Commit Graph

412 Commits

Author SHA1 Message Date
Rafael Espindola
ac69459e0f Replace the F_Binary flag with a F_Text one.
After this I will set the default back to F_None. The advantage is that
before this patch forgetting to set F_Binary would corrupt a file on windows.
Forgetting to set F_Text produces one that cannot be read in notepad, which
is a better failure mode :-)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@202052 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-24 18:20:12 +00:00
Rafael Espindola
9f5066c72c Make DisableIntegratedAS a TargetOption.
This replaces the old NoIntegratedAssembler with at TargetOption. This is
more flexible and will be used to forward clang's -no-integrated-as option.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@201836 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-21 03:13:54 +00:00
Rafael Espindola
ed6718d228 One last pass of DataLayout variable renaming.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@201834 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-21 02:01:42 +00:00
Eli Bendersky
cf42174647 Refactor TargetOptions initialization into a single place.
The same code (~20 lines) for initializing a TargetOptions object from CodeGen
cmdline flags is duplicated 4 times in 4 different tools. This patch moves it
into a utility function.

Since the CodeGen/CommandFlags.h file defines cl::opt flags in a header, it's
a bit of a touchy situation because we should only link them into tools. So this
patch puts the init function in the header.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@201699 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-19 17:09:35 +00:00
Rafael Espindola
f39297678b Remove support for not using .loc directives.
Clang itself was not using this. The only way to access it was via llc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200862 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-05 18:00:21 +00:00
Juergen Ributzka
9ce88db752 Add target analysis passes to the codegen pipeline for MCJIT.
This patch adds the target analysis passes (usually TargetTransformInfo) to the
codgen pipeline. We also expose now the AddAnalysisPasses method through the C
API, because the optimizer passes would also benefit from better target-specific
cost models.

Reviewed by Andrew Kaylor

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199926 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-23 19:23:28 +00:00
Chandler Carruth
9f20a4c6ce Re-sort #include lines again, prior to moving headers around.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199080 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-13 08:04:33 +00:00
Chandler Carruth
8a5351f0f5 [PM] Rename the IR printing pass header to a more generic and correct
name to match the source file which I got earlier. Update the include
sites. Also modernize the comments in the header to use the more
recommended doxygen style.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199041 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-12 11:10:32 +00:00
Chandler Carruth
bc65a8d518 Move the LLVM IR asm writer header files into the IR directory, as they
are part of the core IR library in order to support dumping and other
basic functionality.

Rename the 'Assembly' include directory to 'AsmParser' to match the
library name and the only functionality left their -- printing has been
in the core IR library for quite some time.

Update all of the #includes to match.

All of this started because I wanted to have the layering in good shape
before I started adding support for printing LLVM IR using the new pass
infrastructure, and commandline support for the new pass infrastructure.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@198688 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-07 12:34:26 +00:00
Nico Weber
94d1e67cb5 Port r198087 and r198089 (strip dead code by default) from make to cmake.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@198198 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-30 03:36:05 +00:00
Nico Weber
5e7c782b29 Strip dead code when linking by default with BFD ld (linux, ...) and ld64 (os x).
This reduces the size of clang-format from 22 MB to 1.8 MB, diagtool goes from
21 MB to 2.8 MB, libclang.so goes from 29 MB to 20 MB, etc.  The size of the
bin/ folder shrinks from 270 MB to 200 MB.

Targets that support plugins and don't already use EXPORTED_SYMBOL_FILE
(which libclang and libLTO already do) can set NO_DEAD_STRIP to opt out.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@198087 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-27 22:38:59 +00:00
NAKAMURA Takumi
3aabdebde2 [CMake] Update LLVM_LINK_COMPONENTS for each CMakeLists.txt.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@196908 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-10 11:13:32 +00:00
Will Dietz
7437feefbb Export symbols in tools that support loading plugins.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@196447 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-05 01:01:58 +00:00
Bill Wendling
61fc8d670f Use function attributes to indicate that we don't want to realign the stack.
Function attributes are the future! So just query whether we want to realign the
stack directly from the function instead of through a random target options
structure.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@187618 91177308-0d34-0410-b5e6-96231b3b80d8
2013-08-01 21:42:05 +00:00
Bill Wendling
f245ae5a4a Replace the "NoFramePointerElimNonLeaf" target option with a function attribute.
There's no need to specify a flag to omit frame pointer elimination on non-leaf
nodes...(Honestly, I can't parse that option out.) Use the function attribute
stuff instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@187093 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-25 00:34:29 +00:00
Bill Wendling
0dcba2fadb Recommit r186217 with testcase fix:
Use the function attributes to pass along the stack protector buffer size.

