similar errors omitted>
fixedbugs/issue4776.go:9: error: program must start with package clause
fixedbugs/issue4776.go:9: error: expected ';' or newline after package clause
fixedbugs/issue4813.go:31: error: index must be integer
<similar errors omitted>
fixedbugs/issue4847.go:22: error: initialization expression for 'matchAny' depends upon itself
fixedbugs/issue5089.go:13: error: redefinition of 'bufio.Buffered': receiver name changed
fixedbugs/issue5089.go:11: note: previous definition of 'bufio.Buffered' was here
fixedbugs/issue5172.go:17: error: reference to undefined field or method 'bar'
fixedbugs/issue5172.go:18: error: reference to undefined field or method 'bar'
fixedbugs/issue5172.go:12: error: use of undefined type 'bar'
fixedbugs/issue5358.go:16: error: argument 2 has incompatible type
fixedbugs/issue5581.go:29: error: use of undefined type 'Blah'
funcdup.go:10: error: redefinition of 'i'
funcdup.go:10: note: previous definition of 'i' was here
<similar errors omitted>
funcdup2.go:10: error: redefinition of 'i'
funcdup2.go:10: note: previous definition of 'i' was here
<similar errors omitted>
slice3err.go:20: error: middle index required in 3-index slice
<similar errors omitted>
slice3err.go:20: error: final index required in 3-index slice
<similar errors omitted>
slice3err.go:21: error: final index required in 3-index slice
slice3err.go:46: error: invalid 3-index slice of string
<similar errors omitted>
slice3err.go:57: error: inverted slice range
<similar errors omitted>
slice3err.go:62: error: invalid slice index: capacity less than length
slice3err.go:64: error: invalid slice index: capacity less than start
slice3err.go:65: error: invalid slice index: capacity less than start
slice3err.go:66: error: invalid slice index: capacity less than start
slice3err.go:68: error: invalid slice index: capacity less than length
slice3err.go:70: error: invalid slice index: capacity less than start
slice3err.go:80: error: invalid slice index: capacity less than length
slice3err.go:82: error: invalid slice index: capacity less than start
slice3err.go:83: error: invalid slice index: capacity less than start
slice3err.go:84: error: invalid slice index: capacity less than start
slice3err.go:86: error: invalid slice index: capacity less than length
slice3err.go:88: error: invalid slice index: capacity less than start
slice3err.go:99: error: array index out of bounds
<similar errors omitted>
slice3err.go:106: error: invalid slice index: capacity less than length
slice3err.go:107: error: invalid slice index: capacity less than start
slice3err.go:118: error: invalid slice index: capacity less than length
slice3err.go:119: error: invalid slice index: capacity less than start
syntax/semi1.go:10: error: missing '{' after if clause
syntax/semi1.go:10: error: reference to undefined name 'x'
syntax/semi1.go:10: error: reference to undefined name 'y'
syntax/semi1.go:12: error: reference to undefined name 'z'
syntax/semi2.go:10: error: missing '{' after switch clause
syntax/semi2.go:10: error: reference to undefined name 'x'
syntax/semi3.go:10: error: missing '{' after for clause
syntax/semi3.go:10: error: reference to undefined name 'x'
syntax/semi3.go:10: error: reference to undefined name 'y'
syntax/semi3.go:10: error: reference to undefined name 'z'
syntax/semi3.go:12: error: reference to undefined name 'z'
syntax/semi4.go:11: error: missing '{' after for clause
syntax/semi4.go:10: error: reference to undefined name 'x'
syntax/semi4.go:12: error: reference to undefined name 'z'
typecheck.go:12: error: reference to undefined name 'b'
typecheck.go:17: error: reference to undefined name 'c'
typecheck.go:11: error: use of undefined type 'b'
typecheck.go:16: error: not enough arguments
typecheck.go:17: error: not enough arguments
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/41520044
There is no necessary relationship between the imports of the
packages a and b, and gccgo happens to import them in a
different order, leading to different output. This ordering
is not the purpose of the test in any case.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/40400043
Types in function scope can have methods on them if they embed another type, but we didn't make the name unique, meaning that 2 identically named types in different functions would conflict with eachother.
Fixes#6269.
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13326045
The compiler computes initialization order by finding
a spanning tree between a package's global variables.
