similar to: [LLVMdev] creating a vreg in eliminateFrameIndex()

On 1/29/15 2:00 PM, Steve King wrote: > Hello LLVM, > The ARM target sometimes adds an instruction with a virtual register > in eliminateFrameIndex(): > > https://github.com/llvm-mirror/llvm/blob/master/lib/Target/ARM/ARMBaseRegisterInfo.cpp > > This looks late for a virtual register to appear. Where is this vreg made real? The register scavenger should take care of such

Thanks Jon and Hal for the helpful pointers. By returning true from requiresRegisterScavenging() and requiresFrameIndexScavenging(), LLVM handled all the scavenging effort. That is nearly painless for the target, so why do some targets seem to do scavenging on their own? When the scavenged register is loaded with a simple immediate, is it safe to search the BB and replace other uses of the same

----- Original Message ----- > From: "Steve King" <steve at metrokings.com> > To: jonathan at codesourcery.com > Cc: llvmdev at cs.uiuc.edu > Sent: Friday, January 30, 2015 10:52:19 AM > Subject: Re: [LLVMdev] creating a vreg in eliminateFrameIndex() > > Thanks Jon and Hal for the helpful pointers. By returning true from > requiresRegisterScavenging() and

On Fri, Jan 30, 2015 at 8:59 AM, Hal Finkel <hfinkel at anl.gov> wrote: >> When the scavenged register is loaded with a simple immediate, is it >> safe to search the BB and replace other uses of the same immediate >> with the reg? In my case this gives a code size reduction. The >> search would stop on finding another use of the scavenged register. > > Are you

I'm confused as to the logic used in the register scavenger when it cannot find a free register. I would think that it would want to free up the emergency spill slot immediately after it's use, because otherwise there is a chance of needing to use the emergency slot again and not be able to. Instead it tries to restore it only right before register it is freeing up. Maybe I'm

Hi Reed, the register scavenger (RS) also keeps track of live registers. This way it "knows" that the register that was spilled/restored far apart is available. Let say you had the following code. You need to find a register to keep vreg1 and vreg2 in. R1 = .... // <- RS current liveness state; we have called RS->forward(It) where It points to here vreg1 = add SP, 1000 ... =

I'm investigating a bug associated with debug information that manifests itself in the XCore backend (PR11105). I'd like to understand what the expected behavior of eliminateFrameIndex() is when it is called on a dbg_value machine instruction. Currently the XCore target replaces the frame index with the frame register and sets the next operand to the byte offset from the frame

Hello all, For my custom processor backend I am trying add some instruction using BuildMI() inside eliminateFrameIndex(). I tried RegScavenger like this: unsigned RegUnused0 = RS->FindUnusedReg(&LASER::GNPRegsRegClass); if (!RegUnused0) RegUnused0 = RS->scavengeRegister(&LASER::GNPRegsRegClass, II, SPAdj); assert(RegUnused0 && "Register scavenger failed");

On Oct 10, 2011, at 10:26 AM, Richard Osborne wrote: > I'm investigating a bug associated with debug information that manifests > itself in the XCore backend (PR11105). I'd like to understand what the > expected behavior of eliminateFrameIndex() is when it is called on a > dbg_value machine instruction. That is up to the target. The TII::emitFrameIndexDebugValue() hook is

Hi! I started writing a LLVM backend for a custom architecture. I have some register and instruction .td files and some other files/classes like a MCStreamer for assembler output. At the moment I can compile some empty programs so far. I implemented the method ::eliminateFrameIndex() similar to the Sparc and ARM backend. The method looks like this: // frame pointer is in reg of class

You mean when I "explicity" use it by calling methods of register scavenger? Right now I'm just allocating virtual registers that will be resolved by the use of register scavenger and I'm also providing an override of the virtual method saveScavengerRegister. In Mips16, I have an extra mips 32 register (not usually very useful since it can only be used in a move instruction)

On 10/10/11 19:19, Jakob Stoklund Olesen wrote: > On Oct 10, 2011, at 10:26 AM, Richard Osborne wrote: >> I'm investigating a bug associated with debug information that manifests >> itself in the XCore backend (PR11105). I'd like to understand what the >> expected behavior of eliminateFrameIndex() is when it is called on a >> dbg_value machine instruction. >

Hi, right now, LLVM does register spilling by: 1. Creating stack object 2. Passing index of that stack object to MRegisterInfo::storeRegToStackSlot 3. At later stage, frame indices are replaced by calling to MRegisterInfo::eliminateFrameIndex. This works for me, but there's slight problem. The target does not have "register + contant" addressing mode, so accessing frame index

I ran into another issue with register scavenger. In my case, I don't need a place on the stack for an emergency spill slot. I have these free mips32 registers, that are not in general very useful for other things, for the emergency spill slot. I can move to and from mips16 (subset of mips32) registers and mips32 registers. I also have a situation where I need two free registers so then

Hello, for an architecture that doesn't have a good way to load/store a given register class to memory, is it instead easy to spill/fill from another register class instead? e.g. - storeRegToStack/loadRegFromStack use a pseudo instruction and add virtual register operand is not supported (spill optimization doesn't seem to like this). - AMDGPU backend seems to do sth. similar? The only

Ok, thanks. Except the question was meant slightly different. Less w.r.t. organizing the register classes, and more w.r.t. implementation. I've noticed for instance that when trying to model this straight forwardly by writing a vreg from spills and reading this from fills (not further elaborated here), that the spiller can't handle vreg def-use pairs: there are assertions making sure a

I almost change all the instruction formats. It was a huge work. I am going to compile and run it now. Best Regards, A. Yazdanbakhsh >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> PhD. Student School of Electrical and Computer Engineering University of Wisconsin-Madison E-mail: yazdanbakhsh

The docs for the above-mentioned function say: This method may modify or replace the specified instruction, as long as it keeps the iterator pointing the the finished product. What does it mean to "keep an interator". Was "invalidates the iterator" intended, so something else. Another question, is how do I really replace the instruction. The operator= is private

> I almost change all the instruction formats. It was a huge work. I am going > to compile and run it now. We have done the similar work[1] on this topic by gcc and we have start migrate our platform to LLVM. In my experience, you need to take care the follow part: * ARMBaseRegisterInfo::getRegPressureLimit * ARMBaseRegisterInfo::getRawAllocationOrder * CalleeSavedRegs *

Hi! I'm making a strange observation in my backend, that ends in illegal code: Version 1: - I lower FrameIndex to TargetFrameIndex (nothing special) - I generate a special address-register ADD instruction in eliminateFrameIndex() to write FramePointer + offset into a new address-register - I use explicit load and store and address-registers in my target instruction patterns: eg (store