Per suggestion in CL 202578, this CL drops the purego build tag used
within this package.
Change-Id: I33626c73d6602e321528544ee601741f7e634c1b
Reviewed-on: https://go-review.googlesource.com/c/go/+/230677
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
Adds a few instructions to ppc64enc.s that were missing from the
previous update.
Change-Id: Ieafce39e905cdf4da3bfb00fdd5a39ab28089cb3
Reviewed-on: https://go-review.googlesource.com/c/go/+/230437
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This updates the PPC64.rules file to use the MOD instructions
that are available in power9. Prior to power9 this is done
using a longer sequence with multiply and divide.
Included in this change is removal of the REM* opcode variations
that set the CC or OV bits since their settings are based
on the DIV and are not appropriate for the REM.
Change-Id: Iceed9ce33e128e1911c15592ee674276ce8ba3fa
Reviewed-on: https://go-review.googlesource.com/c/go/+/229761
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Prior to this commit, NewXxx could panic when passed an image.Rectangle
with one of width or height being negative. But it might not panic if
both were negative, because (bpp * w * h) could still be positive. After
this commit, it will panic if both are negative.
With overflow, NewXxx might not have panicked if (bpp * w * h), the
length passed to "make([]uint8, length)", was still non-negative (after
truncation), but even if w and h were valid (non-negative), the overall
byte slice wasn't long enough. Iterating over the pixels would possibly
panic later with index out of bounds. This change moves the panic
earlier, closer to where the mistake is.
Change-Id: I011feb2d53515fc3f0fe72bb6c23b3953772c577
Reviewed-on: https://go-review.googlesource.com/c/go/+/230220
Reviewed-by: Rob Pike <r@golang.org>
Convert some optimizations rules to strongly-typed versions. Similar to
CL 230338, this CL only converts rules that need no additional changes
(i.e: only need to change '->' to '=>').
This CL covers the rules from line 800 - 1219.
Passes toolstash-check
Change-Id: I94181a809fa38918b78301f1c0c680b7a8ab552f
Reviewed-on: https://go-review.googlesource.com/c/go/+/230738
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Previously, looking up only IPv4 or IPv6 addresses was only possible
with DefaultResolver via ResolveIPAddr. Add this functionality to the
Resolver type with a new method, LookupIP. This largely brings Resolver
functionally to parity with the global functions. The name LookupIP is
used over ResolveIPAddr to be consistent with the other Resolver
methods.
There are two main benefits to (*Resolver).LookupIP over
(*Resolver).LookupHost. First is an ergonomic benefit. Wanting a
specific family of address is common enough to justify a method, evident
by the existence of ResolveIPAddr. Second, this opens the possibility of
not performing unnecessary DNS requests when only a specific family of
addresses are needed. This optimization is left to follow up work.
Updates #30452
Change-Id: I241f61019588022a39738f8920b0ddba900cecdd
Reviewed-on: https://go-review.googlesource.com/c/go/+/228641
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Allocate the runcases slice on the stack if the number
of select cases is small (up to 4).
Found while looking at production profiles of common
proto based RPC server framework code in Google which do
not have a large number of cases.
name old time/op new time/op delta
Select/1 147ns ± 2% 120ns ± 6% -18.32% (p=0.000 n=7+10)
Select/4 316ns ± 5% 249ns ± 2% -21.23% (p=0.000 n=10+10)
Select/8 516ns ± 3% 515ns ± 3% ~ (p=0.858 n=10+9)
name old alloc/op new alloc/op delta
Select/1 96.0B ± 0% 64.0B ± 0% -33.33% (p=0.000 n=10+10)
Select/4 336B ± 0% 208B ± 0% -38.10% (p=0.000 n=10+10)
Select/8 672B ± 0% 672B ± 0% ~ (all equal)
name old allocs/op new allocs/op delta
Select/1 4.00 ± 0% 3.00 ± 0% -25.00% (p=0.000 n=10+10)
Select/4 7.00 ± 0% 6.00 ± 0% -14.29% (p=0.000 n=10+10)
Select/8 11.0 ± 0% 11.0 ± 0% ~ (all equal)
Change-Id: I1687e74fc8e86606a27f03fa8a561bcfb68775d6
Reviewed-on: https://go-review.googlesource.com/c/go/+/230657
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Opening a connection with Connect should still create a derived
context with a timeout because some clients will not use a timeout
and the connection pool may open a connection asynchronously.
