llvm-project/llvm/lib/Transforms
Tim Northover ed9117f88d GlobalOpt: do not promote globals used atomically to constants.
Some atomic loads are implemented as cmpxchg (particularly if large or
floating), and that usually requires write access to the memory involved
or it will segfault.

We can still propagate the constant value to users we understand though.

llvm-svn: 360662
2019-05-14 11:03:13 +00:00
..
AggressiveInstCombine Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
Coroutines [coroutines] Fix spills of static array allocas 2019-05-13 23:58:24 +00:00
Hello Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
IPO GlobalOpt: do not promote globals used atomically to constants. 2019-05-14 11:03:13 +00:00
InstCombine [InstCombine] try harder to form rotate (funnel shift) (PR20750) 2019-05-13 17:28:19 +00:00
Instrumentation [MemorySanitizer] getMMXVectorTy - assert valid element size. NFCI. 2019-05-14 10:29:18 +00:00
ObjCARC [ObjC][ARC] Let ARC optimizer bail out if the number of pointer states 2019-04-25 19:42:55 +00:00
Scalar Revert "[LSR] Tweak setup cost depth threshold to 10." 2019-05-13 15:37:18 +00:00
Utils [DebugInfo] Use zero linenos for debug intrinsics when promoting dbg.declare 2019-05-10 10:03:41 +00:00
Vectorize [SLP] Refactor VectorizableTree to use unique_ptr. 2019-05-10 18:55:17 +00:00
CMakeLists.txt Another try to commit 323321 (aggressive instruction combine). 2018-01-25 12:06:32 +00:00
LLVMBuild.txt Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00