Funcdepth was already int32. Make Escloopdepth
and Decldepth also int32 instead of int.
No functional changes for non-absurd code. Passes toolstash -cmp.
Change-Id: I47e145dd732b6a73cfcc6d45956df0dbccdcd999
Reviewed-on: https://go-review.googlesource.com/10129
Reviewed-by: Russ Cox <rsc@golang.org>
This reverts commit 5726af54eb.
It broke all the builds.
Change-Id: I4b1dde86f9433717d303c1dabd6aa1a2bf97fab2
Reviewed-on: https://go-review.googlesource.com/10143
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This CL was generated by updating Val in go.go
and then running:
sed -i "" 's/\.U\.[SBXFC]val = /.U = /' *.go
sed -i "" 's/\.U\.Sval/.U.\(string\)/g' *.go *.y
sed -i "" 's/\.U\.Bval/.U.\(bool\)/g' *.go *.y
sed -i "" 's/\.U\.Xval/.U.\(\*Mpint\)/g' *.go *.y
sed -i "" 's/\.U\.Fval/.U.\(\*Mpflt\)/g' *.go *.y
sed -i "" 's/\.U\.Cval/.U.\(\*Mpcplx\)/g' *.go *.y
No functional changes. Passes toolstash -cmp.
This reduces the size of gc.Node from 424 to 392 bytes.
This in turn reduces the permanent (pprof -inuse_space)
memory usage while compiling the test/rotate?.go tests:
test old(MB) new(MB) change
rotate0 379.49 364.78 -3.87%
rotate1 373.42 359.07 -3.84%
rotate2 381.17 366.24 -3.91%
rotate3 374.30 359.95 -3.83%
CL 8445 was similar to this; gri asked that Val's implementation
be hidden first. CLs 8912, 9263, and 9267 have at least
isolated the changes to the cmd/internal/gc package.
Updates #9933.
Change-Id: I83ddfe003d48e0a73c92e819edd3b5e620023084
Reviewed-on: https://go-review.googlesource.com/10059
Reviewed-by: Russ Cox <rsc@golang.org>
This lets us abstract away which arguments can be constants and so on
and lets the back ends reverse the order of arguments if that helps.
Change-Id: I283ec1d694f2dd84eba22e5eb4aad78a2d2d9eb0
Reviewed-on: https://go-review.googlesource.com/9810
Reviewed-by: David Chase <drchase@google.com>
This includes the following information in the per-function summary:
outK = paramJ encoded in outK bits for paramJ
outK = *paramJ encoded in outK bits for paramJ
heap = paramJ EscHeap
heap = *paramJ EscContentEscapes
Note that (currently) if the address of a parameter is taken and
returned, necessarily a heap allocation occurred to contain that
reference, and the heap can never refer to stack, therefore the
parameter and everything downstream from it escapes to the heap.
The per-function summary information now has a tuneable number of bits
(2 is probably noticeably better than 1, 3 is likely overkill, but it
is now easy to check and the -m debugging output includes information
that allows you to figure out if more would be better.)
A new test was added to check pointer flow through struct-typed and
*struct-typed parameters and returns; some of these are sensitive to
the number of summary bits, and ought to yield better results with a
more competent escape analysis algorithm. Another new test checks
(some) correctness with array parameters, results, and operations.
The old analysis inferred a piece of plan9 runtime was non-escaping by
counteracting overconservative analysis with buggy analysis; with the
bug fixed, the result was too conservative (and it's not easy to fix
in this framework) so the source code was tweaked to get the desired
result. A test was added against the discovered bug.
The escape analysis was further improved splitting the "level" into
3 parts, one tracking the conventional "level" and the other two
computing the highest-level-suffix-from-copy, which is used to
generally model the cancelling effect of indirection applied to
address-of.
With the improved escape analysis enabled, it was necessary to
modify one of the runtime tests because it now attempts to allocate
too much on the (small, fixed-size) G0 (system) stack and this
failed the test.
