summaryrefslogtreecommitdiffstats
path: root/llvm/lib/CodeGen
Commit message (Collapse)AuthorAgeFilesLines
* Speed up cycle checking significantly by caching results.David Greene2010-02-231-3/+9
| | | | llvm-svn: 96956
* Dead code eliminationJakob Stoklund Olesen2010-02-232-14/+0
| | | | llvm-svn: 96837
* Updated version of r96634 (which was reverted due to failing 176.gcc andJim Grosbach2010-02-221-3/+4
| | | | | | | | 126.gcc nightly tests. These failures uncovered latent bugs that machine DCE could remove one half of a stack adjust down/up pair, causing PEI to assert. This update fixes that, and the tests now pass. llvm-svn: 96822
* Fix various doxygen warnings.Dan Gohman2010-02-221-2/+2
| | | | llvm-svn: 96779
* It turned out that we failed to emit proper symbol stubs on non-x86/darwin ↵Anton Korobeynikov2010-02-211-1/+29
| | | | | | | | | for ages (we emitted a reference to a stub, but no stub was emitted). The code inside x86-32/macho target objfile lowering should actually be the generic one - move it there. This (I really, really hope) should fix EH issues on ppc/darwin and arm/darwin. llvm-svn: 96755
* Enable assertion to detect cyclic valno references.Jakob Stoklund Olesen2010-02-211-1/+1
| | | | | | This changes the stack overflow in PR6363 to an assertion failure. llvm-svn: 96744
* Revert 96634. It causes assertion failures for 126.gcc and 176.gcc inBob Wilson2010-02-191-4/+3
| | | | | | the armv6 nightly tests. llvm-svn: 96691
* Add support for the 'alignstack' attribute to the x86 backend. Fixes PR5254.Charles Davis2010-02-191-0/+3
| | | | | | Also, FileCheck'ize a test. llvm-svn: 96686
* Revert Anton's most recent EH patch (r96637), since it breaks a lot ofBob Wilson2010-02-191-17/+0
| | | | | | ARM and Thumb tests. llvm-svn: 96680
* Revert commits 96556 and 96640, because commit 96556 breaks theDuncan Sands2010-02-192-80/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | dragonegg self-host build. I reverted 96640 in order to revert 96556 (96640 goes on top of 96556), but it also looks like with both of them applied the breakage happens even earlier. The symptom of the 96556 miscompile is the following crash: llvm[3]: Compiling AlphaISelLowering.cpp for Release build cc1plus: /home/duncan/tmp/tmp/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:4982: void llvm::SelectionDAG::ReplaceAllUsesWith(llvm::SDNode*, llvm::SDNode*, llvm::SelectionDAG::DAGUpdateListener*): Assertion `(!From->hasAnyUseOfValue(i) || From->getValueType(i) == To->getValueType(i)) && "Cannot use this version of ReplaceAllUsesWith!"' failed. Stack dump: 0. Running pass 'X86 DAG->DAG Instruction Selection' on function '@_ZN4llvm19AlphaTargetLowering14LowerOperationENS_7SDValueERNS_12SelectionDAGE' g++: Internal error: Aborted (program cc1plus) This occurs when building LLVM using LLVM built by LLVM (via dragonegg). Probably LLVM has miscompiled itself, though it may have miscompiled GCC and/or dragonegg itself: at this point of the self-host build, all of GCC, LLVM and dragonegg were built using LLVM. Unfortunately this kind of thing is extremely hard to debug, and while I did rummage around a bit I didn't find any smoking guns, aka obviously miscompiled code. Found by bisection. r96556 | evancheng | 2010-02-18 03:13:50 +0100 (Thu, 18 Feb 2010) | 5 lines Some dag combiner goodness: Transform br (xor (x, y)) -> br (x != y) Transform br (xor (xor (x,y), 1)) -> br (x == y) Also normalize (and (X, 1) == / != 1 -> (and (X, 1)) != / == 0 to match to "test on x86" and "tst on arm" r96640 | evancheng | 2010-02-19 01:34:39 +0100 (Fri, 19 Feb 2010) | 16 lines Transform (xor (setcc), (setcc)) == / != 1 to (xor (setcc), (setcc)) != / == 1. e.g. On x86_64 %0 = icmp eq i32 %x, 0 %1 = icmp eq i32 %y, 0 %2 = xor i1 %1, %0 br i1 %2, label %bb, label %return => testl %edi, %edi sete %al testl %esi, %esi sete %cl cmpb %al, %cl je LBB1_2 llvm-svn: 96672
* Transform (xor (setcc), (setcc)) == / != 1 toEvan Cheng2010-02-191-3/+13
| | | | | | | | | | | | | | | | | | | (xor (setcc), (setcc)) != / == 1. e.g. On x86_64 %0 = icmp eq i32 %x, 0 %1 = icmp eq i32 %y, 0 %2 = xor i1 %1, %0 br i1 %2, label %bb, label %return => testl %edi, %edi sete %al testl %esi, %esi sete %cl cmpb %al, %cl je LBB1_2 llvm-svn: 96640
* Use the same encoding for EH stuff uniformly on all MachO targets.Anton Korobeynikov2010-02-191-0/+17
| | | | | | This hopefulyl should unbreak EH on PPC/Darwin. llvm-svn: 96637
* Radar 7636153. In the presence of large call frames, it's not sufficientJim Grosbach2010-02-191-3/+4
