[llvm-dev] [GSoC 2016] Interprocedural Register Allocation - Introduction and Feedback
Mehdi Amini via llvm-dev
llvm-dev at lists.llvm.org
Tue May 24 22:16:14 PDT 2016
> On May 24, 2016, at 10:08 PM, vivek pandya <vivekvpandya at gmail.com> wrote:
>
>
>
> On Wed, May 25, 2016 at 10:08 AM, Mehdi Amini <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>> wrote:
>
>> On May 24, 2016, at 9:17 PM, vivek pandya <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>> wrote:
>>
>> Dear Mentors,
>>
>> Please help me to understand our plan to implement Interprocedural Register allocator by propogating register usage info. While writing this mail I am considering all previous discussion over llvm-dev and IRC.
>>
>> 1) A MachineFunction pass to be executed POST-RA to collect the information about the used Registers.
>> 2) An Immutable pass which will store reg usage info collected by previous pass and return it whenever queried.
>> 3) A Target specific MachineFucntion pass that will use the register usage info for available for call instrction to achive IPRA. This pass should run at PRE-RA.
>>
>> Relation among above passes:
>>
>> 1) pass will store info to 2) pass as well use info for call instruction found while processing.
>>
>> 3) pass only requires to query information from 2) pass.
>>
>>
>> Questions
>> =========
>>
>> Which pass is responsible for load/store of callee saved register, at the begining of each function call? And how does it uses RegMask of call instruction to generate load/store. I think Intra-procedural register allocator is not responsible to generate load/store around the call site.
>>
>> /- - -> (A) - - -> (D)
>> /
>> (K)- - ->(T)- - -> (B) - - -> (E)
>> \
>> \- - -> (C) - - -> (F)
>>
>> So as per our discussion we would require following passes:
>>
>> Suppose in given example call graph , register allocation for D is completed now we have that information available So 3) pass while processing A , it would collect reg usage info for all callees and OR them and then it should update A's regmask by going to parant procedure that actually calls A ??
>
> No, Pass 3) is only looking for every call MI in A and updating the associated regmask by replacing it with the information stored in the immutable pass.
> What is the use of that ? Please be more specific.
I'm not sure I get the question. You describe 3 different passes at the beginning, and it seems fairly obvious what is the responsibility of each.
This seems quite straightforward to me, so I'm not sure what to explain, here is the logical sequence on D and A:
1) Codegen function D, MachineFunction post-RA: collect the register usage for D. Store the new regmask in the immutable pass
2) Move to function A, perform ISel, then run the MachineFunction pre-RA that finds the call to D. Query the immutable pass and get the regmask for D. Update the regmask associated with the call.
Done.
>
>
>> How reg mask details of call D would be used by Register allocator while allocating register for A and also not generating load/store for register being used by A in body of D as we have callee saved convention.
>
> I expect all of that to be handled automatically when updating the regmask.
> Please elaborate your both comments, what ever you have think of it to be.
>
>
>>
>> How the pass responsible for generating load/store will optimize for the child node of call graph where it does not require to load/store because caller will not use register used by callee ? I mean how our IPRA will take care of this?
>
> I don't understand that.
> In above example call graph when pass 3) is executed for D it does not do any work because there is not call to any other function in body of D but actually in presence of IPRA there is no need to preserve (store/load) any register in function D ( and also in other such leaf node of call graph). The reason for the same is A now aware of register usage in D so it will not use any register which intersect with D's register usage thus D need not to worry about any thing.
> Is this make sense?
No. D has to obey the calling convention if it is not internal. So any callee-saved register has to be ... saved.
--
Mehdi
>
>
>>
>> In short I am not much clear with the method for using information to get effect of IPRA without modifying Register allocator them self(i.e by updating regMask of call instructions).
