Summary:
- [python] decode strings coming from `os.*` commands
- [python] decode strings coming from the command-line
- [python] encode a few remaining unicodes into strings
- [java] replace lex/yacc parser for javac verbose output by regex-based matching to handle unicode in paths
- [make] random fix of `make test` to have `make clean test` work
- [integration tests] add e2e build integration tests for utf8 in the PWD
Closes#76
Reviewed By: martinoluca
Differential Revision: D3240809
fb-gh-sync-id: 8c2e1ed
fbshipit-source-id: 8c2e1ed
Summary:
Results of AbsInt checkers are node id -> abstract state maps.
It's hard to compare/combine the results of multiple analyses if the node id types are different.
Needed for the upcoming improvements of the preanalysis.
Reviewed By: jvillard
Differential Revision: D3235669
fb-gh-sync-id: c5251cf
fbshipit-source-id: c5251cf
Summary:
The case where the right hand side of the `Letderef` expression is an identifier was missing. With this diff, the following example is now working as expected:
class A {
public Object foo() {
return new Object();
}
}
class B extends A {
public Object foo() {
return null;
}
}
public class Test {
static Object bar(A a) {
return a.foo();
}
static void shoulReport() {
B b = new B();
bar(b).toString();
}
}
using the command:
INFER_LAZY_DYNAMIC_DISPATCH=1 infer -- javac Test.java
Reviewed By: sblackshear
Differential Revision: D3238986
fb-gh-sync-id: d6059fb
fbshipit-source-id: d6059fb
Summary:
The philosophy of the tracing mode reporting is to not report the errors in a method if reaching this error does depend on information that can be false at call site. Typically with:
void foo(Object obj, int x) {
if (x == 3) {
obj.toString();
}
}
it may be that we always call `foo` with a non-null parameter or `x != 3`.
Thechnically, the reporting code matches the pairs of the form (precondition, error) and filtering out the cases where the precondtions was not imposing constraints on the calling context, and report the other cases. So the NPE could be reported in the following case:
void bar() {
foo(null, 3);
}
However, we were missing the case where there was anyway no way to call a method in a safe way, i.e. all the preconditions were of the form: (precondition, error), for example:
void baz(boolean b) {
if (b) {
foo(null, 3);
} else {
foo(null, 3);
}
}
In that case, the summary is of the form
PRE (1): b = false
POST: NullPointerException
PRE (2): b = true
POST: NullPointerException
In which case it is legit to report `NullPointerException` in `baz`.
Reviewed By: sblackshear, jberdine
Differential Revision: D3220501
fb-gh-sync-id: 7fb7d70
fbshipit-source-id: 7fb7d70
Summary: So that we no longer have to run `Tabulation.prop_is_exn` before running `Tabulation.prop_get_exn_name`.
Reviewed By: jberdine
Differential Revision: D3222545
fb-gh-sync-id: a7faa06
fbshipit-source-id: a7faa06
Summary:
As suggested in the discussion https://github.com/facebook/infer/issues/326 this pull request implements
```ocaml
get_overriden_method : Tenv.t -> Procname.java -> Procname.t
```
to get the method of a superclass that is being overridden by a specific java pname.
I thought of unit test this, but unfortunately I wasn't able to figure out how to create the proper context with OUnit2. Perhaps the easiest way to test this will be integration tests.
Feel free to reject the pull request if unit tests are mandatory (or for any other reason, of course).
Closes https://github.com/facebook/infer/pull/341
Reviewed By: jeremydubreil
Differential Revision: D3221254
Pulled By: sblackshear
fb-gh-sync-id: 9c26258
fbshipit-source-id: 9c26258
Summary: Example of dynamic dispatch with interfaces were already working. Adding some tests now so that we don't break this.
Reviewed By: sblackshear
Differential Revision: D3220360
fb-gh-sync-id: 11395dd
fbshipit-source-id: 11395dd
Summary:
I ran perf on rocksdb analysis and found out that ~40% of time is spent inside ocaml GC originating
from Prop.typ_normalize.
After this change, profile shows that GC is ~2% and Prop.typ_normalize takes 50% of the time.
Reviewed By: jberdine
Differential Revision: D3219113
fb-gh-sync-id: 27c34d9
fbshipit-source-id: 27c34d9
Summary: Provide possibility to replace clang internal headers path if they are overwritten by `-isystem`. User needs to specify path to wrong headers to be replaced with infer's clang.
