similar to: Memory leaks in LegacyPassManager depending on order of addRequired passes

With the intrinsic support for ‘memcpy’ and ‘memset’ the operands also have associated alignment operands. I think that ‘memcmp’ should also provide the alignment information for each of the source operands (when statically known). In some cases this will lead to more optimal alignment aware lowering, and for targets for which unaligned access is costly or fatal, it can be lowered safely.

Improving lowering for memcmp is definitely something we should do for all targets. Doing it in a target specific way is decidedly non-ideal. It looks like we already have some code in SelectionDAGBuilder which tries to optimize the lowering for the memcpy library call. I am a bit confused by the problem you are trying to solve. Are you specifically interested in lowering for constant

Can I make another suggestion: create an intrinsic for memory equality, e.g. llvm.memcmp_eq.p0i8.p0i8.i64(i8*a, i8*b, i64 len). This intrinsic would return zero if the memory regions are equal, and nonzero otherwise. However, it does NOT return any notion of "greater" or "less". Many applications require only determining equality, rather than a total ordering. Given that

Hello, I've made changes to the prologue to not spill callee saved gprs to the stack but rather spill them to unused vector registers. I'm not sure how to handle this in the cfi directives. Originally, we would use cfi_offset to give an offset of where it is saved on the stack. I tried to instead use the cfi_restore directive. As the docs say ".cfi_restore says that the rule

Currently on PowerPC, calls to memcmp are not expanded and are left as library calls. In certain conditions, expansion can improve performance rather than calling the library function as done for functions like memcpy, memmove, etc. This patch (https://reviews.llvm.org/D28163) is an initial implementation for PowerPC to expand memcmp when the size is an 8 byte multiple. The approach currently

So I got very mixed results. With the CallGraphSCCPass, both `addRequired<DominatorTreeWrapperPass>` and `addRequired<MemoryDependenceAnalysis>` fail at runtime. The LLVM core has just two CallGraphSCCPasses and neither uses neither analyses, so it's hard to find a valid example. I transformed the pass into a ModulePass, using scc_iterator as shown in CGPassManager to process

Hey, I am trying to use the BlockFrequencyInfoWrapperPass to obtain hotness information in my LLVM pass. Attached is a minimal example of code which creates a SIGSEGV. The pass calls AU.addRequired<BlockFrequencyInfoWrapperPass>(); in getAnalysisUsage(..). The problem exists with changed and unchanged IR. The binary is instrumented like this: clang input.bc -fprofile-generate -o

Hi all, I am experiencing a problem managing the command line option of a set of passes in my LLVM project. Attached you find a toy project the triggers the problem. The project is made up of two passes: "AnalysisPass" and "TransformPass". "TransformPass" requires "AnalysisPass" and they both share a common integer command line option called

Hi I am modifying LLVM to enable runtime feedback directed optimizations. I have modified the instrumentation part of LLVM to keep instrumentation data in some pre-setup structures in memory and use the JIT to recompile some of the hot functions with profiling information. I need an analysis pass which loads the profiling data in memory and can be used by different optimization passes in LLVM,

Thanks John. Does this solve the problem of analysis availability though? If I still have to run the function analyses manually, I might as well keep rolling with the CallGraphSCCPass. (I probably should have mentioned that this is what I’m using right now.) Félix > Le 2015-05-19 à 10:12:32, John Criswell <jtcriswel at gmail.com> a écrit : > > On 5/18/15 10:45 PM, Félix Cloutier

Thank You. Actually i pass polly canonaclize IR to my new created polly pass called "mypass". Mypass should first detect scops then find depedndencies as the mechanism conventional approach. Now i know how to write llvm pass here i am writing pass as loadable module first afterwards i will integrate it with opt in the end. I tried writing following code. Could you please help me on

i am organizing my LLVM Passes into an Analysis Pass and a few Optimization Passes, which takes the result from analysis pass to guide their transformations. I found a surprising pass type dependency between the analysis pass and optimization passes. E.g. Analysis-Pass Type Optimization-Pass Type Status in Optimization Pass when adding the Analysis Pass

i want to emit binary code for the following implemented vector assembly instructions. P_256B_LOAD_DWORD R_0_R2048b_0, pword ptr [rip + b] P_256B_LOAD_DWORD R_0_R2048b_1, pword ptr [rip + c] P_256B_VADD R_0_R2048b_0, R_0_R2048b_1, R_0_R2048b_0 P_256B_STORE_DWORD pword ptr [rip + a], R_0_R2048b_0 I added the following lines in X86MCInstLower.cpp; unsigned NewOpc; switch (OutMI.getOpcode())

I have modified the code as follows; now i am using both scopdetection and scopinformation before dependency check but i think link is missing... virtual bool runOnFunction(Function &F) { std::unique_ptr<ScopInfo> Result; std::unique_ptr<ScopDetection> Result2; //polly::ScopDetection pl; auto &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo(); auto

patchpoint is intended to be used in VMs that do their own linking, and so you wouldn't expect the function parameter to be resolved by LLVM. Presumably, in VMKit, you could just plant a pointer constant to the function you wish to call initially? -Filip On Dec 17, 2013, at 2:42 PM, Gaël Thomas <gael.thomas00 at gmail.com> wrote: > Hi all, > > I'm trying to play with

Hi, I'm gettign an assertion fail/crash in X86FrameLowering::GetFrameIndexReference when compiling the following bitcode: https://gist.github.com/carlokok/868cddebeb9acc8ccbac6253de0480b0 I tried removing the llvm.frameaddres calls but that's not it, where can I start looking for what my mistake here is? Code seems to verify just fine. ; #0 0x00e1afe8

Hi all, I'm trying to play with patchoint (with MCJIT and VMKit) and I don't understand something. I generate this call for my first patch point. Basically, I want to call f(0). %5 = call i64 (i64, i32, i8*, i32, ...)* @llvm.experimental.patchpoint.i64( i64 42, ;; patch point id is 42 i32 0, ;; 0 bytes for the padding i8* bitcast (i32 (i32)* @f to i8*), ;; my function f i32 1,

HI Hameeza, what do mean by link is still missing? Currently it is not possible to use polyhedral information from within in-tree LLVM passes. This is because Polly is not currently part of the LLVM tree, and is only linked into the tools as an external project. I.e., you can't depend on Polly passes in an in-tree pass. What you can do, though, is use Polly from an out-of-tree context. If

Hello, I have implemented a pass to count total number of instructions in LLVM 8.0. Its source code is attached here. The pass runs fine with LLVM 4.0. But with LLVM 8.0 following error is shown; $ $LLVMopt_BIN/opt -load $LLVM_SO/LLVMStatic-Info.so -one output-simple.bc WARNING: You're attempting to print out a bitcode file. This is inadvisable as it may cause display problems. If you REALLY

I think r262546 introduced the assumption that allocas are used exactly once with catchpad. It seems easy to fix, though. On Mon, Oct 17, 2016 at 11:51 PM, Carlo Kok via llvm-dev < llvm-dev at lists.llvm.org> wrote: > This turned out to be related to reusing the local used in the catchpad, > for example given the following c++ code: > > extern void rthrow(); > > int