Makes the error message more consistent between OAS and OAS2.
Fixes#26616.
Change-Id: I07ab46c5ef8a37efb2cb557632697f5d1bf789f7
Reviewed-on: https://go-review.googlesource.com/131280
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Fence-post implications of the form "x-1 >= w && x > min ⇒ x > w"
were not correctly handling unsigned domain, by always checking signed
limits.
This bug was uncovered once we taught prove that len(x) is always
>= 0 in the signed domain.
In the code being miscompiled (s[len(s)-1]), prove checks
whether len(s)-1 >= len(s) in the unsigned domain; if it proves
that this is always false, it can remove the bound check.
Notice that len(s)-1 >= len(s) can be true for len(s) = 0 because
of the wrap-around, so this is something prove should not be
able to deduce.
But because of the bug, the gate condition for the fence-post
implication was len(s) > MinInt64 instead of len(s) > 0; that
condition would be good in the signed domain but not in the
unsigned domain. And since in CL105635 we taught prove that
len(s) >= 0, the condition incorrectly triggered
(len(s) >= 0 > MinInt64) and things were going downfall.
Fixes#27251Fixes#27289
Change-Id: I3dbcb1955ac5a66a0dcbee500f41e8d219409be5
Reviewed-on: https://go-review.googlesource.com/132495
Reviewed-by: Keith Randall <khr@golang.org>
Nil check is special in that it has no use but we must keep it.
Count it as a use of the auto.
Fixes#27278.
Change-Id: I857c3d0db2ebdca1bc342b4993c0dac5c01e067f
Reviewed-on: https://go-review.googlesource.com/131955
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
In the compiler frontend, walkinrange indiscriminately calls Int64()
on const CTINT nodes, even though Int64's return value is undefined
for anything over 2⁶³ (in practise, it'll return a negative number).
This causes the introduction of bad constants during rewrites of
unsigned expressions, which make the compiler reject valid Go
programs.
This change introduces a preliminary check that Int64() is safe to
call on the consts on hand. If it isn't, walkinrange exits without
doing any rewrite.
Fixes#27143
Change-Id: I2017073cae65468a521ff3262d4ea8ab0d7098d9
Reviewed-on: https://go-review.googlesource.com/130735
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Consolidate decision about whether -race and -msan options are
supported in cmd/internal/sys. Use consolidated functions in
cmd/compile and cmd/go. Use a copy of them in cmd/dist; cmd/dist can't
import cmd/internal/sys because Go 1.4 doesn't have it.
Fixes#24315
Change-Id: I9cecaed4895eb1a2a49379b4848db40de66d32a9
Reviewed-on: https://go-review.googlesource.com/121816
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Gccgo produced incorrect order of evaluation for expressions
involving &&, || subexpressions. The fix is CL 125299.
Updates #26495.
Change-Id: I18d873281709f3160b3e09f0b2e46f5c120e1cab
Reviewed-on: https://go-review.googlesource.com/125301
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Code fix was in CL 122556. This is a corresponding test case.
Fixes#26426
Change-Id: Ib8769f367aed8bead029da0a8d2ddccee1d1dccb
Reviewed-on: https://go-review.googlesource.com/124535
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
If one tries to use promoted fields in a struct literal, the compiler
errors correctly. However, if the embedded fields are of struct pointer
type, the field.Type.Sym.Name expression below panics.
This is because field.Type.Sym is nil in that case. We can simply use
field.Sym.Name in this piece of code though, as it only concerns
embedded fields, in which case what we are after is the field name.
Added a test mirroring fixedbugs/issue23609.go, but with pointer types.
Fixes#26416.
Change-Id: Ia46ce62995c9e1653f315accb99d592aff2f285e
Reviewed-on: https://go-review.googlesource.com/124395
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
The arm64 backend generates "TST" for "if uint32(a)&uint32(b) == 0",
which should be "TSTW".
fixes#26438
Change-Id: I7d64c30e3a840b43486bcd10eea2e3e75aaa4857
Reviewed-on: https://go-review.googlesource.com/124637
Run-TryBot: Ben Shi <powerman1st@163.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Autos must be kept if their address reaches the control value of a
block. We didn't see this before because it is rare for an auto's
address to reach a control value without also reaching a phi or
being written to memory. We can probably optimize away the
comparisons that lead to this scenario since autos cannot alias
with pointers from elsewhere, however for now we take the
conservative approach and just ensure the auto is properly
initialised if its address reaches a control value.
Fixes#26407.
Change-Id: I02265793f010a9e001c3e1a5397c290c6769d4de
Reviewed-on: https://go-review.googlesource.com/124335
Reviewed-by: David Chase <drchase@google.com>
If both branches of a write barrier test go to the same block,
then there's no unsafe points.
