Change-Id: I2e4d74a86faa95321e847a061e06c3efff7f20df
Reviewed-on: https://go-review.googlesource.com/c/go/+/605775
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
I was not sure this was correct so I exhaustively checked all possibilities:
https://go.dev/play/p/hjmCLm4Iagzhttps://go.dev/play/p/R9RuRGKwCbN
Change-Id: I85f053df825a4d77f978de42f8a1fcaf4b881def
Reviewed-on: https://go-review.googlesource.com/c/go/+/605696
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Change-Id: I419faa781db085b98ea25008ca127d0317fb34e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/605695
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
I didn't implemented negative limits since prove is most useful for BCE which
should never be negative in the first place.
Change-Id: I302ee462cdc20bd4edff0618f7e49ff66fc2a007
Reviewed-on: https://go-review.googlesource.com/c/go/+/605136
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Change-Id: Id522bde5bba627d9cdc8c3d8e907bdc168e5b13c
Reviewed-on: https://go-review.googlesource.com/c/go/+/605157
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
addLocalFacts loop already ft.update which sets up limits correctly, but doing this in flowLimit help us since other values might depend on this limit.
Updates #68857
We could improve this further:
- remove mod alltogheter when we can prove a < b.
- we could do more adhoc computation in flowLimit to set umax and umin tighter
Change-Id: I5184913577b6a51a07cb53a6e6b73552a982de0b
Reviewed-on: https://go-review.googlesource.com/c/go/+/605156
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change-Id: I6902c405cab7bd573f6a721a6ca7c783713ea39a
Reviewed-on: https://go-review.googlesource.com/c/go/+/604456
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
This help to optimize code like this:
func f(buckets *[512]bucket, v value) {
a, b := v.computeSomething()
// assume a and b are proved < 512
b := &buckets[a ^ b] // pick a random bucket
b.store(v)
}
Change-Id: I1acf702f5a8137f9ded49081b4703922879b0288
Reviewed-on: https://go-review.googlesource.com/c/go/+/604455
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
See the cases above all of theses are implemented (except XOR which has a blank case with comments).
Change-Id: I9e2994490dac89e86ba70c1abeb1af1cbcf032e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/604416
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
y := bits.TrailingZeros(x)
if y > bits.Len(x.umax)-1 {
then must always be true 1 << y > x.umax which is impossible
}
Change-Id: Iab4fce1c2ef828bee3a8a4a977cbadb5f9333136
Reviewed-on: https://go-review.googlesource.com/c/go/+/603996
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Change-Id: Ie3c7e5eaba6a9a29389018625c4b784d07c6f173
Reviewed-on: https://go-review.googlesource.com/c/go/+/603537
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Change-Id: Id991ec0826a4e2857f00330b4b7ff2b71907b789
Reviewed-on: https://go-review.googlesource.com/c/go/+/606615
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Get rid of TODO in prove pass.
We currently avoid marking shifts of constants as bounded, where
bounded means we don't have to worry about <0 or >=bitwidth shifts.
We do this because it causes different rule applications during lowering
which cause some codegen tests to fail.
Add some new rules which ensure that we get the right final instruction
sequence regardless of the ordering. Then we can remove this special case.
Change-Id: I4e962d4f09992b42ab47e123de5ded3b8b8fb205
Reviewed-on: https://go-review.googlesource.com/c/go/+/602935
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Local facts can get us to unsatisfiable because there is an
unconditional panic in the block. That shouldn't declare the whole
block as unreachable, because we do still need to enter it to get
that panic.
Fixes#68816
Change-Id: I9220edb46089690702d2eb61d112815c7ac91f16
Reviewed-on: https://go-review.googlesource.com/c/go/+/604118
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
I think I introduced #68809 when rewriting the prove pass, by
introducing an off-by-one error here:
https://go-review.googlesource.com/c/go/+/599096/5/src/cmd/compile/internal/ssa/prove.go
lines 872-874.
The min++ is already handled in one of the two following cases with
the (r&eq==0) condition. Move the min++ to just the other case.
Fixes#68809
Change-Id: Iffae99e29219c94aaf584cd7ae247289fa604a00
Reviewed-on: https://go-review.googlesource.com/c/go/+/604100
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
We don't need noLimit checks in a bunch of places.
Also simplify folding of provable constant results.
