I was playing around with souper recently on some wasms I had lying
around and these are some optimization opportunities that popped out
which seemed easy-enough to add to the egraph-based optimizations.
In the provided test case in #5716, the result of a call was then
added to 0. We have a rewrite rule that sets the remat-bit on any add
of a value and a constant, because these frequently appear (e.g. from
address offset calculations) and this can frequently reduce register
pressure (one long-lived base vs. many long-lived base+offset values).
Separately, we have an algebraic rule that `x+0` rewrites to `x`.
The result of this was that we had an eclass with the remat bit set on
the add, but the add was also union'd into the call. We pick the
latter during extraction, because it's cheaper not to do the add at
all; but we still get the remat bit, and try to remat a call (!),
which blows up later.
This PR fixes the logic to look up the "best value" for a value (i.e.,
whatever extraction determined), and look up the remat bit on *that*
node, not the canonical node.
(Why did the canonical node become the iadd and not the call? Because
the former had a lower value-number, as an accident of IR
construction; we don't impose any requirements on the input CLIF's
value-number ordering, and I don't think this breaks any of the
important acyclic properties, even though there is technically a
dependence from a lower-numbered to a higher-numbered node. In essence
one can think of them as having "virtual numbers" in any true
topologically-sorted order, and the only place the actual integer
indices matter should be in choosing the "canonical ID", which is just
used for dedup'ing, modulo this bug.)
Fixes#5716.
* Remove trailing whitespace in `lower.isle` files
* Legalize the `band_not` instruction into simpler form
This commit legalizes the `band_not` instruction into `band`-of-`bnot`,
or two instructions. This is intended to assist with egraph-based
optimizations where the `band_not` instruction doesn't have to be
specifically included in other bit-operation-patterns.
Lowerings of the `band_not` instruction have been moved to a
specialization of the `band` instruction.
* Legalize `bor_not` into components
Same as prior commit, but for the `bor_not` instruction.
* Legalize bxor_not into bxor-of-bnot
Same as prior commits. I think this also ended up fixing a bug in the
s390x backend where `bxor_not x y` was actually translated as `bnot
(bxor x y)` by accident given the test update changes.
* Simplify not-fused operands for riscv64
Looks like some delegated-to rules have special-cases for "if this
feature is enabled use the fused instruction" so move the clause for
testing the feature up to the lowering phase to help trigger other rules
if the feature isn't enabled. This should make the riscv64 backend more
consistent with how other backends are implemented.
* Remove B{and,or,xor}Not from cost of egraph metrics
These shouldn't ever reach egraphs now that they're legalized away.
* Add an egraph optimization for `x^-1 => ~x`
This adds a simplification node to translate xor-against-minus-1 to a
`bnot` instruction. This helps trigger various other optimizations in
the egraph implementation and also various backend lowering rules for
instructions. This is chiefly useful as wasm doesn't have a `bnot`
equivalent, so it's encoded as `x^-1`.
* Add a wasm test for end-to-end bitwise lowerings
Test that end-to-end various optimizations are being applied for input
wasm modules.
* Specifically don't self-update rustup on CI
I forget why this was here originally, but this is failing on Windows
CI. In general there's no need to update rustup, so leave it as-is.
* Cleanup some aarch64 lowering rules
Previously a 32/64 split was necessary due to the `ALUOp` being different
but that's been refactored away no so there's no longer any need for
duplicate rules.
* Narrow a x64 lowering rule
This previously made more sense when it was `band_not` and rarely used,
but be more specific in the type-filter on this rule that it's only
applicable to SIMD types with lanes.
* Simplify xor-against-minus-1 rule
No need to have the commutative version since constants are already
shuffled right for egraphs
* Optimize band-of-bnot when bnot is on the left
Use some more rules in the egraph algebraic optimizations to
canonicalize band/bor/bxor with a `bnot` operand to put the operand on
the right. That way the lowerings in the backends only have to list the
rule once, with the operand on the right, to optimize both styles of
input.
* Add commutative lowering rules
* Update cranelift/codegen/src/isa/x64/lower.isle
Co-authored-by: Jamey Sharp <jamey@minilop.net>
---------
Co-authored-by: Jamey Sharp <jamey@minilop.net>
* Cranelift: Introduce the `tail` calling convention
This is an unstable-ABI calling convention that we will eventually use to
support Wasm tail calls.
Co-Authored-By: Jamey Sharp <jsharp@fastly.com>
* Cranelift: Introduce the `return_call` and `return_call_indirect` instructions
These will be used to implement tail calls for Wasm and any other language
targeting CLIF. The `return_call_indirect` instruction differs from the Wasm
instruction of the same name by taking a native address callee rather than a
Wasm function index.