This can only happen if the resulting memory state is somehow dead,
which can only occur in degenerate cases, like infinite loops. No
point in cleaning up the useless branch in these situations.
Fixes#26024.
Change-Id: I93a7df9fdf2fc94c6c4b1fe61180dc4fd4a0871f
Reviewed-on: https://go-review.googlesource.com/123655
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Lack of a well-defined order between VarDef and related
address operations sometimes causes problems with store order
and write barrier transformations; glitches in the order are
made irreparable (by later optimizations) if the two parts of
the glitch straddle a split in the original block caused by
insertion of a write barrier diamond.
Fix this by creating a LocalAddr for addresses of locals
(what VarDef matters for) that takes a memory input to
help make the order explicit. Addr is modified to only
be legal for SB operand, so there is no overlap between
Addr and LocalAddr uses (there may be some downstream
cleanup from this).
Changes to generic.rules and rewrite.go ensure that codegen
tests continue to pass; CSE of LocalAddr is impaired, not
quite sure of the cost.
Fixes#26105.
Change-Id: Id4192b4440aa4e9d7ba54a465c456df9b530b515
Reviewed-on: https://go-review.googlesource.com/122483
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
For golang.org/cl/74110, I forgot that you can use range-based for
loops to extract key values from a map value.
This wasn't a problem for the binary format importer, because it was
more tolerant about missing inline function bodies. However, the
indexed importer is more particular about this.
We could potentially just make it more lenient like the binary
importer, but tweaking the logic here is easy enough and seems like
the preferable solution.
Fixes#26341.
Change-Id: I54564dcd0be60ea393f8a0f6954b7d3d61e96ee5
Reviewed-on: https://go-review.googlesource.com/123475
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Fix the panic message produced for an interface conversion error to
only say "types from different packages" if they are definitely from
different packges. If they may be from the same package, say "types
from different scopes."
Updates #18911Fixes#26094
Change-Id: I0cea50ba31007d88e70c067b4680009ede69bab9
Reviewed-on: https://go-review.googlesource.com/123395
Reviewed-by: Austin Clements <austin@google.com>
Functions exported on behalf of other packages need to have their
argument stack maps specified explicitly. They don't get an implicit
map because they are not in the local package, and if they get defer'd
they need argument maps.
Fixes#24419
Change-Id: I35b7d8b4a03d4770ba88699e1007cb3fcb5397a9
Reviewed-on: https://go-review.googlesource.com/122676
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
When DWARF is disabled, some alg functions were not generated.
Make sure they are generated when we about to generate calls to
them.
Fixes#23546.
Change-Id: Iecfa0eea830e42ee92e55268167cefb1540980b2
Reviewed-on: https://go-review.googlesource.com/122403
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
We need to make sure that the terminating comparison has the right
sense given the increment direction. If the increment is positive,
the terminating comparsion must be < or <=. If the increment is
negative, the terminating comparison must be > or >=.
Do a few cleanups, like constant-folding entry==0, adding comments,
removing unused "exported" fields.
Fixes#26116
Change-Id: I14230ee8126054b750e2a1f2b18eb8f09873dbd5
Reviewed-on: https://go-review.googlesource.com/121940
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Expanding interface method sets is handled during width calculation,
which can't be performed concurrently. Make sure that we eagerly
expand interfaces in the frontend when importing them, even if they're
not actually used by code, because we might need to generate a type
description of them.
Fixes#25055.
Change-Id: I6fd2756de2c7d5dbc33056f70b3028ca3aebab41
Reviewed-on: https://go-review.googlesource.com/122517
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Late opt pass may generate dead stores, which messes up store
chain calculation in later passes. Run generic deadcode even
in -N mode to remove them.
Fixes#26163.
Change-Id: I8276101717bb978d5980e6c7998f53fd8d0ae10f
Reviewed-on: https://go-review.googlesource.com/121856
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
If the address of an auto reaches a phi then any further stores to
the pointer represented by the phi probably need to be kept. This
is because stores to the other arguments to the phi may be visible
to the program.
Fixes#26153.
Change-Id: Ic506c6c543bf70d792e5b1a64bdde1e5fdf1126a
Reviewed-on: https://go-review.googlesource.com/121796
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
This reverts commit 1a27f048ad.
Reason for revert: Broke the ssacheck and -N-l builders, and the -N-l fix looks like it will take some time and take a different route entirely.