Compiling src/std after touching src/runtime/*.go with -m logging
turned on shows 420 fewer heap allocation sites (10538 vs 10968).
Profiling allocations in src/html/template with
for i in {1..5} ;
do go tool 6g -memprofile=mastx.${i}.prof -memprofilerate=1 *.go;
go tool pprof -alloc_objects -text mastx.${i}.prof ;
done
showed a 15% reduction in allocations performed by the compiler.
Update #3753
Update #4720Fixes#10466
Change-Id: I0fd97d5f5ac527b45f49e2218d158a6e89951432
Reviewed-on: https://go-review.googlesource.com/8202
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
To avoid confusion with the runtime concept of copying stack.
Change-Id: I33442377b71012c2482c2d0ddd561492c71e70d0
Reviewed-on: https://go-review.googlesource.com/8639
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Russ Cox <rsc@golang.org>
This cleanup is in anticipation of implementing
jump-free booleans (CL 2284) and zero-aware
comparisons (issue 10381).
No functional changes. Passes toolstash -cmp.
Change-Id: I50f394c60fa2927e177d7fc85b75085060a9e912
Reviewed-on: https://go-review.googlesource.com/8738
Reviewed-by: Russ Cox <rsc@golang.org>
Val is used to hold constant values.
Reg was the odd duck out.
Generated using eg.
No functional changes. Passes toolstash -cmp.
Change-Id: Ic1de769a1f92bb02e09a4428d998b716f307e2f6
Reviewed-on: https://go-review.googlesource.com/8912
Reviewed-by: Russ Cox <rsc@golang.org>
Fixes#9855
Use an architectural zero register as the source for zeroing, if available.
Change-Id: Ie5b4ba4e3d356c6f892bfd1cebd14d5152bdeeb0
Reviewed-on: https://go-review.googlesource.com/8722
Reviewed-by: Keith Randall <khr@golang.org>
No functional changes. Passes toolstash -cmp.
Change-Id: I4fba0c248645c3910ee3f7fc99dacafb676c5dc2
Reviewed-on: https://go-review.googlesource.com/8911
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Found with https://github.com/opennota/check.
Change-Id: I50c173382782fb16b15100e02c1c85610bc233a0
Reviewed-on: https://go-review.googlesource.com/7130
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Minux Ma <minux@golang.org>
The original implementation used 16 int "words" but only 29 bits per word
for a total of 16*29 = 464 bits, with a space consumption of 16*64 = 1024
bits on a 64 bit machine. Switching to 512 bits increases precision while
still using (in the worst case) half the amount of memory per mp value on
a 64 bit machine.
Also: Decreased permitted number of least-significant mantissa bits which
may be incorrect when considering if a precise floating-point constant is
an integer from 29 to 16 bits.
Change-Id: Iee9287056f0e9aa4f06ceac0724ff4674f710c53
Reviewed-on: https://go-review.googlesource.com/8429
Reviewed-by: Russ Cox <rsc@golang.org>
All multi-precision arithmetic is now based on math/big.
- passes all.bash
- added test cases for fixed bugs
Fixes#7740.
Fixes#6866.
Change-Id: I67268b91766970ced3b928260053ccdce8753d58
Reviewed-on: https://go-review.googlesource.com/7912
Reviewed-by: Russ Cox <rsc@golang.org>
This change cleans up some of the uglyness introduced in 8fc73a39ef
by moving the gc.Use_sse into the gc.Arch struct and adjusting its
zero value to be more useful.
Change-Id: I26ff5d9ac57b3f25e936519e443de6583cdafa56
Reviewed-on: https://go-review.googlesource.com/7994
Reviewed-by: Russ Cox <rsc@golang.org>
Adjust Thearch.FREG_MIN/MAX when using non sse2 mode in 8g.
Also, gc.Use_sse is treated as a bool, so make it a bool.
Change-Id: I840411605344bb31c32f492b3e6729166c084f0c
Reviewed-on: https://go-review.googlesource.com/7993
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>
Run-TryBot: Dave Cheney <dave@cheney.net>
These can be implemented with just a compare and a move instruction.
