summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Revert "[DebugInfo] Remove some users of DBG_VALUEs IsIndirect field"Jeremy Morse2020-02-1223-120/+97
| | | | | | | | | | | | | | | | | | | | | | | This reverts commit ed29dbaafa49bb8c9039a35f768244c394411fea. I'm backing out D68945, which as the discussion for D73526 shows, doesn't seem to handle the -O0 path through the codegen backend correctly. I'll reland the patch when a fix is worked out, apologies for all the churn. The two parent commits are part of this revert too. Conflicts: llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp llvm/test/DebugInfo/X86/dbg-addr-dse.ll SelectionDAGBuilder conflict is due to a nearby change in e39e2b4a79c6 that's technically unrelated. dbg-addr-dse.ll conflicted because 41206b61e30c (legitimately) changes the order of two lines. There are further modifications to dbg-value-func-arg.ll: it landed after the patch being reverted, and I've converted indirection to be represented by the isIndirect field rather than DW_OP_deref. (cherry picked from commit 6531a78ac4b5b229bce272706593a0bc873877d7)
* Revert "[DebugInfo][DAG] Distinguish different kinds of location indirection"Jeremy Morse2020-02-123-106/+7
| | | | | | | | | | | | | | This reverts commit 3137fe4d23eeb8df08c03e9111465325eeafe08e. I'm backing out D68945, which this patch is a follow up for. It'll be re-landed when D68945 is fixed. The changes to dbg-value-func-arg.ll occur because our handling of certain kinds of location now mixes up indirection that happens at different points in a DIExpression. While this is a regression, it's a return to the prior behaviour while a better patch is sought. (cherry picked from commit ece761427f63de96ee52bbd6be1c61b07967a917)
* [DebugInfo] Re-instate LiveDebugVariables scope trimmingJeremy Morse2020-02-123-4/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | This patch reverts part of r362750 / D62650, which stopped LiveDebugVariables from trimming leading variable location ranges down to only covering those instructions that are in scope. I've observed some circumstances where the number of DBG_VALUEs in a function can be amplified in an un-necessary way, to cover more instructions that are out of scope, leading to very slow compile times. Trimming the range of instructions that the variables cover solves the slow compile times. The specific problem that r362750 tries to fix is addressed by the assignment to RStart that I've added. Any variable location that begins at the first instruction of a block will now be considered to begin at the start of the block. While these sound the same, the have different SlotIndexes, and the register allocator may shoehorn additional instructions in between the two. The test added in the past (wrong_debug_loc_after_regalloc.ll) still works with this modification. live-debug-variables.ll has a range trimmed to not cover the prologue of the function, while dbg-addr-dse.ll has a DBG_VALUE sink past one instruction with no DebugLoc, which is expected behaviour. Differential Revision: https://reviews.llvm.org/D73691 (cherry picked from commit 41206b61e30c3d84188cb17b91c2c0c800982dd1)
* [clang-tidy] Added check to disable bugprone-infinite-loop on known false ↵Nathan James2020-02-122-0/+19
| | | | | | | | | | | | | | | | | | condition Summary: Addresses [[ https://bugs.llvm.org/show_bug.cgi?id=44816 | bugprone-infinite-loop false positive with CATCH2 ]] by disabling the check on loops where the condition is known to always eval as false, in other words not a loop. Reviewers: aaron.ballman, alexfh, hokein, gribozavr2, JonasToth Reviewed By: gribozavr2 Subscribers: xazax.hun, cfe-commits Tags: #clang, #clang-tools-extra Differential Revision: https://reviews.llvm.org/D74374 (cherry picked from commit c69ec6476806147e46bf09b693acb24177982dc2)
* [OpenCL] Restrict addr space conversions in nested pointersAnastasia Stulova2020-02-128-17/+86
| | | | | | | | | | | | | | | | | | | | | | Address space conversion changes pointer representation. This commit disallows such conversions when they are not legal i.e. for the nested pointers even with compatible address spaces. Because the address space conversion in the nested levels can't be generated to modify the pointers correctly. The behavior implemented is as follows: - Any implicit conversions of nested pointers with different address spaces is rejected. - Any conversion of address spaces in nested pointers in safe casts (e.g. const_cast or static_cast) is rejected. - Conversion in low level C-style or reinterpret_cast is accepted but with a warning (this aligns with OpenCL C behavior). Fixes PR39674 Differential Revision: https://reviews.llvm.org/D73360 (cherry picked from commit 6064f426a18304e16b51cc79e74c9c2d55ef5a9c)
* Fix MSVC build with C++ EH enabledReid Kleckner2020-02-121-1/+1