Change-Id: Ie0ac5e86ab7d72a303dfbbc48dfdf1e092d4f61a
Reviewed-on: https://go-review.googlesource.com/121715
Reviewed-by: David Chase <drchase@google.com>
Given a carefully constructed input, writebarrier would
split a block with the OpAddr in the first half and the
VarDef in the second half which ultimately leads to a
compiler crash because the scheduler is no longer able
to put them in the proper order.
To fix, recognize the implicit dependence of OpAddr on
the VarDef of the same symbol if any exists.
This fix was chosen over making OpAddr take a memory
operand to make the dependence explicit, because this
change is less invasive at this late part of the 1.11
release cycle.
Fixes#26105.
Change-Id: I9b65460673af3af41740ef877d2fca91acd336bc
Reviewed-on: https://go-review.googlesource.com/121436
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
SSA can handle 1-element array, but only when the element type
is SSAable. When building SSA for INDEX of 1-element array, we
did not check the element type is SSAable. And when it's not,
it resulted in an unhandled SSA op.
Fixes#26120.
Change-Id: Id709996b5d9d90212f6c56d3f27eed320a4d8360
Reviewed-on: https://go-review.googlesource.com/121496
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Code generation for OpAMD64CMOV[WLQ]EQF uses AX as a scratch register,
but only CMOVQEQF, correctly lets compiler know. Mark other 2 as
clobbering AX.
Fixes#26097
Change-Id: I2a65bd67bf18a540898b4a0ae6c8766e0b767b19
Reviewed-on: https://go-review.googlesource.com/121336
Run-TryBot: Ilya Tocar <ilya.tocar@intel.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Giovanni Bajo <rasky@develer.com>
On Wasm, the offset was not folded into LoweredAddr, so it was
not rematerializeable. This led to the address-taken operation
in some cases generated too early, before the local variable
becoming live. The liveness code thinks the variable live when
the address is taken, then backs it up to live at function
entry, then complains about it, because nothing other than
arguments should be live on entry.
This CL folds the offset into the address operation, so it is
rematerializeable and so generated right before use, after the
variable actually becomes live.
It might be possible to relax the liveness code not to think a
variable live when its address being taken, but until the address
actually being used. But it would be quite complicated. As we're
late in Go 1.11 freeze, it would be better not to do it. Also,
I think the address operation is rematerializeable now on all
architectures, so this is probably less necessary.
This may also be a slight optimization, as the address+offset is
now rematerializeable, which can be generated on the Wasm stack,
without using any "registers" which are emulated by local
variables on Wasm. I don't know how to do benchmarks on Wasm. At
least, cmd/go binary size shrinks 9K.
Fixes#25966.
Change-Id: I01e5869515d6a3942fccdcb857f924a866876e57
Reviewed-on: https://go-review.googlesource.com/120599
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Richard Musiol <neelance@gmail.com>
For non-unit increment, loopbce checks to see if the
increment evenly divides the difference between (constant)
loop start and end. This test panics when the increment
is zero.
Fix: check for zero, if found, don't optimize the loop.
Also added missing copyright notice to loopbce.go.
Fixes#26043.
Change-Id: I5f460104879cacc94481949234c9ce8c519d6380
Reviewed-on: https://go-review.googlesource.com/120759
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
If expanding an inline function body required lazily expanding a
package-scoped type whose identifier was shadowed within the function
body, the lazy expansion would instead overwrite the local symbol
definition instead of the package-scoped symbol. This was due to
importsym using s.Def instead of s.PkgDef.
Unfortunately, this is yet another consequence of the current awkward
scope handling code.
Passes toolstash-check.
Fixes#25984.
Change-Id: Ia7033e1749a883e6e979c854d4b12b0b28083dd8
Reviewed-on: https://go-review.googlesource.com/120456
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
MOVWconst's AuxInt is Int32. SSA check complains if the AuxInt
does not fit in int32. Convert uint32 to int32 to make it happy.
The generated code is unchanged. MOVW only cares low 32 bits.
Passes "toolstash -cmp" std cmd for ARM.
Fixes#25993.
Change-Id: I2b6532c9c285ea6d89652505fb7c553f85a98864
Reviewed-on: https://go-review.googlesource.com/120335
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Adds the appropriate check to inl.go.
Includes tests of both -race+go:norace and plain go:norace.
Fixes#24651.
Change-Id: Id806342430c20baf4679a985d12eea3b677092e0
Reviewed-on: https://go-review.googlesource.com/119195
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Inlining of switch statements into a RETURNed expression
can sometimes lead to the switch being walked twice, which
results in a miscompiled switch statement. The bug depends
on:
1) multiple results
2) named results
3) a return statement whose expression includes a call to a
function containing a switch statement that is inlined.
It may also be significant that the default case of that
switch is a panic(), though that's not proven.