Co-Authored-By: Jamey Sharp <jsharp@fastly.com>
* Cranelift: Implement verification rules for `return_call[_indirect]`
They must:
* have the same return types between the caller and callee,
* have the same calling convention between caller and callee,
* and that calling convention must support tail calls.
Co-Authored-By: Jamey Sharp <jsharp@fastly.com>
* cargo fmt
---------
Co-authored-by: Jamey Sharp <jsharp@fastly.com>
This rewrite was introduced in #5676 and then reverted in #5682 due to a footgun
where we accidentally weren't actually checking the `y == !z` precondition. This
commit fixes the precondition check. It also fixes the arithmetic to be
correctly masked to the value type's width.
This reverts commit 268f6bfc1d.
Most of these optimizations are in the egraph `cprop.isle` rules now,
making a separate crate unnecessary.
Also I think the `udiv` optimizations here are straight-up wrong (doing
signed instead of unsigned division, and panicking instead of preserving
traps on division by zero) so I'm guessing this crate isn't seriously
used anywhere.
At the least, bjorn3 confirms that cg_clif doesn't use this, and I've
verified that Wasmtime doesn't either.
Closes#1090.
Improve the generated code for unordered floating point comparisons by negating the comparison and inveritng the branches. This allows us to pick the unordered versions, which generate significantly better code.
Add a conditional branch instruction with two targets: brif. This instruction will eventually replace brz and brnz, as it encompasses the behavior of both.
This PR also changes the InstructionData layout for instruction formats that hold BlockCall values, taking the same approach we use for Value arguments. This allows branch_destination to return a slice to the BlockCall values held in the instruction, rather than requiring that we pattern match on InstructionData to fetch the then/else blocks.
Function generation for fuzzing has been updated to generate uses of brif, and I've run the cranelift-fuzzgen target locally for hours without triggering any new failures.
Ideally these pairs of CLIF instructions should emit a single x86
instruction, but they don't today. This test will tell us if somebody
fixes that.
Similar tests might make sense for imul/umulhi as well as signed
versions, but I haven't tried that.
This PR follows up on #5382 and #5391, which rebuilt the egraph-based optimization framework to be more performant, by enabling it by default.
Based on performance results in #5382 (my measurements on SpiderMonkey and bjorn3's independent confirmation with cg_clif), it seems that this is reasonable to enable. Now that we have been fuzzing compiler configurations with egraph opts (#5388) for 6 weeks, having fixed a few fuzzbugs that came up (#5409, #5420, #5438) and subsequently received no further reports from OSS-Fuzz, I believe it is stable enough to rely on.
This PR enables `use_egraphs`, and also normalizes its meaning: previously it forced optimization (it basically meant "turn on the egraph optimization machinery"), now it runs egraph opts if the opt level indicates (it means "use egraphs to optimize if we are going to optimize"). The conditionals in the top-level pass driver are a little subtle, but will get simpler once we can remove the non-egraph path (which we plan to do eventually!).
Fixes#5181.
* Support mergeable-but-side-effectful (idempotent) operations in general in the egraph's GVN.
This mirrors the similar change made in #5534.
* Add tests for egraph case.
* Adding in the foundations for Winch `filetests`
This commit adds two new crates into the Winch workspace:
`filetests` and `test-macros`. The intent is to mimic the
structure of Cranelift `filetests`, but in a simpler way.
* Updates to documentation
This commits adds a high level document to outline how to test Winch
through the `winch-tools` utility. It also updates some inline
documentation which gets propagated to the CLI.
* Updating test-macro to use a glob instead of only a flat directory
* add clif-util compile option to output object file
* switch from a box to a borrow
* update objectmodule tests to use borrowed isa
* put targetisa into an arc
* Switch duplicate loads w/ dynamic memories test to `min_size = 0`
This test was accidentally hitting a special case for bounds checks for when we
know that `offset + access_size < min_size` and can skip some steps. This
commit changes the `min_size` of the memory to zero so that we are forced to do
fully general bounds checks.
