summaryrefslogtreecommitdiffstats
path: root/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
Commit message (Collapse)AuthorAgeFilesLines
...
* Fixed a bug in SelectionDAG.cpp.Elena Demikhovsky2012-01-031-6/+50
| | | | | | | | | | | | The failure seen on win32, when i64 type is illegal. It happens on stage of conversion VECTOR_SHUFFLE to BUILD_VECTOR. The failure message is: llc: SelectionDAG.cpp:784: void VerifyNodeCommon(llvm::SDNode*): Assertion `(I->getValueType() == EltVT || (EltVT.isInteger() && I->getValueType().isInteger() && EltVT.bitsLE(I->getValueType()))) && "Wrong operand type!"' failed. I added a special test that checks vector shuffle on win32. llvm-svn: 147445
* Revert 147399. It broke CodeGen/ARM/vext.ll.Rafael Espindola2012-01-011-39/+5
| | | | llvm-svn: 147400
* Fixed a bug in SelectionDAG.cpp.Elena Demikhovsky2012-01-011-5/+39
| | | | | | | | | | | | The failure seen on win32, when i64 type is illegal. It happens on stage of conversion VECTOR_SHUFFLE to BUILD_VECTOR. The failure message is: llc: SelectionDAG.cpp:784: void VerifyNodeCommon(llvm::SDNode*): Assertion `(I->getValueType() == EltVT || (EltVT.isInteger() && I->getValueType().isInteger() && EltVT.bitsLE(I->getValueType()))) && "Wrong operand type!"' failed. I added a special test that checks vector shuffle on win32. llvm-svn: 147399
* Initial CodeGen support for CTTZ/CTLZ where a zero input produces anChandler Carruth2011-12-131-3/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | undefined result. This adds new ISD nodes for the new semantics, selecting them when the LLVM intrinsic indicates that the undef behavior is desired. The new nodes expand trivially to the old nodes, so targets don't actually need to do anything to support these new nodes besides indicating that they should be expanded. I've done this for all the operand types that I could figure out for all the targets. Owners of various targets, please review and let me know if any of these are incorrect. Note that the expand behavior is *conservatively correct*, and exactly matches LLVM's current behavior with these operations. Ideally this patch will not change behavior in any way. For example the regtest suite finds the exact same instruction sequences coming out of the code generator. That's why there are no new tests here -- all of this is being exercised by the existing test suite. Thanks to Duncan Sands for reviewing the various bits of this patch and helping me get the wrinkles ironed out with expanding for each target. Also thanks to Chris for clarifying through all the discussions that this is indeed the approach he was looking for. That said, there are likely still rough spots. Further review much appreciated. llvm-svn: 146466
* Remove dead llvm.eh.sjlj.dispatchsetup intrinsic.Bill Wendling2011-11-281-2/+0
| | | | llvm-svn: 145263
* Add a couple asserts so it will be easier to debug if we accidentally pass ↵Eli Friedman2011-11-161-0/+4
| | | | | | indexed loads/stores to the legalizer. llvm-svn: 144767
* Some cleanup and bulletproofing for node replacement in LegalizeDAG. To ↵Eli Friedman2011-11-111-57/+59
| | | | | | | | maintain LegalizeDAG invariants, whenever we a node is replaced, we must attempt to delete it, and if it still has uses after it is replaced (which can happen in rare cases due to CSE), we must revisit it. llvm-svn: 144432
* Added invariant field to the DAG.getLoad method and changed all calls.Pete Cooper2011-11-081-20/+25
| | | | | | When this field is true it means that the load is from constant (runt-time or compile-time) and so can be hoisted from loops or moved around other memory accesses llvm-svn: 144100
* Add a bunch of calls to RemoveDeadNode in LegalizeDAG, so legalization ↵Eli Friedman2011-11-081-3/+19
| | | | | | | | doesn't get confused by CSE later on. Fixes PR11318. Re-commit of r144034, with an extra fix so that RemoveDeadNode doesn't blow up. llvm-svn: 144055
