Commit Graph

15 Commits

Author SHA1 Message Date
Jakob Stoklund Olesen
03dee5e442 Replace some uses of layout::Cursor with FuncCursor.
The layout::Cursor is unfortunate because it doesn't reference the whole
function.
2017-09-21 11:27:15 -07:00
Dan Gohman
74845fed6c Use more helper functions on CursorBase to reduce cursor repositioning. 2017-09-20 11:19:35 -07:00
Dan Gohman
b888894fbb Use CursorBase utility functions to reduce repositioning clutter. 2017-09-20 09:06:17 -07:00
Dan Gohman
bbe056bf9d Make passes assert their dependencies consistently. (#156)
* Make passes assert their dependencies consistently.

This avoids ambiguity about whose responsibility it is to run
to compute cfg, domtree, and loop_analysis data.

* Reset the `valid` flag in DominatorTree's `clear()`.

* Remove the redundant assert from DominatorTree::with_function.

* Remove the message strings from obvious asserts.

This avoids having them spill out into multiple lines.

* Refactor calls to `compute` on `Context` objects into helper functions.
2017-09-14 14:38:53 -07:00
Dan Gohman
e2f0fe58d1 Move logic out of cton-util wasm.
Give LoopAnalysis `is_valid` and `ensure` functions similar to
DominatorTree and others, so that it can be computed on demand in the
same way.

This removes the last need for src/wasm.rs to have embedded knowledge
of the dependencies of the passes it's running.
2017-09-12 16:32:47 -07:00
Dan Gohman
9d0d6b5a1b Move verify calls out of Context.
Also, move flowgraph() calls out of filetest and into the passes that
need them so that filetest doesn't have embedded knowledge of these
dependencies.

This resolves a TODO about the way Context was running the verifier, and
it makes the Context functions and the filetest runners more transparent.

This also fixes simple_gvn to use the existing dominator tree rather
than computing its own.
2017-09-12 16:32:47 -07:00
Dan Gohman
69f8174c03 Move ensure_domtree out of Context and into DominatorTree.
This also moves the calls to it out of Context and into the passes that
actually need it, so that Context's functions don't have any logic of
their own.
2017-09-12 16:32:47 -07:00
Dan Gohman
de02976162 Minor comment cleanups. 2017-08-31 17:02:26 -07:00
Dan Gohman
2efdc0ed37 Update rustfmt to 0.9.0. 2017-08-31 10:44:59 -07:00
Jakob Stoklund Olesen
9cb0529be4 Move EntityList and SparseMap into the entity module.
These data structures are dependent on EntityRef and EntityMap, so it
makes sense to keep them in the same module.
2017-08-18 16:14:06 -07:00
Jakob Stoklund Olesen
dba0df787c Move most Cursor methods into a CursorBase trait.
The Cursor navigation methods all just depend on the cursor's position
and layout reference. Make a CursorBase trait that provides access to
this information with methods and implement the navigation methods on
top of that.

This makes it possible to have multiple types implement the cursor
interface.
2017-08-04 09:08:24 -07:00
Dan Gohman
0c7316ae28 Lint fixes (#99)
* Replace a single-character string literal with a character literal.

* Use is_some() instead of comparing with Some(_).

* Add code-quotes around type names in comments.

* Use !...is_empty() instead of len() != 0.

* Tidy up redundant returns.

* Remove redundant .clone() calls.

* Remove unnecessary explicit lifetime parameters.

* Tidy up unnecessary '&'s.

* Add parens to make operator precedence explicit.

* Use debug_assert_eq instead of debug_assert with ==.

* Replace a &Vec argument with a &[...].

* Replace `a = a op b` with `a op= b`.

* Avoid unnecessary closures.

* Avoid .iter() and .iter_mut() for iterating over containers.

* Remove unneeded qualification.
2017-06-19 16:24:10 -07:00
Jakob Stoklund Olesen
ea8a8a95a8 Remove the ebb_dominates function.
This is now subsumed by the generic 'dominates' function.
2017-06-12 14:59:34 -07:00
Denis Merigoux
81284dbd93 LICM pass: small changes after code review (#94) 2017-06-08 09:41:57 -07:00
Denis Merigoux
e47f4a49fb LICM pass (#87)
* LICM pass

* Uses loop analysis to detect loop tree
* For each loop (starting with the inner ones), create a pre-header and move there loop-invariant instructions
* An instruction is loop invariant if it does not use as argument a value defined earlier in the loop
* File tests to check LICM's correctness
* Optimized pre-header creation
If the loop already has a natural pre-header, we use it instead of creating a new one.
The natural pre-header of a loop is the only predecessor of the header it doesn't dominate.
2017-06-07 11:27:22 -07:00