 Now that we have robust function attributes, don't use a command line option to
 specify the stack protecto buffer size.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186863 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-22 20:15:21 +00:00
Rafael Espindola
c1b49b56d4 Add a wrapper for open.
This centralizes the handling of O_BINARY and opens the way for hiding more
differences (like how open behaves with directories).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186447 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-16 19:44:17 +00:00
Chandler Carruth
12c74dc2c2 Revert commit r186217 -- this is breaking bots:
http://lab.llvm.org:8013/builders/clang-x86_64-darwin11-nobootstrap-RAincremental/builds/4328

Original commit log:
  Use the function attributes to pass along the stack protector buffer
  size.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186234 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-13 01:00:17 +00:00
Bill Wendling
8a50013cc2 Use the function attributes to pass along the stack protector buffer size.
Now that we have robust function attributes, don't use a command line option to
specify the stack protecto buffer size.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186217 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-12 22:25:20 +00:00
Wan Xiaofei
1b11a1925e Test commit access.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184278 91177308-0d34-0410-b5e6-96231b3b80d8
2013-06-19 02:26:00 +00:00
Andy Gibbs
200241e4de Replace uses of the deprecated std::auto_ptr with OwningPtr.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179373 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-12 10:56:28 +00:00
Chandler Carruth
0f3e4b1495 Manually update the dependencies in the Makefiles. It turns out that all
that work on the LLVMBuild based dependency specification didn't
actually work, we just now maintain dependencies in *3* places instead
of 2. Yay.

There may still be some missing dependencies, I'm still sifting through
the bots and my builds, but this is a step in the right direction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@177988 91177308-0d34-0410-b5e6-96231b3b80d8
2013-03-26 03:45:47 +00:00
Chandler Carruth
7fc162f893 Split out the IRReader header and the utility functions it provides into
its own library. These functions are bridging between the bitcode reader
and the ll parser which are in different libraries. Previously we didn't
have any good library to do this, and instead played fast and loose with
a "header only" set of interfaces in the Support library. This really
doesn't work well as evidenced by the recent attempt to add timing logic
to the these routines.

As part of this, make them normal functions rather than weird inline
functions, and sink the implementation into the library. Also clean up
the header to be nice and minimal.

This requires updating lots of build system dependencies to specify that
the IRReader library is needed, and several source files to not
implicitly rely upon the header file to transitively include all manner
of other headers.

If you are using IRReader.h, this commit will break you (the header
moved) and you'll need to also update your library usage to include
'irreader'. I will commit the corresponding change to Clang momentarily.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@177971 91177308-0d34-0410-b5e6-96231b3b80d8
2013-03-26 02:25:37 +00:00
Chandler Carruth
aeef83c6af Switch TargetTransformInfo from an immutable analysis pass that requires
a TargetMachine to construct (and thus isn't always available), to an
analysis group that supports layered implementations much like
AliasAnalysis does. This is a pretty massive change, with a few parts
that I was unable to easily separate (sorry), so I'll walk through it.

The first step of this conversion was to make TargetTransformInfo an
analysis group, and to sink the nonce implementations in
ScalarTargetTransformInfo and VectorTargetTranformInfo into
a NoTargetTransformInfo pass. This allows other passes to add a hard
requirement on TTI, and assume they will always get at least on
implementation.