| | | | | | | | | for ARM to just check if a function has a FP to determine if it's safe to simplify the stack adjustment pseudo ops prior to eliminating frame indices. Allow targets to override the default behavior and does so for ARM and Thumb2. llvm-svn: 96634
* Always normalize spill weights, also for intervals created by spilling.Jakob Stoklund Olesen2010-02-182-4/+10
| | | | | | | | | | | | | | Moderate the weight given to very small intervals. The spill weight given to new intervals created when spilling was not normalized in the same way as the original spill weights calculated by CalcSpillWeights. That meant that restored registers would tend to hang around because they had a much higher spill weight that unspilled registers. This improves the runtime of a few tests by up to 10%, and there are no significant regressions. llvm-svn: 96613
* Make CodePlacementOpt detect special EH control flow byDan Gohman2010-02-181-7/+8
| | | | | | | | | | checking whether AnalyzeBranch disagrees with the CFG directly, rather than looking for EH_LABEL instructions. EH_LABEL instructions aren't always at the end of the block, due to FP_REG_KILL and other things. This fixes an infinite loop compiling MultiSource/Benchmarks/Bullet. llvm-svn: 96611
* Uniformize the way these options are printed. Requested byDuncan Sands2010-02-182-2/+2
| | | | | | Russell Wallace. llvm-svn: 96580
* Remove terminating dot in description. Inconsistency pointedDuncan Sands2010-02-181-1/+1
| | | | | | out by Russell Wallace. llvm-svn: 96579
* Some dag combiner goodness:Evan Cheng2010-02-182-6/+70
| | | | | | | | Transform br (xor (x, y)) -> br (x != y) Transform br (xor (xor (x,y), 1)) -> br (x == y) Also normalize (and (X, 1) == / != 1 -> (and (X, 1)) != / == 0 to match to "test on x86" and "tst on arm" llvm-svn: 96556
* Make the non-temporal bit "significant" in MemSDNodes so they aren'tDavid Greene2010-02-171-8/+17
| | | | | | CSE'd or otherwise combined with temporal MemSDNodes. llvm-svn: 96505
* Dead code elimination.Jakob Stoklund Olesen2010-02-172-38/+0
| | | | llvm-svn: 96496
* move isOnlyReachableByFallthrough out of MachineBasicBlock into AsmPrinter,Chris Lattner2010-02-172-31/+34
| | | | | | | and add a sparc implementation that knows about delay slots. Patch by Nathan Keynes! llvm-svn: 96492
* sink special case "cannotyetselect" for intrinsics out of the Chris Lattner2010-02-171-0/+5
| | | | | | tblgen splatted code into the implementation. llvm-svn: 96460
* Removed an early out which was causing the PBQP allocator to not compute ↵Lang Hames2010-02-171-4/+0
| | | | | | live-in sets or run the rewriter. llvm-svn: 96450
* Before setting scope end marker, pay attention to scope begin marker and ↵Devang Patel2010-02-171-1/+12
| | | | | | existing scope end marker, if any. Scope must begin before it ends and nested inlined scope do not truncate surrounding scope. llvm-svn: 96445
* There are two ways of checking for a given type, for example isa<PointerType>(T)Duncan Sands2010-02-163-11/+11
| | | | | | | and T->isPointerTy(). Convert most instances of the first form to the second form. Requested by Chris. llvm-svn: 96344
* Handle DBG_VALUE mixed with labels when doing PHIDale Johannesen2010-02-161-1/+18
| | | | | | | elimination. Before a DBG_VALUE could affect codegen. The solution here is imperfect and not final. llvm-svn: 96318
* Reapply 96294; now that I've gotten around to lookingDale Johannesen2010-02-161-4/+26
| | | | | | | at older buildbot messages, I see the failure predates this patch. llvm-svn: 96307
* Something broke. Hard to believe it was this patchDale Johannesen2010-02-161-26/+4
| | | | | | | but it's harder to believe it's the other candidate, so reverting. Temporarily I hope. llvm-svn: 96303
* Fix a memory leak. Patch by Nicolas Geoffray.Evan Cheng2010-02-151-0/+1
| | | | llvm-svn: 96295
* More handling of DBG_VALUE.Dale Johannesen2010-02-151-4/+26
| | | | llvm-svn: 96294
* Add missed entry to cmake build list fileAnton Korobeynikov2010-02-151-0/+1
| | | | llvm-svn: 96292
* Use ttype encoding consistentlyAnton Korobeynikov2010-02-151-1/+1
| | | | llvm-svn: 96290
* Move TLOF implementations to libCodegen to resolve layering violation.Anton Korobeynikov2010-02-151-0/+874
| | | | llvm-svn: 96288
* It turns out that we emitted completely wrong common EH frame since the ↵Anton Korobeynikov2010-02-151-7/+9
| | | | | | early beginning! The "z" CIE augmentation mark indicates that there is an extra word containing the size of augmentation section. However, even for frames w/o any augmentation data we emitted size uleb! (this is the case of "unwinding through C code"). Thus frame moves information was emitterd completely wrong. llvm-svn: 96287