Likewise, if a connection close makes a network operation it should
provide some type of sane timeout for the operation.
Fixes#38185
Change-Id: I9b7ce2996c81c486170dcc84b12672a99610fa27
Reviewed-on: https://go-review.googlesource.com/c/go/+/230438
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
CL 220499 started marking readonly syms as SRODATA earlier,
so we can use that in the writebarrier pass now.
Passes toolstash-check.
Change-Id: Ic4d49714b8bffbe03c8e9a75ca96df4475bae732
Reviewed-on: https://go-review.googlesource.com/c/go/+/230559
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
The span set data structure may leak blocks due to a race in the logic
to check whether it's safe to free a block. The simplest example of this
race is between two poppers:
1. Popper A claims slot spanSetEntries-2.
2. Popper B claims slot spanSetEntries-1.
3. Popper A gets descheduled before it subtracts from block.used.
4. Popper B subtracts from block.used, sees that claimed
spanSetEntries-1, but also that block.used != 0, so it returns.
5. Popper A comes back and subtracts from block.used, but it didn't
claim spanSetEntries-1 so it also returns.
The spine is left with a stale block pointer and the block later gets
overwritten by pushes, never to be re-used again.
The problem here is that we designate the claimer of slot
spanSetEntries-1 to be the one who frees the block, but that may not be
the thread that actually does the last subtraction from block.used.
Fixing this problem is tricky, and the fundamental problem there is that
block.used is not stable: it may be observed to be zero, but that
doesn't necessarily mean you're the last popper!
Do something simpler: keep a counter of how many pops have happened to a
given block instead of block.used. This counter monotonically increases
when a pop is _completely done_. Because this counter is monotonically
increasing, and only increases when a popper is done, then we know for
sure whichever popper is the last to increase it (i.e. its value is
spanSetBlockEntries) is also the last popper in the block. Because the
race described above still exists, the last popper may not be the one
which claimed the last slot in the block, but we know for certain nobody
else is popping from that block anymore so we can safely free it.
Finally, because pops serialize with pushes to the same slot, we need
not worry about concurrent pushers at all.
Updates #37487.
Change-Id: I6697219372774c8ca7d8ee6895eaa230a64ce9e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/230497
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Currently mcaches are flushed to mcentral after a bunch of memstats have
already been read. This is not safe (in the sense that it doesn't ensure
consisent memstats) since memstats may in general change when mcentral
data structures are manipulated.
Note that prior to the new mcentral implementation this was not a
problem because mcentral operations happened to never modify certain
memstats. As of the new mcentral implementation, we might for example
persistentalloc when uncaching a span, which would change memstats. This
can cause a skew between the value of sys (which currently is calculated
before mcaches are flushed) and the value of gc_sys and other_sys.
Fix this by moving mcache flushing to the very top of updatememstats.
Also leave a comment explaining that this must be done first, in
general, because mcentrals make no guarantee that they will not
influence memstats (and doing so would be unnecessarily restrictive).
Fixes#38712.
Change-Id: I15bacb313c54a46e380a945a71bb75db67169c1b
Reviewed-on: https://go-review.googlesource.com/c/go/+/230498
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
This is a second attempt at CL 230024, with
cmd/go/testdata/script/mod_retention.txt updated to perform a
version-independent comparison on the 'go' version added to a go.mod
file that lacks one.
Fixes#38708
Change-Id: I15dcd83b51ed5ec57946b419bcbaec41e85a46f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/230382
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Convert some optimizations rules to strongly-typed versions. So far, I
have only converted the rules that need no additional changes (i.e: only
need to change '->' to "=>").
This CL covers the rules from line 478 - line 800 in S390X.rules file.
Some compare and branch rules also fall in this range, but they were
already done previously in another CL.
Passes toolstash-check.
Change-Id: I9167c5f1a32f4fd6c29bacc13fff95e83b0533e0
Reviewed-on: https://go-review.googlesource.com/c/go/+/230338
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The compiler has better error messages for methods called without a
pointer receiver when one is expected. This change is similar to
CL 229801, but for method calls.
Also, added better error messages for functions called with the wrong
capitalization. I left the third TODO in this switch statement almost
as-is because I'm not sure that the extra complexity is worth it -
I adjusted the error to look like the one the compiler reports.