Do so, avoiding the overhead of a call into the runtime.
These assertions are a significant cost in Go code that uses interface{}
as a safe alternative to C's void* (or unsafe.Pointer), such as the
current version of the Go compiler.
*T here includes pointer to T but also any Go type represented as
a single pointer (chan, func, map). It does not include [1]*T or struct{*int}.
That requires more work in other parts of the compiler; there is a TODO.
Change-Id: I7ff681c20d2c3eb6ad11dd7b3a37b1f3dda23965
Reviewed-on: https://go-review.googlesource.com/7862
Reviewed-by: Rob Pike <r@golang.org>
This CL moves the bulk of the code that has been copy-and-pasted
since the initial 386 port back into a shared place, cutting 5 copies to 1.
The motivation here is not cleanup per se but instead to reduce the
cost of introducing changes in shared concepts like regalloc or general
expression evaluation. For example, a change after this one will
implement x.(*T) without a call into the runtime. This CL makes that
followup work 5x easier.
The single copy still has more special cases for architecture details
than I'd like, but having them called out explicitly like this at least
opens the door to generalizing the conditions and smoothing out
the distinctions in the future.
This is a LARGE CL. I started by trying to pull in one function at a time
in a sequence of CLs and it became clear that everything was so
interrelated that it had to be moved as a whole. Apologies for the size.
It is not clear how many more releases this code will matter for;
eventually it will be replaced by Keith's SSA work. But as noted above,
the deduplication was necessary to reduce the cost of working on
the current code while we have it.
Passes tests on amd64, 386, arm, and ppc64le.
Can build arm64 binaries but not tested there.
Being able to build binaries means it is probably very close.
Change-Id: I735977f04c0614f80215fb12966dfe9bbd1f5861
Reviewed-on: https://go-review.googlesource.com/7853
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
- renamed (existing) Mpint -> Mpfix
- defined (new) Mpint using big.Int
- modified funcs mpxxx operating on new Mpint
- renamed funcs mpxxx -> _mpxxx if still needed with Mpfix
- left old (possibly unused) code in place for comparison
Passes all.bash.
Change-Id: I1fc7bba7dc4b6386f2f0950d745cec17c1e67615
cmd/internal/gc: renamed Mpint -> Mpfix
Change-Id: Ia06aeae1081ef29d5ad9b711fb57e4c5579ce29b
Reviewed-on: https://go-review.googlesource.com/7830
Reviewed-by: Russ Cox <rsc@golang.org>
Also clean up code a little.
Change-Id: I23b7d2b7871b31e0974f1305e54f0c18dcab05d9
Reviewed-on: https://go-review.googlesource.com/7746
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Rob Pike <r@golang.org>
The ProgInfo is loaded many times during each analysis pass.
Load it once at the beginning (in Flowstart if using that, or explicitly,
as in plive.go) and then refer to the cached copy.
Removes many calls to proginfo.
Makes Prog a little bigger, but the previous CL more than compensates.
Change-Id: If90a12fc6729878fdae10444f9c3bedc8d85026e
Reviewed-on: https://go-review.googlesource.com/7745
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Change-Id: I3096a7497955bc475739739ee23be387e9162867
Reviewed-on: https://go-review.googlesource.com/7210
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
namebuf was a global char buffer in the C version of gc, which was
useful for providing common storage for constructing symbol and file
names. However, now that it's just a global Go string and the string
data is dynamically allocated anyway, it doesn't serve any purpose
except to force extra write barriers everytime it's assigned to.
Also, introduce Lookupf(fmt, args...) as shorthand for
Lookup(fmt.Sprintf(fmt, args...)), which was a very common pattern for
using namebuf.
Passes "go build -toolexec 'toolstash -cmp' -a std".