The TargetTransformInfo analysis group leverages the delegation chaining
trick that AliasAnalysis uses, where the base class for the analysis
group delegates to the previous analysis *pass*, allowing all but tho
NoFoo analysis passes to only implement the parts of the interfaces they
support. It also introduces a new trick where each pass in the group
retains a pointer to the top-most pass that has been initialized. This
allows passes to implement one API in terms of another API and benefit
when some other pass above them in the stack has more precise results
for the second API.

The second step of this conversion is to create a pass that implements
the TargetTransformInfo analysis using the target-independent
abstractions in the code generator. This replaces the
ScalarTargetTransformImpl and VectorTargetTransformImpl classes in
lib/Target with a single pass in lib/CodeGen called
BasicTargetTransformInfo. This class actually provides most of the TTI
functionality, basing it upon the TargetLowering abstraction and other
information in the target independent code generator.

The third step of the conversion adds support to all TargetMachines to
register custom analysis passes. This allows building those passes with
access to TargetLowering or other target-specific classes, and it also
allows each target to customize the set of analysis passes desired in
the pass manager. The baseline LLVMTargetMachine implements this
interface to add the BasicTTI pass to the pass manager, and all of the
tools that want to support target-aware TTI passes call this routine on
whatever target machine they end up with to add the appropriate passes.

The fourth step of the conversion created target-specific TTI analysis
passes for the X86 and ARM backends. These passes contain the custom
logic that was previously in their extensions of the
ScalarTargetTransformInfo and VectorTargetTransformInfo interfaces.
I separated them into their own file, as now all of the interface bits
are private and they just expose a function to create the pass itself.
Then I extended these target machines to set up a custom set of analysis
passes, first adding BasicTTI as a fallback, and then adding their
customized TTI implementations.

The fourth step required logic that was shared between the target
independent layer and the specific targets to move to a different
interface, as they no longer derive from each other. As a consequence,
a helper functions were added to TargetLowering representing the common
logic needed both in the target implementation and the codegen
implementation of the TTI pass. While technically this is the only
change that could have been committed separately, it would have been
a nightmare to extract.

The final step of the conversion was just to delete all the old
boilerplate. This got rid of the ScalarTargetTransformInfo and
VectorTargetTransformInfo classes, all of the support in all of the
targets for producing instances of them, and all of the support in the
tools for manually constructing a pass based around them.

Now that TTI is a relatively normal analysis group, two things become
straightforward. First, we can sink it into lib/Analysis which is a more
natural layer for it to live. Second, clients of this interface can
depend on it *always* being available which will simplify their code and
behavior. These (and other) simplifications will follow in subsequent
commits, this one is clearly big enough.

Finally, I'm very aware that much of the comments and documentation
needs to be updated. As soon as I had this working, and plausibly well
commented, I wanted to get it committed and in front of the build bots.
I'll be doing a few passes over documentation later if it sticks.

Commits to update DragonEgg and Clang will be made presently.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171681 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-07 01:37:14 +00:00
Chandler Carruth
7bdf6b00e0 Convert the TargetTransformInfo from an immutable pass with dynamic
interfaces which could be extracted from it, and must be provided on
construction, to a chained analysis group.

The end goal here is that TTI works much like AA -- there is a baseline
"no-op" and target independent pass which is in the group, and each
target can expose a target-specific pass in the group. These passes will
naturally chain allowing each target-specific pass to delegate to the
generic pass as needed.

In particular, this will allow a much simpler interface for passes that
would like to use TTI -- they can have a hard dependency on TTI and it
will just be satisfied by the stub implementation when that is all that
is available.

This patch is a WIP however. In particular, the "stub" pass is actually
the one and only pass, and everything there is implemented by delegating
to the target-provided interfaces. As a consequence the tools still have
to explicitly construct the pass. Switching targets to provide custom
passes and sinking the stub behavior into the NoTTI pass is the next
step.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171621 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-05 11:43:11 +00:00
Chandler Carruth
0b8c9a80f2 Move all of the header files which are involved in modelling the LLVM IR
into their new header subdirectory: include/llvm/IR. This matches the
directory structure of lib, and begins to correct a long standing point
of file layout clutter in LLVM.