Reviewed By: martinoluca
Differential Revision: D3212850
fb-gh-sync-id: be3d51c
fbshipit-source-id: be3d51c
Summary:Open-source clang has caught up a bit with apple's clang, so we don't need to
filter as many compilation flags as we used to.
Reviewed By: akotulski, martinoluca
Differential Revision: D3212553
fb-gh-sync-id: 5638dc8
fbshipit-source-id: 5638dc8
Summary:This enables controlling the encoding chosen by infer via the usual environment
variables. For instance:
```
LC_ALL="C" infer ... # sets LOCALE to "ascii"
LC_ALL="en_US.UTF-8" infer ... # sets LOCALE to "UTF-8"
```
This gives an easy solution to #320: run `LC_ALL="en_US.UTF-8" infer ...`.
Right now the only solution is to edit the Python scripts by hand instead!
Reviewed By: jberdine
Differential Revision: D3207573
fb-gh-sync-id: 62d5b98
fbshipit-source-id: 62d5b98
Summary:stdlibc++ headers didn't like the fact that hash<unique_ptr> didn't have defined operator() directly.
Do that and provide empty body. Keep inheritance in case it helps compilation to succeed.
Reviewed By: dulmarod
Differential Revision: D3207721
fb-gh-sync-id: 8c950da
fbshipit-source-id: 8c950da
Summary:Local variable created by conditional operator translation is now declared in scope of whole
procedure. Semantically there is no difference, hopefuly backend will not complain about this
change. Also, nullifying that variable is deferred to preanalysis instead of calling it manually
Reviewed By: jvillard
Differential Revision: D3155733
fb-gh-sync-id: 6cec8fc
fbshipit-source-id: 6cec8fc
Summary:`exc.output` can be (always is?) `None`. Other places in the code only print
using `traceback.print_exc()` and this was the only place trying to print this
extra info (`git grep CalledProcessError`). This caused `utils.stdout()` to
raise an exception, which was further confusing.
closes#330
Reviewed By: jberdine
Differential Revision: D3203896
fb-gh-sync-id: d2988d8
fbshipit-source-id: d2988d8
Summary:It turns out, apple clang turns off cxx-modules under the hood. Open source clang doesn't do it by default
and we need to do it ourselves.
Reviewed By: jvillard, martinoluca
Differential Revision: D3201604
fb-gh-sync-id: 82cea0f
fbshipit-source-id: 82cea0f
Summary: For performance critical sections of the code, this checker detects memory allocations or calls to methods annotated as expensive. However, such cases of memory allocations or expensive calls are acceptable is occuring in rare cases. This diff adds supports for the "unlikely" branch prediction method and does not track expensive calls in unlikely branches.
Reviewed By: sblackshear
Differential Revision: D3193473
fb-gh-sync-id: ea87e49
fbshipit-source-id: ea87e49
Summary:InferPrint has a special case code path that does not add results_dir to
the specs_library if it is the default. This seems to be unnecessary.
Reviewed By: jeremydubreil
Differential Revision: D3195088
fb-gh-sync-id: 67e968a
fbshipit-source-id: 67e968a
Summary:BinaryConditionalOperator should evaluate condition expression once, but we used to evaluate it twice.
Fix translation to account for it.
Reviewed By: dulmarod
Differential Revision: D3179803
fb-gh-sync-id: a801a7e
fbshipit-source-id: a801a7e
Summary:Improve the error traces so that:
- the error get reported on the first offending call, which is more intuitive for inline comments
- the traces now jump from call location to callee definition and so forth until the end of the call stack
Reviewed By: sblackshear
Differential Revision: D3183756
fb-gh-sync-id: 089ddaf
fbshipit-source-id: 089ddaf
Summary:This diff translate cpp lambdas. For the moment it does not take care of
captured variables. Captured variables will come in the next diff.
Reviewed By: dulmarod
Differential Revision: D3114790
fb-gh-sync-id: bf36450
fbshipit-source-id: bf36450
Summary:Make node ids be `private int` to make sure we don't mix them with random
integers from other sources.
Reviewed By: sblackshear, cristianoc
Differential Revision: D3179670
fb-gh-sync-id: 4bcf4f0
fbshipit-source-id: 4bcf4f0
Summary:This wasn't used anywhere. Frontends that wish to do something like goto can
just set the targets of the goto as successors of the current node, no need for
a special instruction to do that.
