aboutsummaryrefslogtreecommitdiff
path: root/lib/Target/ARM/ARMISelLowering.cpp
AgeCommit message (Collapse)Author
2011-11-26Move code into anonymous namespaces.Benjamin Kramer
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145154 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-16Fix ARM SjLj-EH dispatch setup code. <rdar://problem/10444602>Bob Wilson
The EmitBasePointerRecalculation function has 2 problems, one minor and one fatal. The minor problem is that it inserts the code at the setjmp instead of in the dispatch block. The fatal problem is that at the point where this code runs, we don't know whether there will be a base pointer, so the entire function is a no-op. The base pointer recalculation needs to be handled as it was before, by inserting a pseudo instruction that gets expanded late. Most of the support for the old approach is still here, but it no longer has any connection to the eh_sjlj_dispatchsetup intrinsic. Clean up the parts related to the intrinsic and just generate the pseudo instruction directly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144781 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-15Remove some unnecessary includes of PseudoSourceValue.h.Jay Foad
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144631 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-15Add vmov.f32 to materialize f32 immediate splats which cannot be handled byEvan Cheng
integer variants. rdar://10437054 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144608 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-11Make sure to expand SIGN_EXTEND_INREG for NEON vectors. PR11319, round 3.Eli Friedman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144361 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-09Make sure we correctly unroll conversions between v2f64 and v2i32 on ARM.Eli Friedman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144241 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-08Lower mem-ops to unaligned i32/i16 load/stores on ARM where supported.Lang Hames
Add support for trimming constants to GetDemandedBits. This fixes some funky constant generation that occurs when stores are expanded for targets that don't support unaligned stores natively. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144102 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-08Added invariant field to the DAG.getLoad method and changed all calls.Pete Cooper
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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144100 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-08Make sure to mark vector extload's as expand on ARM. Fixes PR11319.Eli Friedman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144057 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-03Reapply r143206, with fixes. Disallow physical register lifetimesDan Gohman
across calls, and only check for nested dependences on the special call-sequence-resource register. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143660 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-02Fixed parameter name.Lang Hames
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143594 91177308-0d34-0410-b5e6-96231b3b80d8
2011-11-02Try to lower memset/memcpy/memmove to vector instructions on ARM where the ↵Lang Hames
alignment permits. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143582 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-29Revert r143206, as there are still some failing tests.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143262 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-28Reapply r143177 and r143179 (reverting r143188), with schedulerDan Gohman
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. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143206 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-28Speculatively disable Dan's commits 143177 and 143179 to see ifDuncan Sands
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. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143188 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-28Eliminate LegalizeOps' LegalizedNodes map and have it just call RAUWDan Gohman
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. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143177 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-26Make sure short memsets on ARM lower to stores, even when optimizing for size.Lang Hames
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143055 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-26Revert r142530 at least temporarily while a discussion is had on ↵James Molloy
llvm-commits regarding exactly how much optsize should optimize for size over performance. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143023 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-26Use a worklist to prevent the iterator from becoming invalidated because of ↵Bill Wendling
the 'removeSuccessor' call. Noticed in a Release+Asserts+Check buildbot. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143018 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-26Revert part of r142530. The patch potentially hurts performance especiallyEvan Cheng
on Darwin platforms where -Os means optimize for size without hurting performance. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143002 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-24Don't crash on variable insertelement on ARM. PR10258.Eli Friedman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142871 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-24Change this overloaded use of Sched::Latency to be an overloadedDan Gohman
use of Sched::ILP instead, as Sched::Latency is going away. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142813 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-22The different flavors of ARM have different valid subsets of registers. CheckBill Wendling
that the set of callee-saved registers is correct for the specific platform. <rdar://problem/10313708> & ctor_dtor_count & ctor_dtor_count-2 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142706 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-20Add missing operand. <rdar://problem/10313323>Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142615 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-19Use literal pool loads instead of MOVW/MOVT for materializing global ↵James Molloy