But it does so by walking both variables and functions
and stops detecting cycles between variables when they
mix with a cycle of mutually recursive functions.
Fixes#4847.
R=golang-dev, daniel.morsing, rsc
CC=golang-dev
https://golang.org/cl/9663047
Individual variables bigger than 10 MB are now
moved to the heap, as if they had escaped on
their own.
This avoids ridiculous stacks for programs that
do things like
x := [1<<30]byte{}
... use x ...
If 10 MB is too small, we can raise the limit.
Fixes#6077.
R=ken2
CC=golang-dev
https://golang.org/cl/12650045
Backends do not exactly expect receiving binary operators with
constant operands or use workarounds to move them to
register/stack in order to handle them.
Fixes#5841.
R=golang-dev, daniel.morsing, rsc
CC=golang-dev
https://golang.org/cl/11107044
clearfat (used to zero initialize structures) will use AX for x86 block ops. If we write to AX while calculating the dest pointer, we will fill the structure with incorrect values.
Since 64-bit arithmetic uses AX to synthesize a 64-bit register, getting an adress by indexing with 64-bit ops can clobber the register.
Fixes#5820.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/11383043
Deferred functions are not run by a call instruction. They are run by
the runtime editing registers to make the call start with a caller PC
returning to a
CALL deferreturn
instruction.
That instruction has always had the line number of the function's
closing brace, but that instruction's line number is irrelevant.
Stack traces show the line number of the instruction before the
return PC, because normally that's what started the call. Not so here.
The instruction before the CALL deferreturn could be almost anywhere
in the function; it's unrelated and its line number is incorrect to show.
Fix the line number by inserting a true hardware no-op with the right
line number before the returned-to CALL instruction. That is, the deferred
calls now appear to start with a caller PC returning to the second instruction
in this sequence:
NOP
CALL deferreturn
The traceback will show the line number of the NOP, which we've set
to be the line number of the function's closing brace.
The NOP here is not the usual pseudo-instruction, which would be
elided by the linker. Instead it is the real hardware instruction:
XCHG AX, AX on 386 and amd64, and AND.EQ R0, R0, R0 on ARM.
Fixes#5856.
R=ken2, ken
CC=golang-dev
https://golang.org/cl/11223043
Escape analysis needs the right curfn value on a dclfunc node, otherwise it will not analyze the function.
When generating method value wrappers, we forgot to set the curfn correctly.
Fixes#5753.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/10383048
A struct with a single field was considered as equivalent to the
field type, which is incorrect is the field is blank.
Fields with padding could make the compiler think some
types are comparable when they are not.
Fixes#5698.
R=rsc, golang-dev, daniel.morsing, bradfitz, gri, r
CC=golang-dev
https://golang.org/cl/10271046
Design doc at golang.org/s/go12slice.
This is an experimental feature and may not be included in the release.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/10743046
Exported inlined functions that perform a string conversion
using a non-exported named type may miss it in export data.
Fixes#5755.
R=rsc, golang-dev, ality, r
CC=golang-dev
https://golang.org/cl/10464043
The existing compilers convert empty strings to empty
but non-nil byte and rune slices. The spec required
a nil byte and rune slice in those cases. That seems
an odd additional requirement. Adjust the spec to
match the reality.
Also, removed over-specification for conversions of
nil []byte and []rune: such nil slices already act
like empty slices and thus don't need extra language.
Added extra examples instead.
Fixes#5704.
R=rsc, r, iant
CC=golang-dev
https://golang.org/cl/10440045
This avoids problems with inlining in genwrappers, which
occurs after functions have been compiled. Compiling a
function may cause some unused local vars to be removed from
the list. Since a local var may be unused due to
optimization, it is possible that a removed local var winds up
beingused in the inlined version, in which case hilarity
ensues.
Fixes#5515.
R=golang-dev, khr, dave
CC=golang-dev
https://golang.org/cl/10210043
It works on i386, but fails on amd64 and arm.
««« original CL description
runtime: prevent the GC from seeing the content of a frame in runfinq()
Fixes#5348.
R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/8954044
»»»
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/8695051
They caused internal compiler errors and they're expensive enough that inlining them doesn't make sense.
Fixes#5259.