Reviewed By: sblackshear
Differential Revision: D3179826
fb-gh-sync-id: 572a6f2
fbshipit-source-id: 572a6f2
Summary:public
When a conditional is the last instruction, there will be a join node leading directly to the exit node.
Some instructions, such as nullification of dead variables, and abstraction, are added to the control flow graph automatically. But, join nodes cannot contain instructions. So when a procedure ends with a conditional, there might be no place to store these instructions.
This diff adds one extra node between the join and the exit node in that situation.
Reviewed By: jvillard
Differential Revision: D3179056
fb-gh-sync-id: 2b9cd7e
fbshipit-source-id: 2b9cd7e
Summary:public
This extends infer/src/Makefile with a mod_dep.dot target that builds a
dot graph of module dependencies.
This also adds ocamldot, which is in the public domain and available
from http://trevorjim.com/projects/ocamldot/ocamldot.tar .
Reviewed By: cristianoc
Differential Revision: D3168488
fb-gh-sync-id: 267fb0e
fbshipit-source-id: 267fb0e
Summary:public
The code:
DataInputStream in = new DataInputStream(new BufferedInputStream(new FileInputStream(file)));
creates a resource with `FileInputStream()` and wraps it twice as a field of `BufferedInputStream` and then as a field of `DataInputStream`. Then calling:
in.close();
needs to go down the wrappers hierachy: `DataInputStream.close()` -> `FilterInputStream.close()` which then calls `BufferedInputStream.close()` -> `FilterInputStream.close()` -> `FileInputStream.close()`.
Going down the wrapper was not working before because `FilterInputStream.close()` was only going further when the type of field `in` was `FileInputStream` wheras it should also continue when the type of the field is any subtype of `FilterInputStream`, e.g. `DataInputStream` and `BufferedInputStream` like in the test example. This diff fixes this last aspect.
Reviewed By: sblackshear
Differential Revision: D3174822
fb-gh-sync-id: 3adbb7e
fbshipit-source-id: 3adbb7e
Summary:public
Report statistics on consumed time and memory in results_dir/perf_stats.json.
Reviewed By: martinoluca
Differential Revision: D3162381
fb-gh-sync-id: e802faa
fbshipit-source-id: e802faa
Summary:public
Instead of translating code from headers blindly, translate only gets transitively referenced from source code.
It won't translate functions from system headers, but in the future we could do that as well
since most of them aren't used and it shouldn't add much overhead.
For now this functionality is hidden behind --cxx-experimental flag
Reviewed By: dulmarod
Differential Revision: D3163519
fb-gh-sync-id: 0c53b10
fbshipit-source-id: 0c53b10
Summary:public
Refactor Utils.SymOp into a separate module, bringing the failure_kind
type and associated operations.
Reviewed By: cristianoc
Differential Revision: D3161640
fb-gh-sync-id: be3d7c9
fbshipit-source-id: be3d7c9
Summary:public
Eliminate the use of the -open Utils command line option passed to the compiler in favor of `open! Utils` in each source file. While slightly convenient, this option causes more headaches than it is worth with other tools e.g. merlin.
Reviewed By: jvillard
Differential Revision: D3168193
fb-gh-sync-id: 4285ef6
fbshipit-source-id: 4285ef6
Summary:public
This will allow us to run translation of more than one function at the same time.
Reviewed By: dulmarod
Differential Revision: D3167965
fb-gh-sync-id: 41e9935
fbshipit-source-id: 41e9935
Summary:public
Also build checkcopyright in the byte code test build.
Reviewed By: martinoluca
Differential Revision: D3161885
fb-gh-sync-id: cf07a59
fbshipit-source-id: cf07a59
Summary:public
This information is no longer needed - ASTExporter always exports pointer to parent for C++/objC methods.
This code is from time it was not true.
Reviewed By: jvillard
Differential Revision: D3162455
fb-gh-sync-id: 35570da
fbshipit-source-id: 35570da
Summary:public
Refactor the ml_loc type and associated operations from Utils to Logging. Seems a better fit, and reduces dependencies.
Reviewed By: cristianoc
Differential Revision: D3161440
fb-gh-sync-id: 2e09c25
fbshipit-source-id: 2e09c25
Summary:public
Instead of using location of init_stmt, use location of variable when translating initialization.