Notably, this CL shrinks 6g's text section by ~15kB:
$ size toolstash/6g tool/linux_amd64/6g
text data bss dec hex filename
4600805 605968 342988 5549761 54aec1 toolstash/6g
4585547 605968 342956 5534471 547307 tool/linux_amd64/6g
Change-Id: I98abb44fc7f43a2e2e48425cc9f215cd0be37442
Reviewed-on: https://go-review.googlesource.com/7080
Reviewed-by: Ian Lance Taylor <iant@golang.org>
More cleanups to gc.Node
- make Node.Local a boolean
- make Type.Local a boolean
- reduce the size of Node.Esc to a uint8
Reducing the size of Node.Esc shaves ~45mb off the RSS compiling cmd/internal/gc on amd64
before:
Maximum resident set size (kbytes): 659496
after:
Maximum resident set size (kbytes): 612196
- declare gc.Funcdepth as int32
- declare Node.Funcdepth as int32
In both cases, these were previously machine specific int types. This doesn't result in
any memory saving at the moment due to struct padding.
Change-Id: Iabef8da15e962fe8b79d7fd3d402fb26ce7ec31c
Reviewed-on: https://go-review.googlesource.com/7261
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Dave Cheney <dave@cheney.net>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Convert Node.Isddd to a boolean and simplify usage.
- Node.Isddd converted to bool
- Type.Isddd converted to bool
- mkinlcall converted to take isddd as a bool
- typecheckaste converted to take isddd as a bool
- ascompatte converted to take isddd as a bool
Change-Id: I52586145619c44182bb0c2c5d80a0a3fe3e50a07
Reviewed-on: https://go-review.googlesource.com/7172
Reviewed-by: Keith Randall <khr@golang.org>
It appears that c2go dropped comments inside struct { ... } and enum { ... }.
Restore them.
Identified missing comments by checking for comments present
in the C code but not the Go code, made a list, and then reapplied
with some mechanical help.
Missing comment finder: http://play.golang.org/p/g6qNUAo1Y0
Change-Id: I323ab45c7ef9d51e28eab3b699eb14bee1eef66b
Reviewed-on: https://go-review.googlesource.com/6899
Reviewed-by: Rob Pike <r@golang.org>
Minor comments added. More to come.
Change-Id: I97511db54d59e1009ef934da38f306a2dc83a6e9
Reviewed-on: https://go-review.googlesource.com/6898
Reviewed-by: Rob Pike <r@golang.org>
Remove the per-achitecture formatter for Prog and replace it with
a global String method. Clean up and regularize the output. Update
tests affected by the format; some tests are made correct now when
they were broken before (and known to be).
Also, related: Change the encoding of the (R1+R2) syntax on ppc64
to be equivalent to (R1)(R2*1), which means it needs no special
handling.
Delete the now unused STRINGSZ constant.
Change-Id: I7f6654d11f80065f3914a3f19353f2f12edfe310
Reviewed-on: https://go-review.googlesource.com/6931
Reviewed-by: Russ Cox <rsc@golang.org>
Node.Needzero only has two values and acts as a bool, so make it a bool.
Change-Id: Ica46e5ebafbe478017ea52ce6bb335f404059677
Reviewed-on: https://go-review.googlesource.com/6800
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Dave Cheney <dave@cheney.net>
TryBot-Result: Gobot Gobot <gobot@golang.org>
- use Bvec, not *Bvec, and bulk allocate backing store
- use range loops
- put Bvecs in BasicBlock struct instead of indexing into parallel slices
Change-Id: I5cb30f50dccb4d38cc18fae422f7f132c52876be
Reviewed-on: https://go-review.googlesource.com/6602
Reviewed-by: Rob Pike <r@golang.org>
The C version of the compiler had just one hash table,
indexed by a (name string, pkg *Pkg) pair.
Because we always know the pkg during a lookup,
replace the one table with a per-Pkg map[string]*Sym.
This also lets us do non-allocating []byte key lookups.
This CL *does* change the generated object files.
In the old code, export data and init calls were emitted
in "hash table order". Now they are emitted in the order
in which they were added to the table.
