wangtao
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
240,792 Commits 29 Branches 0 Tags 2.2 GiB
Commit Graph

32 Commits

Author SHA1 Message Date
Samuel Antao d06239d359 [CUDA][OpenMP] Create generic offload action 
Summary:
This patch replaces the CUDA specific action by a generic offload action. The offload action may have multiple dependences classier in “host” and “device”. The way this generic offloading action is used is very similar to what is done today by the CUDA implementation: it is used to set a specific toolchain and architecture to its dependences during the generation of jobs.

This patch also proposes propagating the offloading information through the action graph so that that information can be easily retrieved at any time during the generation of commands. This allows e.g. the "clang tool” to evaluate whether CUDA should be supported for the device or host and ptas to easily retrieve the target architecture.

This is an example of how the action graphs would look like (compilation of a single CUDA file with two GPU architectures)
```
0: input, "cudatests.cu", cuda, (host-cuda)
1: preprocessor, {0}, cuda-cpp-output, (host-cuda)
2: compiler, {1}, ir, (host-cuda)
3: input, "cudatests.cu", cuda, (device-cuda, sm_35)
4: preprocessor, {3}, cuda-cpp-output, (device-cuda, sm_35)
5: compiler, {4}, ir, (device-cuda, sm_35)
6: backend, {5}, assembler, (device-cuda, sm_35)
7: assembler, {6}, object, (device-cuda, sm_35)
8: offload, "device-cuda (nvptx64-nvidia-cuda:sm_35)" {7}, object
9: offload, "device-cuda (nvptx64-nvidia-cuda:sm_35)" {6}, assembler
10: input, "cudatests.cu", cuda, (device-cuda, sm_37)
11: preprocessor, {10}, cuda-cpp-output, (device-cuda, sm_37)
12: compiler, {11}, ir, (device-cuda, sm_37)
13: backend, {12}, assembler, (device-cuda, sm_37)
14: assembler, {13}, object, (device-cuda, sm_37)
15: offload, "device-cuda (nvptx64-nvidia-cuda:sm_37)" {14}, object
16: offload, "device-cuda (nvptx64-nvidia-cuda:sm_37)" {13}, assembler
17: linker, {8, 9, 15, 16}, cuda-fatbin, (device-cuda)
18: offload, "host-cuda (powerpc64le-unknown-linux-gnu)" {2}, "device-cuda (nvptx64-nvidia-cuda)" {17}, ir
19: backend, {18}, assembler
20: assembler, {19}, object
21: input, "cuda", object
22: input, "cudart", object
23: linker, {20, 21, 22}, image
```
The changes in this patch pass the existent regression tests (keeps the existent functionality) and resulting binaries execute correctly in a Power8+K40 machine.

Reviewers: echristo, hfinkel, jlebar, ABataev, tra

Subscribers: guansong, andreybokhanko, tcramer, mkuron, cfe-commits, arpith-jacob, carlo.bertolli, caomhin

Differential Revision: https://reviews.llvm.org/D18171

llvm-svn: 275645
 2016-07-15 23:13:27 +00:00
Nico Weber 5a459f8249 Rename Action::begin() to Action::input_begin(). 
Also introduce inputs() that reutnrs an llvm::iterator_range.
Iterating over A->inputs() is much less mysterious than
iterating over *A.  No intended behavior change.

llvm-svn: 261674
 2016-02-23 19:30:43 +00:00
Benjamin Kramer c02670ed50 [libclang] Add entry points that take a full command line including argv[0]. 
This provides both a more uniform interface and makes libclang behave like
clang tooling wrt relative paths against argv[0]. This is necessary for
finding paths to a c++ standard library relative to a clang binary given
in a compilation database. It can also be used to find paths relative to
libclang.so if the full path to it is passed in.

Differential Revision: http://reviews.llvm.org/D14695

llvm-svn: 253466
 2015-11-18 16:14:27 +00:00
Artem Belevich 0ff05cd165 [cuda] Driver changes to compile and stitch together host and device-side CUDA code. 
NOTE: reverts r242077 to reinstate r242058, r242065, 242067
        and includes fix for OS X test failures.

  - Changed driver pipeline to compile host and device side of CUDA
    files and incorporate results of device-side compilation into host
    object file.

  - Added a test for cuda pipeline creation in clang driver.

  New clang options:
  --cuda-host-only   - Do host-side compilation only.
  --cuda-device-only - Do device-side compilation only.