At this point in the CL stack, compilebench reports no performance
changes. The only thing of note is that binaries got a bit smaller.
name old text-bytes new text-bytes delta
HelloSize 960kB ± 0% 952kB ± 0% -0.83% (p=0.000 n=10+10)
CmdGoSize 12.3MB ± 0% 12.1MB ± 0% -1.53% (p=0.000 n=10+10)
Change-Id: Id4be75eec0f8c93f2f3b93a8521ce2278ee2ee2c
Reviewed-on: https://go-review.googlesource.com/c/go/+/599197
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Move some code from when we learn that we take a branch, to when
we learn that a boolean is true or false. It is more consistent
this way (and may lead to a few more cases where we can derive
useful relations).
Change-Id: Iea7b2d6740e10c9c71c4b1546881f501da81cd21
Reviewed-on: https://go-review.googlesource.com/c/go/+/599098
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
The constant lattice for these types is pretty simple.
We no longer need the old-style facts table, as the ordering
table now has all that information.
Change-Id: If0e118c27a4de8e9bfd727b78942185c2eb50c4b
Reviewed-on: https://go-review.googlesource.com/c/go/+/599097
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Handles a lot more cases where constant ranges can eliminate
various (mostly bounds failure) paths.
Fixes#66826Fixes#66692Fixes#48213
Update #57959
TODO: remove constant logic from poset code, no longer needed.
Change-Id: Id196436fcd8a0c84c7d59c04f93bd92e26a0fd7e
Reviewed-on: https://go-review.googlesource.com/c/go/+/599096
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Here begins a pretty major rewrite of the prove pass. The fundamental
observation is that although keeping facts about relations between
two SSA values could use O(n^2) space, keeping facts about relations
between an SSA value and constants needs only O(n) space. We can just
keep track of min/max for every SSA value at little cost.
Redo the limit table to just keep track of limits for all SSA values.
Use just a slice instead of a map. It may use more space (but still
just O(n) space), but accesses are a lot faster. And with the cache
in the compiler, that space will be reused quickly.
This is part of my planning to add lots more constant limits in the
prove pass.
Change-Id: Ie36819fad5631a8b79c3630fe0e819521796551a
Reviewed-on: https://go-review.googlesource.com/c/go/+/599255
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
I wrote theses checks because I got bad panics on some innocent functions,
turns out I was working around #63955 but I was not aware of that at the time.
The proper fix was included in CL 539977 this is now doing nothing.
Change-Id: I89329329933527b6f3cb817dc1e039a38f58da9a
Reviewed-on: https://go-review.googlesource.com/c/go/+/560975
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Fixes#61629
This reduce the pressure on regalloc because then the loop only keep alive
one value (the iterator) instead of the iterator and the upper bound since
the comparison now acts against an immediate, often zero which can be skipped.
This optimize things like:
for i := 0; i < n; i++ {
Or a range over a slice where the index is not used:
for _, v := range someSlice {
Or the new range over int from #61405:
for range n {
It is hit in 975 unique places while doing ./make.bash.
Change-Id: I5facff8b267a0b60ea3c1b9a58c4d74cdb38f03f
Reviewed-on: https://go-review.googlesource.com/c/go/+/512935
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Jorropo <jorropo.pgm@gmail.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
Fix spelling errors discovered using https://github.com/codespell-project/codespell. Errors in data files and vendored packages are ignored.
Change-Id: I83c7818222f2eea69afbd270c15b7897678131dc
GitHub-Last-Rev: 3491615b1b
GitHub-Pull-Request: golang/go#60758
Reviewed-on: https://go-review.googlesource.com/c/go/+/502576
Auto-Submit: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
For now, only apply the rule if either of arguments are constants. That
would catch a lot of real user code, without slowing down the compiler
with code generated for string comparison (experience in CL 410336).
Updates #57959Fixes#45928
Change-Id: Ie2e830d6d0d71cda3947818b22c2775bd94f7971
Reviewed-on: https://go-review.googlesource.com/c/go/+/483359
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Currently, the prove pass can get knowledge from some specific logic
operators only before the CFG is explored, which means that the bounds
information of the branch will be ignored.
This CL updates the facts table by the logic operators in every
branch. Combined with the branch information, this will be helpful for
BCE in some circumstances.