>>
>> Also 1) pass and 3) pass are seem to intersecting for their work, for example consider while scanning register usage info for T function the final register usage info should be <all regs used by T> OR < reg usage info A > OR <reg usage info B > OR < reg usage info C > because K should not use any register which is used by T, A ,D, B, C, E, F with out load/store the relevant paper also discuss this situation and suggest to fall back to load/store approach. So as we move to upper region of the call graph it is very likely that enough regiseters are not there to allocate.
>
> The calling convention (for anything else than internal function) will always have some callee-saved registers.
> If we have a deep call-graph of internal-only function, then we may generate a lot of spill at the top. Somehow we may have to think about driving some heuristic with PGO.
>
> This kind of problem is addressed by this paper Register allocation across procedure and module boundaries - http://dl.acm.org/citation.cfm?id=93551 <http://dl.acm.org/citation.cfm?id=93551>
> In this paper authors have tried to eliminate load/store only in call intensive regions ( contrast to simple bottom up ) by analyzing call frequency at compile time ( also has facility to plug in Profile info) and rest of the call graph will follow normal register allocator. But yes we will think about this later.
>
> I suspect we can think about that a bit later. Let's focus on the simple for now.
>
> --
> Mehdi
>
>
>
>>
>> Please bear with my silly questions.
>>
>> Sincerely,
>> Vivek
>>
>>
>> On Wed, May 25, 2016 at 8:46 AM, vivek pandya <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>> wrote:
>>
>>
>> On Wed, May 25, 2016 at 8:44 AM, Hal Finkel <hfinkel at anl.gov <mailto:hfinkel at anl.gov>> wrote:
>>
>> From: "vivek pandya" <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>>
>> To: "Hal Finkel" <hfinkel at anl.gov <mailto:hfinkel at anl.gov>>
>> Cc: "llvm-dev" <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>>, "Matthias Braun" <matze at braunis.de <mailto:matze at braunis.de>>, "Mehdi Amini" <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>>, "Quentin Colombet" <qcolombet at apple.com <mailto:qcolombet at apple.com>>
>> Sent: Tuesday, May 24, 2016 9:34:29 PM
>>
>> Subject: Re: [GSoC 2016] Interprocedural Register Allocation - Introduction and Feedback
>>
>>
>>
>> On Wed, May 25, 2016 at 3:53 AM, Hal Finkel <hfinkel at anl.gov <mailto:hfinkel at anl.gov>> wrote:
>>
>> From: "vivek pandya" <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>>
>> To: "Quentin Colombet" <qcolombet at apple.com <mailto:qcolombet at apple.com>>
>> Cc: "Hal Finkel" <hfinkel at anl.gov <mailto:hfinkel at anl.gov>>, "llvm-dev" <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>>, "Matthias Braun" <matze at braunis.de <mailto:matze at braunis.de>>, "Mehdi Amini" <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>>
>> Sent: Tuesday, May 24, 2016 1:00:58 PM
>> Subject: Re: [GSoC 2016] Interprocedural Register Allocation - Introduction and Feedback
>>
>> Hello,
>>
>> I have written following code to check each register if it is used by machineFunction or not :
>>
>> MachineRegisterInfo *MRI = &MF.getRegInfo();
>> TargetRegisterInfo *TRI = (TargetRegisterInfo *)MF.getSubtarget().getRegisterInfo();
>> Some reason you can't use a const pointer here?
>> MCRegisterInfo is just used to get conventional name of register for given target like AX, BX on X86.
>>
>> const TargetMachine &TM = MF.getTarget();
>> const MCRegisterInfo *MCRI = TM.getMCRegisterInfo();
>> DEBUG(dbgs() << "Function Name : " << MF.getName() << "\n");
>>
>> for(TargetRegisterInfo::regclass_iterator i = (*TRI).regclass_begin(), e = (*TRI).regclass_end(); i != e; i++ ) {
>> for(TargetRegisterClass::iterator pregi = (*i)->begin(), prege = (*i)->end(); pregi != prege; pregi++ ) {
>> DEBUG( dbgs() << "Physical Register : " << MCRI->getName(*pregi) << " is modified "<< MRI->isPhysRegModified(*pregi) << " \n");
>> Try isPhysRegUsed.