Most of the time it change anything with some exceptions:
// example1 - C/C++/objC
int x = // now: assignment happens in this line
3; // past: assignment happens in this line
// example2: valid in C++11 only
struct X {
int x = 0; // now: one assignment here
int y = 2; // now: one assigmnent here
X() = default; // before: 2 assignments in this line
};
Reviewed By: dulmarod
Differential Revision: D3155870
fb-gh-sync-id: f38c78c
fbshipit-source-id: f38c78c
Summary:public
The merge option -m would link every file in a captured results dir.
This includes the .start file, which would compromise the subsequent checks for modified files during reactive analysis.
Now only link files inside directories inside the results dir (don't link any files in the first 2 levels).
Reviewed By: jberdine
Differential Revision: D3155819
fb-gh-sync-id: 8ad180f
fbshipit-source-id: 8ad180f
Summary:public
This may be unnecessary, but I found stale files while debugging and had to ask if they were the problem.
Reviewed By: jvillard
Differential Revision: D3130087
fb-gh-sync-id: d3222c3
fbshipit-source-id: d3222c3
Summary:public
Add modeling of methods that can raise exceptions when parameters are
nil, and that return nil when passed nil.
Reviewed By: dulmarod
Differential Revision: D3101739
fb-gh-sync-id: 76af5a2
fbshipit-source-id: 76af5a2
Summary:public
Use getconf to compute NCPU as it is uniform across platforms.
Use result in several places to avoid fork bombs. I routinely see
errors about fork running out of memory when building e.g. the clang
plugin.
Reviewed By: jvillard
Differential Revision: D3148970
fb-gh-sync-id: 6d071c9
fbshipit-source-id: 6d071c9
Summary:public
Remove option to filter source files from being analysed from clang wrapper.
The option is unused and there should be better mechanism to do so if we need it.
This will slightly simplify the script and allow to analyze files that are from
outside of "blessed" extensions (such as .cxx or .hpp)
Reviewed By: dulmarod
Differential Revision: D3126330
fb-gh-sync-id: 1628913
fbshipit-source-id: 1628913
Summary:public
clang-format changed filename inside __has_include(<FILENAME>), turn off
clang format for that part of the code.
Reviewed By: jberdine
Differential Revision: D3133593
fb-gh-sync-id: c601514
fbshipit-source-id: c601514
Summary:public
In order to make infer more resiliant to compilation failure,
make static_assert to do nothing.
As a bonus, set _FORTIFY_SOURCE inside same file instead of command line
Reviewed By: jvillard
Differential Revision: D3133446
fb-gh-sync-id: 590f4ad
fbshipit-source-id: 590f4ad
Summary:public
Will be needed later to avoid circular dependencies between dom.ml and upcoming
numericalDomain.ml.
Reviewed By: jberdine
Differential Revision: D3126697
fb-gh-sync-id: 678d49f
fbshipit-source-id: 678d49f
Summary:public
Models in headers are purely C++ concept and it's
useless if C++ mode is turned off. Since those
models can lead to compilation errors, turn them
off for non-C++ analysis
Reviewed By: martinoluca
Differential Revision: D3126294
fb-gh-sync-id: 0912e7b
fbshipit-source-id: 0912e7b
Summary:public
This fixes the compilation of Infer from a path containing symbolic links.
Reviewed By: jberdine
Differential Revision: D3123244
fb-gh-sync-id: 09dbd13
fbshipit-source-id: 09dbd13
Summary:public
1. Make detection of libc++/stdlibc++ headers more robust
2. Turn on c++11 only for newer versions of stdlibc++
Reviewed By: jvillard
Differential Revision: D3121187
fb-gh-sync-id: c2e5be5
fbshipit-source-id: c2e5be5
Summary:public
It could happen that the modification times of symbolic links and
original files checked by reactive mode could be exactly the same. In
this case, some files would perpetually be re-analyzed.
This diff creates symlinks and sets their accessed and modified times
to 1 second in the future so that strict timestamp checking is robust.
Reviewed By: cristianoc
Differential Revision: D3098451
fb-gh-sync-id: 3724468
fbshipit-source-id: 3724468
Summary:public It's useful for the transfer functions to understand what the current procedure is. Applications include debugging, implementing context-sensitivity, asking which program variables are parameters, and the list goes on.
In the future, we'll almost certainly want to pass the tenv to the transfer functions as well.