| | | | | | | | | | | | | | | | | Mark the CrashRecoveryContextImpl constructor noexcept, so that MSVC won't emit an unwind helper to clean up the allocation from `new` if the constructor throws an exception. Otherwise, MSVC complains: llvm\lib\Support\CrashRecoveryContext.cpp(220): error C2712: \ Cannot use __try in functions that require object unwinding The other simple fix would be to wrap `new` in a static helper or lambda. Users have reported that Tensorflow builds LLVM with /EHsc. (cherry picked from commit a349c09162a8260bdf691c4f7ab72a16c33975f6)
* [SystemZ] Bugfix in emitSelect()Jonas Paulsson2020-02-122-2/+46
| | | | | | | | | | | | | | When more than one SelectPseudo instruction is handled a new MBB is returned. This must not be done if that would result in leaving an undhandled isel pseudo behind in the original MBB. Fixes https://bugs.llvm.org/show_bug.cgi?id=44849. Review: Ulrich Weigand Differential Revision: https://reviews.llvm.org/D74352 (cherry picked from commit 0311e28e9cc01a244faa774b8cab337b45404fa9)
* [Clang][Driver] After default -fintegrated-cc1, make ↵Alexandre Ganea2020-02-128-9/+59
| | | | | | | | | | | | | llvm::report_fatal_error() generate preprocessed source + reproducer.sh again. Added a test for #pragma clang __debug llvm_fatal_error to test for the original issue. Added llvm::sys::Process::Exit() and replaced ::exit() in places where it was appropriate. This new function would call the current CrashRecoveryContext if one is running on the same thread; or call ::exit() otherwise. Fixes PR44705. Differential Revision: https://reviews.llvm.org/D73742 (cherry picked from commit faace365088a2a3a4cb1050a9facfc34a7a56577)
* Add SystemZ release notesUlrich Weigand2020-02-111-0/+13
|
* [Support] Don't modify the current EH context during stack unwindingReid Kleckner2020-02-111-1/+7
| | | | | | | | | | | | | Copy it instead. Otherwise, key registers (such as RBP) may get zeroed out by the stack unwinder. Fixes CrashRecoveryTest.DumpStackCleanup with MSVC in release builds. Reviewed By: stella.stamenova Differential Revision: https://reviews.llvm.org/D73809 (cherry picked from commit b074acb82f7e75a189fa7933b09627241b166121)
* IR Linking: Support merging Warning+Max module metadata flagsDavid Blaikie2020-02-114-24/+65
| | | | | | | | | | | | | | | | | Summary: Debug Info Version was changed to use "Max" instead of "Warning" per the original design intent - but this maxes old/new IR unlinkable, since mismatched merge styles are a linking failure. It seems possible/maybe reasonable to actually support the combination of these two flags: Warn, but then use the maximum value rather than the first value/earlier module's value. Reviewers: tejohnson Differential Revision: https://reviews.llvm.org/D74257 (cherry picked from commit ba9cae58bbdd41451ee67773c9d0f90a01756f12)
* Prefer __vector over vector keyword for altivecserge-sans-paille2020-02-111-2/+2
| | | | | | | | | | | `vector' uses the keyword-and-predefine mode from gcc, while __vector is reliably supported. As a side effect, it also makes the code consistent in its usage of __vector. Differential Revision: https://reviews.llvm.org/D74129 (cherry picked from commit 3185c30c54d0af5bffbff3bcfd721668d086ff10)
* ReleaseNotes / external projects: ZigHans Wennborg2020-02-101-1/+11
|
* Re-land "[Clang][Driver] Remove -M group options ..." and "[Clang] Avoid ↵Alexandre Ganea2020-02-108-17/+118
| | | | | | | | | | | crashing when generating crash diagnostics when '#pragma clang __debug ..." This re-lands commits f41ec709d9d388dc43469e6ac7f51b6313f7e4af (https://reviews.llvm.org/D74076) and commit 5fedc2b410853a6aef05e8edf19ebfc4e071e28f (https://reviews.llvm.org/D74070) The previous build break was caused by '#pragma clang __debug llvm_unreachable' used in a non-assert build. Move it to a separate test in crash-report-with-asserts.c. (cherry picked from commit 75f09b54429bee17a96e2ba7a2ac0f0a8a7f7e74)
* [Support] When using SEH, create a impl instance for CrashRecoveryContext. NFCI.Alexandre Ganea2020-02-101-25/+35
| | | | | | | | | | Previously, the SEH codepath in CrashRecoveryContext didn't create a CrashRecoveryContextImpl. The other codepaths (VEH and Unix) were creating it. When running with -fintegrated-cc1, this is needed to handle exit() as a jump to CrashRecoveryContext's exception filter, through a call to RaiseException. In that situation, we need a user-defined exception code, which is later interpreted as an exit() by the exception filter. This in turn needs to set RetCode accordingly, *inside* the exception filter, and *before* calling HandleCrash(). Differential Revision: https://reviews.llvm.org/D74078 (cherry picked from commit 2a3fa0fc5cd7d3398c0293915b0e569eaa0be24b)
* [analyzer] Add 10.0.0 release notes.Hans Wennborg2020-02-101-4/+18
| | | | | | Differential revision: https://reviews.llvm.org/D73966 By Kristóf Umann!
