Summary:
- A device functions could be used as a non-type template parameter in a
global/host function template. However, we should not try to retrieve that
device function and reference it in the host-side debug info as it's
only valid at device side.
Subscribers: aprantl, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D58992
llvm-svn: 355551
The issue is the following code:
__cn1->__add(*__ip);
(*__ip)->__c_ = __cn1;
`__ip` points into the array of iterators for container `__cn2`. This code adds
the iterator to the array of iterators for `__cn1`, and updates the iterator to
point to the new container.
This code works fine, except when `__cn1` and `__cn2` are the same container.
`__cn1->__add()` might need to grow the array of iterators, and when it does,
`__ip` becomes invalid, so the second line becomes a use-after-free error.
Simply swapping the order of the above two lines is not sufficient, because of
the memmove() below. The easiest and most performant solution is just to skip
touching any iterators if the containers are the same.
Differential Revision: https://reviews.llvm.org/D58926
llvm-svn: 355550
Summary: Not having this seems like an oversight, and makes stage2 builds odd.
Reviewers: ddunbar, dexonsmith
Subscribers: mgorny, jkorous, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D59032
llvm-svn: 355547
Summary:
Currently, libc++'s `<stdexcept>` doesn't play nice with `vcruntime`. Specifically:
* `logic_error` and `runtime_error` have a different layout.
* libc++'s `logic_error` and `runtime_error` override `what()` but `vcruntime` does not.
* `vcruntime` uses weak vtables for `<stdexcept>` types.
* libc++'s `<stdexcept>` constructors and assignment operators may have different manglings than `vcruntimes`.
This patch makes libc++'s declarations in `<stdexcept>` match those provided by MSVC's STL as closely as possible.
If MSVC doesn't declare a special member, then neither do we. This ensures that the implicit definitions have the same linkage, visibility, triviality, and noexcept-ness.
Reviewers: thomasanderson, ldionne, smeenai
Reviewed By: thomasanderson
Subscribers: jdoerfert, libcxx-commits
Differential Revision: https://reviews.llvm.org/D58945
llvm-svn: 355546
When mismatched #pragma detect_mismatch declarations occur, now print the conflicting OBJs.
lld-link: error: /failifmismatch: mismatch detected for 'TEST':
>>> test.obj has value 1
>>> test2.obj has value 2
Fixes PR38579
Differential Revision: https://reviews.llvm.org/D58910
llvm-svn: 355543
Part 4 of CSPGO changes:
(1) add support in cmake for cspgo build.
(2) fix an issue in big endian.
(3) test cases.
Differential Revision: https://reviews.llvm.org/D54175
llvm-svn: 355541
Restore a reverted commit, with the silly mistake fixed. Sorry for the previous breakage.
Be consistent about how we treat atomics in non-zero address spaces. If we get to the backend, we tend to lower them as if in address space 0. Do the same if we need to insert a libcall instead.
Differential Revision: https://reviews.llvm.org/D58760
llvm-svn: 355540
This is a new attempt for bringing TSan libdispatch support to Linux.
The main issue with the last patch (https://reviews.llvm.org/D53171) was
that we want to avoid building a separate library.
The updated plan is as follows:
1) Hide libdispatch support behind a flag: true on Darwin, false
elsewhere. If flag is specified, assume that libdispatch header and
-flbocks is available for building. This way we can directly include
the libdispatch header and rely on blocks runtime for our
implementation.
2) Optionally/weakly intercept libdispatch API functions.
This patch accomplishes 1). It compiles (without the flag enabled) on
Linux. Follow-up patches will provide 2) and enabling of tests on Linux.
Reviewed By: dvyukov
Differential Revision: https://reviews.llvm.org/D58935
llvm-svn: 355538
Move the x86 combine from D58974 into the DAGCombine VSELECT code and update the SELECT version to use the isBooleanFlip helper as well.
Requested by @spatel on D59006
llvm-svn: 355533
In file PPCBranchSelector.cpp we tend to over estimate code size due to large
alignment and inline assembly. Usually it causes larger computed branch offset,
it is not big problem. But sometimes it may also causes smaller computed branch
offset than actual branch offset. If the offset is close to the limit of
encoding, it may cause problem at run time.
Following is a simplified example.
actual estimated
address address
...
bne Far 100 10c
.p2align 4
Near: 110 110
...
