Commit message (Collapse) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
* | Remove unused MachineLoopRanges analysis. | Jakub Staszak | 2012-11-27 | 4 | -149/+0 | |
| | | | | llvm-svn: 168659 | |||||
* | Make comment names match function names. | Eric Christopher | 2012-11-27 | 1 | -3/+3 | |
| | | | | llvm-svn: 168644 | |||||
* | Add in sections for the fission case (no change so incorrect) and | Eric Christopher | 2012-11-27 | 1 | -1/+32 | |
| | | | | | | add a TODO for starting. llvm-svn: 168643 | |||||
* | Reorder section output ordering. | Eric Christopher | 2012-11-27 | 1 | -25/+27 | |
| | | | | llvm-svn: 168638 | |||||
* | Whitespace cleanup. | Eric Christopher | 2012-11-27 | 1 | -27/+27 | |
| | | | | llvm-svn: 168637 | |||||
* | Add an assertion to ensure freezeReservedRegs() is only ever called once. | Chad Rosier | 2012-11-26 | 1 | -0/+2 | |
| | | | | llvm-svn: 168633 | |||||
* | Now that the X86 Maximal Stack Alignment Check pass has been removed (i.e., | Chad Rosier | 2012-11-26 | 2 | -3/+0 | |
| | | | | | | | | | r168627), we no longer need to call the freezeReservedRegs() function a second time. Previously, this pass was conservatively adding the FP to the set of reserved registers, requiring the second update to the reserved registers. rdar://12719844 llvm-svn: 168631 | |||||
* | Now that the X86 Maximal Stack Alignment Check pass has been removed (i.e., | Chad Rosier | 2012-11-26 | 1 | -1/+0 | |
| | | | | | | | | | r168627), we no longer need to call the freezeReservedRegs() function a second time. Previously, this pass was conservatively adding the FP to the set of reserved registers, requiring the second update to the reserved registers. rdar://12719844 llvm-svn: 168630 | |||||
* | Don't use iterator after being erased. | Jakub Staszak | 2012-11-26 | 1 | -1/+1 | |
| | | | | llvm-svn: 168622 | |||||
* | Remove unneeded #includes. | Jakub Staszak | 2012-11-26 | 1 | -4/+0 | |
| | | | | llvm-svn: 168608 | |||||
* | Refactor to make helper method static. | Craig Topper | 2012-11-25 | 2 | -29/+14 | |
| | | | | llvm-svn: 168557 | |||||
* | Remove duplicate check of LimitFloatPrecision. It was already checked ↵ | Craig Topper | 2012-11-25 | 1 | -1/+1 | |
| | | | | | | earlier before IsExp10 could be set to true. llvm-svn: 168553 | |||||
* | Factor common code out of individual if blocks into common tail. | Craig Topper | 2012-11-25 | 1 | -24/+12 | |
| | | | | llvm-svn: 168551 | |||||
* | Remove redundant calls to getCurDebugLoc in visitIntrinsicCall. It's already ↵ | Craig Topper | 2012-11-24 | 1 | -7/+4 | |
| | | | | | | called at the start of the function and captured in a local variable. llvm-svn: 168548 | |||||
* | Refactor a bit to make some helper methods static. | Craig Topper | 2012-11-24 | 2 | -39/+20 | |
| | | | | llvm-svn: 168546 | |||||
* | Factor some common code out of individual if blocks. | Craig Topper | 2012-11-24 | 1 | -52/+27 | |
| | | | | llvm-svn: 168538 | |||||
* | Refactor a bit to make some helper functions static. | Craig Topper | 2012-11-23 | 2 | -54/+24 | |
| | | | | llvm-svn: 168524 | |||||
* | Cleanup: Simplify loop end logic in computeRegisterProperties(). | Patrik Hägglund | 2012-11-23 | 1 | -5/+4 | |
| | | | | llvm-svn: 168507 | |||||
* | Fix 80-col violation | Eli Bendersky | 2012-11-22 | 1 | -1/+2 | |
| | | | | llvm-svn: 168498 | |||||
* | llvm.fmuladd.* lowering should be checking isOperationLegalOrCustom, rather than | Lang Hames | 2012-11-22 | 1 | -1/+1 | |
| | | | | | | isOperationLegal. Thanks to Craig Topper for pointing this out. llvm-svn: 168485 | |||||
* | Pull some code out into functions to make rearranging them a bit easier. | Eric Christopher | 2012-11-22 | 2 | -46/+83 | |
| | | | | llvm-svn: 168481 | |||||
* | Whitespace. | Eric Christopher | 2012-11-21 | 1 | -29/+29 | |
| | | | | llvm-svn: 168402 | |||||
* | Update for some of the coding standard before rearranging functions | Eric Christopher | 2012-11-21 | 3 | -53/+53 | |
| | | | | | | around. llvm-svn: 168401 | |||||
* | Update some comments. | Eric Christopher | 2012-11-21 | 1 | -1/+2 | |
| | | | | llvm-svn: 168400 | |||||
* | Update and add some comments. | Eric Christopher | 2012-11-21 | 1 | -0/+8 | |
| | | | | llvm-svn: 168399 | |||||
* | Whitespace. | Eric Christopher | 2012-11-21 | 1 | -26/+26 | |
| | | | | llvm-svn: 168398 | |||||
* | Remove constness from this, it modifies the output stream as does | Eric Christopher | 2012-11-20 | 2 | -2/+2 | |
| | | | | | | everything else underneath. llvm-svn: 168395 | |||||
* | Remove unused function argument, add a bit to the comment. | Eric Christopher | 2012-11-20 | 2 | -5/+6 | |
| | | | | llvm-svn: 168387 | |||||
