Commit Graph

196 Commits

Author SHA1 Message Date
Ariel Ben-Yehuda
48655ea661 add a bunch of debug logging to transform::inline 2017-11-09 23:03:41 +02:00
Mikhail Modin
2d71c5f10c add TerminatorKind::FalseEdges and use it in matches 2017-11-02 09:43:36 +03:00
Douglas Campos
081b29c1f6 stop using monomorphize::resolve() 2017-09-29 22:34:48 -04:00
Douglas Campos
8a5800e1d4 take ParamEnv into account when resolving 2017-09-29 22:34:47 -04:00
Douglas Campos
24fc50fffa make it not to be a method 2017-09-29 22:34:46 -04:00
Douglas Campos
5eb302fba8 always use resolve 2017-09-29 22:34:46 -04:00
Douglas Campos
3bd09dee45 re-enable mir inlining across trait methods
this fixes #44389
2017-09-29 22:34:46 -04:00
bors
909733286f Auto merge of #44383 - qmx:gh/40473/no-inline-trait-method, r=nikomatsakis
MIR: should not inline trait method

Fixes #40473.

The idea here is bailing out of inlining if we're talking about a trait method.
2017-09-11 04:59:28 +00:00
Douglas Campos
10a5b539c7 simplify 2017-09-08 11:06:28 -04:00
bors
dead08cb33 Auto merge of #44142 - alexcrichton:dllimport-query, r=nikomatsakis
Migrate a slew of metadata methods to queries

This PR intends to make more progress on #41417, knocking off some low-hanging fruit.

Closes #44190
cc #44137
2017-09-08 09:47:58 +00:00
Douglas Campos
784e4316eb ugly, but works! 2017-09-06 23:06:14 -04:00
Oliver Schneider
5bb870faca Reintroduce the early returns 2017-09-06 12:25:46 +02:00
Oliver Schneider
02fb1b0b72 Fix a bug in the inliner 2017-09-06 10:33:53 +02:00
Alex Crichton
a2e2aba07d rustc: Migrate lang items to a query
This commit moves the calculation of the `LanguageItems` structure into a
query rather than being calculated before the `TyCtxt` exists, with the eventual
end goal of removing some `CrateStore` methods.
2017-09-05 07:37:57 -07:00
Eduard-Mihai Burtescu
c911925243 rustc_mir: implement visit_local instead/along visit_lvalue where possible. 2017-09-03 19:14:31 +03:00
John Kåre Alsaker
d29af37999 Merge branch 'master' of https://github.com/rust-lang/rust into gen 2017-08-28 02:41:16 +02:00
Alex Crichton
82bad957c0 rustc: Add a FIXME for try_get in MIR inlining
It sounds like this is being handled elsewhere, so for now just preserve the
existing behavior of ignoring th error.
2017-08-25 16:08:35 -07:00
Alex Crichton
1210ebff43 Merge remote-tracking branch 'origin/master' into gen 2017-08-16 10:03:47 -07:00
Zack M. Davis
1b6c9605e4 use field init shorthand EVERYWHERE
Like #43008 (f668999), but _much more aggressive_.
2017-08-15 15:29:17 -07:00
John Kåre Alsaker
df608e710c Rename suspend to yield 2017-07-28 15:46:24 +02:00
John Kåre Alsaker
d861982ca6 Generator literal support 2017-07-28 15:46:23 +02:00
Ralf Jung
66fce33ecb overload the mir ty methods to make them more ergonomic to use 2017-07-12 12:59:05 -07:00
Ralf Jung
62b2e5446d Refactor: {Lvalue,Rvalue,Operand}::ty only need the locals' types, not the full &Mir 2017-07-11 16:38:16 -07:00
Eduard-Mihai Burtescu
33ecf72e8e rustc: move the PolyFnSig out of TyFnDef. 2017-06-27 16:39:52 +03:00
Niko Matsakis
541523d10d rewrite layout to take a (param-env, ty) pair instead of infcx 2017-06-01 12:56:30 -04:00
Niko Matsakis
6c4b961eff move projection mode into parameter environment 2017-06-01 12:56:30 -04:00
Niko Matsakis
45dd0632bc rename parameter_environment to param_env 2017-05-22 15:55:56 -04:00
Niko Matsakis
226358e2fb rename ParameterEnvironment to ParamEnv 2017-05-22 15:54:16 -04:00
Niko Matsakis
1b7acb6f34 centralize the caching for is-copy, is-sized, and is-freeze
Use the trait-environment+type as the key. Note that these
are only invoked on types that live for the entire compilation
(no inference artifacts). We no longer need the various special-case
bits and caches that were in place before.
2017-05-22 14:55:03 -04:00
Eduard-Mihai Burtescu
bd8246be74 rustc: uniformly compute ParameterEnvironment's "free outlive scope". 2017-05-13 17:42:59 +03:00
Niko Matsakis
0afcfce8d8 update comment about heuristics 2017-05-02 16:21:57 -04:00
Niko Matsakis
393fa4f1b7 rename from item_mir to optimized_mir 2017-05-02 16:21:57 -04:00
Niko Matsakis
1dd9c3e52a support inlining by asking for optimizer mir for callees
I tested this with it enabled 100% of the time, and we were able to run
mir-opt tests successfully.
2017-05-02 16:21:56 -04:00
Niko Matsakis
9c154a67bf rip out everything but MirPass, move the logic into suites 2017-05-02 16:21:55 -04:00
Niko Matsakis
c253df5249 remove Pass and (temporarily) drop Inline 2017-05-02 16:21:55 -04:00
Niko Matsakis
a26e966307 convert the inline pass to use the new multi result
This involves changing various details about that system,
though the basic shape remains the same.
2017-05-02 14:01:36 -04:00
Niko Matsakis
2b32cb90c7 retool MIR passes completely
The new setup is as follows. There is a pipeline of MIR passes that each
run **per def-id** to optimize a particular function. You are intended
to request MIR at whatever stage you need it. At the moment, there is
only one stage you can request:

- `optimized_mir(def_id)`

This yields the final product. Internally, it pulls the MIR for the
given def-id through a series of steps. Right now, these are still using
an "interned ref-cell" but they are intended to "steal" from one
another:

- `mir_build` -- performs the initial construction for local MIR
- `mir_pass_set` -- performs a suite of optimizations and transformations
- `mir_pass` -- an individual optimization within a suite

So, to construct the optimized MIR, we invoke:

    mir_pass_set((MIR_OPTIMIZED, def_id))

which will build up the final MIR.
2017-05-02 14:01:01 -04:00
Niko Matsakis
46b342fbc0 simplify the MirPass traits and passes dramatically
Overall goal: reduce the amount of context a mir pass needs so that it
resembles a query.

- The hooks are no longer "threaded down" to the pass, but rather run
  automatically from the top-level (we also thread down the current pass
  number, so that the files are sorted better).
  - The hook now receives a *single* callback, rather than a callback per-MIR.
- The traits are no longer lifetime parameters, which moved to the
  methods -- given that we required
  `for<'tcx>` objecs, there wasn't much point to that.
- Several passes now store a `String` instead of a `&'l str` (again, no
  point).
2017-05-02 14:01:01 -04:00
Niko Matsakis
11b6b0663a rework MirPass API to be stateless and extract helper fns 2017-05-02 14:01:00 -04:00
Niko Matsakis
0e5e2f3634 introduce mir_keys()
Each MIR key is a DefId that has MIR associated with it
2017-05-02 14:01:00 -04:00
Ariel Ben-Yehuda
e1377a4f47 avoid calling mk_region unnecessarily
this improves typeck & trans performance by 1%. This looked hotter on
callgrind than it is on a CPU.
2017-04-22 21:00:50 +03:00
Eduard-Mihai Burtescu
0adfd810f8 rustc: combine type_needs_drop_given_env and may_drop into needs_drop. 2017-04-20 14:44:43 +03:00
Ariel Ben-Yehuda
384ec80d2b store Spans for all MIR locals 2017-04-11 23:52:51 +03:00
James Miller
3eb26d1f2b Only run inlining if mir opts are enabled 2017-03-10 03:54:38 +02:00
James Miller
f55e92b2a7 Add dep-graph tasks where needed 2017-03-10 03:54:38 +02:00
James Miller
71d0d921c2 Initial implementation of inlining for MIR
Fairly basic implementation of inlining for MIR. Uses conservative
heuristics for inlining.
2017-03-10 03:54:26 +02:00