Far: 8108 8108
Actual offset: 0x8108 - 0x100 = 0x8008
Computed offset: 0x8108 - 0x10c = 0x7ffc
The computed offset is at most ((1 << alignment) - 4) bytes smaller than actual
offset. So we add this number to the offset for safety.
Differential Revision: https://reviews.llvm.org/D57718
llvm-svn: 355529
This was reverted because it breaks the GreenDragon bot, but
the reason for the breakage is lost, so I'm resubmitting this
now so we can find out what the problem is.
llvm-svn: 355528
Emit an error for an unsupported relocation. mach-o relocations can't
encode the form -SYM + cst.
Differential Revision: https://reviews.llvm.org/D58944
llvm-svn: 355527
Core files need to know the size of the PRSTATUS header so that we can grab the register values that follow it. The code that figure out this size was using a hard coded list of architecture cores instead of relying on 32 or 64 bit for most cores.
The fix here fixes core files for 32 bit ARM. Prior to this the PRSTATUS header size was being returned as zero and the register values were being taken from the first bytes of the PRSTATUS struct (signo, etc).
Differential Revision: https://reviews.llvm.org/D58985
llvm-svn: 355526
Summary: This tests a fix in the ASTImpoter.cpp to ensure that we import built-in correctly,
see differential: https://reviews.llvm.org/D58743
Once this change is merged this test should pass and should catch regressions in this feature.
Differential Revision: https://reviews.llvm.org/D58790
llvm-svn: 355525
Use this feature to fix a bug on ARM where 4 byte alignment is
incorrectly assumed.
Differential Revision: https://reviews.llvm.org/D57335
llvm-svn: 355522
Summary:
This breaks ABI for folks using -fshort-enums, and does not really buy
us anything.
http://llvm.org/PR40977
Reviewers: mclow.lists, EricWF
Subscribers: christof, jkorous, dexonsmith, libcxx-commits, zoecarver
Differential Revision: https://reviews.llvm.org/D59029
llvm-svn: 355521
Summary:
This adds support for 64 bit buffer atomic arithmetic instructions but does not include
cmpswap as that depends on a fix to the way the register pairs are handled
Change-Id: Ib207ea65fb69487ccad5066ea647ae8ddfe2ce61
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58918
llvm-svn: 355520
On Windows, lldb::thread_t is just a void*, so the we will try to
allocate an object of type void when deserializing. Undef this for now
until we support void* arguments.
llvm-svn: 355519
This allows us to store more info about where we're emitting the remarks
without cluttering LLVMContext. This is needed for future support for
the remark section.
Differential Revision: https://reviews.llvm.org/D58996
Original llvm-svn: 355507
llvm-svn: 355514
Summary:
In r354298 a DominatorTree construction was added via new function
combineToUSubWithOverflow, which was subsequently restructured into
replaceMathCmpWithIntrinsic in r354689. We are hitting a very long
compile time due to this repeated construction, once per math cmp in
the function.
We shouldn't need to build the DominatorTree more than once per
function, except when a transformation invalidates it. There is already
a boolean flag that is returned from these methods indicating whether
the DT has been modified. We can simply build the DT once per
Function walk in CodeGenPrepare::runOnFunction, since any time a change
is made we break out of the Function walk and restart it.
I modified the code so that both replaceMathCmpWithIntrinsic as well as
mergeSExts (which was also building a DT) use the DT constructed by the
run method.
From -mllvm -time-passes:
Before this patch: CodeGen Prepare user time is 328s
With this patch: CodeGen Prepare user time is 21s
Reviewers: spatel
Subscribers: jdoerfert, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58995
llvm-svn: 355512
Summary:
This file implements some general purpose data structures, and so it
belongs to the Utility module.
Reviewers: zturner, jingham, JDevlieghere, clayborg, espindola
Subscribers: emaste, mgorny, javed.absar, arichardson, MaskRay, lldb-commits
Differential Revision: https://reviews.llvm.org/D58970
llvm-svn: 355509
This allows us to store more info about where we're emitting the remarks
without cluttering LLVMContext. This is needed for future support for
the remark section.
Differential Revision: https://reviews.llvm.org/D58996
llvm-svn: 355507
Summary:
This allows generic code to query these parameters, and is a common
practice in a lot of other template classes.
Reviewers: zturner, Bigcheese
Subscribers: kristina, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58969
llvm-svn: 355504
Peter Collingbourne