* Preliminary patch to improve dwarf EH generation - Hooks to return ↵Anton Korobeynikov2010-02-155-93/+109
| | | | | | Personality / FDE / LSDA / TType encoding depending on target / options (e.g. code model / relocation model) - MCIzation of Dwarf EH printer to use encoding information - Stub generation for ELF target (needed for indirect references) - Some other small changes here and there llvm-svn: 96285
* Fix PR6300.Jakob Stoklund Olesen2010-02-151-3/+2
| | | | | | | A virtual register can be used before it is defined in the same MBB if the MBB is part of a loop. Teach the implicit-def pass about this case. llvm-svn: 96279
* Split SelectionDAGISel::IsLegalAndProfitableToFold toEvan Cheng2010-02-151-8/+13
| | | | | | | | IsLegalToFold and IsProfitableToFold. The generic version of the later simply checks whether the folding candidate has a single use. This allows the target isel routines more flexibility in deciding whether folding makes sense. The specific case we are interested in is folding constant pool loads with multiple uses. llvm-svn: 96255
* Add non-temporal flags and remove an assumption of default arguments.David Greene2010-02-1510-174/+295
| | | | llvm-svn: 96240
* Add non-temporal flags to MachineMemOperand.David Greene2010-02-151-2/+3
| | | | llvm-svn: 96226
* Uniformize the names of type predicates: rather than having isFloatTy andDuncan Sands2010-02-153-8/+8
| | | | | | isInteger, we now have isFloatTy and isIntegerTy. Requested by Chris! llvm-svn: 96223
* Ignore DBG_VALUE in a couple more places.Dale Johannesen2010-02-151-1/+4
| | | | llvm-svn: 96207
* Fix PR6283.Jakob Stoklund Olesen2010-02-132-19/+38
| | | | | | | | | | | When coalescing with a physreg, remember to add imp-def and imp-kill when dealing with sub-registers. Also fix a related bug in VirtRegRewriter where substitutePhysReg may reallocate the operand list on an instruction and invalidate the reg_iterator. This can happen when a register is mentioned twice on the same instruction. llvm-svn: 96072
* Teach MachineFrameInfo to track maximum alignment while stack objects are beingEvan Cheng2010-02-131-13/+2
| | | | | | | | | | created. This ensures it's updated at all time. It means targets which perform dynamic stack alignment would know whether it is required and whether frame pointer register cannot be made available register allocation. This is a fix for rdar://7625239. Sorry, I can't create a reasonably sized test case. llvm-svn: 96069
* Besides removing phi cycles that reduce to a single value, also remove deadBob Wilson2010-02-132-21/+72
| | | | | | | | phi cycles. Adjust a few tests to keep dead instructions from being optimized away. This (together with my previous change for phi cycles) fixes Apple radar 7627077. llvm-svn: 96057
* give MCCodeEmitters access to the current MCContext.Chris Lattner2010-02-121-1/+1
| | | | llvm-svn: 96038
* Rewrite handling of DBG_VALUE; previous algorithmDale Johannesen2010-02-121-16/+24
| | | | | | | | | | | didn't handle X = Y<dead> = use X DBG_VALUE(X) I was hoping to avoid this approach as it's slower, but I don't think it can be done. llvm-svn: 95996
* Reapply the new LoopStrengthReduction code, with compile time andDan Gohman2010-02-121-0/+4
| | | | | | | | | | bug fixes, and with improved heuristics for analyzing foreign-loop addrecs. This change also flattens IVUsers, eliminating the stride-oriented groupings, which makes it easier to work with. llvm-svn: 95975
* * Updated the cost matrix normalization proceedure to better handle infinite ↵Lang Hames2010-02-122-13/+21
| | | | | | | | | | costs. * Enabled R1/R2 application for nodes with infinite spill costs in the Briggs heuristic (made safe by the changes to the normalization proceedure). * Removed a redundant header. llvm-svn: 95973
* Add a new pass on machine instructions to optimize away PHI cycles that Bob Wilson2010-02-123-0/+143
| | | | | | | | | | | reduce down to a single value. InstCombine already does this transformation but DAG legalization may introduce new opportunities. This has turned out to be important for ARM where 64-bit values are split up during type legalization: InstCombine is not able to remove the PHI cycles on the 64-bit values but the separate 32-bit values can be optimized. I measured the compile time impact of this (running llc on 176.gcc) and it was not significant. llvm-svn: 95951
* Reapply coalescer fix for better cross-class coalescing.Jakob Stoklund Olesen2010-02-111-1/+1
| | | | | | This time with fixed test cases. llvm-svn: 95938
OpenPOWER on IntegriCloud