* | Formatting. | Eric Christopher | 2012-11-20 | 1 | -10/+11 | |
| | | | | llvm-svn: 168384 | |||||
* | Whitespace. | Eric Christopher | 2012-11-20 | 1 | -12/+8 | |
| | | | | llvm-svn: 168383 | |||||
* | Fix physical register liveness calculations: | Tim Northover | 2012-11-20 | 2 | -8/+13 | |
| | | | | | | | + Take account of clobbers + Give outputs priority over inputs since they happen later. llvm-svn: 168360 | |||||
* | Remove a function argument and propagate const around accordingly. | Eric Christopher | 2012-11-19 | 2 | -7/+11 | |
| | | | | llvm-svn: 168338 | |||||
* | Whitespace and 80-col. | Eric Christopher | 2012-11-19 | 1 | -49/+49 | |
| | | | | llvm-svn: 168337 | |||||
* | Make AsmPrinter::EmitTTypeReference() more robust - put | Anton Korobeynikov | 2012-11-19 | 3 | -19/+11 | |
| | | | | | | the zero GV check inside, so we won't forget it at the caller side. llvm-svn: 168328 | |||||
* | Factor out type info emission into separate routine. | Anton Korobeynikov | 2012-11-19 | 3 | -2/+64 | |
| | | | | | | | It turned out that ARM wants different layout of type infos. This is yet another patch in attempt to fix PR7187 llvm-svn: 168325 | |||||
* | Move section label emission to module end. Nothing should be | Eric Christopher | 2012-11-19 | 1 | -3/+3 | |
| | | | | | | | depending on them being emitted before the text and/or data sections and testing didn't uncover any. llvm-svn: 168321 | |||||
* | Handle mixed normal and early-clobber defs on inline asm. | Jakob Stoklund Olesen | 2012-11-19 | 1 | -2/+10 | |
| | | | | | | PR14376. llvm-svn: 168320 | |||||
* | Move else onto line with preceding closing brace. | Craig Topper | 2012-11-19 | 1 | -2/+1 | |
| | | | | llvm-svn: 168294 | |||||
* | Broaden isSchedulingBoundary to check aliases of SP. | Andrew Trick | 2012-11-17 | 1 | -1/+2 | |
| | | | | | | | | On PPC the stack pointer is X1, but ADJCALLSTACK writes R1. Fixes PR14315: Register regmask dependency problem with misched. llvm-svn: 168248 | |||||
* | Mark FP_EXTEND form v2f32 to v2f64 as "expand" for ARM NEON. Patch by Pete ↵ | Eli Friedman | 2012-11-17 | 1 | -0/+1 | |
| | | | | | | Couperus. llvm-svn: 168240 | |||||
* | Use array_pod_sort instead of std::sort. | Andrew Trick | 2012-11-16 | 1 | -41/+36 | |
| | | | | llvm-svn: 168203 | |||||
* | Remove conditions from 'else if' that were guaranteed by preceding 'if'. | Craig Topper | 2012-11-16 | 1 | -12/+12 | |
| | | | | llvm-svn: 168191 | |||||
* | Factor out the final FADD that's common to multiple code paths in the ↵ | Craig Topper | 2012-11-16 | 1 | -45/+30 | |
| | | | | | | visitLog* functions. llvm-svn: 168183 | |||||
* | Factor some common code to reduce compile size. | Craig Topper | 2012-11-16 | 1 | -37/+20 | |
| | | | | llvm-svn: 168143 | |||||
* | Mark FP_ROUND for converting NEON v2f64 to v2f32 as expand. Add a missing | Eli Friedman | 2012-11-15 | 1 | -0/+1 | |
| | | | | | | | | case to vector legalization so this actually works. Patch by Pete Couperus. Fixes PR12540. llvm-svn: 168107 | |||||
* | Use std::stable_sort instead of std::sort when sorting stack slots | Ulrich Weigand | 2012-11-15 | 1 | -1/+3 | |
| | | | | | | to guarantee deterministic code generation. llvm-svn: 168074 | |||||
* | [reg scavenger] Fix the isUsed/isAliasUsed functions so as to not report a false | Chad Rosier | 2012-11-15 | 1 | -1/+1 | |
| | | | | | | | | | | | | | | | | positive. In this particular case, R6 was being spilled by the register scavenger when it was in fact dead. The isUsed function reported R6 as used because the R6_R7 alias was reserved (due to the fact that we've reserved R7 as the FP). The solution is to only check if the original register (i.e., R6) isReserved and not the aliases. The aliases are only checked to make sure they're available. The test case is derived from one of the nightly tester benchmarks and is rather intractable and difficult to reproduce, so I haven't included it. rdar://12592448 llvm-svn: 168054 | |||||
* | Fix indeterminism in MI scheduler DAG construction. | Sergei Larin | 2012-11-15 | 1 | -15/+15 | |
| | | | | | | | Similarly to several recent fixes throughout the code replace std::map use with the MapVector. Add find() method to the MapVector. llvm-svn: 168051 | |||||
* | Add llvm.ceil, llvm.trunc, llvm.rint, llvm.nearbyint intrinsics. | Craig Topper | 2012-11-15 | 1 | -0/+20 | |
| | | | | llvm-svn: 168025 | |||||
* | Fix an obvious merge bug in -join-globalcopies (disabled). | Andrew Trick | 2012-11-15 | 1 | -1/+3 | |
| | | | | | | | | Jakub Staszak spotted this in review. I don't notice these things until I manually rerun benchmarks. But reducing unit tests is a very high priority. llvm-svn: 168021 |