summaryrefslogtreecommitdiffstats
path: root/llvm
Commit message (Collapse)AuthorAgeFilesLines
* [WinEH] Update CoreCLR EH state numberingJoseph Tremoulet2016-01-044-183/+731
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: Fix the CLR state numbering to generate correct tables, and update the lit test to verify them. The CLR numbering assigns one state number to each catchpad and cleanuppad. It also computes two tree-like relations over states: 1) Each state has a "HandlerParentState", which is the state of the next outer handler enclosing this state's handler (same as nearest ancestor per the ParentPad linkage on EH pads, but skipping over catchswitches). 2) Each state has a "TryParentState", which: a) for a catchpad that's not the last handler on its catchswitch, is the state of the next catchpad on that catchswitch. b) for all other pads, is the state of the pad whose try region is the next outer try region enclosing this state's try region. The "try regions are not present as such in the IR, but will be inferred based on the placement of invokes and pads which reach each other by exceptional exits. Catchswitches do not get their own states, but each gets mapped to the state of its first catchpad. Table generation requires each state's "unwind dest" state to have a lower state number than the given state. Since HandlerParentState can be computed as a function of a pad's ParentPad, and TryParentState can be computed as a function of its unwind dest and the TryParentStates of its children, the CLR state numbering algorithm first computes HandlerParentState in a top-down pass, then computes TryParentState in a bottom-up pass. Also reword some comments/names in the CLR EH table generation to make the distinction between the different kinds of "parent" clear. Reviewers: rnk, andrew.w.kaylor, majnemer Subscribers: AndyAyers, llvm-commits Differential Revision: http://reviews.llvm.org/D15325 llvm-svn: 256760
* AMDGPU: Avoid assertions after SGPR spilling failedNicolai Haehnle2016-01-042-10/+11
| | | | | | | | | | | | | | | | | | | | Summary: The comment explains it: emitError does not necessarily exit the compilation process, and then using NoRegister leads to assertions later on. This generates incorrect code, of course, but the user should know to not use the result when an error has been emitted. It would be nice to have a test-case for this inside the LLVM repository, but llc exits on error. shader-db tests trigger the underlying issue at least on Tonga. Reviewers: arsenm, tstellarAMD, mareko Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D15826 llvm-svn: 256757
* [AVX512] add PSRAD and PSRAQ IntrinsicMichael Zuckerman2016-01-044-1/+251
| | | | | | Differential Revision: http://reviews.llvm.org/D15851 llvm-svn: 256754
* [AVX512] add PSRAW Intrinsic Michael Zuckerman2016-01-044-0/+147
| | | | | | Differential Revision: http://reviews.llvm.org/D15850 llvm-svn: 256751
* [MC] Fix file name in file headerJeroen Ketema2016-01-041-1/+1
| | | | llvm-svn: 256749
* [AVX512] add PSRLV Intrinsic Michael Zuckerman2016-01-045-0/+173
| | | | | | Differential Revision: http://reviews.llvm.org/D15838 llvm-svn: 256747
* Fix a horrible infloop in value tracking in the face of dead code.Chandler Carruth2016-01-041-1/+6
| | | | | | | | | | | | | | | Amazingly, we just never triggered this without: 1) Moving code around for MetadataTracking so that a certain *different* amount of inlining occurs in the per-TU compile step. 2) Then you LTO opt or clang with a bootstrap, and get inlining, loop opts, and GVN line up everything *just* right. I don't really know how we didn't hit this before. We really need to be fuzz testing stuff, it shouldn't be hard to trigger. I'm working on crafting a reduced nice test case, and will submit that when I have it, but I want to get LTO build bots going again. llvm-svn: 256735
* [TableGen] Fix a typo in r256733.Craig Topper2016-01-041-1/+1
| | | | llvm-svn: 256734
* [TableGen] Use some free space in Init to store the opcode for ↵Craig Topper2016-01-042-20/+23
| | | | | | UnOpInit/BinOpInit/TernOpInit allowing those types to be a little smaller. NFC llvm-svn: 256733
* [TableGen] Call llvm_shutdown on exit so that all the ManagedStatic objects ↵Craig Topper2016-01-041-0/+3
| | | | | | in the support library will be deleted. llvm-svn: 256732
* [X86] Make hasFP constant timeDavid Majnemer2016-01-046-25/+33
| | | | | | | | | | We need a frame pointer if there is a push/pop sequence after the prologue in order to unwind the stack. Scanning the instructions to figure out if this happened made hasFP not constant-time which is a violation of expectations. Let's compute this up-front and reuse that computation when we need it. llvm-svn: 256730
* [LICM] Make instruction sinking funclet-awareDavid Majnemer2016-01-044-18/+147