R=golang-dev, r, iant, remyoudompheng
CC=golang-dev
https://golang.org/cl/8636043
The offset of an embedded field s.X must be relative to s
and not to the implicit s.Field of which X is a direct field.
Moreover, no indirections may happen on the path.
Fixes#4909.
R=nigeltao, ality, daniel.morsing, iant, gri, r
CC=golang-dev
https://golang.org/cl/8287043
Reusing it when multiple comparisons occurred in the same
function call led to bad overwriting.
Fixes#5162.
R=golang-dev, daniel.morsing
CC=golang-dev
https://golang.org/cl/8174047
Usually, there is no esc info when inlining, but there will be when generating inlined wrapper functions.
If we don't use this information, we get invalid addresses on the stack.
Fixes#5056.
R=golang-dev, rsc
CC=golang-dev, remyoudompheng
https://golang.org/cl/7850045
It used to not mark parameters as escaping if only one of the
fields it points to leaks out of the function. This causes
problems when importing from another package.
Fixes#4964.
R=rsc, lvd, dvyukov, daniel.morsing
CC=golang-dev
https://golang.org/cl/7648045
Composite literals using the &T{} form were incorrectly
exported, leading to weird errors at import time.
Fixes#4879.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/7395054
A new comment directive //go:noescape instructs the compiler
that the following external (no body) func declaration should be
treated as if none of its arguments escape to the heap.
Fixes#4099.
R=golang-dev, dave, minux.ma, daniel.morsing, remyoudompheng, adg, agl, iant
CC=golang-dev
https://golang.org/cl/7289048
Was not re-walking the new AND node, so that its ullman
count was wrong, so that the code generator attempted to
store values in registers across the call.
Fixes#4752.
R=ken2
CC=golang-dev
https://golang.org/cl/7288054
For consistency with conversions that look like function calls,
conversions that don't look like function calls now allow an
optional trailing comma.
That is, int(x,) has always been syntactically valid.
Now []int(x,) is valid too.
Fixes#4162.
R=ken2
CC=golang-dev
https://golang.org/cl/7288045
Expressions involving nil, even if they can be evaluated
at compile time, do not count as Go constants and cannot
be used in const initializers.
Fixes#4673.
Fixes#4680.
R=ken2
CC=golang-dev
https://golang.org/cl/7278043
The test case of issue 4585 was not passing due to
miscalculation of memequal args, and the previous fix
does not handle padding at the end of a struct.
Handling of padding at end of structs also fixes the case
of [n]T where T is such a padded struct.
Fixes#4585.
(again)
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/7133059
The linker split PKGDEF into (prefix, name, def) pairs,
and defines def to begin after a space following the identifier.
This is totally wrong for the following export data:
func "".FunctionName()
var SomethingCompletelyUnrelated int
The linker would parse
name=`"".FunctionName()\n\tvar`
def=`SomethingCompletelyUnrelated int`
since there is no space after FunctionName.
R=minux.ma, rsc
CC=golang-dev
https://golang.org/cl/7068051
A constant node of type uintptr with a nil literal could
happen in two cases: []int(nil)[1:] and
uintptr(unsafe.Pointer(nil)).
Fixes#4614.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/7059043
Unnamed types like structs with embedded fields can have methods.
These methods are generated on-the-fly by the compiler and
it may happen for identical types in different packages.
The linker must accept these multiple definitions.
Fixes#4590.
R=golang-dev, rsc
CC=golang-dev, remy
https://golang.org/cl/7030051
Before this CL, defining the variable worked fine, but then when
the implicit package-level init func was created, that caused a
name collision and a confusing error about the redeclaration.
Also add a test for issue 3705 (func init() needs body).
Fixes#4517.
R=ken2
CC=golang-dev
https://golang.org/cl/7008045
Ordinary variable load was assumed to be not worth saving,
but not if one of the function calls later might change
its value.
Fixes#4313.
R=ken2
CC=golang-dev
https://golang.org/cl/6997047
The patch makes the compile user an ordinary package-local
symbol for the name of embedded fields of builtin type.
This is incompatible with the fix delivered for issue 2687
(revision 3c060add43fb) but fixes it in a different way, because
the explicit symbol on the field makes the typechecker able to
find it in lookdot.
Fixes#3552.