Reviewed By: jberdine
Differential Revision: D3104997
fb-gh-sync-id: 1c0df8f
fbshipit-source-id: 1c0df8f
Summary:public
This create `make install` and `make uninstall` targets that can be used to
install Infer on a system. All the files needed are installed under
"/usr/local/lib/infer" (by default, but can be configured with ./configure
flags as usual), pretty much like their git infer/ counterparts. Moreover,
"infer" and "inferTraceBugs" are installed in /usr/local/bin (similarly
configurable).
As a bonus, `opam install infer` and `opam remove infer` now work. To use
these, it's better to pin the git repo with "-k git":
cd ~/infer/ # where the repo is
opam pin add -k git --yes infer .
The reason is that otherwise, and only if the repo is not clean, opam will
convert symlinks into regular files, which is not good (compilation will fail).
Reviewed By: jberdine
Differential Revision: D3109337
fb-gh-sync-id: 96c9fd1
fbshipit-source-id: 96c9fd1
Summary:public
This is needed to have a working `opam install infer`. Actual working `opam
install infer` in a follow-up diff.
Reviewed By: jberdine
Differential Revision: D3109308
fb-gh-sync-id: 49d7276
fbshipit-source-id: 49d7276
Summary:public
Simplifies the way to configure cmake to run infer's fake compiler commands.
Instead of `CC=/path/to/infer/infer/lib/capture/clang cmake .`, which is what
is advised on #25 and is now outdated because the paths to infer's clang has
changed since, simply run `infer -- cmake .`. The only caveat is that infer
tries to analyze the end result, and prints "No issues found". This could be
fixed later.
Reviewed By: jeremydubreil
Differential Revision: D3093162
fb-gh-sync-id: 99df50a
fbshipit-source-id: 99df50a
Summary:public
Rename functions and arguments to be more uniform, and change several to simplify types by using Builtin.t.
Reviewed By: cristianoc
Differential Revision: D3107836
fb-gh-sync-id: 8445f79
fbshipit-source-id: 8445f79
Summary:public
Allow the use of `infer -- ./configure`. This can be useful if the full path to
the compiler is recorded by the `./configure` command. That is the case for the
samba source tree for instance (because `./configure` calls `waf configure`
under the hood).
Reviewed By: jeremydubreil
Differential Revision: D3093065
fb-gh-sync-id: 2663418
fbshipit-source-id: 2663418
Summary:public
This will allow SymExec to depend on Inferconfig with introducing
circular dependencies, as Inferconfig calls Builtin.is_registered.
Reviewed By: jeremydubreil
Differential Revision: D3100614
fb-gh-sync-id: 786cf62
fbshipit-source-id: 786cf62
Summary:public
To ensure that e.g. Config.results_dir is not used before being set.
Reviewed By: cristianoc
Differential Revision: D3102724
fb-gh-sync-id: ced0974
fbshipit-source-id: ced0974
Summary:public
Eradicate need the procedure attributes for callees.
It relies on the java front-end to create proc descs for callees that are declared but not defined.
This diff remove that needs, and when a callee without prodedure attributes is found, it creates one on the fly. The attribute created is similar to what the Java front-end would do, except
that the number and types of arguments are part of the call instruction, so they can
be used to create the formal parameters.
Reviewed By: jeremydubreil
Differential Revision: D3073904
fb-gh-sync-id: 381ff67
fbshipit-source-id: 381ff67
Summary:public
Allow the value of the INFER_ONDEMAND_FILE environment variable to be
either an absolute path, or relative to the project root.
Likewise, allow the entries in the file determined by
INFER_ONDEMAND_FILE to be either absolute or relative to the project
root. Note, however, that if they are absolute but not under the
project root, they will not be found.
Reviewed By: cristianoc
Differential Revision: D3098489
fb-gh-sync-id: e861300
fbshipit-source-id: e861300
Summary:public
In Python 2, `shutil.rmtree()`, `os.walk()`, `os.path.join()`, etc. are not
happy when the locale cannot decode the filenames they have to deal with.
Decrease the likelihood of this happening by making the file names generated by
infer ascii-only.
Also ignore character decoding errors optimistically when reading the json
report file.