Change-Id: I5a48d5c9add996dc43ad04a905641d901522de0b
Reviewed-on: https://go-review.googlesource.com/6600
Reviewed-by: Rob Pike <r@golang.org>
Strlit was just a poor excuse for a Go string.
Use a Go string.
In the one case where it was a string-or-nil (Type.Note), use a *string.
Zconv was a poor excuse for %q. Use %q.
The only important part about Zconv's implementation
was that the compiler and linker agreed on the quoting rules.
Now they both use %q instead of having two Zconvs.
This CL *does* change the generated object files, because the
quoted strings end up in symbol names.
For example the string "\r\n" used to be named go.string."\r\n"
and is now go.string."\x0d\n".
Change-Id: I5c0d38e1570ffc495f0db1a20273c9564104a7e8
Reviewed-on: https://go-review.googlesource.com/6519
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
This avoids the argument appearing to escape
(due to the fact that proginfo is always called
via a function pointer).
Change-Id: Ib9351ba18c80fd89e6a1d4f19dea386d4c657337
Reviewed-on: https://go-review.googlesource.com/6518
Reviewed-by: Rob Pike <r@golang.org>
No functional changes.
This diff was generated as follows:
* Manually edit cmd/internal/gc/go.go to update types and group variables.
* Manually edit initialization in cmd/internal/gc/align.go--localized s/1/true.
* Manually fix the handling of sign in cmd/internal/gc/walk.go in func bounded (near line 4000).
* Manually update go.y and regenerate y.go.
* Run gofmt -r many times to do the rest, using https://gist.github.com/josharian/0f61dbb2dff81f938e70.
toolstash -cmp on the stdlib comes back green.
Change-Id: I19766ed551714e51b325133e7138818d117b3a9a
Reviewed-on: https://go-review.googlesource.com/6530
Reviewed-by: Russ Cox <rsc@golang.org>
Inlining refuses to inline bodies containing an actual function call, so that
if that call or a child uses runtime.Caller it cannot observe
the inlining.
However, inlining was also refusing to inline bodies that contained
function calls that were themselves inlined away. For example:
func f() int {
return f1()
}
func f1() int {
return f2()
}
func f2() int {
return 2
}
The f2 call in f1 would be inlined, but the f1 call in f would not,
because f1's call to f2 blocked the inlining, despite itself eventually
being inlined away.
Account properly for this kind of transitive inlining and enable.
Also bump the inlining budget a bit, so that the runtime's
heapBits.next is inlined.
This reduces the time for '6g *.go' in html/template by around 12% (!).
(For what it's worth, closing Chrome reduces the time by about 17%.)
Change-Id: If1aa673bf3e583082dcfb5f223e67355c984bfc1
Reviewed-on: https://go-review.googlesource.com/5952
Reviewed-by: Austin Clements <austin@google.com>
The dummy implementation was causing lots of argument lists
to be prepared and thrown away.
Change-Id: Id0040dec6b0937f3daa8a8d8911fa3280123e863
Reviewed-on: https://go-review.googlesource.com/5700
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reconvert using rsc.io/c2go rev 27b3f59.
(Same as last conversion, but C sources have changed
due to merging master into this branch.)
Change-Id: Ib314bb9ac14a726ceb83e2ecf4d1ad2d0b331c38
Reviewed-on: https://go-review.googlesource.com/5471
Reviewed-by: Rob Pike <r@golang.org>
Reconvert using rsc.io/c2go rev 27b3f59.
Changes to converter:
- fatal does not return, so no fallthrough after fatal in switch
- many more function results and variables identified as bool
- simplification of negated boolean expressions
Change-Id: I3bc67da5e46cb7ee613e230cf7e9533036cc870b
Reviewed-on: https://go-review.googlesource.com/5171
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
First draft of converted Go compiler, using rsc.io/c2go rev 83d795a.
Change-Id: I29f4c7010de07d2ff1947bbca9865879d83c32c3
Reviewed-on: https://go-review.googlesource.com/4851
Reviewed-by: Rob Pike <r@golang.org>