* [Clang] Remove unused #pragma clang __debug handle_crashAlexandre Ganea2020-02-103-17/+0
| | | | | | | | | As discussed in D70568, remove this because it isn't used anywhere, and I think it's better to go through real crashes for testing (#pragma clang __debug crash). Also remove the support function llvm::CrashRecoveryContext::HandleCrash() which was added at the same time by @ddunbar. Differential Revision: https://reviews.llvm.org/D74063 (cherry picked from commit 8ecde3ac34bbb5a8d53d8ec5cd32867658646df1)
* [AArch64] Add option to enable/disable load-store renaming.Florian Hahn2020-02-109-15/+28
| | | | | | | | This patch adds a new option to enable/disable register renaming in the load-store optimizer. Defaults to disabled, as there is a potential mis-compile caused by this. (cherry picked from commit 8e3f59b45ae185cc9b4e3a817d7ac958f1d55976)
* AMDGPU/EG,CM: Implement fsqrt using recip(rsqrt(x)) instead of x * rsqrt(x)Jan Vesely2020-02-104-18/+34
| | | | | | | | | | | | The old version might be faster on EG (RECIP_IEEE is Trans only), but it'd need extra corner case checks. This gives correct corner case behaviour and saves a register. Fixes OCL CTS sqrt test (1-thread, scalar) on Turks. Reviewer: arsenm Differential Revision: https://reviews.llvm.org/D74017 (cherry picked from commit e6686adf8a743564f0c455c34f04752ab08cf642)
* [clang-tidy] Fix false positive for cppcoreguidelines-init-variablesNathan James2020-02-102-5/+15
| | | | | | | | | | | | | | | | Summary: Fixes [[ https://bugs.llvm.org/show_bug.cgi?id=44746 | False positive for cppcoreguidelines-init-variables in range based for loop in template function ]] Reviewers: aaron.ballman, alexfh, hokein, JonasToth, gribozavr2 Reviewed By: aaron.ballman Subscribers: merge_guards_bot, xazax.hun, wuzish, nemanjai, kbarton, cfe-commits Tags: #clang, #clang-tools-extra Differential Revision: https://reviews.llvm.org/D73843 (cherry picked from commit efcd09cea9a51c522954aa24e4b5513266daf6c3)
* [X86] Use MVT::i8 instead of MVT::i64 for shift amount in BuildSDIVPow2Craig Topper2020-02-102-1/+25
| | | | | | | | | | | | X86 uses i8 for shift amounts. This code can fail on a 32-bit target if it runs after type legalization. This code was copied from AArch64 and modified for X86, but the shift amount wasn't changed to the correct type for X86. Fixes PR44812 (cherry picked from commit ec9a94af4d5fb3270f2451fcbec5a3a99f4ac03a)
* [BPF] disable ReduceLoadWidth during SelectionDag phaseYonghong Song2020-02-102-0/+169
| | | | | | | | | | | | | | | | | | | | | | | | | The compiler may transform the following code ctx = ctx + reloc_offset ... (*(u32 *)ctx) & 0x8000 ... to ctx = ctx + reloc_offset ... (*(u8 *)(ctx + 1)) & 0x80 ... where reloc_offset will be replaced with a constant during AsmPrinter phase. The above transformed code will be rejected the kernel verifier as it does not allow *(type *)((ctx + non_zero_offset1) + non_zero_offset2) style access pattern. It is hard at SelectionDag phase to identify whether a load is related to context or not. Sometime, interprocedure analysis may be needed. So let us simply prevent such optimization from happening. Differential Revision: https://reviews.llvm.org/D73997 (cherry picked from commit d96c1bbaa03574daf759e5e9a6c75047c5e3af64)
* [analyzer] Fix a couple of bugs in HTML report generation.Artem Dergachev2020-02-105-1/+121
| | | | | | | | It should now produce valid HTML again. Differential Revision: https://reviews.llvm.org/D73993 (cherry picked from commit 482e236e569e8324f70778af1eb756923cd490dc)
* [cmake] Fix clang builds with BUILD_SHARED=ON and CLANG_LINK_CLANG_DYLIB=ONTom Stellard2020-02-101-1/+16