Fixes#57243
Change-Id: I0bd164f1b47804ccfc37879abe9788740b016fd5
Reviewed-on: https://go-review.googlesource.com/c/go/+/419555
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Eric Fang <eric.fang@arm.com>
Reviewed-by: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Change-Id: Ica8d5e5799a4de532764ae86cdb623508d3a8e18
GitHub-Last-Rev: 3e97cca9de
GitHub-Pull-Request: golang/go#58689
Reviewed-on: https://go-review.googlesource.com/c/go/+/471021
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
This reverts commit 3680b5e9c4.
Reason for revert: causes long compile times on certain functions. See issue #57959
Change-Id: Ie9e881ca8abbc79a46de2bfeaed0b9d6c416ed42
Reviewed-on: https://go-review.googlesource.com/c/go/+/463295
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Change-Id: I69065f8adf101fdb28682c55997f503013a50e29
Reviewed-on: https://go-review.googlesource.com/c/go/+/449757
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Joedian Reid <joedian@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Joedian Reid <joedian@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
If x+delta cannot overflow/underflow, we can derive:
x+delta < x if delta<0 (this CL included)
x+delta > x if delta>0 (this CL not included due to
a recursive stack overflow)
Remove 95 bounds checks during ./make.bat
Fixes#51622
Change-Id: I60d9bd84c5d7e81bbf808508afd09be596644f09
Reviewed-on: https://go-review.googlesource.com/c/go/+/406175
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Not used any more.
Fixes#53860
Change-Id: Id0b1c3ed30b576d6c5f08f064d1262de337262b3
Reviewed-on: https://go-review.googlesource.com/c/go/+/418374
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
For this code:
z &= 63
_ = x<<z | x>>(64-z)
Now can prove 'x<<z' in bound. In ppc64 lowering pass, it will not
produce an extra '(ANDconst <typ.Int64> [63] z)' causing
codegen/rotate.go failed. Just remove the type check in rewrite rules
as the workaround.
Removes 32 bounds checks during make.bat.
Fixes#52563.
Change-Id: I14ed2c093ff5638dfea7de9bc7649c0f756dd71a
Reviewed-on: https://go-review.googlesource.com/c/go/+/404315
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
It is hit ~70k times building go.
This make the go binary, 0.04% smaller.
I didn't included benchmarks because this is just constant foldings
and is hard to mesure objectively.
For example, this enable rewriting things like:
if x == 20 {
return x + 30 + z
}
Into:
if x == 20 {
return 50 + z
}
It's not just fixing programer's code,
the ssa generator generate code like this sometimes.
Change-Id: I0861f342b27f7227b5f1c34d8267fa0057b1bbbc
GitHub-Last-Rev: 4c2f9b5216
GitHub-Pull-Request: golang/go#52669
Reviewed-on: https://go-review.googlesource.com/c/go/+/403735
Reviewed-by: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reorganize the way we rewrite expression switches on strings, so that
jump tables are naturally used for the outer switch on the string length.
The changes to the prove pass in this CL are required so as to not repeat
the test for string length in each case.
name old time/op new time/op delta
SwitchStringPredictable 2.28ns ± 9% 2.08ns ± 5% -9.04% (p=0.000 n=10+10)
SwitchStringUnpredictable 10.5ns ± 1% 9.5ns ± 1% -9.08% (p=0.000 n=9+10)
Update #5496
Update #34381
Change-Id: Ie6846b1dd27f3e472f7c30dfcc598c68d440b997
Reviewed-on: https://go-review.googlesource.com/c/go/+/395714
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@google.com>
[This CL is part of a sequence implementing the proposal #51082.
The design doc is at https://go.dev/s/godocfmt-design.]
Run the updated gofmt, which reformats doc comments,
on the main repository. Vendored files are excluded.
For #51082.
Change-Id: I7332f099b60f716295fb34719c98c04eb1a85407
Reviewed-on: https://go-review.googlesource.com/c/go/+/384268
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
A run of lines that are indented with any number of spaces or tabs
format as a <pre> block. This commit fixes various doc comments
that format badly according to that (standard) rule.
For example, consider:
// - List item.
// Second line.
// - Another item.
Because the - lines are unindented, this is actually two paragraphs
separated by a one-line <pre> block. This CL rewrites it to:
// - List item.
// Second line.
// - Another item.
Today, that will format as a single <pre> block.
In a future release, we hope to format it as a bulleted list.
Various other minor fixes as well, all in preparation for reformatting.
For #51082.
