T0b1/wasmtime
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

egraph-based midend: draw the rest of the owl (productionized). (#4953)

Browse Source 
* egraph-based midend: draw the rest of the owl.

* Rename `egg` submodule of cranelift-codegen to `egraph`.

* Apply some feedback from @jsharp during code walkthrough.

* Remove recursion from find_best_node by doing a single pass.

Rather than recursively computing the lowest-cost node for a given
eclass and memoizing the answer at each eclass node, we can do a single
forward pass; because every eclass node refers only to earlier nodes,
this is sufficient. The behavior may slightly differ from the earlier
behavior because we cannot short-circuit costs to zero once a node is
elaborated; but in practice this should not matter.

* Make elaboration non-recursive.

Use an explicit stack instead (with `ElabStackEntry` entries,
alongside a result stack).

* Make elaboration traversal of the domtree non-recursive/stack-safe.

* Work analysis logic in Cranelift-side egraph glue into a general analysis framework in cranelift-egraph.

* Apply static recursion limit to rule application.

* Fix aarch64 wrt dynamic-vector support -- broken rebase.

* Topo-sort cranelift-egraph before cranelift-codegen in publish script, like the comment instructs me to!

* Fix multi-result call testcase.

* Include `cranelift-egraph` in `PUBLISHED_CRATES`.

* Fix atomic_rmw: not really a load.

* Remove now-unnecessary PartialOrd/Ord derivations.

* Address some code-review comments.

* Review feedback.

* Review feedback.

* No overlap in mid-end rules, because we are defining a multi-constructor.

* rustfmt

* Review feedback.

* Review feedback.

* Review feedback.

* Review feedback.

* Remove redundant `mut`.

* Add comment noting what rules can do.

* Review feedback.

* Clarify comment wording.

* Update `has_memory_fence_semantics`.

* Apply @jameysharp's improved loop-level computation.

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Fix suggestion commit.

* Fix off-by-one in new loop-nest analysis.

* Review feedback.

* Review feedback.

* Review feedback.

* Use `Default`, not `std::default::Default`, as per @fitzgen

Co-authored-by: Nick Fitzgerald <fitzgen@gmail.com>

* Apply @fitzgen's comment elaboration to a doc-comment.

Co-authored-by: Nick Fitzgerald <fitzgen@gmail.com>

* Add stat for hitting the rewrite-depth limit.

* Some code motion in split prelude to make the diff a little clearer wrt `main`.

* Take @jameysharp's suggested `try_into()` usage for blockparam indices.

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Take @jameysharp's suggestion to avoid double-match on load op.

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Fix suggestion (add import).

* Review feedback.

* Fix stack_load handling.

* Remove redundant can_store case.

* Take @jameysharp's suggested improvement to FuncEGraph::build() logic

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Tweaks to FuncEGraph::build() on top of suggestion.

* Take @jameysharp's suggested clarified condition

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Clean up after suggestion (unused variable).

* Fix loop analysis.

* loop level asserts

* Revert constant-space loop analysis -- edge cases were incorrect, so let's go with the simple thing for now.

* Take @jameysharp's suggestion re: result_tys

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Fix up after suggestion

* Take @jameysharp's suggestion to use fold rather than reduce

Co-authored-by: Jamey Sharp <jamey@minilop.net>

* Fixup after suggestion

* Take @jameysharp's suggestion to remove elaborate_eclass_use's return value.

* Clarifying comment in terminator insts.

Co-authored-by: Jamey Sharp <jamey@minilop.net>
Co-authored-by: Nick Fitzgerald <fitzgen@gmail.com>
This commit is contained in:
Chris Fallin
2022-10-11 18:15:53 -07:00
committed by GitHub
parent e2f1ced0b6
commit 2be12a5167
59 changed files with 5125 additions and 1580 deletions
Download Patch File Download Diff File Expand all files Collapse all files

