similar to: Re-computing Live-in/Live-out Physical Registers for Basic Blocks Using LivePhysRegs

Tim Northover <t.p.northover at gmail.com> writes: >> /tmp/foo.s:1:2: error: instruction requires: distinct-ops >> sllk %r2,%r3,1 >> ^ > > That seems like it would be a good improvement for all targets. Thanks, sounds like it might be more acceptable than I thought :-) >> ARM seems to rely on the current MatchOperandParserImpl() behaviour,

I think what's happening is BranchFolder::MaintainLiveIns is using a forward analysis on top of these missing kill flags, and updating the BB-live-ins/live-outs with an incorrect set of registers. Then when the stackmaps liveness analysis happens, it's not doing anything wrong, but it starts with the wrong set of live registers and will propagate those to the point of the

On 9/17/2015 7:04 AM, Sky Flyer via llvm-dev wrote: > It seems like d0 is always 14! > I check it with ARMGenAsmMatcher.inc it was the same! > How is it possible? because it should give the same register value that > matches the underlying platform not any autogenerated value!? The returned number is the register id as defined in <YourTarget>GenRegisterInfo.inc. These numbers

I spent some time digging into a Valgrind report of uninitialized values in LLVM r287520 built using itself. (One of quite a few such reports that comes up during a "make check".) I could use another set of eyes on the issue if someone has time. This command gives me an error: valgrind -q ./bin/llc < /home/regehr/llvm/test/CodeGen/Hexagon/hwloop-dbg.ll -march=hexagon

On Apr 14, 2010, at 1:55 PMPDT, Villmow, Micah wrote: > I’ve recently sync’d to a newer version of LLVM(Apple branch 2326 from Apple branch 2323.8) that changed the interface to addCommonCodeGenPasses which caused the default implementation to be executed instead of my overriding implementation. This default implementation has DeadMachineInstructionElim pass enabled, which is causing havoc

I've recently sync'd to a newer version of LLVM(Apple branch 2326 from Apple branch 2323.8) that changed the interface to addCommonCodeGenPasses which caused the default implementation to be executed instead of my overriding implementation. This default implementation has DeadMachineInstructionElim pass enabled, which is causing havoc with my backend. Before entering this pass, everything

Just want to emphasize that on x86-64 and using Valgrind: LLVM compiled with LLVM gets 360 unexpected test fails LLVM compiled with GCC gets 22 unexpected test fails Of course I don't know how many of these are caused by this bitfield speculation issue. John On 11/21/2016 10:48 PM, regehr via llvm-dev wrote: > Alright, here's what seems to be happening... > > The testcase

Dale, Yeah that is correct, so that isn't the problem, not sure why I was thinking it is. The !MRI->use_no_dbg_empty(Reg) seems to be correct and the problem is LivePhysRegs[Reg] always returning false. I've looked into this more and there is a part where I'm working with physical registers. If there is a function call, there is a copy from virtual register space to the functions

Hi, I'm a bit unsure of the semantics of undef physical registers. The explanations I've seen in the code and in the langref seems to pertain more to constant values and virtual registers. What I really want to achieve is a push-pop of a register to have a temporary to work with, without having to check if this register is defined or not. However, whenever the reg is not defined before

Hi all, I've run into a couple bugs recently that affected stackmap liveness analysis in various ways: http://llvm.org/bugs/show_bug.cgi?id=19224 - function arguments stay live unnecessarily http://llvm.org/bugs/show_bug.cgi?id=21265 - eflags can end up as a live out http://llvm.org/bugs/show_bug.cgi?id=21266 - %rip can end up as a live out The first two have nothing to do with stackmaps

> The right way to update the CFG very much depends on how you're > transforming it. I would like to export the CFG for control flow checking. Theoretically, it should be possible for a compiler to know every target of every control flow instruction, except for computed targets that are not known at compile-time. When a machine basic block is split between two branches, as shown below:

> /tmp/foo.s:1:2: error: instruction requires: distinct-ops > sllk %r2,%r3,1 > ^ That seems like it would be a good improvement for all targets. > ARM seems to rely on the current MatchOperandParserImpl() behaviour, > so I'm not going to suggest changing it unconditionally. Presumably you switched it and looked at what fell over; do you remember what kind

Dear all, in my TestRegisterInfo.td file, I defined a register like this: class TestReg<bits<6> enc, string name> : Register<name> { let HWEncoding{5-0} = enc; let Namespace = "TEST"; } def D0 : TestReg<0x01, "d0">, DwarfRegNum<[1]>; but when I compile, the result I have in TestGenAsmMatcher.inc is this: case 'd': // 7

All, ARM LLVM afficianados: I'm trying to understand how LiveVariables computes liveness for CPSR. In particular, I would like to tell whether a particular instruction is a use/kill/def for doing some within block liveness checking I need. Thanks in advance, David -------------- next part -------------- An HTML attachment was scrubbed... URL:

I'm trying to add some instructions that are only available on certain processors. These instructions use context-dependent parsers. Everything works fine for the valid cases, but if you try to use an instruction on processors that don't support it, the asm parser says: /tmp/foo.s:1:2: error: invalid operands for instruction sllk %r2,%r3,1 ^ rather than:

Thank you for your reply! > Every MachineBasicBlock has a list of successors; you can access it with > the successors() accessor. That's what you should be using for any CFG > analysis. I am aware of these methods of class MachineBasicBlock, which allows one to access a MachineBasicBlock's successors and predecessors in the CFG. But the CFG itself may no longer be valid if a

I searched source code of LLVM/Clang 4.0.0 for 'random_seed' with grep. It seems the -frandom-seed option is not supported. The -rng-seed option appears to be defined in ./lib/Support/RandomNumberGenerator.cpp, which is source code for class RandomNumberGenerator. The constructor of class RandomNumberGenerator is private and is only called by Module::createRNG (defined in

Evan & llvmdev, I'm seeing a case where ARM Load/Store optimizer is breaking code. I have not had any luck trying to come up with a minimal example; it is breaking in our stage 2 LLVM build. But here's what I'm seeing in the debug output: # Before ARMLoadStoreOptimizer: BB#21: derived from LLVM BB %cond.end Live Ins: %LR %R0 %R1 %R7 %R10 %R11 Predecessors according to

On Jul 17, 2013, at 10:26 AM, Richard Sandiford <rsandifo at linux.vnet.ibm.com> wrote: > Tim Northover <t.p.northover at gmail.com> writes: >>> /tmp/foo.s:1:2: error: instruction requires: distinct-ops >>> sllk %r2,%r3,1 >>> ^ >> >> That seems like it would be a good improvement for all targets. > > Thanks, sounds

Ping. Anyone have any idea on how to fix this? Thanks, Micah From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Villmow, Micah Sent: Wednesday, April 14, 2010 2:28 PM To: llvmdev at cs.uiuc.edu Subject: Re: [LLVMdev] Default behavior of DeadMachineInstructionElim deletes all instructions Dale, Yeah that is correct, so that isn't the problem, not sure