| | | | | | | | | | | | | | | | | | | Summary: We were linking all the clang objects and shared libraries into libclang-cpp.so, which was causing the command line options to be registered twice. Reviewers: beanz, mgorny Reviewed By: beanz, mgorny Subscribers: merge_guards_bot, mgorny, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D68520 (cherry picked from commit ebcf25ea8100fc9987fd1edd1975194addc2fc05)
* [ARM] Fix non-determenistic behaviourDiogo Sampaio2020-02-101-3/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | Summary: ARM Type Promotion pass does not clear the container that defines if one variable was visited or not, missing optimization opportunities by luck when two llvm:Values from different functions are allocated at the same memory address. Also fixes a comment and uses existing method to pop and obtain last element of the worklist. Reviewers: samparker Reviewed By: samparker Subscribers: kristof.beyls, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D73970 (cherry picked from commit 8ba2b6281075c65c1a47abed57810e1201942533)
* PowerPC release notesHans Wennborg2020-02-101-1/+23
| | | | By Anil Mahmud.
* [InstCombine] Fix infinite min/max canonicalization loop (PR44541)Nikita Popov2020-02-102-0/+31
| | | | | | | | | | | | While D72944 also fixes https://bugs.llvm.org/show_bug.cgi?id=44541, it does so in a more roundabout manner and there might be other loopholes to trigger the same issue. This is a more direct fix, that prevents the transform if the min/max is based on a non-canonical sub X, 0 instruction. Differential Revision: https://reviews.llvm.org/D73849 (cherry picked from commit a148b9e9909db6a592609eb35b4de38c9e67cb8b)
* [InstCombine] Support disabling expensive combines in optNikita Popov2020-02-102-2/+3
| | | | | | | | | | | | | | | Currently, there is no way to disable ExpensiveCombines when doing a standalone opt -instcombine run, as that's the default, and the opt option can currently only be used to force enable, not to force disable. The only way to disable expensive combines is via -O1 or -O2, but that of course also runs the rest of the kitchen sink... This patch allows using opt -instcombine -expensive-combines=0 to run InstCombine without ExpensiveCombines. Differential Revision: https://reviews.llvm.org/D72861 (cherry picked from commit 2ca092f3209579fde7a38ade511c1bbcef213c36)
* [InstCombine] Add test for -expensive-combines option; NFCNikita Popov2020-02-101-0/+28
| | | | | | This shows that -expensive-combines=0 is ignored. (cherry picked from commit 2d0d4235a282e0f900d31ac1054aafc0c526245c)
* [InstCombine] Fix infinite loop in min/max load/store bitcast combine (PR44835)Nikita Popov2020-02-102-0/+34
| | | | | | | | | | | | | | | | | | Fixes https://bugs.llvm.org/show_bug.cgi?id=44835. Skip the transform if it wouldn't actually do anything (apart from removing and reinserting the same instructions). Note that the test case doesn't loop on current master anymore, only on the LLVM 10 release branch. The issue is already mitigated on master due to worklist order fixes, but we should fix the root cause there as well. As a side note, we should probably assert in combineLoadToNewType() that it does not combine to the same type. Not doing this here, because this assertion would also be triggered in another place right now. Differential Revision: https://reviews.llvm.org/D74278 (cherry picked from commit 23db9724d0e5490fa5a2a726acf015f84e2c87cf)
* Revert "[ARM] Improve codegen of volatile load/store of i64"Victor Campos2020-02-087-342/+6
| | | | This reverts commit 60e0120c913dd1d4bfe33769e1f000a076249a42.