R=lvd, rsc, daniel.morsing
CC=golang-dev
https://golang.org/cl/6866047
The typechecking code was doing an extra, unnecessary
indirection.
Fixes#4458.
R=golang-dev, daniel.morsing, rsc
CC=golang-dev
https://golang.org/cl/6998051
A fatal error used to happen when escassign-ing a multiple
function return to a single node. However, the situation
naturally appears when using "go f(g())" or "defer f(g())",
because g() is escassign-ed to sink.
Fixes#4529.
R=golang-dev, lvd, minux.ma, rsc
CC=golang-dev
https://golang.org/cl/6920060
I just committed a patch to gccgo that notices that float was
never defined, causing an additional unmatched error message.
Rename the type to avoid that message.
R=golang-dev, minux.ma
CC=golang-dev
https://golang.org/cl/6947049
This check for BADWIDTH might happen while in defercheckwidth, making it raise errors for non-erroneous situations.
Fixes#4495.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6927043
Used to say:
issue4251.go:12: inverted slice range
issue4251.go:12: constant -1 overflows uint64
issue4251.go:16: inverted slice range
issue4251.go:16: constant -1 overflows uint64
issue4251.go:20: inverted slice range
issue4251.go:20: constant -1 overflows uint64
With this patch, only gives the "inverted slice range" errors.
R=golang-dev, daniel.morsing
CC=golang-dev
https://golang.org/cl/6871058
Fixes#4396.
For fixed arrays larger than the unmapped page, agenr would general a nil check by loading the first word of the array. However there is no requirement for the first element of a byte array to be word aligned, so this check causes a trap on ARMv5 hardware (ARMv6 since relaxed that restriction, but it probably still comes at a cost).
Switching the check to MOVB ensures alignment is not an issue. This check is only invoked in a few places in the code where large fixed arrays are embedded into structs, compress/lzw is the biggest offender, and switching to MOVB has no observable performance penalty.
Thanks to Rémy and Daniel Morsing for helping me debug this on IRC last night.
R=remyoudompheng, minux.ma, rsc
CC=golang-dev
https://golang.org/cl/6854063
The current spec says that when calling make, if both len and
cap are constant, it is an error if len > cap. The gc
compiler does not yet implement that, but when it does this
will need to change. Changing it now for the benefit of
gccgo.
R=gri
CC=golang-dev
https://golang.org/cl/6867064
The compiler was confused when inlining a T.Method(f()) call
where f returns multiple values: support for this was marked
as TODO.
Variadic calls are not supported but are not inlined either.
Add a test preventively for that case.
Fixes#4167.
R=golang-dev, rsc, lvd
CC=golang-dev
https://golang.org/cl/6871043
Bools from comparisons can be assigned to all bool types, but this idealness would propagate through logical operators when the result should have been lowered to a non-ideal form.
Fixes#3924.
R=golang-dev, remyoudompheng, r, rsc, mtj
CC=golang-dev
https://golang.org/cl/6855061
The stack overflow checker in the linker uses the spadj field
to determine whether stack space will be large enough or not.
When spadj=0, the checker treats the function as a nosplit
and emits an error although the program is correct.
Also enable the stack checker in 8l.
Fixes#4316.
R=rsc, golang-dev
CC=golang-dev
https://golang.org/cl/6855088
When exporting a body containing
x, ok := v.(Type)
the definition for Type was not being included, so when the body
was actually used, it would cause an "unknown type" compiler error.
Fixes#4370.
R=ken2
CC=golang-dev
https://golang.org/cl/6827064
The old code worked with gc, I assume because the linker
unified identical strings, but it failed with gccgo.
R=rsc
CC=gobot, golang-dev
https://golang.org/cl/6826063
Avoids problems with local declarations shadowing other names.
We write a more explicit form than the incoming program, so there
may be additional type annotations. For example:
int := "hello"
j := 2
would normally turn into
var int string = "hello"
var j int = 2
but the int variable shadows the int type in the second line.
This CL marks all local variables with a per-function sequence number,
so that this would instead be:
var int·1 string = "hello"
var j·2 int = 2
Fixes#4326.
R=ken2
CC=golang-dev
https://golang.org/cl/6816100
When local declarations needed unexported types, these could
be missing in the export data.
Fixes build with -gcflags -lll, except for exp/gotype.