Fixesgolang/go#38658
Change-Id: Ie0ca2503e12f3659f112f0135cc27db1b027fdcb
Reviewed-on: https://go-review.googlesource.com/c/go/+/230380
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Linux 4.5 introduced (and Linux 5.3 refined) the copy_file_range
system call, which allows file systems the opportunity to implement
copy acceleration techniques. This commit adds support for
copy_file_range(2) to the os package.
Introduce a new ReadFrom method on *os.File, which makes *os.File
implement the io.ReaderFrom interface. If dst and src are both files,
this enables io.Copy(dst, src) to call dst.ReadFrom(src), which, in
turn, will call copy_file_range(2) if possible. If copy_file_range(2)
is not supported by the host kernel, or if either of dst or src
refers to a non-regular file, ReadFrom falls back to the regular
io.Copy code path.
Add internal/poll.CopyFileRange, which acquires locks on the
appropriate poll.FDs and performs the actual work, as well as
internal/syscall/unix.CopyFileRange, which wraps the copy_file_range
system call itself at the lowest level.
Rework file layout in internal/syscall/unix to accomodate the
additional system call numbers needed for copy_file_range.
Merge these definitions with the ones used by getrandom(2) into
sysnum_linux_$GOARCH.go files.
A note on additional optimizations: if dst and src both refer to pipes
in the invocation dst.ReadFrom(src), we could, in theory, use the
existing splice(2) code in package internal/poll to splice directly
from src to dst. Attempting this runs into trouble with the poller,
however. If we call splice(src, dst) and see EAGAIN, we cannot know
if it came from src not being ready for reading or dst not being
ready for writing. The write end of src and the read end of dst are
not under our control, so we cannot reliably use the poller to wait
for readiness. Therefore, it seems infeasible to use the new ReadFrom
method to splice between pipes directly. In conclusion, for now, the
only optimization enabled by the new ReadFrom method on *os.File is
the copy_file_range optimization.
Fixes#36817.
Change-Id: I696372639fa0cdf704e3f65414f7321fc7d30adb
Reviewed-on: https://go-review.googlesource.com/c/go/+/229101
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
IntSize is an untyped constant that does not need explicit conversion.
Annotating IntSize as an int and running github.com/mdempsky/unconvert
reveals these two cases.
Fixes#38682.
Change-Id: I014646b7457ddcde32474810153229dcf0c269c6
Reviewed-on: https://go-review.googlesource.com/c/go/+/230306
Run-TryBot: Akhil Indurti <aindurti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Currently allocToCache assumes it can move the search address past the
block it allocated the cache from, which violates the property that
searchAddr should always point to mapped memory (i.e. memory represented
by pageAlloc.inUse).
This bug was already fixed once for pageAlloc.alloc in the Go 1.14
release via CL 216697, but that changed failed to take into account
allocToCache.
Fixes#38605.
Change-Id: Id08180aa10d19dc0f9f551a1d9e327a295560dff
Reviewed-on: https://go-review.googlesource.com/c/go/+/229577
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Originally, we use an assembly function that returns a boolean result to
tell whether the machine has vector facility or not. It is now no longer
needed when we can directly use cpu.S390X.HasVX variable. This CL
also removes the last occurence of hasVectorFacility function on s390x.
Change-Id: Id20cb746c21eacac5e13344b362e2d87adfe4317
Reviewed-on: https://go-review.googlesource.com/c/go/+/230337
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Originally, we use an assembly function that returns a boolean result to
tell whether the machine has vector facility or not. It is now no longer
needed when we can directly use cpu.S390X.HasVX variable.
Change-Id: Ic1dae851982532bcfd9a9453416c112347f21d87
Reviewed-on: https://go-review.googlesource.com/c/go/+/230318
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Originally, we use an assembly function that returns a boolean result to
tell whether the machine has vector facility or not. It is now no longer
needed when we can directly use cpu.S390X.HasVX variable.
Change-Id: Ic3ffeb9e63238ef41406d97cdc42502145ddb454
Reviewed-on: https://go-review.googlesource.com/c/go/+/230319
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This CL allows the usage of KDSA instruction when it is available. The
instruction is designed to be resistant to side channel attacks and
offers performance improvement for ed25519.