>> ok
>>
>> }
>> }
>> DEBUG(dbgs() << "\n");
>>
>> The pass which is executing this code is schedule POST-RA stage but this gives me true for all registers i.e in each function all registers are being used except EBP and some other similar, Is this a correct way to get register usage information ? I think I have made some mistake please help.
>>
>> You might look at the implementation of these functions in lib/CodeGen/MachineRegisterInfo.cpp and figure out if they're returning true because UsedPhysRegMask.test(PhysReg) is true or because reg_nodbg_empty(*AliasReg) is true.
>> Yes that helped now I am getting actual register which have been used by given function, but a little problem
>> The updated code is as shown below :
>> for(TargetRegisterInfo::regclass_iterator i = (*TRI).regclass_begin(), e = (*TRI).regclass_end(); i != e; i++ ) {
>> for(TargetRegisterClass::iterator pregi = (*i)->begin(), prege = (*i)->end(); pregi != prege; pregi++ ) {
>> for (MCRegAliasIterator AliasReg(*pregi, TRI, true); AliasReg.isValid(); ++AliasReg) {
>> if (!MRI->reg_nodbg_empty(*AliasReg)) {
>> DEBUG( dbgs() << "Physical Register : " << MCRI->getName(*pregi) << " is used "<< MRI->isPhysRegUsed(*pregi) << " \n");
>> break; // no need to process more alias
>> }
>> }
>> }
>> }
>> But here some registers are getting processed with in different classes (unnecessary processing) Is this only way to iterate through all used register (using RegClass iterator) ? Is there any way to avoid duplicate regs?
>> Of course currently I am just printing but next I am thinking to use a map to track usage info , in that only distinct register info will be stored but still due to loop structure I need to iterate through a single register 3 - 4 times making it time consuming.
>> Yes, I believe you can just do:
>>
>> for (unsigned Reg = 0; Reg < TRI->getNumRegs(); ++Reg) {
>> Oh yes thanks I just forgot that PhyReg starts at 0.
>>
>>
>> -Hal
>>
>>
>> -Vivek
>>
>>
>> -Hal
>>
>>
>>
>> Vivek
>>
>> On Wed, May 18, 2016 at 11:42 PM, Quentin Colombet <qcolombet at apple.com <mailto:qcolombet at apple.com>> wrote:
>>
>> On May 18, 2016, at 11:00 AM, vivek pandya <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>> wrote:
>>
>>
>>
>> Vivek Pandya
>>
>>
>> On Wed, May 18, 2016 at 11:25 PM, Quentin Colombet <qcolombet at apple.com <mailto:qcolombet at apple.com>> wrote:
>>
>> On May 18, 2016, at 10:46 AM, vivek pandya <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>> wrote:
>>
>>
>>
>> Vivek Pandya
>>
>>
>> On Wed, May 11, 2016 at 4:01 PM, Hal Finkel <hfinkel at anl.gov <mailto:hfinkel at anl.gov>> wrote:
>>
>> From: "vivek pandya" <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>>
>> To: "Mehdi Amini" <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>>
>> Cc: "Hal Finkel" <hfinkel at anl.gov <mailto:hfinkel at anl.gov>>, "Quentin Colombet" <qcolombet at apple.com <mailto:qcolombet at apple.com>>, "llvm-dev" <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>>, "Matthias Braun" <matze at braunis.de <mailto:matze at braunis.de>>
>> Sent: Wednesday, May 11, 2016 3:15:03 AM
>> Subject: Re: [GSoC 2016] Interprocedural Register Allocation - Introduction and Feedback
>>
>>
>>
>>
>> Vivek Pandya
>>
>>
>> On Wed, May 11, 2016 at 10:02 AM, vivek pandya <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>> wrote:
>>
>>
>> Vivek Pandya
>>
>>
>> On Wed, May 11, 2016 at 9:43 AM, Mehdi Amini <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>> wrote:
>>
>> On May 10, 2016, at 6:06 PM, Hal Finkel <hfinkel at anl.gov <mailto:hfinkel at anl.gov>> wrote:
>>
>>
>>
>> From: "vivek pandya" <vivekvpandya at gmail.com <mailto:vivekvpandya at gmail.com>>
>> To: "llvm-dev" <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>>, "Tim Amini Golling" <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>>, "Hal Finkel" <hfinkel at anl.gov <mailto:hfinkel at anl.gov>>
>> Cc: "Quentin Colombet" <qcolombet at apple.com <mailto:qcolombet at apple.com>>
>> Sent: Tuesday, May 10, 2016 2:59:16 PM
>> Subject: [GSoC 2016] Interprocedural Register Allocation - Introduction and Feedback
>>
>> Hello LLVM Community,
>>
>> Sorry for delay as I was busy in final exams.