* StringRef.h: __builtin_strlen seems to exist in VS 2017 MSVC 19.16 or laterHans Wennborg2020-02-071-1/+1
| | | | | | | This is a follow-up to ff837aa63cd, as discussed on the llvm-commits thread for that one. (cherry picked from commit 1b3d1661bbeb2c90f8f3ef6e2b77a70bc148731e)
* [Concepts] Fix incorrect check when instantiating abbreviated template ↵Saar Raz2020-02-063-9/+47
| | | | | | | | | | | | | type-constraints We would incorrectly check whether the type-constraint had already been initialized, causing us to ignore the invented template type constraints entirely. Also, TemplateParameterList would store incorrect information about invented type parameters when it observed them before their type-constraint was initialized, so we recreate it after initializing the function type of an abbreviated template. (cherry picked from commit 38fd69995fc5a6f16e0aa132a46e5ccdbc2eebb3)
* [LV] Fix predication for branches with matching true and false succs.Florian Hahn2020-02-062-1/+80
| | | | | | | | | | | | | | | | | | | | | Currently due to the edge caching, we create wrong predicates for branches with matching true and false successors. We will cache the condition for the edge from the true successor, and then lookup the same edge (src and dst are the same) for the edge to the false successor. If both successors match, the condition should always be true. At the moment, we cannot really create constant VPValues, but we can just create a true condition as X | !X. Later passes will clean that up. Fixes PR44488. Reviewers: rengolin, hsaito, fhahn, Ayal, dorit, gilr Reviewed By: Ayal Differential Revision: https://reviews.llvm.org/D73079 (cherry picked from commit f14f2a856802e086662d919e2ead641718b27555)
* Fix x86 32bits MLIR build (NFC)Dimitry Andric2020-02-061-1/+1
| | | | | | | | | | This is fixing a build error: error: non-constant-expression cannot be narrowed from type 'unsigned int' to 'Region::iterator::difference_type' (aka 'int') in initializer list Fix pr44767 (cherry picked from commit 31fd112eb4a90600e0f340f19053e5715e92ec4c)
* [LLDB] Fix compilation with GCC 5Martin Storsjö2020-02-064-3/+12
| | | | | | Differential Revision: https://reviews.llvm.org/D74084 (cherry picked from commit 5bbaf543585c54868f8a2bdd9e74edcf395b24b3)
* [clangd] Add workaround for GCC5 host compilers. NFC.Michael Liao2020-02-061-1/+7
| | | | (cherry picked from commit 40514a7d7a3b745ba43c2d014e54a0d78d65d957)
* Make llvm::crc32() work also for input sizes larger than 32 bits.Hans Wennborg2020-02-062-1/+29
| | | | | | | | | | | | The problem was noticed by the Chrome OS toolchain folks (crbug.com/1048445) because llvm-objcopy --add-gnu-debuglink would insert the wrong checksum when processing a binary larger than 4 GB. That use case regressed in 1e1e3ba2526 when we started using llvm::crc32() in more places. Differential revision: https://reviews.llvm.org/D74039 (cherry picked from commit 6c4a8bc0a9f6a466d90d542bef66d69550c1b041)
* [TSan] Ensure we can compile the runtime with older SDKsJulian Lettner2020-02-051-1/+8
| | | | | | | | | | | | | | | | | | | | | One of my changes [1] included in this release silently bumped the minimal macOS SDK required for building the TSan runtime to SDK 10.12. Let's ensure release 10 does not unexpectedly break builders with old SDKs and add proper minimal SDK checking in CMake for subsequent releases. This fix `#ifdef`s out interceptors for newer APIs. Note that the resulting TSan runtime is less complete: when these newer APIs are used TSan will report false positives. Fixes llvm 10 release blocker: #44682 https://bugs.llvm.org/show_bug.cgi?id=44682 [1] 894abb46f891cba2e0ef581650f27f512a7824b4 Reviewed By: dmajor Differential Revision: https://reviews.llvm.org/D74059
* [libcxx] Add release notes for the 10.x branch for things I've doneMartin Storsjö2020-02-051-0/+5
| | | | Differential Revision: https://reviews.llvm.org/D74029
* [clang] Add release notes for the 10.x branch for things I've doneMartin Storsjö2020-02-051-0/+6
| | | | Differential Revision: https://reviews.llvm.org/D74028
* [docs] Add LLVM/LLDB release notes for the 10.x branch for things I've doneMartin Storsjö2020-02-051-0/+9