* Cranelift: Mark `uadd_overflow_trap` as okay for GVN
Although this improves our test sequence for duplicate loads with dynamic
memories, it unfortunately doesn't have any effect on sightglass benchmarks:
```
instantiation :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[34448 35607.23 37158] gvn_uadd_overflow_trap.so
[34566 35734.05 36585] main.so
instantiation :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[44101 60449.62 92712] gvn_uadd_overflow_trap.so
[44011 60436.37 92690] main.so
instantiation :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[35595 36675.72 38153] gvn_uadd_overflow_trap.so
[35440 36670.42 37993] main.so
compilation :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[17370195 17405125.62 17471222] gvn_uadd_overflow_trap.so
[17369324 17404859.43 17470725] main.so
execution :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[7055720520 7055886880.32 7056265930] gvn_uadd_overflow_trap.so
[7055719554 7055843809.33 7056193289] main.so
compilation :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[683589861 683767276.00 684098366] gvn_uadd_overflow_trap.so
[683590024 683767998.02 684097885] main.so
execution :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[46436883 46437135.10 46437823] gvn_uadd_overflow_trap.so
[46436883 46437087.67 46437785] main.so
compilation :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[126522461 126565812.58 126647044] gvn_uadd_overflow_trap.so
[126522176 126565757.75 126647522] main.so
execution :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[653010531 653010533.03 653010544] gvn_uadd_overflow_trap.so
[653010531 653010533.18 653010537] main.so
```
* cranelift-codegen-meta: Rename `side_effects_okay_for_gvn` to `side_effects_idempotent`
* cranelift-filetests: Ensure there is a trailing newline for blessed Wasm tests
* Cranelift: Make spectre guards GVN-able
While these instructions have a side effect that is otherwise invisible to the
optimizer, the side effect in question is idempotent, so it can be de-duplicated
by GVN.
* Cranelift: Run redundant load replacement and GVN twice
This allows us to actually replace redundant Wasm loads with dynamic memories.
While this improves our hand-crafted test sequences, it doesn't seem to have any
improvement on sightglass benchmarks run with dynamic memories, however it also
isn't a hit to compilation times, so seems generally good to land anyways:
```
$ cargo run --release -- benchmark -e ~/scratch/once.so -e ~/scratch/twice.so -m insts-retired --processes 20 --iterations-per-process 3 --engine-flags="--static-memory-maximum-size 0" -- benchmarks/default.suite
compilation :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[683595240 683768610.53 684097577] once.so
[683597068 700115966.83 1664907164] twice.so
instantiation :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[44107 60411.07 92785] once.so
[44138 59552.32 92097] twice.so
compilation :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[17369916 17404839.78 17471458] once.so
[17369935 17625713.87 30700150] twice.so
compilation :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[126523640 126566170.80 126648265] once.so
[126523076 127174580.30 163145149] twice.so
instantiation :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[34569 35686.25 36513] once.so
[34651 35749.97 36953] twice.so
instantiation :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[35146 36639.10 37707] once.so
[34472 36580.82 38431] twice.so
execution :: instructions-retired :: benchmarks/spidermonkey/benchmark.wasm
No difference in performance.
[7055720115 7055841324.82 7056180024] once.so
[7055717681 7055877095.85 7056225217] twice.so
execution :: instructions-retired :: benchmarks/pulldown-cmark/benchmark.wasm
No difference in performance.
[46436881 46437081.28 46437691] once.so
[46436883 46437127.68 46437766] twice.so
execution :: instructions-retired :: benchmarks/bz2/benchmark.wasm
No difference in performance.
[653010530 653010533.27 653010539] once.so
[653010531 653010532.95 653010538] twice.so
```
Rework the compilation of fcmp in the riscv64 backend to be in ISLE, removing the need for the dedicated Fcmp instruction. This change is motivated by #5500, which showed that the riscv64 backend was generating branch instructions in the middle of a basic block.
We can't remove lower_br_fcmp quite yet as it's used in a few places in the emit module, but it's now no longer reachable from the ISLE lowerings.
Fixes#5500
* cranelift: Add `iabs.i128` runtest
* riscv64: Fix incorrect extension in iabs
When lowering iabs, we were accidentally comparing the unextended value
this caused the instruction to misbehave with certain top bits.
This commit also adds a zbb lowering that does not use jumps.
When lowering `select+icmp` we have an optimization that allows us to
avoid materializing the icmp result.
We were accidentally not masking the high bits for i8 and i16 in this case.
Issue #5498 reported this as an illegal instruction but what was happening
there was that the invalid select caused a division by zero.
We had a off-by-one bounds check error when checking if we should
jump to the default block in a br-table. Instead of always jumping
to the default block when we have a jump table with 0 targets we
would try to compute an offset past the end of the table.
This sometimes would not crash, but it would crash if the there was
no block after the br_table, thus adding a cold block would cause a
segfault.
The actual fix is quite simple, do not count the default block
as a jump table entry when computing the limits.
This commit also does a bunch of cleanup and adding some comments
to the br_table emission code.
In #5031, we removed `bool` types from CLIF, using integers instead for
"truthy" values. This greatly simplified the IR, and was generally an
improvement.
However, because x86's `SETcc` instruction sets only the low 8 bits of a
register, we chose to use `i8` types as the result of `icmp` and `fcmp`,
to avoid the need for a masking operation when materializing the result.
Unfortunately this means that uses of truthy values often now have
`uextend` operations, especially when coming from Wasm (where truthy
values are naturally `i32`-typed). For example, where we previously had
`(brz (icmp ...))`, we now have `(brz (uextend (icmp ...)))`.
