Commit Graph

18 Commits

Author SHA1 Message Date
Lang Hames 98c2ac13d3 [Orc] Add support for emitting indirect stubs directly into the JIT target's
memory, rather than representing the stubs in IR. Update the CompileOnDemand
layer to use this functionality.

Directly emitting stubs is much cheaper than building them in IR and codegen'ing
them (see below). It also plays well with remote JITing - stubs can be emitted
directly in the target process, rather than having to send them over the wire.

The downsides are:

(1) Care must be taken when resolving symbols, as stub symbols are held in a
    separate symbol table. This is only a problem for layer writers and other
    people using this API directly. The CompileOnDemand layer hides this detail.

(2) Aliases of function stubs can't be symbolic any more (since there's no
    symbol definition in IR), but must be converted into a constant pointer
    expression. This means that modules containing aliases of stubs cannot be
    cached. In practice this is unlikely to be a problem: There's no benefit to
    caching such a module anyway.

On balance I think the extra performance is more than worth the trade-offs: In a
simple stress test with 10000 dummy functions requiring stubs and a single
executed "hello world" main function, directly emitting stubs reduced user time
for JITing / executing by over 90% (1.5s for IR stubs vs 0.1s for direct
emission).

llvm-svn: 250712
2015-10-19 17:43:51 +00:00
Duncan P. N. Exon Smith 1275bffa96 OrcJIT: Remove implicit ilist iterator conversions, NFC
llvm-svn: 250192
2015-10-13 18:10:59 +00:00
Lang Hames 44780acd91 [Orc] Teach the CompileOnDemand layer to clone aliases.
This allows modules containing aliases to be lazily jit'd. Previously these
failed with missing symbol errors because the aliases weren't cloned from the
original module.

llvm-svn: 249481
2015-10-06 22:55:05 +00:00
Lang Hames 2e36ddf671 [Orc] Fix a bug in the CompileOnDemand layer where stub decls were not cloned
into partitions. Also, add an option to clone stub definitions (not just decls)
into partitions: these definitions could be inlined in some places to avoid the
overhead of calling via the stub.

Found by inspection - no test case yet, although I plan to add a unit test for
this once the CompileOnDemand layer refactoring settles down.

llvm-svn: 239640
2015-06-12 21:31:15 +00:00
Lang Hames cd68eba3b9 [Orc] Reapply r236465 with fixes for the MSVC bots.
llvm-svn: 236506
2015-05-05 17:37:18 +00:00
Lang Hames ac31a1f141 [Orc] Revert r236465 - It broke the Windows bots.
Looks like the usual missing explicit move-constructor issue with MSVC. I should
have a fix shortly.

llvm-svn: 236472
2015-05-04 23:30:01 +00:00
Lang Hames a68970dfd5 [Orc] Refactor the compile-on-demand layer to make module partitioning lazy,
and avoid cloning unused decls into every partition.

Module partitioning showed up as a source of significant overhead when I
profiled some trivial test cases. Avoiding the overhead of partitionging
for uncalled functions helps to mitigate this.

This change also means that it is no longer necessary to have a
LazyEmittingLayer underneath the CompileOnDemand layer, since the
CompileOnDemandLayer will not extract or emit function bodies until they are
called.

llvm-svn: 236465
2015-05-04 22:03:10 +00:00
Lang Hames dc4260db2a [Orc] Make the makeStub function propagate argument attributes onto the call to
the function body.

This is necessary for correctness when lazily compiling.

Also, flesh out the Orc unit test infrastructure slightly, and add a unit test
for this.

llvm-svn: 235347
2015-04-20 20:41:45 +00:00
Lang Hames c6de458f7c [Orc] During module partitioning, rename anonymous and asm-private globals.
If they're not (re)named, these globals will fail to resolve when the
partitioned modules are linked.

llvm-svn: 234707
2015-04-12 20:05:51 +00:00
Lang Hames eb9bdb5d16 [Orc] Tidy up IndirectionUtils API a little, add some comments. NFC.
llvm-svn: 234669
2015-04-11 00:23:49 +00:00
Lang Hames 2f50744fe7 [Orc] Fix local-linkage handling in the CompileOnDemand layer.
llvm-svn: 233895
2015-04-02 05:28:10 +00:00
Benjamin Kramer 0a446fd56c Add missing includes. make_unique proliferated everywhere.
llvm-svn: 230909
2015-03-01 21:28:53 +00:00
Lang Hames d1c2082c39 [Orc] Remove redundant using directive.
llvm-svn: 230154
2015-02-22 01:48:23 +00:00
Lang Hames 53ccf88893 [Orc] Add header comment to IndirectionUtils.cpp.
llvm-svn: 230153
2015-02-22 01:45:31 +00:00
Lang Hames e738061c49 [Orc] Move Orc code into a namespace (llvm::orc), update Kaleidoscope code.
NFC.

llvm-svn: 230143
2015-02-21 20:44:36 +00:00
Lang Hames 2754714fb9 [Orc] Update the Orc indirection utils and refactor the CompileOnDemand layer.
This patch replaces most of the Orc indirection utils API with a new class:
JITCompileCallbackManager, which creates and manages JIT callbacks.
Exposing this functionality directly allows the user to create callbacks that
are associated with user supplied compilation actions. For example, you can
create a callback to lazyily IR-gen something from an AST. (A kaleidoscope
example demonstrating this will be committed shortly).

This patch also refactors the CompileOnDemand layer to use the
JITCompileCallbackManager API.

llvm-svn: 229461
2015-02-17 01:18:38 +00:00
Chandler Carruth 71f308adb7 Re-sort #include lines using my handy dandy ./utils/sort_includes.py
script. This is in preparation for changes to lots of include lines.

llvm-svn: 229088
2015-02-13 09:09:03 +00:00
Lang Hames 93de2a12a3 [Orc] New JIT APIs.
This patch adds a new set of JIT APIs to LLVM. The aim of these new APIs is to
cleanly support a wider range of JIT use cases in LLVM, and encourage the
development and contribution of re-usable infrastructure for LLVM JIT use-cases.

These APIs are intended to live alongside the MCJIT APIs, and should not affect
existing clients.

Included in this patch:

1) New headers in include/llvm/ExecutionEngine/Orc that provide a set of
   components for building JIT infrastructure.
   Implementation code for these headers lives in lib/ExecutionEngine/Orc.

2) A prototype re-implementation of MCJIT (OrcMCJITReplacement) built out of the
   new components.

3) Minor changes to RTDyldMemoryManager needed to support the new components.
   These changes should not impact existing clients.

4) A new flag for lli, -use-orcmcjit, which will cause lli to use the
   OrcMCJITReplacement class as its underlying execution engine, rather than
   MCJIT itself.

Tests to follow shortly.

Special thanks to Michael Ilseman, Pete Cooper, David Blaikie, Eric Christopher,
Justin Bogner, and Jim Grosbach for extensive feedback and discussion.

llvm-svn: 226940
2015-01-23 21:25:00 +00:00