| | | | | | | | | | | We had two bugs here: - We might try to sink into a catchswitch, causing verifier failures. - We will succeed in sinking into a cleanuppad but we didn't update the funclet operand bundle. This fixes PR26000. llvm-svn: 256728
* [TableGen] Change TGParser::SetValue to take an ArrayRef instead of ↵Craig Topper2016-01-042-15/+10
| | | | | | std::vector reference. Use None in many places where a default constructed vector was being passed. NFC llvm-svn: 256726
* [TableGen] Fix a bug that caused the wrong name for a record built from a ↵Craig Topper2016-01-042-7/+12
| | | | | | | | | | | | multiclass containing a defm called NAME that references another multiclass that contains a defm that uses NAME concatenated with other strings. It would end up doing the concatenations from the second multiclass twice. This occured because SetValue detected a self assignment when trying to set the value of NAME to a VarInit called NAME. NAME is special here and it will get cleaned up later. So add a flag to suppress the self assignment check for this case. Strangely the self-assignment error was returning false indicating it wasn't an error, but it wasn't doing the right thing. So this also changes it to report an error. This fixes the names of some AVX512 FMA instructions that showed this double expansion. llvm-svn: 256725
* llvm/include/llvm/ProfileData/InstrProfData.inc: Add c++-mode in the header.NAKAMURA Takumi2016-01-041-1/+1
| | | | llvm-svn: 256721
* Fix one file that I didn't convert properly in r256707.Dimitry Andric2016-01-031-1/+1
| | | | llvm-svn: 256720
* Use std::is_sorted and std::none_of instead of manual loops. NFCCraig Topper2016-01-034-25/+23
| | | | llvm-svn: 256719
* [PGO] Cleanup: remove reduncant calls in loweringXinliang David Li2016-01-031-2/+0
| | | | | | | | CoverageMapping data's section and alignment is already set during creation. No need to call it again during lowering. llvm-svn: 256716
* [PGO] Cleanup: Use covmap header definition in the template fileXinliang David Li2016-01-031-4/+4
| | | | | | | | | | | | | | This is one last remaining instrumentatation related structure that needs to be migrate to use the centralized template definition. With this change, instrumentation code related to coverage module header will be kept in sync with the coverage mapping reader. The remaining code which makes implicit assumption about covmap control structure layout in the the lowering pass will cleaned up in a different patch. This patch is not intended to have no functional change. llvm-svn: 256715
* [X86][MMX] Regenerated vector insertion test.Simon Pilgrim2016-01-031-10/+24
| | | | | | Shows the true horror of what is going on.... llvm-svn: 256713
* [PGO] Code refactoring to use header struct def /NFCXinliang David Li2016-01-031-5/+10
| | | | llvm-svn: 256712
* [PGO] Introduce coverage map header structure in PGO data template file /NFCXinliang David Li2016-01-032-1/+26
| | | | llvm-svn: 256710
* [SelectionDAG] Pulled out common code for CONCAT_VECTORS node creationSimon Pilgrim2016-01-031-39/+55
| | | | | | Pulled out the similar CONCAT_VECTORS creation code from the 2/3 operand getNode() calls (to handle all UNDEF and all BUILD_VECTOR cases). Added a similar handler to the general getNode() call as well. llvm-svn: 256709
* [X86][SSE] Added tests for insertion of zero elements into vectorsSimon Pilgrim2016-01-031-0/+539
| | | | | | Many of these could be much better if we just lowered them all as shuffles - especially for the 256-bit vectors. llvm-svn: 256708
* Fix several accidental DOS line endings in source filesDimitry Andric2016-01-0346-1688/+1688
| | | | | | | | | | | | | | | Summary: There are a number of files in the tree which have been accidentally checked in with DOS line endings. Convert these to native line endings. There are also a few files which have DOS line endings on purpose, and I have set the svn:eol-style property to 'CRLF' on those. Reviewers: joerg, aaron.ballman Subscribers: aaron.ballman, sanjoy, dsanders, llvm-commits Differential Revision: http://reviews.llvm.org/D15848 llvm-svn: 256707
* [X86][SSE41] Added test cases for improving insertps shufflesSimon Pilgrim2016-01-031-0/+111
| | | | | | As mentioned on D14261, an upcoming patch will improve combines of insertps instructions. llvm-svn: 256706
* [X86][SSE] Added v4f32 shuffle with zero testsSimon Pilgrim2016-01-031-0/+139
| | | | | | This is mainly test cases for improvements to insertps matching, but pre-SSE41 shuffles could be improved as well llvm-svn: 256705
* [TableGen] Simplify some code slightly. No need to check if the arrays are ↵Craig Topper2016-01-031-18/+10
| | | | | | empty before printing. The loop can be made to print the same thing if the loop is empty. NFC llvm-svn: 256703
* Use an ArrayRef to simplify repeated calculation of the array end. NFCCraig Topper2016-01-031-5/+6
| | | | llvm-svn: 256702
* Use std::is_sorted instead of manual loops. NFCCraig Topper2016-01-032-8/+9
| | | | llvm-svn: 256701