| | | | Differential Revision: https://reviews.llvm.org/D74027
* [LLD] Add release notes for MinGW for the 10.x branchMartin Storsjö2020-02-051-1/+27
| | | | Differential Revision: https://reviews.llvm.org/D74026
* [OpenMP][OMPT] fix reduction test for 32-bit x86protze@itc.rwth-aachen.de2020-02-051-3/+7
| | | | | | | | | | | | | | Fixes [[ https://bugs.llvm.org/show_bug.cgi?id=44733 | TEST 'libomp :: ompt/synchronization/reduction/tree_reduce.c' FAILED on 32-bit x86 ]] For 32-bit we need at least 3 variables to avoid atomic reduction to be choosen by runtime function `__kmp_determine_reduction_method`. This patch adds reduction variables to the testcase. Reviewers: mgorny, Hahnfeld Differential Revision: https://reviews.llvm.org/D73850 (cherry picked from commit 90e4ebdce55fd3c1f8877f19784a5339b9890f98)
* Don't warn about missing declarations for partial template specializationsAaron Puchert2020-02-052-0/+3
| | | | | | | | | | Summary: Just like templates, they are excepted from the ODR rule. Reviewed By: aaron.ballman, rsmith Differential Revision: https://reviews.llvm.org/D68923 (cherry picked from commit 27684ae66d5545f211c0ac4393d0ba2bf3b5b47c)
* [X86] -fpatchable-function-entry=N,0: place patch label after ENDBR{32,64}Fangrui Song2020-02-053-16/+77
| | | | | | | | | | | | | | | | Similar to D73680 (AArch64 BTI). A local linkage function whose address is not taken does not need ENDBR32/ENDBR64. Placing the patch label after ENDBR32/ENDBR64 has the advantage that code does not need to differentiate whether the function has an initial ENDBR. Also, add 32-bit tests and test that .cfi_startproc is at the function entry. The line information has a general implementation and is tested by AArch64/patchable-function-entry-empty.mir Reviewed By: nickdesaulniers Differential Revision: https://reviews.llvm.org/D73760 (cherry picked from commit 8ff86fcf4c038c7156ed4f01e7ed35cae49489e2)
* Update for Clang 10 release notes in order to have reference to D66404.Andi-Bogdan Postelnicu2020-02-051-1/+18
| | | | | | | | | | | | Summary: Since `D66404` adds some significat modifications to the `CFG` we should include it in the release notes. Reviewers: hans Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D74031
* [ARM][VecReduce] Force expand vector_reduce_fminDavid Green2020-02-054-6/+2273
| | | | | | | | | | | | | | Under MVE, we do not have any lowering for fminimum, which a vector_reduce_fmin without NoNan will be expanded into. As with the other recent patches, force this to expand in the pre-isel pass. Note that Neon lowering would be OK because the scalar fminimum uses the vector VMIN instruction, but is probably better to just rely on the scalar operations, which is what is done here. Also fixes what appears to be the reversal of INF vs -INF in the vector_reduce_fmin widening code. (cherry picked from commit 362d00e0510ee75750499e2993a782428e377215)
* [ARM] Expand vector reduction intrinsics on soft floatNikita Popov2020-02-052-1/+71
| | | | | | | | | | | | Followup to D73135. If the target doesn't have hard float (default for ARM), then we assert when trying to soften the result of vector reduction intrinsics. This patch marks these for expansion as well. (A bit odd to use vectors on a target without hard float ... but that's where you end up if you expose target-independent vector types.) Differential Revision: https://reviews.llvm.org/D73854 (cherry picked from commit 1cc4f8d17247cd9be88addd75d060f9321b6f8b0)
* [AArch64][ARM] Always expand ordered vector reductions (PR44600)Nikita Popov2020-02-057-2/+687
| | | | | | | | | | | | | | | | | | fadd/fmul reductions without reassoc are lowered to VECREDUCE_STRICT_FADD/FMUL nodes, which don't have legalization support. Until that is in place, expand these intrinsics on ARM and AArch64. Other targets always expand the vector reduction intrinsics. Additionally expand fmax/fmin reductions without nonan flag on AArch64, as the backend asserts that the flag is present when lowering VECREDUCE_FMIN/FMAX. This fixes https://bugs.llvm.org/show_bug.cgi?id=44600. Differential Revision: https://reviews.llvm.org/D73135 (cherry picked from commit 70d345e687caba4ac1f95655c6924dfa91e0083f)
OpenPOWER on IntegriCloud