134
cranelift/codegen/src/opts/cprop.isle Normal file
View File

@@ -0,0 +1,134 @@
;; Constant propagation.
(rule (simplify
(iadd (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_add k1 k2)))))
(rule (simplify
(isub (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_sub k1 k2)))))
(rule (simplify
(imul (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_mul k1 k2)))))
(rule (simplify
(sdiv (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(if-let d (u64_sdiv k1 k2))
(subsume (iconst ty (imm64 d))))
(rule (simplify
(udiv (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(if-let d (u64_udiv k1 k2))
(subsume (iconst ty (imm64 d))))
(rule (simplify
(bor (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_or k1 k2)))))
(rule (simplify
(band (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_and k1 k2)))))
(rule (simplify
(bxor (fits_in_64 ty)
(iconst ty (u64_from_imm64 k1))
(iconst ty (u64_from_imm64 k2))))
(subsume (iconst ty (imm64 (u64_xor k1 k2)))))
(rule (simplify
(bnot (fits_in_64 ty)
(iconst ty (u64_from_imm64 k))))
(subsume (iconst ty (imm64 (u64_not k)))))
;; Canonicalize via commutativity: push immediates to the right.
;;
;; (op k x) --> (op x k)
(rule (simplify
(iadd ty k @ (iconst ty _) x))
(iadd ty x k))
;; sub is not commutative, but we can flip the args and negate the
;; whole thing.
(rule (simplify
(isub ty k @ (iconst ty _) x))
(ineg ty (isub ty x k)))
(rule (simplify
(imul ty k @ (iconst ty _) x))
(imul ty x k))
(rule (simplify
(bor ty k @ (iconst ty _) x))
(bor ty x k))
(rule (simplify
(band ty k @ (iconst ty _) x))
(band ty x k))
(rule (simplify
(bxor ty k @ (iconst ty _) x))
(bxor ty x k))
;; Canonicalize via associativity: reassociate to a right-heavy tree
;; for constants.
;;
;; (op (op x k) k) --> (op x (op k k))
(rule (simplify
(iadd ty (iadd ty x k1 @ (iconst ty _)) k2 @ (iconst ty _)))
(iadd ty x (iadd ty k1 k2)))
;; sub is not directly associative, but we can flip a sub to an add to
;; make it work:
;; - (sub (sub x k1) k2) -> (sub x (add k1 k2))
;; - (sub (sub k1 x) k2) -> (sub (sub k1 k2) x)
;; - (sub (add x k1) k2) -> (sub x (sub k2 k1))
;; - (add (sub x k1) k2) -> (add x (sub k2 k1))
;; - (add (sub k1 x) k2) -> (sub (add k1 k2) x)
(rule (simplify (isub ty
(isub ty x (iconst ty (u64_from_imm64 k1)))
(iconst ty (u64_from_imm64 k2))))
(isub ty x (iconst ty (imm64 (u64_add k1 k2)))))
(rule (simplify (isub ty
(isub ty (iconst ty (u64_from_imm64 k1)) x)
(iconst ty (u64_from_imm64 k2))))
(isub ty (iconst ty (imm64 (u64_sub k1 k2))) x))
(rule (simplify (isub ty
(iadd ty x (iconst ty (u64_from_imm64 k1)))
(iconst ty (u64_from_imm64 k2))))
(isub ty x (iconst ty (imm64 (u64_sub k1 k2)))))
(rule (simplify (iadd ty
(isub ty x (iconst ty (u64_from_imm64 k1)))
(iconst ty (u64_from_imm64 k2))))
(iadd ty x (iconst ty (imm64 (u64_sub k2 k1)))))
(rule (simplify (iadd ty
(isub ty (iconst ty (u64_from_imm64 k1)) x)
(iconst ty (u64_from_imm64 k2))))
(isub ty (iconst ty (imm64 (u64_add k1 k2))) x))
(rule (simplify
(imul ty (imul ty x k1 @ (iconst ty _)) k2 @ (iconst ty _)))
(imul ty x (imul ty k1 k2)))
(rule (simplify
(bor ty (bor ty x k1 @ (iconst ty _)) k2 @ (iconst ty _)))
(bor ty x (bor ty k1 k2)))
(rule (simplify
(band ty (band ty x k1 @ (iconst ty _)) k2 @ (iconst ty _)))
(band ty x (band ty k1 k2)))
(rule (simplify
(bxor ty (bxor ty x k1 @ (iconst ty _)) k2 @ (iconst ty _)))
(bxor ty x (bxor ty k1 k2)))
;; TODO: fadd, fsub, fmul, fdiv, fneg, fabs