  --cuda-gpu-arch=<ARCH> - specify GPU architecture for device-side
    compilation. E.g. sm_35, sm_30. Default is sm_20. May be used more
    than once in which case one device-compilation will be done per
    unique specified GPU architecture.

  Differential Revision: http://reviews.llvm.org/D9509

llvm-svn: 242085
 2015-07-13 23:27:56 +00:00
Rafael Espindola abbd6d6824 This reverts commit r242058, r242065, r242067. 
The tests were failing on OS X.

Revert "[cuda] Driver changes to compile and stitch together host and device-side CUDA code."
Revert "Fixed regex to properly match '64' in the test case."
Revert "clang/test/Driver/cuda-options.cu REQUIRES clang-driver, at least."

llvm-svn: 242077
 2015-07-13 22:26:30 +00:00
Artem Belevich cd42e7f77a [cuda] Driver changes to compile and stitch together host and device-side CUDA code. 
- Changed driver pipeline to compile host and device side of CUDA
    files and incorporate results of device-side compilation into host
    object file.

  - Added a test for cuda pipeline creation in clang driver.

  New clang options:
  --cuda-host-only   - Do host-side compilation only.
  --cuda-device-only - Do device-side compilation only.

  --cuda-gpu-arch=<ARCH> - specify GPU architecture for device-side
    compilation. E.g. sm_35, sm_30. Default is sm_20. May be used more
    than once in which case one device-compilation will be done per
    unique specified GPU architecture.

  Differential Revision: http://reviews.llvm.org/D9509

llvm-svn: 242058
 2015-07-13 20:21:06 +00:00
Justin Bogner aab9792b6c Driver: Use pointee_iterator rather than iterating over unique_ptrs 
There's probably never a good reason to iterate over unique_ptrs. This
lets us use range-for and say Job.foo instead of (*it)->foo in a few
places.

llvm-svn: 218938
 2014-10-03 01:04:53 +00:00
David Blaikie c11bf80265 unique_ptrify JobList::Jobs 
llvm-svn: 217168
 2014-09-04 16:04:28 +00:00
Alp Toker f994cef836 Track IntrusiveRefCntPtr::get() changes from LLVM r212366 
llvm-svn: 212369
 2014-07-05 03:08:06 +00:00
Craig Topper 49a2790fb3 [C++11] Use 'nullptr'. Frontend edition. 
llvm-svn: 209389
 2014-05-22 04:46:25 +00:00
Alp Toker 1761f11801 Eliminate DefaultImageName from the Driver constructor 
All callers were passing in "a.out" or garbage so a sensible default works fine
here as a cleanup.

This also brings about the possibility of adapting the value based on the
driver's compatibility mode in future.

The setting can still be changed via Driver::DefaultImageName as needed.

llvm-svn: 208926
 2014-05-15 22:26:36 +00:00
Ahmed Charles b89843299a Replace OwningPtr with std::unique_ptr. 
This compiles cleanly with lldb/lld/clang-tools-extra/llvm.

llvm-svn: 203279
 2014-03-07 20:03:18 +00:00
Ahmed Charles 9a16beb8bc Change OwningPtr::take() to OwningPtr::release(). 
This is a precursor to moving to std::unique_ptr.

llvm-svn: 203275
 2014-03-07 19:33:25 +00:00
Hans Wennborg b212b34f19 Move Compilation::PrintJob and PrintDiagnosticJob into Job::Print. 
This moves the code to Job.cpp, which seems like a more natural fit,
and replaces the "is this a JobList? is this a Command?" logic with
a virtual function call.

It also removes the code duplication between PrintJob and
PrintDiagnosticJob and simplifies the code a little.

There's no functionality change here, except that the Executable is
now always printed within quotes, whereas it would previously not be
quoted in crash reports, which I think was a bug.

Differential Revision: http://llvm-reviews.chandlerc.com/D1653

llvm-svn: 190620
 2013-09-12 18:23:34 +00:00
Reid Kleckner 898229ab4b [Driver] Refactor clang driver to use LLVM's Option library 
The big changes are:
- Deleting Driver/(Arg|Opt)*
- Rewriting includes to llvm/Option/ and re-sorting
- 'using namespace llvm::opt' in clang::driver
- Fixing the autoconf build by adding option everywhere

As discussed in the review, this change includes using directives in
header files.  I'll make follow up changes to remove those in favor of
name specifiers.