* Revert r144034 while I try to track down a crash.Eli Friedman2011-11-071-19/+3
| | | | llvm-svn: 144044
* Add a bunch of calls to RemoveDeadNode in LegalizeDAG, so legalization ↵Eli Friedman2011-11-071-3/+19
| | | | | | doesn't get confused by CSE later on. Fixes PR11318. llvm-svn: 144034
* Reapply r143206, with fixes. Disallow physical register lifetimesDan Gohman2011-11-031-515/+257
| | | | | | | across calls, and only check for nested dependences on the special call-sequence-resource register. llvm-svn: 143660
* Revert r143206, as there are still some failing tests.Dan Gohman2011-10-291-257/+515
| | | | llvm-svn: 143262
* Reapply r143177 and r143179 (reverting r143188), with schedulerDan Gohman2011-10-281-515/+257
| | | | | | | | | fixes: Use a separate register, instead of SP, as the calling-convention resource, to avoid spurious conflicts with actual uses of SP. Also, fix unscheduling of calling sequences, which can be triggered by pseudo-two-address dependencies. llvm-svn: 143206
* Speculatively disable Dan's commits 143177 and 143179 to see ifDuncan Sands2011-10-281-257/+515
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | it fixes the dragonegg self-host (it looks like gcc is miscompiled). Original commit messages: Eliminate LegalizeOps' LegalizedNodes map and have it just call RAUW on every node as it legalizes them. This makes it easier to use hasOneUse() heuristics, since unneeded nodes can be removed from the DAG earlier. Make LegalizeOps visit the DAG in an operands-last order. It previously used operands-first, because LegalizeTypes has to go operands-first, and LegalizeTypes used to be part of LegalizeOps, but they're now split. The operands-last order is more natural for several legalization tasks. For example, it allows lowering code for nodes with floating-point or vector constants to see those constants directly instead of seeing the lowered form (often constant-pool loads). This makes some things somewhat more complicated today, though it ought to allow things to be simpler in the future. It also fixes some bugs exposed by Legalizing using RAUW aggressively. Remove the part of LegalizeOps that attempted to patch up invalid chain operands on libcalls generated by LegalizeTypes, since it doesn't work with the new LegalizeOps traversal order. Instead, define what LegalizeTypes is doing to be correct, and transfer the responsibility of keeping calls from having overlapping calling sequences into the scheduler. Teach the scheduler to model callseq_begin/end pairs as having a physical register definition/use to prevent calls from having overlapping calling sequences. This is also somewhat complicated, though there are ways it might be simplified in the future. This addresses rdar://9816668, rdar://10043614, rdar://8434668, and others. Please direct high-level questions about this patch to management. Delete #if 0 code accidentally left in. llvm-svn: 143188
* Delete #if 0 code accidentally left in.Dan Gohman2011-10-281-17/+0
| | | | llvm-svn: 143179
* Eliminate LegalizeOps' LegalizedNodes map and have it just call RAUWDan Gohman2011-10-281-514/+273
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | on every node as it legalizes them. This makes it easier to use hasOneUse() heuristics, since unneeded nodes can be removed from the DAG earlier. Make LegalizeOps visit the DAG in an operands-last order. It previously used operands-first, because LegalizeTypes has to go operands-first, and LegalizeTypes used to be part of LegalizeOps, but they're now split. The operands-last order is more natural for several legalization tasks. For example, it allows lowering code for nodes with floating-point or vector constants to see those constants directly instead of seeing the lowered form (often constant-pool loads). This makes some things somewhat more complicated today, though it ought to allow things to be simpler in the future. It also fixes some bugs exposed by Legalizing using RAUW aggressively. Remove the