R=golang-dev, rsc, lvd
CC=golang-dev
https://golang.org/cl/6813067
I fixed a bug in gccgo that was causing it to only give an
error for the first package that was imported and not used.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6813058
Someone new to the language may not know the connection between ints and arrays, which was the only thing that the previous error told you anything about.
Fixes#4256.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6739048
Since this patch changes the way complex literals are written
in export data, there are a few other glitches.
Fixes#4159.
R=golang-dev, rsc
CC=golang-dev, remy
https://golang.org/cl/6674047
The other tests either need a complex procedure
or are architecture- or OS-dependent.
Update #4139.
R=golang-dev, daniel.morsing, iant
CC=golang-dev
https://golang.org/cl/6618062
Also mention that ignoring second blank identifier of range is required by the spec in the code.
Fixes#4173.
R=daniel.morsing, remyoudompheng, r
CC=golang-dev
https://golang.org/cl/6594043
const1.go:31:12: error: integer constant overflow
const1.go:31:12: error: integer constant overflow
const1.go:33:12: error: integer constant overflow
const1.go:33:12: error: integer constant overflow
const1.go:34:14: error: integer constant overflow
const1.go:35:17: error: integer constant overflow
const1.go:35:17: error: integer constant overflow
const1.go:35:17: error: integer constant overflow
const1.go:35:17: error: integer constant overflow
const1.go:35:17: error: integer constant overflow
const1.go:36:19: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:37:24: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:37:24: error: integer constant overflow
const1.go:37:15: error: integer constant overflow
const1.go:38:12: error: integer constant overflow
const1.go:38:12: error: integer constant overflow
const1.go:38:12: error: integer constant overflow
const1.go:38:12: error: integer constant overflow
const1.go:41:20: error: integer constant overflow
const1.go:41:20: error: integer constant overflow
const1.go:42:20: error: integer constant overflow
const1.go:42:20: error: integer constant overflow
const1.go:44:28: error: integer constant overflow
const1.go:44:28: error: integer constant overflow
const1.go:45:14: error: integer constant overflow
const1.go:49:14: error: integer constant overflow
const1.go:50:14: error: integer constant overflow
const1.go:51:14: error: integer constant overflow
const1.go:54:23: error: integer constant overflow
const1.go:54:23: error: integer constant overflow
const1.go:54:23: error: integer constant overflow
const1.go:54:23: error: integer constant overflow
const1.go:56:14: error: integer constant overflow
const1.go:57:24: error: integer constant overflow
const1.go:57:24: error: integer constant overflow
const1.go:58:24: error: integer constant overflow
const1.go:58:24: error: integer constant overflow
const1.go:59:22: error: integer constant overflow
const1.go:59:22: error: integer constant overflow
const1.go:61:24: error: integer constant overflow
const1.go:62:20: error: division by zero
const1.go:65:19: error: floating point constant overflow
const1.go:65:19: error: floating point constant overflow
const1.go:66:28: error: floating point constant overflow
const1.go:66:28: error: floating point constant overflow
const1.go:67:19: error: floating point constant overflow
const1.go:67:19: error: floating point constant overflow
const1.go:68:19: error: division by zero
const1.go:33:14: error: integer constant overflow
const1.go:35:19: error: integer constant overflow
const1.go:42:22: error: integer constant overflow
const1.go:53:17: error: integer constant overflow
const1.go:55:14: error: integer constant overflow
const1.go:59:24: error: integer constant overflow
const1.go:69:20: error: expected integer type
const1.go:75:4: error: argument 1 has incompatible type (cannot use type int8 as type int)
const1.go:76:4: error: argument 1 has incompatible type (cannot use type int8 as type int)
const1.go:77:4: error: argument 1 has incompatible type (cannot use type uint8 as type int)
const1.go:79:4: error: argument 1 has incompatible type (cannot use type float32 as type int)
const1.go:80:4: error: argument 1 has incompatible type (cannot use type float64 as type int)