There are still more header files to move here, but I wanted to handle
them in separate commits to make tracking what files make sense at each
layer easier.

The only really questionable files here are the target intrinsic
tablegen files. But that's a battle I'd rather not fight today.

I've updated both CMake and Makefile build systems (I think, and my
tests think, but I may have missed something).

I've also re-sorted the includes throughout the project. I'll be
committing updates to Clang, DragonEgg, and Polly momentarily.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171366 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-02 11:36:10 +00:00
Chandler Carruth
7f00f87767 Sort a few more #include lines in tools/... unittests/... and utils/...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171363 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-02 10:26:28 +00:00
NAKAMURA Takumi
f2a68db426 llvm/tools: Add #include "llvm/TargetTransformInfo.h"
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@169817 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-11 05:53:37 +00:00
Chandler Carruth
f010c464a1 Sort the #include lines for tools/...
Again, tools are trickier to pick the main module header for than
library source files. I've started to follow the pattern of using
LLVMContext.h when it is included as a stub for program source files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@169252 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-04 10:44:52 +00:00
Jakob Stoklund Olesen
882092808b Add a -time-compilations=<N> option to llc.
This causes llc to repeat the module compilation N times, making it
possible to get more accurate information from -time-passes when
compiling small modules.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@169040 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-30 21:42:47 +00:00
Pedro Artigas
6eda081345 One more step towards making doInitialization and doFinalization useful for
start up and clean up module passes, now that ASAN and TSAN are fixed the
tests pass



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168905 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-29 17:47:05 +00:00
Owen Anderson
9780d352b9 Revert r168635 "Step towards implementation of pass manager with doInitialization and doFinalization per module detangled from runOn?? calls, still has temporary code not to break ASAN to be removed when that pass conforms to the proposed model".
It appears to have broken at least one buildbot.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168654 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-27 00:53:24 +00:00
Owen Anderson
0d30d30d58 Step towards implementation of pass manager with doInitialization and doFinalization per module detangled from runOn?? calls, still has temporary code not to break ASAN to be removed when that pass conforms to the proposed model
Patch by Pedro Artigas, with feedback from by Chandler Carruth.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168635 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-26 23:54:47 +00:00
Owen Anderson
40b6fdb81e Add doInitialization and doFinalization methods to ModulePass's, to allow them to be re-initialized and reused on multiple Module's.
Patch by Pedro Artigas.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168008 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-15 00:14:15 +00:00
Nadav Rotem
cbd9a19b5d Reapply the TargerTransformInfo changes, minus the changes to LSR and Lowerinvoke.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166248 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-18 23:22:48 +00:00
Bob Wilson
3b9a911efc Temporarily revert the TargetTransform changes.
The TargetTransform changes are breaking LTO bootstraps of clang.  I am
working with Nadav to figure out the problem, but I am reverting it for now
to get our buildbots working.

This reverts svn commits: 165665 165669 165670 165786 165787 165997
and I have also reverted clang svn 165741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166168 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-18 05:43:52 +00:00
Nadav Rotem
e3d0e86919 Add a new interface to allow IR-level passes to access codegen-specific information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165665 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-10 22:04:55 +00:00
Micah Villmow
791cfc211a Move TargetData to DataLayout.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165403 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-08 16:39:34 +00:00
Chad Rosier
35907e9862 Add support for the --param ssp-buffer-size= driver option.
PR9673