part of LegalizeOps that attempted to patch up invalid chain operands on libcalls generated by LegalizeTypes, since it doesn't work with the new LegalizeOps traversal order. Instead, define what LegalizeTypes is doing to be correct, and transfer the responsibility of keeping calls from having overlapping calling sequences into the scheduler. Teach the scheduler to model callseq_begin/end pairs as having a physical register definition/use to prevent calls from having overlapping calling sequences. This is also somewhat complicated, though there are ways it might be simplified in the future. This addresses rdar://9816668, rdar://10043614, rdar://8434668, and others. Please direct high-level questions about this patch to management. llvm-svn: 143177
* Move the legalization of vector loads and stores into LegalizeVectorOps. In someNadav Rotem2011-10-151-183/+4
| | | | | | cases we need the second type-legalization pass in order to support all cases. llvm-svn: 142060
* Add support for legalization of vector SHL/SRA/SRL instructionsNadav Rotem2011-10-111-0/+25
| | | | llvm-svn: 141667
* Add support for legalization of vector trunc-store where the saved scalar ↵Nadav Rotem2011-10-111-13/+13
| | | | | | type is illegal (for example, v2i16 on systems where the smallest store size is i32) llvm-svn: 141661
* Cleanup the trunc-store legalization code and add asserts.Nadav Rotem2011-10-111-68/+87
| | | | llvm-svn: 141659
* Moved type construction out of the loop and added an assert on the legality ↵Nadav Rotem2011-10-011-10/+10
| | | | | | of the type. Formatted lines to the 80 char limit. llvm-svn: 140952
* Revert r140463; The patch assumes that <4 x i1> is saved to memory as 4 x i8,Nadav Rotem2011-09-271-10/+1
| | | | | | while the decision is to bit-pack small values. llvm-svn: 140601
* [Vector-Select] Address one of the problems in 10902.Nadav Rotem2011-09-241-1/+10
| | | | | | | | | | When generating the trunc-store of i1's, we need to use the vector type and not the scalar type. This patch fixes the assertion in CodeGen/Generic/bool-vector.ll when running with -promote-elements. llvm-svn: 140463
* Some legalization fixes for atomic load and store.Eli Friedman2011-09-151-1/+1
| | | | llvm-svn: 139851
* Split the init.trampoline intrinsic, which currently combines GCC'sDuncan Sands2011-09-061-1/+2
| | | | | | | | | | | | | | | | | | | | init.trampoline and adjust.trampoline intrinsics, into two intrinsics like in GCC. While having one combined intrinsic is tempting, it is not natural because typically the trampoline initialization needs to be done in one function, and the result of adjust trampoline is needed in a different (nested) function. To get around this llvm-gcc hacks the nested function lowering code to insert an additional parent variable holding the adjust.trampoline result that can be accessed from the child function. Dragonegg doesn't have the luxury of tweaking GCC code, so it stored the result of adjust.trampoline in the memory GCC set aside for the trampoline itself (this is always available in the child function), and set up some new memory (using an alloca) to hold the trampoline. Unfortunately this breaks Go which allocates trampoline memory on the heap and wants to use it even after the parent has exited (!). Rather than doing even more hacks to get Go working, it seemed best to just use two intrinsics like in GCC. Patch mostly by Sanjoy Das. llvm-svn: 139140
* Revert r131152, r129796, r129761. This code is currently consideredDan Gohman2011-09-011-52/+43
| | | | | | | | to be unreliable on platforms which require memcpy calls, and it is complicating broader legalize cleanups. It is hoped that these cleanups will make memcpy byval easier to implement in the future. llvm-svn: 138977
* Atomic load/store on ARM/Thumb.Eli Friedman2011-08-261-0/+26
