| Age | Commit message (Collapse) | Author | Lines |
|
|
|
|
|
Add `-Z dump-mir-dataflow` flag for dumping dataflow results visualization
Previously, to visualize the results of a MIR dataflow pass, one had to add a `#[rustc_mir(borrowck_graphviz_postflow)]` attribute to functions of interest. However, there is no way to specify this attribute on closures and generators, so it was impossible to view results for these MIR bodies.
This PR adds a flag, `-Z dump-mir-dataflow`, which will output the dataflow results for any functions specified in `-Z dump-mir` to the output directory specified by `-Z dump-mir-dir`. This behavior is modeled on the `-Z dump-mir-graphviz` flag.
|
|
|
|
|
|
|
|
|
|
|
|
Index impl, remove body fn
rustc_codegen_ssa: Fix BodyAndCache reborrow to Body and change instances of body() call to derefence
rustc_mir: Fix BodyAndCache reborrow to Body and change intances of body() call to derefence
|
|
rename all body_cache back to body
|
|
|
|
(lifetime errors still exist)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The commit should have changed comments as well.
At the time of writting, it passes the tidy and check tool.
Revisions asked by eddyb :
- Renamed of all the occurences of {visit/super}_mir
- Renamed test structures `CachedMir` to `Cached`
Fixing the missing import on `AggregateKind`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Extend `dump_mir` and functions it calls in order to allow callers to
add custom information. We do this by adding an enum `PassWhere` and
an extra argument of type `FnMut(PassWhere, &mut Write) ->
io::Result<()>`. This callback is responsible for printing the extra
information when MIR is dumped at various stages.
For the "nll" pass, use the new mechanism to dump the `Region`
information after the header, but before the control flow graph for
every function.
In the interest of keeping the output somewhat concise, implement
a custom Debug impl for `Region`
Open Questions:
* What should we call what has been called `PassWhere` so far?
|
|
|
|
This involves changing various details about that system,
though the basic shape remains the same.
|
|
This is a more principled version of the `RefCell` we were using
before. We now allocate a `Steal<Mir<'tcx>>` for each intermediate MIR
pass; when the next pass steals the entry, any later attempts to use it
will panic (there is no way to *test* if MIR is stolen, you're just
supposed to *know*).
|
|
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.
|
|
this temporary disables `inline`
|
|
|
|
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).
|
|
|
|
Each MIR key is a DefId that has MIR associated with it
|
|
|
|
|
|
This reduces the peak RSS for a cut-down version of the program
in #36799 by 10%, from 951MB to 856MB.
|
|
|
|
|
|
|
|
|
|
the *only* place where critical edges need to be broken is on Call
instructions, so only break them there.
|