Add tests that we are able to run the analysis and report the bug on a function
with a utf8 name, and that we are able to remove the previous results
directory.
closes#287
Reviewed By: cristianoc
Differential Revision: D3058858
fb-gh-sync-id: b88cd35
shipit-source-id: b88cd35
Summary:public
see the file for some example usage
Reviewed By: sblackshear, jeremydubreil
Differential Revision: D3058440
fb-gh-sync-id: c891dfe
shipit-source-id: c891dfe
Summary:public
D2987288 introduced a regression where usage messages were no longer printed, and instead exceptions went uncaught.
Reviewed By: jeremydubreil
Differential Revision: D3089620
fb-gh-sync-id: 1cca0c6
shipit-source-id: 1cca0c6
Summary:public
It seems that creating the procedure description of the callees this should no longer be useful to run the checkers
Reviewed By: sblackshear
Differential Revision: D3083523
fb-gh-sync-id: 040b104
shipit-source-id: 040b104
Summary:public
The "fake" procedure description of the callees, containing info about the formal parameters, is current used by Eradicate. This diff forces the creation of the callee procedure descriptions when running Infer with Buck.
Reviewed By: sblackshear
Differential Revision: D3083452
fb-gh-sync-id: 24a70e6
shipit-source-id: 24a70e6
Summary:public
It seems that the support for using the Buck cache with Infer has been brocken for a while.
Reviewed By: sblackshear
Differential Revision: D3079493
fb-gh-sync-id: fd92d1c
shipit-source-id: fd92d1c
Summary:public
Conversion constructor of unique_ptr should be explicit. Fix it.
Reviewed By: cristianoc
Differential Revision: D3075661
fb-gh-sync-id: e911d8c
shipit-source-id: e911d8c
Summary:public
The option -merge_captured specifies that this is merging the results of capture
using the buck integration. The file specifying the targets is read from `infer-out/infer-deps.txt`.
Each dependency specifies a path in buck-out for one target:
where the results directory after capture is.
The option triggers a merge of the results directories into infer-out.
The merge consists in making a virtual copy, where each file in a
target in `buck-out` gets virtually copied into infer-out by making
one symbolic link per file.
There is a mechanism to detect when the capture of a target has already
been copied: when each source file already exists at the destination.
There's also an option `-modified_targets modified_targets.txt`.
If a target is listed in `modified_targets.txt`, this will force a new creation of links
for that target, whether those links exist or not.
Reviewed By: martinoluca
Differential Revision: D3070318
fb-gh-sync-id: 6d2e7a5
shipit-source-id: 6d2e7a5
Summary:public
TOgether with compiling Infer in debug mode, this allows to get exceptions stack traces when Infer fails
Reviewed By: sblackshear
Differential Revision: D2899992
fb-gh-sync-id: 55b4d3d
shipit-source-id: 55b4d3d
Summary:public
Create a model of std::unique_ptr in similar fashion to what was done to std::shared_ptr.
For now, we are modeling it as container of raw pointer (no ownership concept).
This time unique_ptr is not derived from std__unique_ptr (unlike shared_ptr, it was easier to not do that) and so we need to provide implementations for all non-member functions per C++ reference:
http://en.cppreference.com/w/cpp/memory/unique_ptr
Reviewed By: dulmarod
Differential Revision: D3048209
fb-gh-sync-id: a9a6455
shipit-source-id: a9a6455
Summary:public
Whenever infer-deps.txt and report.json files are encountered after the analysis with Buck
they will be automatically merged and deduplicated with all the other files of the same kind.
This change also emits the results of the analysis to stdout.
Reviewed By: jvillard
Differential Revision: D3064487
fb-gh-sync-id: 3599fba
shipit-source-id: 3599fba
Summary:public
Before this diff, the Java frontend was not adding the definition of the inherited interfaces to the type environment, thus failing to answer questions like "does type X implements Closeable". Infer was therefore missing to detect resource leaks when the resource was indirectly implementing Closeable via an intermediate interface.
Reviewed By: sblackshear
Differential Revision: D3067555
fb-gh-sync-id: 86d0760
shipit-source-id: 86d0760
Summary: public Like the pre-analysis, these should be called for every CFG. This is a stepping stone toward getting rid of remove_tmps and making it part of the liveness analysis.
Reviewed By: jvillard
Differential Revision: D3059021
fb-gh-sync-id: 3a8d818
shipit-source-id: 3a8d818
Sam Blackshear
Jules Villard
Dulma Churchill
Jeremy Dubreil
Luis Cruz
Andrzej Kotulski
Dino Distefano
Peter O'Hearn
Josh Berdine
Cristiano Calcagno
martinoluca