Rearranged the walk case for ORETURN so that double walks are
not possible. Added a test, because this is so fiddly.
Added a check against double walks, verified that it fires
w/o other fix.
Fixes#25776.
Change-Id: I2d594351fa082632512ef989af67eb887059729b
Reviewed-on: https://go-review.googlesource.com/118318
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Ensure that compiler error suggestions after case insensitive
field lookups don't mistakenly reported unexported fields if
those fields aren't in the local package being processed.
Fixes#25727
Change-Id: Icae84388c2a82c8cb539f3d43ad348f50a644caa
Reviewed-on: https://go-review.googlesource.com/117755
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The original fix (https://go-review.googlesource.com/c/go/+/35831)
for this issue was incorrect as it reported cycles in cases where
it shouldn't.
Instead, use a different approach: A type cycle containing aliases
is only a cycle if there are no type definitions. As soon as there
is a type definition, alias expansion terminates and there is no
cycle.
Approach: Split sprint_depchain into two non-recursive and more
easily understandable functions (cycleFor and cycleTrace),
and use those instead for cycle reporting. Analyze the cycle
returned by cycleFor before issueing an alias cycle error.
Also: Removed original fix (main.go) which introduced a separate
crash (#23823).
Fixes#18640.
Fixes#23823.
Fixes#24939.
Change-Id: Ic3707a9dec40a71dc928a3e49b4868c5fac3d3b7
Reviewed-on: https://go-review.googlesource.com/118078
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
The wasm archtecture was missing a rule to handle OffPtr with a
negative offset. This commit makes it so OffPtr always gets lowered
to I64AddConst.
Fixes#25741
Change-Id: I1d48e2954e3ff31deb8cba9a9bf0cab7c4bab71a
Reviewed-on: https://go-review.googlesource.com/116595
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
In the old binary export format, parameter names for parameter lists
which contained only types where never written, so this problem didn't
come up.
Fixes#25101.
Change-Id: Ia8b817f7f467570b05f88d584e86b6ef4acdccc6
Reviewed-on: https://go-review.googlesource.com/116376
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
The stack frame includes the callee args section. At the point where
we were checking the frame size, that part of the frame had not been
computed yet. Move the check later so we can include the callee args size.
Fixes#20780
Update #25507
Change-Id: Iab97cb89b3a24f8ca19b9123ef2a111d6850c3fe
Reviewed-on: https://go-review.googlesource.com/115195
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Before the CL 115277 we did not run the test on Windows,
so let's just go back to not running the test on Windows.
There is nothing OS-specific about this test,
so skipping it on Windows doesn't seem like a big deal.
Updates #25693Fixes#25586
Change-Id: I1eb3e158b322d73e271ef388f8c6e2f2af0a0729
Reviewed-on: https://go-review.googlesource.com/115857
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This CL removes the rundircmpout action completely
because it is not used anywhere.
The run case already looks for output files. Rename the cmpout action
mentioned in tests to the run action and remove "cmpout" from run.go.
Change-Id: I835ceb70082927f8e9360e0ea0ba74f296363ab3
Reviewed-on: https://go-review.googlesource.com/115575
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
To allow testing of fixedbugs/bug345.go in Go,
a new flag -n is introduced. This flag disables setting
of relative path for local imports and imports search path
to current dir, namely -D . -I . are not passed to the compiler.
Error regexps are fixed to allow running the test in temp directory.
This change eliminates the last place where Perl
script "errchk" was used.
Fixes#25586.
Change-Id: If085f466e6955312d77315f96d3ef1cb68495aef
Reviewed-on: https://go-review.googlesource.com/115277
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
When we deadcode-remove a block which is a write barrier test,
remove that block from the list of write barrier test blocks.
Fixes#25516
Change-Id: I1efe732d5476003eab4ad6bf67d0340d7874ff0c
Reviewed-on: https://go-review.googlesource.com/115037
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Extend stack frame limit of 1GB to include large argument/return areas.
Argument/return areas are part of the parent frame, not the frame itself,
so they need to be handled separately.
Fixes#25507.
Change-Id: I309298a58faee3e7c1dac80bd2f1166c82460087
Reviewed-on: https://go-review.googlesource.com/115036
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This change enables bug248 to be tested with Go code.
For that, it adds a flag -1 to error check and run directory
with one package failing compilation prior the last package
which should be run.
Specifically, the "p" package in bug1.go file was renamed into "q"
to compile them in separate steps,
bug2.go and bug3.go files were reordered,
bug2.go was changed into non-main package.
Updates #25586.
Change-Id: Ie47aacd56ebb2ce4eac66c792d1a53e1e30e637c
Reviewed-on: https://go-review.googlesource.com/114818
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
