summaryrefslogtreecommitdiffstats
path: root/llvm/lib
Commit message (Collapse)AuthorAgeFilesLines
...
* ARM: Use a dedicated intrinsic for vector bitwise select.Jim Grosbach2012-09-211-2/+29
| | | | | | | | | | | The expression based expansion too often results in IR level optimizations splitting the intermediate values into separate basic blocks, preventing the formation of the VBSL instruction as the code author intended. In particular, LICM would often hoist part of the computation out of a loop. rdar://11011471 llvm-svn: 164340
* Ignore PHI-defs for -new-coalescer interference checks.Jakob Stoklund Olesen2012-09-201-4/+8
| | | | | | | | A PHI can't create interference on its own. If two live ranges interfere at a PHI, they must also interfere when leaving one of the PHI predecessors. llvm-svn: 164330
* Extend -new-coalescer SSA update to handle mapped values as well.Jakob Stoklund Olesen2012-09-202-9/+62
| | | | | | | | | | | | | The old-fashioned many-to-one value mapping doesn't always work when merging vector lanes. A value can map to multiple different values, and it can even be necessary to insert new PHIs. When a value number is defined by a copy from a value number that required SSa update, include the live range of the copied value number in the SSA update as well. It is not necessarily a copy of the original value number any longer. llvm-svn: 164329
* Only emit DW_AT_object_pointer if this is a definition.Eric Christopher2012-09-201-3/+0
| | | | llvm-svn: 164326
* SimplifyCFG: sink common codes from IF, ELSE blocks down to END block.Manman Ren2012-09-201-0/+173
| | | | | | | | | | | | We already have HoistThenElseCodeToIf, this patch implements SinkThenElseCodeToEnd. When END block has only two predecessors and each predecessor terminates with unconditional branches, we compare instructions in IF and ELSE blocks backwards and check whether we can sink the common instructions down. rdar://12191395 llvm-svn: 164325
* Revert r164308 to fix buildbots.Bill Wendling2012-09-204-12/+11
| | | | llvm-svn: 164309
* Make the 'get*AlignmentFromAttr' functions into member functions within the ↵Bill Wendling2012-09-204-11/+12
| | | | | | Attributes class. llvm-svn: 164308
* Remove more bare uses of the different Attribute enums.Bill Wendling2012-09-201-6/+6
| | | | llvm-svn: 164307
* Make the 'getAsString' function a method of the Attributes class.Bill Wendling2012-09-203-48/+48
| | | | llvm-svn: 164305
* Fix 80-col violations.Nadav Rotem2012-09-201-13/+19
| | | | llvm-svn: 164297
* Change enum type in a static table to uint8_t instead. Saves about 700 ↵Craig Topper2012-09-201-6/+6
| | | | | | hundred bytes of static data. Change unsigned char in same table to uint8_t for explicitness. llvm-svn: 164285
* Re-work X86 code generation of atomic ops with spin-loopMichael Liao2012-09-205-540/+508
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
* Convert some attribute existence queries over to use the predicate methods.Bill Wendling2012-09-194-38/+38
| | | | llvm-svn: 164268
* Add predicates for queries on whether an attribute exists.Bill Wendling2012-09-191-7/+6
| | | | llvm-svn: 164264
* Add in new data types that are used by AMDIL/ANL among others.Micah Villmow2012-09-191-0/+20
| | | | llvm-svn: 164261
* Resolve conflicts involving dead vector lanes for -new-coalescer.Jakob Stoklund Olesen2012-09-191-9/+155
| | | | | | | | | | | | | | | | | | | | | A common coalescing conflict in vector code is lane insertion: %dst = FOO %src = BAR %dst:ssub0 = COPY %src The live range of %src interferes with the ssub0 lane of %dst, but that lane is never read after %src would have clobbered it. That makes it safe to merge the live ranges and eliminate the COPY: %dst = FOO %dst:ssub0 = BAR This patch teaches the new coalescer to resolve conflicts where dead vector lanes would be clobbered, at least as long as the clobbered vector lanes don't escape the basic block. llvm-svn: 164250
* This patch adds memory support functions which will later be used to ↵Andrew Kaylor2012-09-193-90/+325