| | | | | | | | | | | | I don't really like the patterns, but I'm having trouble coming up with a better way to handle them. I plan on making other targets use the same legalization ARM-without-memory-barriers is using... it's not especially efficient, but if anyone cares, it's not that hard to fix for a given target if there's some better lowering. llvm-svn: 138621
* Basic x86 code generation for atomic load and store instructions.Eli Friedman2011-08-241-0/+5
| | | | llvm-svn: 138478
* Code generation for 'fence' instruction.Eli Friedman2011-07-271-0/+2
| | | | llvm-svn: 136283
* Revert r136156, which broke several buildbots.Dan Gohman2011-07-271-1/+14
| | | | llvm-svn: 136206
* Delete unnecessarily cautious LastCALLSEQ code.Dan Gohman2011-07-261-14/+1
| | | | llvm-svn: 136156
* land David Blaikie's patch to de-constify Type, with a few tweaks.Chris Lattner2011-07-181-14/+14
| | | | llvm-svn: 135375
* LegalizeDAG doesn't need its own copy of this enum.Dan Gohman2011-07-151-7/+1
| | | | llvm-svn: 135320
* Delete LegalizeDAG's own version of isTypeLegal and getTypeActionDan Gohman2011-07-151-26/+16
| | | | | | and just use the ones from TargetLowering directly. llvm-svn: 135318
* Delete an unused variable and a redundant assert.Dan Gohman2011-07-151-9/+1
| | | | llvm-svn: 135311
* Modernize comments.Dan Gohman2011-07-151-7/+4
| | | | llvm-svn: 135305
* Add an intrinsic and codegen support for fused multiply-accumulate. The intentCameron Zwarich2011-07-081-0/+4
| | | | | | is to use this for architectures that have a native FMA instruction. llvm-svn: 134742
* Add a testcase for checking the integer-promotion of many different vectorNadav Rotem2011-06-141-0/+167
| | | | | | | | | | | | | types (with power of two types such as 8,16,32 .. 512). Fix a bug in the integer promotion of bitcast nodes. Enable integer expanding only if the target of the conversion is an integer (when the type action is scalarize). Add handling to the legalization of vector load/store in cases where the saved vector is integer-promoted. llvm-svn: 132985
* Refactor getActionType and getTypeToTransformTo ; place all of the 'decision'Nadav Rotem2011-05-271-1/+1
| | | | | | code in one place. Re-apply 131534 and fix the multi-step promotion of integers. llvm-svn: 132217
* Revert commit 131534 since it seems to have broken several buildbots.Duncan Sands2011-05-181-1/+1
| | | | | | | | Original log entry: Refactor getActionType and getTypeToTransformTo ; place all of the 'decision' code in one place. llvm-svn: 131536
* Refactor getActionType and getTypeToTransformTo ; place all of the 'decision'Nadav Rotem2011-05-181-1/+1
| | | | | | code in one place. llvm-svn: 131534
* Misc. code cleanups.Dan Gohman2011-05-171-60/+59
| | | | llvm-svn: 131497
* Delete unused variables.Dan Gohman2011-05-161-7/+5
| | | | llvm-svn: 131430
* Trim #includes.Dan Gohman2011-05-161-7/+0
| | | | llvm-svn: 131429
* Fix whitespace and 80-column violations.Dan Gohman2011-05-161-10/+10
| | | | llvm-svn: 131428
* Since I can't reproduce the failures from 131261, re-trying with aStuart Hastings2011-05-131-2/+6
| | | | | | simplified version. <rdar://problem/9298790> llvm-svn: 131274
* Revert 131266 and 131261 due to buildbot complaints.Stuart Hastings2011-05-131-9/+2
| | | | | | rdar://problem/9298790 llvm-svn: 131269
* Non-fast-isel followup to 129634; correctly handle branches controlledStuart Hastings2011-05-121-2/+9
| | | | | | | | | | | | | by non-CMP expressions. The executable test case (129821) would test this as well, if we had an "-O0 -disable-arm-fast-isel" LLVM-GCC tester. Alas, the ARM assembly would be very difficult to check with FileCheck. The thumb2-cbnz.ll test is affected; it generates larger code (tst.w vs. cmp #0), but I believe the new version is correct. rdar://problem/9298790 llvm-svn: 131261
* Correctly walk through nested and adjacent CALLSEQ_START nodes. NoStuart Hastings2011-05-101-1/+2
| | | | | | | test case; I've only seen this on a release branch, and I can't get it to reproduce on trunk. rdar://problem/7662569 llvm-svn: 131152
OpenPOWER on IntegriCloud