Change-Id: I95cf06040d4186830e571cd50148be3bf8daf189
Reviewed-on: https://go-review.googlesource.com/c/go/+/384257
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Currently, we keep track of slice len by mapping from slice ID to
len/cap SSA value. However, slice len/cap can have multiple SSA values,
so when updating fact table for slice len/cap, we only update in one
place.
Instead, we can take advantage of the transitive relations provided by
poset. So all duplicated slice lens are set as equal to one another.
When updating fact table for one, that fact will be reflected to all
others. The same mechanism is applied for slice cap.
Removes 15 bounds checks from std/cmd.
Fixes#42603
Change-Id: I32c07968824cc33765b1e441b3ae2c4b5f5997c3
Reviewed-on: https://go-review.googlesource.com/c/go/+/273670
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
CL 244579 added guard clauses to prevent a faulty state that was
possible under the incorrect logic of the uniquePred loop in
addLocalInductiveFacts. That faulty state was still making the
intended optimization, but not for the correct reason.
Removing the faulty state also removed the overly permissive application
of the optimization, and therefore made these two tests fail.
We disabled the tests of this optimization in CL 244579 to allow us to
quickly apply the fix in the CL. This CL now corrects the logic of the
uniquePred loop in order to apply the optimization correctly.
The comment above the uniquePred loop says that it will follow unique
predecessors until it reaches a join point. Without updating the child
node on each iteration, it cannot follow the chain of unique
predecessors more than one step. Adding the update to the child node
on each iteration of the loop allows the logic to follow the chain of
unique predecessors until reaching a join point (because a non-unique
predecessor will signify a join point).
Updates #40502.
Change-Id: I23d8367046a2ab3ce4be969631f9ba15dc533e6c
Reviewed-on: https://go-review.googlesource.com/c/go/+/246157
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Currently in addLocalInductiveFacts, we only check whether
direct edge from if block to phi block exists. If not, the
following logic will treat the phi block as the first successor,
which is wrong.
This patch makes prove pass more conservative, so we disable
some cases in test/prove.go. We will do some optimization in
the following CL and enable these cases then.
Fixes#40367.
Change-Id: I27cf0248f3a82312a6f7dabe11c79a1a34cf5412
Reviewed-on: https://go-review.googlesource.com/c/go/+/244579
Reviewed-by: Zach Jones <zachj1@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Taking over Zach's CL 212277. Just cleaned up and added a test.
For a positive, signed integer, an arithmetic right shift of count
(bit-width - 1) equals zero. e.g. int64(22) >> 63 -> 0. This CL makes
prove replace these right shifts with a zero-valued constant.
These shifts may arise in source code explicitly, but can also be
created by the generic rewrite of signed division by a power of 2.
// Signed divide by power of 2.
// n / c = n >> log(c) if n >= 0
// = (n+c-1) >> log(c) if n < 0
// We conditionally add c-1 by adding n>>63>>(64-log(c))
(first shift signed, second shift unsigned).
(Div64 <t> n (Const64 [c])) && isPowerOfTwo(c) ->
(Rsh64x64
(Add64 <t> n (Rsh64Ux64 <t>
(Rsh64x64 <t> n (Const64 <typ.UInt64> [63]))
(Const64 <typ.UInt64> [64-log2(c)])))
(Const64 <typ.UInt64> [log2(c)]))
If n is known to be positive, this rewrite includes an extra Add and 2
extra Rsh. This CL will allow prove to replace one of the extra Rsh with
a 0. That replacement then allows lateopt to remove all the unneccesary
fixups from the generic rewrite.
There is a rewrite rule to handle this case directly:
(Div64 n (Const64 [c])) && isNonNegative(n) && isPowerOfTwo(c) ->
(Rsh64Ux64 n (Const64 <typ.UInt64> [log2(c)]))
But this implementation of isNonNegative really only handles constants
and a few special operations like len/cap. The division could be
handled if the factsTable version of isNonNegative were available.
Unfortunately, the first opt pass happens before prove even has a
chance to deduce the numerator is non-negative, so the generic rewrite
has already fired and created the extra Ops discussed above.
Fixes#36159
By Printf count, this zeroes 137 right shifts when building std and cmd.
Change-Id: Iab486910ac9d7cfb86ace2835456002732b384a2
Reviewed-on: https://go-review.googlesource.com/c/go/+/232857
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