| | | | | | implement section-specific protection handling in MCJIT. llvm-svn: 164249
* Add support for macro parameters/arguments delimited by spaces,Preston Gurd2012-09-193-25/+133
| | | | | | | | | | to improve compatibility with GNU as. Based on a patch by PaX Team. Fixed assertion failures on non-Darwin and added additional test cases. llvm-svn: 164248
* Add support for accessing an MDNode's operands via the C binding. Patch byDuncan Sands2012-09-191-0/+13
| | | | | | Anthony Bryant. llvm-svn: 164247
* Support default parameters/arguments for assembler macros.Preston Gurd2012-09-191-7/+31
| | | | | | | | This patch is based on the one by PaX Team. Patch by Andy Zhang! llvm-svn: 164246
* Enhance unmatched '.endr' directive error message in assembler.Preston Gurd2012-09-191-1/+1
| | | | | | | | The directive can be matched with directives other than '.rept' Patch by Andy Zhang! llvm-svn: 164245
* Unify the logic in SelectAtomicLoadAdd and SelectAtomicLoadArithMichael Liao2012-09-191-176/+140
| | | | | | | | | | | - Merge the processing of LOAD_ADD with other atomic load-arith operations - Separate the logic getting target constant for atomic-load-op and add an optimization for atomic-load-add on i16 with negative value - Optimize a minor case for atomic-fetch-add i16 with negative operand. Test case is revised. llvm-svn: 164243
* Renaming functions to match coding style guidelinesMichael Ilseman2012-09-191-4/+4
| | | | llvm-svn: 164238
* Doxygen-ify commentsMichael Ilseman2012-09-191-8/+16
| | | | llvm-svn: 164235
* Put the * and & next to the variable, rather than the type.Michael Ilseman2012-09-191-73/+73
| | | | llvm-svn: 164232
* GCC doesn't understand that OrigAliasResult having a value is correlated withDuncan Sands2012-09-191-1/+1
| | | | | | | ArePhisAssumedNoAlias, and warns that OrigAliasResult may be used uninitialized. Pacify GCC. llvm-svn: 164229
* Small structs for PPC64 SVR4 must be passed right-justified in registers.Bill Schmidt2012-09-192-61/+143
| | | | | | | | | | | | | | | lib/Target/PowerPC/PPCISelLowering.{h,cpp} Rename LowerFormalArguments_Darwin to LowerFormalArguments_Darwin_Or_64SVR4. Rename LowerFormalArguments_SVR4 to LowerFormalArguments_32SVR4. Receive small structs right-justified in LowerFormalArguments_Darwin_Or_64SVR4. Rename LowerCall_Darwin to LowerCall_Darwin_Or_64SVR4. Rename LowerCall_SVR4 to LowerCall_32SVR4. Pass small structs right-justified in LowerCall_Darwin_Or_64SVR4. test/CodeGen/PowerPC/structsinregs.ll New test. llvm-svn: 164228
* SimplifyCFG: Don't generate invalid code for switch used to initializeHans Wennborg2012-09-191-9/+8
| | | | | | | | | | | | | two variables where the first variable is returned and the second ignored. I don't think this occurs in practice (other passes should have cleaned up the unused phi node), but it should still be handled correctly. Also make the logic for determining if we should return early less sketchy. llvm-svn: 164225
* IntegerDivision: Style cleanups, avoid warning about mixing || and && ↵Benjamin Kramer2012-09-191-7/+5
| | | | | | without parens. llvm-svn: 164216
* Prevent inlining of callees which allocate lots of memory into a recursive ↵Nadav Rotem2012-09-191-14/+58
| | | | | | | | | | | | | | | | | | caller. Example: void foo() { ... foo(); // I'm recursive! bar(); } bar() { int a[1000]; // large stack size } rdar://10853263 llvm-svn: 164207
* CodeGenPrep: turn lookup tables into switches for some targets.Hans Wennborg2012-09-191-4/+114
| | | | | | | | | | | | | | | | | | | This is a follow-up from r163302, which added a transformation to SimplifyCFG that turns some switches into loads from lookup tables. It was pointed out that some targets, such as GPUs and deeply embedded targets, might not find this appropriate, but SimplifyCFG doesn't have enough information about the target to decide this. This patch adds the reverse transformation to CodeGenPrep: it turns loads from lookup tables back into switches for targets where we do not build jump tables (assuming these are also the targets where lookup tables are inappropriate). Hopefully we will eventually get to have target information in SimplifyCFG, and then this CodeGenPrep transformation can be removed. llvm-svn: 164206