Benchmarks:
name old time/op new time/op delta
Signing-8 120µs ±20% 62µs ±12% -48.40% (p=0.000 n=10+10)
Verification-8 325µs ±17% 69µs ±10% -78.80% (p=0.000 n=10+10)
name old alloc/op new alloc/op delta
Signing-8 448B ± 0% 0B -100.00% (p=0.000 n=10+10)
Verification-8 288B ± 0% 0B -100.00% (p=0.000 n=10+10)
name old allocs/op new allocs/op delta
Signing-8 5.00 ± 0% 0.00 -100.00% (p=0.000 n=10+10)
Verification-8 2.00 ± 0% 0.00 -100.00% (p=0.000 n=10+10)
Change-Id: I0330ce83d807370b419ce638bc2cae4cb3c250dc
Reviewed-on: https://go-review.googlesource.com/c/go/+/202578
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <mike.munday@ibm.com>
This reverts CL 33677.
Reason for revert: NetBSD is broken
Updates #38649
Change-Id: Id60e3c97d3cb4fb0053dea03b95dbbb0b850c883
Reviewed-on: https://go-review.googlesource.com/c/go/+/230038
Run-TryBot: Andrew Bonventre <andybons@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Several new ones came from my testing (long, repeated runs) and one (assistQueue ->
spine) came from the staticlockranking builder (filed as issue 38441).
Fixes#38441
Change-Id: I4268da0d8b8cc51251eba6bd936110c8ab4c4e61
Reviewed-on: https://go-review.googlesource.com/c/go/+/229480
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Currently, the small object sweeper will sweep until it finds a free
slot or there are no more spans of that size class to sweep. In dense
heaps, this can cause sweeping for a given size class to take
unbounded time, and gets worse with larger heaps.
This CL limits the small object sweeper to try at most 100 spans
before giving up and allocating a fresh span. Since it's already shown
that 100 spans are completely full at that point, the space overhead
of this fresh span is at most 1%.
This CL is based on an experimental CL by Austin Clements (CL 187817)
and is updated to be part of the mcentral implementation, gated by
go115NewMCentralImpl.
Updates #18155.
Change-Id: I37a72c2dcc61dd6f802d1d0eac3683e6642b6ef8
Reviewed-on: https://go-review.googlesource.com/c/go/+/229998
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Austin Clements <austin@google.com>
Currently mcentral is implemented as a couple of linked lists of spans
protected by a lock. Unfortunately this design leads to significant lock
contention.
The span ownership model is also confusing and complicated. In-use spans
jump between being owned by multiple sources, generally some combination
of a gcSweepBuf, a concurrent sweeper, an mcentral or an mcache.
So first to address contention, this change replaces those linked lists
with gcSweepBufs which have an atomic fast path. Then, we change up the
ownership model: a span may be simultaneously owned only by an mcentral
and the page reclaimer. Otherwise, an mcentral (which now consists of
sweep bufs), a sweeper, or an mcache are the sole owners of a span at
any given time. This dramatically simplifies reasoning about span
ownership in the runtime.
As a result of this new ownership model, sweeping is now driven by
walking over the mcentrals rather than having its own global list of
spans. Because we no longer have a global list and we traditionally
haven't used the mcentrals for large object spans, we no longer have
anywhere to put large objects. So, this change also makes it so that we
keep large object spans in the appropriate mcentral lists.
In terms of the static lock ranking, we add the spanSet spine locks in
pretty much the same place as the mcentral locks, since they have the
potential to be manipulated both on the allocation and sweep paths, like
the mcentral locks.
This new implementation is turned on by default via a feature flag
called go115NewMCentralImpl.
Benchmark results for 1 KiB allocation throughput (5 runs each):
name \ MiB/s go113 go114 gotip gotip+this-patch
AllocKiB-1 1.71k ± 1% 1.68k ± 1% 1.59k ± 2% 1.71k ± 1%
AllocKiB-2 2.46k ± 1% 2.51k ± 1% 2.54k ± 1% 2.93k ± 1%
AllocKiB-4 4.27k ± 1% 4.41k ± 2% 4.33k ± 1% 5.01k ± 2%
AllocKiB-8 4.38k ± 3% 5.24k ± 1% 5.46k ± 1% 8.23k ± 1%
AllocKiB-12 4.38k ± 3% 4.49k ± 1% 5.10k ± 1% 10.04k ± 0%
AllocKiB-16 4.31k ± 1% 4.14k ± 3% 4.22k ± 0% 10.42k ± 0%
AllocKiB-20 4.26k ± 1% 3.98k ± 1% 4.09k ± 1% 10.46k ± 3%
AllocKiB-24 4.20k ± 1% 3.97k ± 1% 4.06k ± 1% 10.74k ± 1%
AllocKiB-28 4.15k ± 0% 4.00k ± 0% 4.20k ± 0% 10.76k ± 1%
Fixes#37487.