>>
>> I am Vivek from India. Thanks for choosing my proposal for Interprocedural Register Allocation (IPRA) in LLVM. Mehdi Amini and Hal Finkel will be mentoring me for this project.
>>
>> IPRA can reduce code size and runtime of programs by allocating register across the module and procedure boundaries.
>>
>> I have identified some old but effective research work on this area.
>> I want community's feedback for feasibility of these approach and I am targeting to implement two of them during this project.
>>
>> Here is list of the papers, I have read first two papers and I would like to discuss those approach first, I will read other two paper then initiate discussion for them as well. All I want is to find out a concrete implementation plan before 23 May, 2016 and for that I need community's help.
>>
>> 1) Compile time ----- Minimizing register usage penalty at procedure calls - http://dl.acm.org/citation.cfm?id=53999 <http://dl.acm.org/citation.cfm?id=53999>
>> ====================================================================In this approach intra-procedural register allocation is used as base but machine code generation order is bottom up traversal of call graph and inter-procedural effect is achieved by propagating register usage information of callee function to caller (i.e child to parent in CallGraph) so that caller can use different registers than callee and can save load store cost at procedure call, this is not trivial as it seems due to recursive calls, library function usage etc. Also for upper region of the graph in this technique available number of registers might become zero in that case it should fall back to normal load store at procedure call. Apart from these difficulties other difficulties have been identified please follow this mail-chain https://groups.google.com/d/topic/llvm-dev/HOYAXv3m1LY/discussion <https://groups.google.com/d/topic/llvm-dev/HOYAXv3m1LY/discussion>
>> My mentor has already provided me a patch that alters code generation order as per bottom up call graph traversal, I am working from that point now. Any other help/suggestion is always welcomed.
>>
>> 2) Link time ----- Global register allocation at link time - http://dl.acm.org/citation.cfm?id=989415 <http://dl.acm.org/citation.cfm?id=989415>
>> ====================================================================In this particular approach (sort of true IPRA) registers will be reallocated (this optimization will be optional if turned off still code will be compiled as per intra-procedural allocation) at link time. Here modules are first complied as per normal compilation but the object code is annotated with details so that linker can build call graph and also calculate usage information at link time. Compiler also write hints in object code that if particular variable is allocated in some other register ( due to new allocation) then how the code should be changed? Thus linker can use these information to decide which variables (global) need to be in same register through out the program execution and also according to register usage information in call graph which procedure will not be active simultaneously so that locals for that procedures can be in same registers with out load store at procedure calls.
>> For these particular method help me to analyze feasibility:
>> 1) Can llvm collects following information at module level in MachineIR? list of procedures in module, list of locals in procedures, list of procedures that a particular procedure can call, and a list of the variables this procedure references. Each entry in the last two lists includes an estimate of the number of times the procedure is called or the variable is referenced in each execution of this procedure
>> 2) Can llvm write informative commands to object files?
>> 3) Can LTO is capable of leveraging those commands?