* Remove code for setting the VEX L-bit as a function of operand size from the ↵Craig Topper2012-09-193-22/+2
| | | | | | code emitters and the disassembler table builder. Fix a couple instructions that were still missing VEX_L. llvm-svn: 164204
* whitespace.Nadav Rotem2012-09-191-9/+0
| | | | llvm-svn: 164203
* Add explicit VEX_L tags to all 256-bit instructions. This will allow us to ↵Craig Topper2012-09-193-240/+254
| | | | | | remove code from the code emitters that examined operands to set the L-bit. llvm-svn: 164202
* De-nest if's and fix mix-upSean Silva2012-09-191-41/+37
| | | | | | | | | | | | | | | | | | Two deeply nested if's obscured that the sense of the conditions was mixed up. Amazingly, TableGen's output is exactly the same even with the sense of the tests fixed; it seems that all of TableGen's conversions are symmetric so that the inverted sense was nonetheless correct "by accident". As such, I couldn't come up with a test case. If there does in fact exist a non-symmetric conversion in TableGen's type system, then a test case should be prepared. Despite the symmetry, both if's are left in place for robustness in the face of future changes. Review by Jakob. llvm-svn: 164195
* Tidy up. Minor formatting.Jim Grosbach2012-09-181-3/+2
| | | | llvm-svn: 164182
* Tidy up. 80 columns.Jim Grosbach2012-09-181-2/+6
| | | | llvm-svn: 164181
* comment typoAndrew Trick2012-09-181-1/+1
| | | | llvm-svn: 164180
* Fix the last crasher I've gotten a reproduction for in SROA. This oneChandler Carruth2012-09-181-0/+2
| | | | | | | | | | | from the dragonegg build bots when we turned on the full version of the pass. Included a much reduced test case for this pesky bug, despite bugpoint's uncooperative behavior. Also, I audited all the similar code I could find and didn't spot any other cases where this mistake cropped up. llvm-svn: 164178
* New utility for expanding integer division for targets that don't support it.Michael Ilseman2012-09-182-0/+307
| | | | | | Implementation derived from compiler-rt's implementation of signed and unsigned integer division. llvm-svn: 164173
* MOVi16 (movw) is only legal on cpus with V6T2 support. rdar://12300648Evan Cheng2012-09-181-2/+4
| | | | llvm-svn: 164169
* Fix the isLocalCall() by checking for linker weakness as well.Roman Divacky2012-09-181-1/+2
| | | | llvm-svn: 164155
* TargetSchedule: cleanup computeOperandLatency logic & diagnostics.Andrew Trick2012-09-181-6/+16
| | | | llvm-svn: 164154
* misched: Make ScheduleDAGInstrs use the TargetSchedule interface.Andrew Trick2012-09-181-16/+14
| | | | llvm-svn: 164153
* Revert r164051.Akira Hatanaka2012-09-181-4/+0
| | | | llvm-svn: 164150
* LSR critical edge splitting fix for PR13756.Andrew Trick2012-09-181-11/+15
| | | | llvm-svn: 164147
* Fix getCommonType in a different way from the way I fixed it whenChandler Carruth2012-09-181-1/+1
| | | | | | | | | | | | working on FCA splitting. Instead of refusing to form a common type when there are uses of a subsection of the alloca as well as a use of the entire alloca, just skip the subsection uses and continue looking for a whole-alloca use with a type that we can use. This produces slightly prettier IR I think, and also fixes the other failure in the test. llvm-svn: 164146
* Fix build for compilers that don't understand injected class names properly.Benjamin Kramer2012-09-181-2/+2
| | | | llvm-svn: 164142
* Avoid symbol name clash when filling TOC.Roman Divacky2012-09-182-3/+14
| | | | | | Patch by Adhemerval Zanella. llvm-svn: 164141
* SROA: Use CRTP for OpSplitter to get rid of virtual dispatch and the ↵Benjamin Kramer2012-09-181-8/+7
| | | | | | virtual-dtor warnings that come with it. llvm-svn: 164140
OpenPOWER on IntegriCloud