It's arguable whether or not we should switch to `i32` truthy values --
in most cases we can avoid materializing a value that's immediately used
for a branch or select, so a mask would in most cases be unnecessary,
and it would be a win at the IR level -- but irrespective of that, this
change *did* regress our generated code quality: our backends had
patterns for e.g. `(brz (icmp ...))` but not with the `uextend`, so we
were *always* materializing truthy values. Many blocks thus ended with
"cmp; setcc; cmp; test; branch" rather than "cmp; branch".
In #5391 we noticed this and fixed it on x64, but it was a general
problem on aarch64 and riscv64 as well. This PR introduces a
`maybe_uextend` extractor that "looks through" uextends, and uses it
where we consume truthy values, thus fixing the regression. This PR
also adds compile filetests to ensure we don't regress again.
The riscv64 backend has not been updated here because doing so appears
to trigger another issue in its branch handling; fixing that is TBD.
Fixes#5199.
Fixes#5200.
Fixes#5452.
Fixes#5453.
On riscv64, there is apparently an autoconversion from `ValueRegs` to
`Reg` that takes just the low register [0], and removing this conversion
causes 48 errors. As a result of this, `select` with an `i128` condition
was silently miscompiling, testing only the low 64 bits. We should
remove this autoconversion to ensure we aren't missing any other silent
truncations, but for now this PR just adds the explicit `I128` logic for
`select` / `select_spectre_guard`.
[0]
d9fdbfd50e/cranelift/codegen/src/isa/riscv64/inst.isle (L1762)
* cranelift-wasm: translate Wasm loads into lower-level CLIF operations
Rather than using `heap_{load,store,addr}`.
* cranelift: Remove the `heap_{addr,load,store}` instructions
These are now legalized in the `cranelift-wasm` frontend.
* cranelift: Remove the `ir::Heap` entity from CLIF
* Port basic memory operation tests to .wat filetests
* Remove test for verifying CLIF heaps
* Remove `heap_addr` from replace_branching_instructions_and_cfg_predecessors.clif test
* Remove `heap_addr` from readonly.clif test
* Remove `heap_addr` from `table_addr.clif` test
* Remove `heap_addr` from the simd-fvpromote_low.clif test
* Remove `heap_addr` from simd-fvdemote.clif test
* Remove `heap_addr` from the load-op-store.clif test
* Remove the CLIF heap runtest
* Remove `heap_addr` from the global_value.clif test
* Remove `heap_addr` from fpromote.clif runtests
* Remove `heap_addr` from fdemote.clif runtests
* Remove `heap_addr` from memory.clif parser test
* Remove `heap_addr` from reject_load_readonly.clif test
* Remove `heap_addr` from reject_load_notrap.clif test
* Remove `heap_addr` from load_readonly_notrap.clif test
* Remove `static-heap-without-guard-pages.clif` test
Will be subsumed when we port `make-heap-load-store-tests.sh` to generating
`.wat` tests.
* Remove `static-heap-with-guard-pages.clif` test
Will be subsumed when we port `make-heap-load-store-tests.sh` over to `.wat`
tests.
* Remove more heap tests
These will be subsumed by porting `make-heap-load-store-tests.sh` over to `.wat`
tests.
* Remove `heap_addr` from `simple-alias.clif` test
* Remove `heap_addr` from partial-redundancy.clif test
* Remove `heap_addr` from multiple-blocks.clif test
* Remove `heap_addr` from fence.clif test
* Remove `heap_addr` from extends.clif test
* Remove runtests that rely on heaps
Heaps are not a thing in CLIF or the interpreter anymore
* Add generated load/store `.wat` tests
* Enable memory-related wasm features in `.wat` tests
* Remove CLIF heap from fcmp-mem-bug.clif test
* Add a mode for compiling `.wat` all the way to assembly in filetests
* Also generate WAT to assembly tests in `make-load-store-tests.sh`
* cargo fmt
* Reinstate `f{de,pro}mote.clif` tests without the heap bits
* Remove undefined doc link
* Remove outdated SVG and dot file from docs
* Add docs about `None` returns for base address computation helpers
* Factor out `env.heap_access_spectre_mitigation()` to a local
* Expand docs for `FuncEnvironment::heaps` trait method
* Restore f{de,pro}mote+load clif runtests with stack memory
* cranelift-filetest: Add the ability to test `.wat` to assembly
* Make the load/store test case generator script use `.wat` tests
And generate tests that exercise both Wasm-to-CLIF lowering and Wasm all the way
to assembly.
* Remove old versions of generated load/store tests
* Add new generated load/store tests
* Fix filename reference in script