addresses when optimizing for size. On spec/gcc, this caused a codesize improvement of ~1.9% for ARM mode and ~4.9% for Thumb(2) mode. This is codesize including literal pools. The pools themselves doubled in size for ARM mode and quintupled for Thumb mode, leaving suggestion that there is still perhaps redundancy in LLVM's use of constant pools that could be decreased by sharing entries. Fixes PR11087. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142530 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-19Make sure we emit the 'movw' and 'movt' only if it's supported. Otherwise, ↵Bill Wendling
use a constant pool. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142485 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-19Remove some dead code.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142484 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Emit the MOVT instruction only if the # LPads is > 64K.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142460 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18For Thumb mode, we need to use a constant pool if the value is too large to beBill Wendling
used with the CMP instruction. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142458 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Use the integer compare when the value is small enough. Use the "move into aBill Wendling
register and then compare against that" method when it's too large. We have to move the value into the register in the "movw, movt" pair of instructions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142440 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Use the integer compare when the value is small enough. Use the "move into aBill Wendling
register and then compare against that" method when it's too large. We have to move the value into the register in the "movw, movt" pair of instructions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142437 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18The value we're comparing against may be too large for the ARM CMPBill Wendling
instruction. Move the value into a register and then use that for the CMP. <rdar://problem/10305266> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142431 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18The immediate may be too large for the CMP instruction. Move it into a registerBill Wendling
and use that in the CMP. <rdar://problem/10305266> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142429 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Use ARM/t2PseudoInst class from ARM/Thumb2 special adds/subs patterns.Andrew Trick
Clean up the patterns, fix comments, and avoid confusing both tools and coders. Note that the special adds/subs SelectionDAG nodes no longer have the dummy cc_out operand. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142397 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Use isIntN and isUIntN to check for valid signed/unsigned numbers.Bob Wilson
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142395 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18whitespaceAndrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142394 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18A landing pad could have more than one predecessor. In that case, we want thatBill Wendling
predecessor to remove the jump to it as well. Delay clearing the 'landing pad' flag until after the jumps have been removed. (There is an implicit assumption in several modules that an MBB which jumps to a landing pad has only two successors.) <rdar://problem/10304224> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142390 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Fix incorrect check for sign-extended constant BUILD_VECTOR.Bob Wilson
<rdar://problem/10298332> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142371 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-18Fix a bunch of unused variable warnings when doing a releaseDuncan Sands
build with gcc-4.6. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142350 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-17Don't renumber the blocks here. This could cause problems later on if anotherBill Wendling
pass renumbers the blocks again. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142258 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-17Add a call to EmitSjLjDispatchBlock.Bill Wendling
Once the intrinsics are marked as having a custom inserter, it will call this method to emit the dispatch table into the machine function. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142245 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-17Add comment explaining that the order of processing doesn't matter here.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142176 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-15ARM cannot select a pattern for trunc-store v4i8; /ARM/vrev.ll fails when ↵Nadav Rotem
promoting elements. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142080 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-15Mark registers as DEAD because they're really just clobbers.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142027 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-14Add missing correctness check to ARMTargetLowering::ReconstructShuffle. ↵Eli Friedman
Fixes PR11129. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142022 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-14Make sure that the register is in the register class before adding it as a ↵Bill Wendling
machine op. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142021 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-14Mark the invoke call instruction as implicitly defining the callee-saved ↵Bill Wendling
registers. The callee-saved registers cannot be live across an invoke call because the control flow may continue along the exceptional edge. When this happens, all of the callee-saved registers are no longer valid. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142018 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-13Simplify and avoid undefined shift. Based on patch by Ahmed Charles.Eli Friedman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@141903 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-10Reapply r141365 now that PR11107 is fixed.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@141591 91177308-0d34-0410-b5e6-96231b3b80d8
2011-10-10Revert r141365. It was causing MultiSource/Benchmarks/MiBench/consumer-lame toBill Wendling
hang, and possibly SPEC/CINT2006/464_h264ref. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@141560 91177308-0d34-0410-b5e6-96231b3b80d8