Change-Id: I92d47355acacf9af2c41bf080c08a8c1638ba210
Reviewed-on: https://go-review.googlesource.com/c/go/+/221182
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
This change implements the spanSet data structure which is based off of
the gcSweepBuf data structure. While the general idea is the same (one
has two of these which one switches between every GC cycle; one to push
to and one to pop from), there are some key differences.
Firstly, we never have a need to iterate over this data structure so
delete numBlocks and block. Secondly, we want to be able to pop from the
front of the structure concurrently with pushes to the back. As a result
we need to maintain both a head and a tail and this change introduces an
atomic headTail structure similar to the one used by sync.Pool. It also
implements popfirst in a similar way.
As a result of this headTail, we need to be able to explicitly reset the
length, head, and tail when it goes empty at the end of sweep
termination, so add a reset method.
Updates #37487.
Change-Id: I5b8ad290ec32d591e3c8c05e496c5627018074f6
Reviewed-on: https://go-review.googlesource.com/c/go/+/221181
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
This change adds a global pool of spanSetBlocks to the spanSet data
structure and adds support for eagerly freeing these blocks back to the
pool if the block goes empty.
This change prepares us to use this data structure in more places in the
runtime by allowing reuse of spanSetBlock.
Updates #37487.
Change-Id: I0752226e3667a9e3e1d87c9b66edaedeae1ac23f
Reviewed-on: https://go-review.googlesource.com/c/go/+/221180
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
This change copies the gcSweepBuf data structure into a new file and
renames it spanSet. It will serve as the basis for a heavily modified
version of the gcSweepBuf data structure for the new mcentral
implementation.
We move it into a separate file now for two reasons:
1. We will need both implementations as they will coexist simultaneously
for a time.
2. By creating it now in a new change it'll make future changes which
modify it easier to review (rather than introducing the new file then).
Updates #37487.
Change-Id: If80603cab6e813a1ee2e5ecd49dcde5d8045a6c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/221179
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Implement multi-control branches for riscv64, switching to using the BNEZ
pseudo-instruction when rewriting conditionals. This will allow for further
branch optimisations to later be performed via rewrites.
Change-Id: I7f2c69f3c77494b403f26058c6bc8432d8070ad0
Reviewed-on: https://go-review.googlesource.com/c/go/+/226399
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
type T [3]string
Prior to this change, we generated this equality alg for T:
func eqT(p, q *T) (r bool) {
for i := range *p {
if len(p[i]) == len(q[i]) {
} else {
return
}
}
for j := range *p {
if runtime.memeq(p[j].ptr, q[j].ptr, len(p[j])) {
} else {
return
}
}
return true
}
That first loop can be profitably eliminated;
it's cheaper to spell out 3 length equality checks.
We now generate:
func eqT(p, q *T) (r bool) {
if len(p[0]) == len(q[0]) &&
len(p[1]) == len(q[1]) &&
len(p[2]) == len(q[2]) {
} else {
return
}
for i := 0; i < len(p); i++ {
if runtime.memeq(p[j].ptr, q[j].ptr, len(p[j])) {
} else {
return
}
}
return true
}
We now also eliminate loops for small float arrays as well,
and for any array of size 1.
These cutoffs were selected to minimize code size on amd64
at this moment, for lack of a more compelling methodology.
Any smallish number would do.
The switch from range loops to plain for loops allowed me
to use a temp instead of a named var, which eliminated
a pointless argument to checkAll.
The code to construct them is also a bit clearer, in my opinion.
Change-Id: I1bdd8ee4a2739d00806e66b17a4e76b46e71231a
Reviewed-on: https://go-review.googlesource.com/c/go/+/230210
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
type T [8]string
Prior to this change, we generated this equality algorithm for T:
func eqT(p, q *T) (r bool) {
for i := range *p {
if p[i] == q[i] {
} else {
return
}
}
return true
}
This change splits this into two loops, so that we can do the
cheap (length) half early and only then do the expensive (contents) half.
We now generate:
func eqT(p, q *T) (r bool) {
for i := range *p {
if len(p[i]) == len(q[i]) {
} else {
return
}
}
for j := range *p {
if runtime.memeq(p[j].ptr, q[j].ptr, len(p[j])) {
} else {
return
}
}
return true
}
The generated code is typically ~17% larger because it contains
two loops instead of one. In the future, we might want to unroll
the first loop when the array is small.