>> In terms of scoping the project for the summer, I definitely recommend that you focus on (1) first. If you finish that, we can certainly move on to other things.
>>
>> I'll add +1 here, but I already wrote the same thing on IRC when discussing with Vivek. True IPRA without a proper MachineModule infrastructure won't be doable in my opinion (even with such infrastructure, it may not be trivial in LLVM in general).
>>
>> Regarding link time, note that any such a design would likely look much different than in David Wall's paper however, because our LTO re-codegens everything anyway. The paper says, "Finally, it keeps us honest as designers of the system; once we postpone anything until link time, the temptation is great to postpone everything, ..." - Well, we've long-since succumb to that temptation when we LTO. C'est la vie.
>>
>> +1 as well, our LTO will benefit naturally from the leaf-to-root information propagation. ThinLTO will be more challenging/interesting though!
>> For the first part a mechanism similar to MachineModulePass would be desirable but that may not be possible during this project, but if we can make some sort of smaller version of that to suit our purpose.
>> I don't think we need to make any kind of MachineModulePass to make this work. Once we alter the visitation order based on the CGSCC iteration scheme, we can keep state in-between functions in the pre-existing hacky way (using static members of the relevant function passes).
>> Sorry my mistake here by first part I mean 1) requirement in the link time approach.
>>
>> I also don't see where/why we need a MachineModule(Pass) for the CGSCC scheme, that said I'd rather avoid using a function pass with static members, if we can have a ModuleAnalysis that is bookkeeping the results for functions in the module and queries by the register allocator somehow.
>> Matthias/Quentin may have other inputs on this aspect.
>>
>> @Hal do you mean to add a simple MachineFunction pass that will just operate on register allocated function and prepare a BitVector to indicate which register is being used by MachineFunction, and then use this pass as analysis pass (i.e just simply return static BitVector for clobbered register when register allocation for next function begins. This part is not much clear to me) this thing can be done by scheduling a pass post register allocation in lib/CodeGen/Passes.cpp
>>
>> void TargetPassConfig::addMachinePasses() {
>> .
>> .
>> .
>> // Run pre-ra passes.
>> addPreRegAlloc();
>>
>> // Run register allocation and passes that are tightly coupled with it,
>> // including phi elimination and scheduling.
>> if (getOptimizeRegAlloc())
>> addOptimizedRegAlloc(createRegAllocPass(true));
>> else
>> addFastRegAlloc(createRegAllocPass(false));
>>
>> // Run post-ra passes.
>> addPostRegAlloc();
>> // Adding a new pass here which keeps register mask information across function calls.
>> .
>> .
>> .
>> }
>>
>> But this also requires current register allocators to use this information in someway because RegMaskBits in LiveIntervalAnalysis.cpp is not static across calls. I mean I am not clear for how to propagate static info to Intra-procedural Register allocators (if possible without disturbing their code )
>> First, my hope is that we won't need to change the register allocators, as such, in order to make use of this information. Instead, we'll simply be able to alter the register masks generated for the call instructions. These masks will indicate fewer clobbers than might otherwise be present based on the ABI because of information gathered during the codegen of the callee. These masks are generally constructed by target based on the calling convention. The PowerPC backend, for example, looks like this:
>>
>> // Add a register mask operand representing the call-preserved registers.
>> const TargetRegisterInfo *TRI = Subtarget.getRegisterInfo();
>> const uint32_t *Mask =
>> TRI->getCallPreservedMask(DAG.getMachineFunction(), CallConv);
>> assert(Mask && "Missing call preserved mask for calling convention");
>> Ops.push_back(DAG.getRegisterMask(Mask));
>>
>> but it can be more complicated. If you look for uses of 'getRegisterMask' in Target/*/*ISelLowering.cpp, you'll see what I mean. Regardless, the code ends up calling some method is the targets TargetRegisterInfo subclass. These methods generally look something like this:
>>
>> const uint32_t *
>> PPCRegisterInfo::getCallPreservedMask(const MachineFunction &MF,
>> CallingConv::ID CC) const {
>> const PPCSubtarget &Subtarget = MF.getSubtarget<PPCSubtarget>();
>> ...