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162284 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-21 16:15:24 +00:00
Bob Wilson
5f91a99427 Add test triples to fix win32 failures. Revert workaround from r161292.
I don't have a win32 system to test, so hopefully I got them all fixed here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161519 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-08 20:31:37 +00:00
NAKAMURA Takumi
17d2853a4b llc: Try to suppress failures since r161262 .
FIXME: Fix several tests on i686-win32 due to lacking of many libraries.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161292 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-04 01:29:37 +00:00
Bob Wilson
772af92cb1 Fix memcmp code-gen to honor -fno-builtin.
I noticed that SelectionDAGBuilder::visitCall was missing a check for memcmp
in TargetLibraryInfo, so that it would use custom code for memcmp calls even
with -fno-builtin.  I also had to add a new -disable-simplify-libcalls option
to llc so that I could write a test for this.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161262 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-03 21:26:18 +00:00
Chad Rosier
7a43e30a9b Tweak prose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@160478 91177308-0d34-0410-b5e6-96231b3b80d8
2012-07-19 00:11:45 +00:00
Evan Cheng
769951f6cc Target option DisableJumpTables is a gross hack. Move it to TargetLowering instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159611 91177308-0d34-0410-b5e6-96231b3b80d8
2012-07-02 22:39:56 +00:00
Bob Wilson
30a507a1f5 Extend TargetPassConfig to allow running only a subset of the normal passes.
This is still a work in progress but I believe it is currently good enough
to fix PR13122 "Need unit test driver for codegen IR passes".  For example,
you can run llc with -stop-after=loop-reduce to have it dump out the IR after
running LSR.  Serializing machine-level IR is not yet supported but we have
some patches in progress for that.

The plan is to serialize the IR to a YAML file, containing separate sections
for the LLVM IR, machine-level IR, and whatever other info is needed.  Chad
suggested that we stash the stop-after pass in the YAML file and use that
instead of the start-after option to figure out where to restart the
compilation.  I think that's a great idea, but since it's not implemented yet
I put the -start-after option into this patch for testing purposes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159570 91177308-0d34-0410-b5e6-96231b3b80d8
2012-07-02 19:48:45 +00:00
Duncan Sands
f6ce8ea20c When users ask for -mcpu=help or -mattr=help, just output the help without
requiring a module.  Original patch by Sunay Ismail, simplified by Arnaud
de Grandmaison, then complicated by me (if a triple was specified on the
command line, output help for that triple, not for the default).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159268 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-27 16:23:48 +00:00
Rafael Espindola
275c85f1a7 Fix llc's -print-before=pass and -print-after=pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159227 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-26 21:33:36 +00:00
Lang Hames
59d454959f Rename fp-op fusion option (yet again) for compatibility with GCC option.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159042 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-22 22:31:00 +00:00
Lang Hames
e023141322 Rename -allow-excess-fp-precision flag to -fuse-fp-ops, and switch from a
boolean flag to an enum: { Fast, Standard, Strict } (default = Standard).

This option controls the creation by optimizations of fused FP ops that store
intermediate results in higher precision than IEEE allows (E.g. FMAs). The
behavior of this option is intended to match the behaviour specified by a
soon-to-be-introduced frontend flag: '-ffuse-fp-ops'.

Fast mode - allows formation of fused FP ops whenever they're profitable.

Standard mode - allow fusion only for 'blessed' FP ops. At present the only
blessed op is the fmuladd intrinsic. In the future more blessed ops may be
added.

Strict mode - allow fusion only if/when it can be proven that the excess
precision won't effect the result.

Note: This option only controls formation of fused ops by the optimizers.  Fused
operations that are explicitly requested (e.g. FMA via the llvm.fma.* intrinsic)
will always be honored, regardless of the value of this option.

Internally TargetOptions::AllowExcessFPPrecision has been replaced by
TargetOptions::AllowFPOpFusion.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158956 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-22 01:09:09 +00:00
Lang Hames
d693cafcfb Add DAG-combines for aggressive FMA formation.
This patch adds DAG combines to form FMAs from pairs of FADD + FMUL or
FSUB + FMUL. The combines are performed when:
(a) Either
      AllowExcessFPPrecision option (-enable-excess-fp-precision for llc)
        OR
      UnsafeFPMath option (-enable-unsafe-fp-math)
    are set, and
(b) TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) is true for the type of
    the FADD/FSUB, and
(c) The FMUL only has one user (the FADD/FSUB).

If your target has fast FMA instructions you can make use of these combines by
overriding TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) to return true for
types supported by your FMA instruction, and adding patterns to match ISD::FMA
to your FMA instructions.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158757 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-19 22:51:23 +00:00