Reviewers: espindola

Differential Revision: http://llvm-reviews.chandlerc.com/D975

llvm-svn: 183989
 2013-06-14 17:17:23 +00:00
Sean Silva f1b49e237f Nuke SetUpBuildDumpLog. 
Also, it was the only reason that `argc` and `argv` were being passed
into createDiagnostics, so remove those parameters and clean up callers.

llvm-svn: 172945
 2013-01-20 01:58:28 +00:00
Chandler Carruth 3a02247dc9 Sort all of Clang's files under 'lib', and fix up the broken headers 
uncovered.

This required manually correcting all of the incorrect main-module
headers I could find, and running the new llvm/utils/sort_includes.py
script over the files.

I also manually added quite a few missing headers that were uncovered by
shuffling the order or moving headers up to be main-module-headers.

llvm-svn: 169237
 2012-12-04 09:13:33 +00:00
Rafael Espindola b0448cd52a Remove some dead code. CLANG_IS_PRODUCTION is now just a build flag and 
is not used in any #ifdef.

llvm-svn: 168703
 2012-11-27 16:10:37 +00:00
Bob Wilson d0de8ceb57 Remove code to fall back to llvm-gcc for i386 kexts. 
More cleanups to follow in separate commits....

llvm-svn: 167566
 2012-11-08 01:03:29 +00:00
Douglas Gregor 811db4eac4 Make DiagnosticOptions intrusively reference-counted, and make sure 
the various stakeholders bump up the reference count. In particular,
the diagnostics engine now keeps the DiagnosticOptions object alive.

llvm-svn: 166508
 2012-10-23 22:26:28 +00:00
Argyrios Kyrtzidis 3169e80603 [driver] When creating the compiler invocation out of command-line 
arguments, force use of clang frontend for the driver.

Fixes rdar://11356765.

llvm-svn: 157205
 2012-05-21 20:11:54 +00:00
Dylan Noblesmith c95d81924d Basic: import IntrusiveRefCntPtr<> into clang namespace 
The class name is long enough without the llvm:: added.
Also bring in RefCountedBase and RefCountedBaseVPTR.

llvm-svn: 150958
 2012-02-20 14:00:23 +00:00
Dylan Noblesmith 2c1dd2716a Basic: import SmallString<> into clang namespace 
(I was going to fix the TODO about DenseMap too, but
that would break self-host right now. See PR11922.)

llvm-svn: 149799
 2012-02-05 02:13:05 +00:00
Dylan Noblesmith e27789991d Basic: import OwningPtr<> into clang namespace 
llvm-svn: 149798
 2012-02-05 02:12:40 +00:00
Dylan Noblesmith e99b27f35e Let CompilerInvocation initialization indicate failure 
This fixes the FIXMEs in ParseAnalyzeArgs. (Also a
precursor to moving the analyzer into an AST plugin.)

For consistency, do the same with AssemblerInvocation.

llvm-svn: 147218
 2011-12-23 03:05:38 +00:00
Sebastian Pop 8188c8a1f3 rename getHostTriple into getDefaultTargetTriple in clang 
llvm-svn: 143503
 2011-11-01 21:33:06 +00:00
Bob Wilson 05de0298d2 Rip out flags for controlling C++ "production mode" separately. 
This is old leftover cruft from the days when C++ was not yet ready
for prime time.

llvm-svn: 141063
 2011-10-04 05:34:14 +00:00
David Blaikie 9c902b5502 Rename Diagnostic to DiagnosticsEngine as per issue 5397 
llvm-svn: 140478
 2011-09-25 23:23:43 +00:00
Chris Lattner 54b1677d23 Move ArrayRef to LLVM.h and eliminate now-redundant qualifiers, patch by Jon Mulder! 
llvm-svn: 135855
 2011-07-23 17:14:25 +00:00
Chris Lattner 0e62c1cc0b remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 
them into the clang namespace.

llvm-svn: 135852
 2011-07-23 10:55:15 +00:00
Eli Friedman e08bca9eb4 Fix the clang part of PR7952: rewrite the specialization of isa<> in DeclBase, 
and stop abusing the multi-level dereference isa<> used to allow.

llvm-svn: 131804
 2011-05-21 19:15:39 +00:00
Argyrios Kyrtzidis 02d32aec00 Add missing file for my previous commit. 
llvm-svn: 128853
 2011-04-04 23:16:36 +00:00