>> return TM.isPPC64() ? (Subtarget.hasAltivec() ? CSR_SVR464_Altivec_RegMask
>> : CSR_SVR464_RegMask)
>> : (Subtarget.hasAltivec() ? CSR_SVR432_Altivec_RegMask
>> : CSR_SVR432_RegMask);
>> }
>>
>> In any case, the fundamental idea here is that, when someone calls getCallPreservedMask in order to set the regmask on a call, we might not have to use the CC at all. Instead, if we've already codegened the function, we might use a cache of 'exact' register masks computed during codegen of the potential callees instead.
>> I am thinking to add a simple Immutable pass MachineRegisterUsageInfo similar to MachineBranchProbabilityInfo that can maintain RegisterUsageInformation per function. Can it be simply done by using UsedPhysRegMask from MachineRegisterInfo ??
>>
>> No, like the comment said, UsedPhysRegMask gives only the registers clobbered by calls:
>> // This bit vector represents all the registers clobbered by function calls.
>>
>> You want to build this information yourself on top of MachineRegisterInfo::isPhysRegModified
>> Ok but then the time complexity will be O(n) n = number of physical register on the target. Am I going correct?
>>
>> Yes, this is correct.
>>
>>
>> Here getCallPreservedMask will call API provided by MachineRegisterUsageInfo to avail the exact register mask but how it can know that the function is already codegen or it will query each time when getCallPreservedMask is called and of available MachineRegisterUsageInfo will return the details otherwise simply return NULL.
>> So changes will be now in TargetRegisterInfo implementation for each target right ??
>>
>>
>> In order to do this, I think we'll need to provide a function callable from the target's getCallPreservedMask implementation, which can return such an 'exact' regmask when available. I think we need to do it this way for two reasons:
>>
>> 1. Not all of the target code calls getCallPreservedMask, but sometimes calls other similar target-specific functions (e.g. getTLSCallPreservedMask).
>> 2. The targets need to opt-in to this behavior because only the target can know that all register uses are really tagged correctly post "pre-emit".
>>
>> Because the target is free to introduce uses of registers at essentially any time, we need to do the scanning for used registers after the "pre-emit" passes run. This can be done by scheduling some simple register-use scanning pass after the call to addPreEmitPass in lib/CodeGen/Passes.cpp.
>>
>>
>> I think this also applies in someway to Mehdi Amini's idea to keep a ModulePass for bookkeeping but then existing register allocators will be required to change so that the code can query the ModulePass for RegMaskBits for particular function.
>> I think that the simplest way to do this is to create an immutable analysis pass (e.g. BasicAA) that keeps the cache of the computed register masks. This is somewhat similar in spirit to how the 'AssumptionCache' analysis works at the IR level. This analysis can then be created by lib/CodeGen/Passes.cpp early, and then queried and passed around later by the CodeGen/Target code. Because it is an immutable analysis, it won't get destroyed until the very end, which is also important because, I imagine, it will need to own the memory associated with the generated register masks.
>>
>> -Hal
>>
>> Vivek
>>
>>
>> Yes for propagating register usage approach we don't need MachineModulePass
>>
>> Vivek
>> --
>> Mehdi
>>
>>
>>
>>
>>
>>
>> --
>> Hal Finkel
>> Assistant Computational Scientist
>> Leadership Computing Facility
>> Argonne National Laboratory
>>
>>
>>
>>
>>
>> --
>> Hal Finkel
>> Assistant Computational Scientist
>> Leadership Computing Facility
>> Argonne National Laboratory
>>
>>
>>
>>
>> --
>> Hal Finkel
>> Assistant Computational Scientist
>> Leadership Computing Facility
>> Argonne National Laboratory
>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20160524/33fa6c66/attachment-0001.html>
More information about the llvm-dev
mailing list