const1.go:81:4: error: floating point constant truncated to integer
const1.go:83:4: error: argument 1 has incompatible type (cannot use type float64 as type int)
const1.go:84:4: error: argument 1 has incompatible type (cannot use type string as type int)
const1.go:85:4: error: argument 1 has incompatible type (cannot use type bool as type int)
const1.go:88:7: error: const initializer cannot be nil
const2.go:14:8: error: expected ‘=’
const5.go:27:7: error: expression is not constant
const5.go:28:7: error: expression is not constant
const5.go:30:7: error: expression is not constant
const5.go:31:7: error: expression is not constant
ddd1.go:57:23: error: invalid use of ‘...’ in type conversion
ddd1.go:59:6: error: invalid use of ‘...’ in type conversion
ddd1.go:60:12: error: use of ‘[...]’ outside of array literal
ddd1.go:21:15: error: argument 1 has incompatible type
ddd1.go:22:10: error: argument 1 has incompatible type
ddd1.go:30:6: error: invalid use of ‘...’ with non-slice
ddd1.go:30:6: error: invalid use of ‘...’ with non-slice
ddd1.go:46:2: error: invalid use of %<...%> with builtin function
ddd1.go:47:2: error: invalid use of %<...%> with builtin function
ddd1.go:49:2: error: invalid use of %<...%> with builtin function
ddd1.go:50:6: error: invalid use of %<...%> with builtin function
ddd1.go:51:6: error: invalid use of %<...%> with builtin function
ddd1.go:53:6: error: invalid use of %<...%> with builtin function
ddd1.go:58:13: error: invalid use of %<...%> with builtin function
ddd1.go:20:10: error: floating point constant truncated to integer
ddd1.go:32:6: error: invalid use of ‘...’ calling non-variadic function
declbad.go:20:3: error: variables redeclared but no variable is new
declbad.go:38:3: error: variables redeclared but no variable is new
declbad.go:44:3: error: variables redeclared but no variable is new
declbad.go:51:3: error: variables redeclared but no variable is new
declbad.go:57:3: error: variables redeclared but no variable is new
declbad.go:63:3: error: variables redeclared but no variable is new
declbad.go:26:3: error: incompatible types in assignment (cannot use type float32 as type int)
declbad.go:32:3: error: incompatible types in assignment (cannot use type int as type float32)
declbad.go:44:3: error: incompatible types in assignment (different number of results)
fixedbugs/bug223.go:21:5: error: initialization expression for ‘m’ depends upon itself
fixedbugs/bug412.go:10:2: error: duplicate field name ‘x’
fixedbugs/bug413.go:11:5: error: initialization expression for ‘i’ depends upon itself
fixedbugs/bug416.go:13:1: error: method ‘X’ redeclares struct field name
fixedbugs/bug435.go:15:49: error: missing ‘)’
fixedbugs/bug435.go:15:2: error: reference to undefined name ‘bar’
fixedbugs/bug451.go:9:9: error: expected package
typeswitch3.go:39:9: error: no new variables on left side of ‘:=’
typeswitch3.go:24:2: error: impossible type switch case (type has no methods)
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6560063
Today, if run.go doesn't understand a test header line it just ignores
the test, making it too easy to write or edit tests that are not actually
being run.
- expand errorcheck to accept flags, so that bounds.go and escape*.go can run.
- create a whitelist of skippable tests in run.go; skipping others is an error.
- mark all skipped tests at top of file.
Update #4139.
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6549054
Revision 63f7abcae015 introduced a bug caused by
code assuming registers started at X5, not X0.
Fixes#4138.
R=rsc
CC=golang-dev, remy
https://golang.org/cl/6558043
Broke tests on 386.
««« original CL description
6l/8l: emit correct opcodes to F(SUB|DIV)R?D.
When the destination was not F0, 6l and 8l swapped FSUBD/FSUBRD and
FDIVD/FDIVRD.
R=golang-dev, dave, rsc
CC=golang-dev
https://golang.org/cl/6498092
»»»
R=golang-dev
CC=golang-dev
https://golang.org/cl/6492100
This fixes a spurious 'invalid recursive type' error, and stops the compiler from emitting errors on uses of the invalid type.
Fixes#3766.
R=golang-dev, dave, minux.ma, rsc
CC=golang-dev
https://golang.org/cl/6443100
CVTSS2SQ's rounding mode is controlled by the RC field of MXCSR;
as we specifically need truncate semantic, we should use CVTTSS2SQ.
Fixes#3804.
R=rsc, r
CC=golang-dev
https://golang.org/cl/6352079