* Use range-based for loop. NFCCraig Topper2016-01-031-2/+2
| | | | llvm-svn: 256700
* [TableGen] Replace a logically negated xor of bools with just an equality ↵Craig Topper2016-01-031-1/+1
| | | | | | comparison for readability. NFC llvm-svn: 256699
* [TableGen] Use std::find_if and a lambda instead of manual loops.Craig Topper2016-01-031-9/+10
| | | | llvm-svn: 256698
* [TableGen] Fix a bug introduced in r256627. If the switch was not emitted we ↵Craig Topper2016-01-031-2/+3
| | | | | | still emitted a closing curly brace. llvm-svn: 256697
* [TableGen] Use range-based for loops. NFCCraig Topper2016-01-031-20/+13
| | | | llvm-svn: 256696
* [PGO] simple refactoring (NFC)Xinliang David Li2016-01-032-4/+10
| | | | llvm-svn: 256695
* WinEHPrepare.cpp: Suppress a warning for -Asserts. [-Wunused-variable]NAKAMURA Takumi2016-01-031-4/+2
| | | | llvm-svn: 256694
* [Verifier] Add braces to satisfy buildbots. NFCJoseph Tremoulet2016-01-021-1/+2
| | | | | | Fix build break introduced by r256691. llvm-svn: 256692
* [WinEH] Verify catchswitch handlersJoseph Tremoulet2016-01-022-0/+16
| | | | | | | | | | | | | | Summary: The handler list must be nonempty and consist solely of CatchPads. Reviewers: rnk, andrew.w.kaylor, majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D15842 llvm-svn: 256691
* [WinEH] Tighten parentPad verifier checksJoseph Tremoulet2016-01-023-7/+62
| | | | | | | | | | | | Summary: A catchswitch cannot be a parent of a cleanuppad or another catchswitch. Reviewers: rnk, andrew.w.kaylor, majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D15841 llvm-svn: 256690
* [WinEH] Update catchrets with cloned successorsJoseph Tremoulet2016-01-022-4/+83
| | | | | | | | | | | | | | | | | | | | Summary: Add a pass to update catchrets when their successors get cloned; the existing pass doesn't catch these because it walks the funclet whose blocks are being cloned but the catchret is in a child funclet. Also update the test for removing incoming PHI values; when the predecessor is a catchret, the relevant color is the catchret's parentPad, not its block's color. Reviewers: andrew.w.kaylor, rnk, majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D15840 llvm-svn: 256689
* Correct misleading formatting of several ifs followed by two statements ↵Yaron Keren2016-01-021-5/+16
| | | | | | | | | | without braces. While the original code would work with or without braces, it makes sense to set HaveSemi to true only if (!HaveSemi), otherwise it's already true, so I put the assignment inside the if block. This addresses PR25998. llvm-svn: 256688
* [WinEH] Add additional verificationDavid Majnemer2016-01-021-4/+7
| | | | | | | | Recolor the IR to make sure our computed colors are not hiding any bugs. Also, verifyFunction if we are running some post-preparation operations; some of these operations can hide latent bugs. llvm-svn: 256687
* [X86] Add intrinsics for reading and writing to the flags registerDavid Majnemer2016-01-0112-34/+157
| | | | | | | | | | | | | | | | | | | | | | | | | | LLVM's targets need to know if stack pointer adjustments occur after the prologue. This is needed to correctly determine if the red-zone is appropriate to use or if a frame pointer is required. Normally, LLVM can figure this out very precisely by reasoning about the contents of the MachineFunction. There is an interesting corner case: inline assembly. The vast majority of inline assembly which will perform a push or pop is done so to pair up with pushf or popf as appropriate. Unfortunately, this inline assembly doesn't mark the stack pointer as clobbered because, well, it isn't. The stack pointer is decremented and then immediately incremented. Because of this, LLVM was changed in r256456 to conservatively assume that inline assembly contain a sequence of stack operations. This is unfortunate because the vast majority of inline assembly will not end up manipulating the stack pointer in any way at all. Instead, let's provide a more principled solution: an intrinsic. FWIW, other compilers (MSVC and GCC among them) also provide this functionality as an intrinsic. llvm-svn: 256685
* [LibCallSimplifier] propagate FMF when shrinking binary callsSanjay Patel2015-12-312-0/+20
| | | | llvm-svn: 256682
* [X86] Remove a return after llvm_unreachable.Craig Topper2015-12-311-1/+0
| | | | llvm-svn: 256681
* [X86] Move shuffle decoding for constant pool into the X86CodeGen library to ↵Craig Topper2015-12-317-181/+238
| | | | | | remove a layering violation in the Util library. llvm-svn: 256680
* [LibCallSimplifier] propagate FMF when shrinking unary callsSanjay Patel2015-12-312-19/+21
| | | | llvm-svn: 256679
* change function names to avoid accidentally matching the substringSanjay Patel2015-12-311-40/+53
| | | | llvm-svn: 256678
* add 'fast' attribute to calls to show that the flag isn't being propagatedSanjay Patel2015-12-311-55/+57
| | | | llvm-svn: 256677
OpenPOWER on IntegriCloud