Change-Id: I26b2793b90ec6aff21766a411b15a4ff1096c03f
Reviewed-on: https://go-review.googlesource.com/c/go/+/230209
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
type T [8]interface{}
Prior to this change, we generated this equality algorithm for T:
func eqT(p, q *T) bool {
for i := range *p {
if p[i] != q[i] {
return false
}
}
return true
}
This change splits this into two loops, so that we can do the
cheap (type) half early and only then do the expensive (data) half.
We now generate:
func eqT(p, q *T) (r bool) {
for i := range *p {
if p[i].type == q[i].type {
} else {
return
}
}
for j := range *p {
if runtime.efaceeq(p[j].type, p[j].data, q[j].data) {
} else {
return
}
}
return true
}
The use of a named return value and a bare return is to work
around some typechecking problems that stymied me.
The structure of using equals and else (instead of not equals and then)
was for implementation convenience and clarity. As a bonus,
it generates slightly shorter code on AMD64, because zeroing a register
to return is cheaper than writing $1 to it.
The generated code is typically ~17% larger because it contains
two loops instead of one. In the future, we might want to unroll
the first loop when the array is small.
Change-Id: I5b2c8dd3384852f085c4f3e1f6ad20bc5ae59062
Reviewed-on: https://go-review.googlesource.com/c/go/+/230208
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
type T struct {
s interface{}
i int
}
Prior to this change, we generated this equality algorithm for T:
func eqT(p, q *T) bool {
return p.s.type == q.s.type &&
runtime.efaceeq(p.s.type, p.s.data, q.s.data) &&
p.i == q.i
}
This change splits the two halves of the interface equality,
so that we can do the cheap (type) half early and the expensive
(data) half late. We now generate:
func eqT(p, q *T) bool {
return p.s.type == q.s.type &&
p.i == q.i &&
runtime.efaceeq(p.s.type, p.s.data, q.s.data)
}
The generated code tends to be a bit smaller. Examples:
go/ast
.eq."".ForStmt 306 -> 304 (-0.65%)
.eq."".TypeAssertExpr 221 -> 219 (-0.90%)
.eq."".TypeSwitchStmt 228 -> 226 (-0.88%)
.eq."".ParenExpr 150 -> 148 (-1.33%)
.eq."".IndexExpr 221 -> 219 (-0.90%)
.eq."".SwitchStmt 228 -> 226 (-0.88%)
.eq."".RangeStmt 334 -> 332 (-0.60%)
Change-Id: Iec9e24f214ca772416202b9fb9252e625c22380e
Reviewed-on: https://go-review.googlesource.com/c/go/+/230207
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Refactor out creating the two Nodes needed to check interface equality.
Preliminary work to other optimizations.
Passes toolstash-check.
Change-Id: Id6b39e8e78f07289193423d0ef905d70826acf89
Reviewed-on: https://go-review.googlesource.com/c/go/+/230206
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
type T struct {
s string
i int
}
Prior to this change, we generated this equality algorithm for T:
func eqT(p, q *T) bool {
return len(p.s) == len(q.s) &&
runtime.memequal(p.s.ptr, q.s.ptr, len(p.s)) &&
p.i == q.i
}
This change splits the two halves of the string equality,
so that we can do the cheap (length) half early and the expensive
(contents) half late. We now generate:
func eqT(p, q *T) bool {
return len(p.s) == len(q.s) &&
p.i == q.i &&
runtime.memequal(p.s.ptr, q.s.ptr, len(p.s))
}
The generated code for these functions tends to be a bit shorter. Examples:
runtime
.eq."".Frame 274 -> 272 (-0.73%)
.eq."".funcinl 249 -> 247 (-0.80%)
.eq."".modulehash 207 -> 205 (-0.97%)
Change-Id: I4efac9f7d410f0a11a94dcee2bf9c0b49b60e301
Reviewed-on: https://go-review.googlesource.com/c/go/+/230205
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Refactor out creating the two Nodes needed to check string equality.
Preliminary work to other optimizations.
Passes toolstash-check.
Change-Id: I72e824dac904e579b8ba9a3669a94fa1471112d2
Reviewed-on: https://go-review.googlesource.com/c/go/+/230204
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
